socioeconomic position and utilisation of preventive health … · 2010-06-09 · preventive health...

Socioeconomic Position and Utilisation of Preventive Health Services among Adults in

the General Population

JIANZHEN ZHANG

BMed, MPH (TH)

A Thesis Submitted in Fulfilment of the Requirements for the Degree of Doctor of Philosophy at the

Queensland University of Technology School of Public Health

2007

ii

KEYWORDS

Socioeconomic position

Preventive health services

Utilisation

Blood pressure check-up

Blood cholesterol check-up

Blood glucose check-up

Determinants

Development of questionnaire

Mail survey

iii

ABSTRACT

Background: International research has shown that socioeconomically

disadvantaged groups experience significantly higher mortality and morbidity rates

than other groups. Both cardiovascular disease (CVD) and diabetes are major

contributors to Australia’s burden of disease, and individuals from lower

socioeconomic groups are more likely to be affected by both, and to have worse

prognoses and outcomes. There is substantial research evidence that a range of

preventive activities can reduce the morbidity and mortality associated with these

conditions. Research in countries with good access to primary health care services

has demonstrated that socioeconomically disadvantaged groups tend to have higher

levels of medical consultations, but make less use of preventive care and screening

services. This fact contributes to their poorer health outcomes, as diagnosis will

typically occur later than for more advantaged individuals, thus leading to a poorer

prognosis. However, to date, there has been little research on the differential

utilisation of preventive health services for CVD and diabetes by different

socioeconomic groups in Australia. To understand socioeconomic influences on the

use of preventive health services, a comprehensive review of the literature of

determinants of health service utilisation was conducted and a number of

explanations for this relationship considered. It was proposed that the following

factors are likely to be important in this relationship: differences in the perception of

the availability of, and accessibility to health care, attitudes and beliefs toward

preventive health care, having a regular source of care, perception of interpersonal

care from general practitioners, and social support. A number of theoretical models

were also reviewed; in particular, the Andersen Behavioural Model of Health Service

Research Utilisation.

Aims: This doctoral research program has described the relationship between

socioeconomic position (SEP) and utilisation of preventive health services in relation

to CVD and diabetes. It aims to improve the understanding of the determinants of

uptake and utilisation of preventive health services in general practice by different

socioeconomic groups in Australia.

iv

Methods: The study was conducted in Brisbane Australia, in 2004, using a cross-

sectional design and a self-administered mailed survey for data collection. A sample

of adults aged 25–64 years was selected randomly from the Brisbane Electoral Roll.

A conceptual model incorporating a range of relevant socio-demographic, risk-factor

and behavioural variables in the relationship between SEP and GP-based use of

preventive health services was used to develop a self-administered questionnaire.

The questionnaire was pilot-tested and then reviewed by a panel of international

experts. A new self-administered questionnaire, the Health Service Utilisation

Questionnaire (HSUQ), was developed. It included 79 items: 12 socio-demographic

items; 10 items assessing health status, disease conditions and smoking status;

20 items assessing use of health services; and 37 items assessing the factors that

might affect use of health services utilisation. The HSUQ was then mailed to 800

randomly selected survey participants. The survey response rate was 65.6 per cent.

After exclusion of those patients with cardiovascular diseases and diabetes, the final

sample size was 381, consisting of 155 males and 226 females. Socioeconomic

indicators were individual education level and family income. Blood pressure, blood

cholesterol and blood glucose check-ups by general practitioners (GPs) were used as

the major outcome variables. Nine scales and two dichotomous variables that

measure those potential factors were derived following Principal Component

Analysis and reliability testing. The data were analysed separately by gender, and

adjusted for age and each of the socioeconomic indicators. Statistical description,

bivariate analysis and multivariable modelling in SPSS were applied for the data

analysis.

Results: The survey results were suggestive of socioeconomically disadvantaged

people being less likely than more advantaged people to utilise preventive health

services for CVD and diabetes. For males, the low socioeconomic groups recorded

the least use of preventive health services among the three education and income

groups, including blood cholesterol and blood glucose check-ups, while the high

socioeconomic group recorded the greatest use of preventive health services. There

was no apparent relationship between education level and blood pressure check-up,

while individuals from low-income families were less likely to go for a blood

pressure check-up. For females, most of the results suggested that the low

socioeconomic groups were less likely than the high socioeconomic groups to have

v

blood cholesterol and blood glucose check-ups. However, this was not the case for

blood pressure check-ups. The results showed that the low and middle

socioeconomic groups were more likely than the high socioeconomic groups to have

BP check-ups. However, the low socioeconomic groups were still less likely than the

middle socioeconomic groups to have a blood pressure check-up. Overall, there was

a similar pattern between education and income and the use of GP-based preventive

health services among both males and females.

The findings from the examination of the mediating factors between SEP and the

GP-based use of preventive health services suggested that socioeconomically

disadvantaged adults (both low level of education and low income) are more

concerned about transport and travel time to health care, and accessibility to health

care in terms of finding a GP who bulk bills, the cost of seeing a GP and having a

choice of GP. They are also less likely to have a regular place of care and social

support. These potential factors are likely to result in a lesser use of preventive health

services than their high-SEP counterparts. In addition, the findings also suggested

that respondents with a low level of education have less-positive attitudes towards

health care, and that those from low-income families do not have a regular care

provider and are less likely to visit their GP for a preventive check-up in relation to

CVD and diabetes in Australia.

Conclusions: Strategies for reducing socioeconomic health inequalities are partly

associated with changing social and economic policies, empowering individuals,

strengthening social and family networks, and improving the equity of the health care

system. Strategies have been recommended for implementation in general practice

that are directed at targeting the needs of disadvantaged groups; for example,

providing longer consultation time and actively offering information on preventive

care. Implementation of health promotion programs is needed in disadvantaged areas

to keep the community informed about the availability of health services and to make

health services more accessible. The health care system needs to be geographically

accessible through improvements to the transport system. In addition, improving

access to a regular source of primary health care is likely to be an important step in

encouraging low-SEP individuals to use preventive health services.

vi

TABLE OF CONTENTS

1 INTRODUCTION .................................................................... 1

1.1 Background to research....................................................................................... 2 1.1.1 Understanding socioeconomic health inequalities...................................................... 2 1.1.2 Socioeconomic position and health ............................................................................ 4

1.1.2.1 Burden of cardiovascular disease and socioeconomic health inequalities......................... 7 1.1.2.2 Burden of Type 2 Diabetes and socioeconomic health inequalities ................................ 12

1.1.3 Socioeconomic position and health service utilisation ............................................. 13

1.2 Rationale and objectives for research program .............................................. 17

1.3 Structure of the thesis ........................................................................................ 18

2 SOCIOECONOMIC POSITION AND UTILISATION OF PREVENTIVE HEALTH SERVICES IN GENERAL PRACTICE: A REVIEW OF THE LITERATURE................................................. 19

2.1 Introduction........................................................................................................ 20

2.2 Primary health care and general practice in Australia .................................. 20

2.3 Overview of prevention and preventive activities ........................................... 24 2.3.1 Understanding of prevention ................................................................................... 24 2.3.2 Guidelines for preventive care for general practice ................................................ 29 2.3.3 The role of the general practitioner in relation to prevention.................................. 32

2.4 Socioeconomic position and use of preventive health services....................... 33 2.4.1 Immunisation........................................................................................................... 34 2.4.2 Dental care............................................................................................................... 35 2.4.3 Screening for cancer prevention.............................................................................. 36 2.4.4 Screening for risk factors for heart disease and diabetes......................................... 38 2.4.5 Preventive checkups and medical examination....................................................... 40

2.5 Determinants of health services utilisation...................................................... 45 2.5.1 Understanding of determinants and multiple levels of influence ............................ 45 2.5.2 Understanding the factors that influence the utilisation of health services ............. 47 2.5.3 Reviewing empirical studies of determinants of preventive health service

utilisation................................................................................................................. 50 2.5.3.1 Individual-level factors.................................................................................................... 51 2.5.3.2 Interpersonal-level factors ............................................................................................... 54 2.5.3.3 Environmental-level factors ............................................................................................ 56 2.5.3.4 System-level factors ........................................................................................................ 60

vii

2.5.3.5 A summary of the potential factors influencing SEP and the use of preventive health services............................................................................................................................ 62

2.6 Models of health services utilisation................................................................. 68 2.6.1 Health Belief Model ................................................................................................ 68 2.6.2 Health services utilisation models ........................................................................... 70 2.6.3 Other models of health services utilisation ............................................................. 77 2.6.4 Summary of the reviewed models ........................................................................... 78

2.7 PhD theoretical framework and research questions....................................... 79

2.8 Summary............................................................................................................. 82

3 STUDY METHODS .............................................................. 84

3.1 Introduction........................................................................................................ 85

3.2 Overview of study design and study population ............................................. 86

3.3 Sample selection ................................................................................................. 87 3.3.1 Sample calculation .................................................................................................. 87 3.3.2 Sample calculation procedure ................................................................................. 87 3.3.3 Sampling frame ....................................................................................................... 88 3.3.4 Sampling procedure................................................................................................. 88

3.4 Survey questionnaire ......................................................................................... 89 3.4.1 Domains for questionnaire ...................................................................................... 89 3.4.2 Identifying questions and scales.............................................................................. 90 3.4.3 Development of questionnaire questions and scales ............................................... 96

3.4.3.1 Formatting questions and scales...................................................................................... 96 3.4.3.2 Structuring of questionnaire ............................................................................................ 96

3.4.4 Expert review of draft questionnaire ..................................................................... 102 3.4.5 Pre-testing of questionnaire................................................................................... 103 3.4.6 Final modification of questionnaire ...................................................................... 105

3.5 Data collection .................................................................................................. 106 3.5.1 Preparation phase .................................................................................................. 106 3.5.2 Mailing contacts .................................................................................................... 107 3.5.3 Data recording and storage.................................................................................... 108

3.6 Data management ............................................................................................ 108 3.6.1 Data coding ........................................................................................................... 109 3.6.2 Data entry .............................................................................................................. 109 3.6.3 Data verification .................................................................................................... 109 3.6.4 Data cleaning......................................................................................................... 110

3.7 Sample exclusion .............................................................................................. 110 3.7.1 Management of missing data................................................................................. 110 3.7.2 Exclusion of respondents with CVD and diabetes conditions............................... 111

3.7.2.1 CVD and diabetes problems and use of preventive health services .............................. 111 3.7.2.2 CVD and diabetes problems and SEP ........................................................................... 113

viii

3.8 Study measurement.......................................................................................... 115 3.8.1 Socioeconomic position ........................................................................................ 115

3.8.1.1 Education....................................................................................................................... 115 3.8.1.2 Family income............................................................................................................... 116

3.8.2 Demographic variables.......................................................................................... 118 3.8.2.1 Age.. .............................................................................................................................. 118 3.8.2.2 Gender ........................................................................................................................... 119

3.8.3 The potential factors of socioeconomic differences in utilisation of preventive health services ....................................................................................................... 119

3.8.3.1 Concerns about availability of health care .................................................................... 124 3.8.3.2 Concerns about accessibility to health care ................................................................... 124 3.8.3.3 Concerns about transport and travel time to health care................................................ 125 3.8.3.4 Perceived interpersonal care.......................................................................................... 126 3.8.3.5 Value of general practitioners ....................................................................................... 126 3.8.3.6 Attitudes towards health care ....................................................................................... 127 3.8.3.7 Value of good health ..................................................................................................... 127 3.8.3.8 Social support................................................................................................................ 128

3.8.4 Other intervening variables: a regular source of care............................................ 129 3.8.5 Outcome variables................................................................................................. 129

3.8.5.1 Blood pressure check-up ............................................................................................... 129 3.8.5.2 Blood cholesterol check-up ........................................................................................... 130 3.8.5.3 Blood glucose check-up ................................................................................................ 130

3.9 Analytical approach and statistical analysis.................................................. 131 3.9.1 Gender separation.................................................................................................. 131 3.9.2 Analytical hypothesis ............................................................................................ 131 3.9.3 Data analysis ......................................................................................................... 132

3.9.3.1 Statistical analysis ......................................................................................................... 132 3.9.3.2 Analytical steps ............................................................................................................. 133 3.9.3.3 Decision-rules for interpretation of study findings........................................................ 134

3.10 Summary ................................................................................................... 135

4 RESULTS: DESCRIPTION OF STUDY SAMPLE AND EXAMINING THE RELATIONSHIP BETWEEN SOCIOECONOMIC POSITIONS AND THE USE OF PREVENTIVE HEALTH SERVICES IN GENERAL PRACTICE............................................................................................................... 137

4.1 Introduction...................................................................................................... 138

4.2 Survey response rate ........................................................................................ 138

4.3 Socio-demographic profile and health status characteristics of respondents............................................................................................................................ 139

4.3.1 Socio-demographic profile .................................................................................... 139 4.3.2 Self-reported preventive health service utilisation ................................................ 141

4.4 Representativeness of sample.......................................................................... 141 4.4.1 Age profile............................................................................................................. 142 4.4.2 Gender profile ....................................................................................................... 142 4.4.3 Education profile ................................................................................................... 143

ix

4.4.4 Household income profile ..................................................................................... 145 4.4.5 Health service utilisation ....................................................................................... 145

4.5 Bivariate associations between SEP, age and the use of preventive health services .............................................................................................................. 146

4.5.1 Blood pressure check-up ....................................................................................... 146 4.5.1.1 Education....................................................................................................................... 146 4.5.1.2 Family income............................................................................................................... 147 4.5.1.3 Age.. .............................................................................................................................. 147

4.5.2 Blood cholesterol check-up ................................................................................... 149 4.5.2.1 Education....................................................................................................................... 149 4.5.2.2 Family income............................................................................................................... 149 4.5.2.3 Age.. .............................................................................................................................. 149

4.5.3 Blood glucose check-up ........................................................................................ 151 4.5.3.1 Education....................................................................................................................... 151 4.5.3.2 Family income............................................................................................................... 151 4.5.3.3 Age.. .............................................................................................................................. 151

4.6 Multivariable relationship between SEP and the use of preventive health services .............................................................................................................. 153

4.6.1 Blood pressure check-up ....................................................................................... 153 4.6.1.1 Education....................................................................................................................... 153 4.6.1.2 Income........................................................................................................................... 154

4.6.2 Blood cholesterol check-up ................................................................................... 156 4.6.2.1 Education....................................................................................................................... 156 4.6.2.2 Income........................................................................................................................... 157

4.6.3 Blood glucose check-up ........................................................................................ 158 4.6.3.1 Education....................................................................................................................... 158 4.6.3.2 Income........................................................................................................................... 160

4.7 Summary........................................................................................................... 161

5 RESULTS: UNDERSTANDING THE RELATIONSHIP BETWEEN SOCIOECONOMIC POSITION AND USE OF PREVENTIVE HEALTH SERVICES ............................................. 163

5.1 Introduction...................................................................................................... 164

5.2 Modelling associations between SEP and the intervening factors............... 164 5.2.1 Concerns about availability of health care ............................................................ 165 5.2.2 Concerns about accessibility to health care........................................................... 166 5.2.3 Concerns about transport and travel time to health care ....................................... 168 5.2.4 Perceived interpersonal care from general practitioners ....................................... 169 5.2.5 Value of general practitioners ............................................................................... 170 5.2.6 Attitudes towards health care ................................................................................ 171 5.2.7 Value of good health ............................................................................................. 172 5.2.8 Social support ........................................................................................................ 173 5.2.9 A regular source of care ........................................................................................ 176

5.3 Modelling associations between the intervening variables and the use of preventive health services................................................................................ 180

5.3.1 Concerns about availability to health care............................................................. 180

x

5.3.2 Concerns about accessibility to health care........................................................... 180 5.3.3 Concerns about transport and travel time to health care ....................................... 181 5.3.4 Perceived interpersonal care.................................................................................. 181 5.3.5 Value of general practitioners ............................................................................... 182 5.3.6 Attitude towards health care.................................................................................. 183 5.3.7 Value of good health ............................................................................................. 183 5.3.8 Social support ........................................................................................................ 184 5.3.9 A regular source of care ........................................................................................ 185

5.4 Modelling associations between SEP and the use of preventive health services adjusting for each of the intervening variables............................... 188

5.4.1 The associations between education and use of GP based preventive health service adjustment for intervening variables among males ............................................... 189


5.4.2 The associations between education and use of GP based preventive health service adjustment for intervening variables among females ............................................ 192


5.4.3 The associations between income and use of GP based preventive health service adjustment for intervening variables among males ............................................... 195


5.4.4 The associations between income and use of GP-based preventive health service adjustment for intervening variables among females ............................................ 199


5.5 Summary........................................................................................................... 203

6 DISCUSSION AND CONCLUSIONS............................... 204

6.1 Introduction...................................................................................................... 205

6.2 Summary of findings....................................................................................... 206 6.2.1 Examining the relationship between socioeconomic position and the use of

preventive health services ..................................................................................... 206 6.2.2 Understanding the relationship between socioeconomic position and use of

preventive health services ..................................................................................... 207

6.3 Discussion of findings....................................................................................... 207 6.3.1 SEP and the use of preventive health services ...................................................... 207 6.3.2 Factors influencing the use of preventive health services by different SEP groups

…………………………………………………………………………………208 6.3.2.1 Concern about transport and travel time to health care ................................................. 208 6.3.2.2 Concern about accessibility to health care .................................................................... 210 6.3.2.3 A regular source of care ................................................................................................ 211 6.3.2.4 Attitudes towards health care ........................................................................................ 213 6.3.2.5 Social support................................................................................................................ 213

xi

6.4 Strengths and limitations................................................................................. 214 6.4.1 Study design and survey implementation.............................................................. 214 6.4.2 Study measurement ............................................................................................... 216 6.4.3 Data analysis ......................................................................................................... 217 6.4.4 Generalisability of study findings ......................................................................... 218

6.5 Study implications............................................................................................ 219 6.5.1 Preventive health service utilisation in general practice ....................................... 219 6.5.2 Health policy and implications for strategies ........................................................ 220 6.5.3 Models of health service utilisation....................................................................... 221 6.5.4 Methodological implications................................................................................. 223

6.6 Conclusions and recommendations for policy and future study ................. 224

APPENDICES ..................................................................................... 226 Appendix 2.1 BEACH study: This was a manuscript that was submitted to BMC: Journal

of Health Services Research.................................................................................. 227 Appendix 3.1 Request letter to Australian Electoral Commission .................................. 253 Appendix 3.2 List of names of experts for questionnaire review.................................... 255 Appendix 3.3 Cover letter for pre-testing the questionnaire ........................................... 256 Appendix 3.4 Ethical clearance for questionnaire pre-testing and mailed survey........... 257 Appendix 3.5 Health Service Utilisation Questionnaire.................................................. 258 Appendix 3.6 Cover letter for the first mailing ............................................................... 269 Appendix 3.7 Thank you and reminder postcard............................................................. 270 Appendix 3.8 Cover letter for the first replacement mailing ........................................... 271 Appendix 3.9 Cover letter for the second replacement mailing ...................................... 272

REFERENCES.................................................................................... 273

xii

LIST OF TABLES

Table 1.1 Prevalence of high blood pressure, high blood cholesterol, overweight and diabetes

by age, 1999/2000, Australia ............................................................................. 10 Table 1.2 Proportion with selected health risk factors, long-term conditions and health

actions by socioeconomic position, 2001 (%) .................................................... 15 Table 2.1 Preventive activity for cardiovascular diseases and diabetes in Australia ............. 31 Table 2.2 List of international studies that have examined the relationship between SEP and

use of preventive health services ......................................................................... 43 Table 2.3 List of Australian studies that have examined the relationship between SEP and

use of preventive health services ......................................................................... 44 Table 2.4 International Literature Review on ‘why do people use health services?’ ............ 49 Table 2.5 Determinants of preventive health services utilisation from the literature review 65 Table 3.1 Domains of the initial questionnaire and their variables and definitions............... 90 Table 3.2 Details of questions and scales used in questionnaire development ...................... 95 Table 3.3 Pilot survey checklist ........................................................................................... 104 Table 3.4 Chi-square test for bivariate associations between chronic conditions and BP, BC

and BG checked by GP in a total sample of 519 respondents ........................... 112 Table 3.5 The logistic regression for multivariable analysis between BP, BC and BG Check-

up and CVD and diabetes conditions adjusted for age and sex......................... 112 Table 3.6 Chi-square test for socioeconomic profile among respondents who had CVD and

diabetes conditions in a total sample of 519 respondents.................................. 113 Table 3.7 The logistic regression for multivariable analysis between CVD and diabetes

problems and SEP variables adjusted for age and sex....................................... 114 Table 3.8 Results of final Principal Component Analysis for the scales ............................. 122 Table 3.9 Mean scores, standard deviations and Cronbach’s Alpha reliability coefficients for

the factorially derived scales ............................................................................. 123 Table 3.10 Classification category for three dependent variables ....................................... 130 Table 4.1 Response rates and useable returns for different mailings................................... 139 Table 4.2 Socio-demographic characteristics of the study sample (n = 381) ...................... 140 Table 4.3 Self-reported preventive health service utilisation of the study sample .............. 141 Table 4.4 Comparison of the age profile of the study sample with the 2001 census data (LGA

of Brisbane) among 25–64-year olds ................................................................ 142 Table 4.5 Comparison of the gender profile of the study sample with 2001 census data (LGA

of Brisbane) among 25–64-year olds ................................................................ 143

xiii

Table 4.6 Comparison of the individual education profile of the study sample with 2001

census data (LGA of Brisbane) among 25–64-year olds................................... 144 Table 4.7 Comparison of respondents in the study sample and other data in terms of health

service utilisation............................................................................................... 146 Table 4.8 Chi-square test for bivariate associations between blood pressure checked by GP

and education, household income and age in a male and female sample.......... 148 Table 4.9 Chi-square test for bivariate associations between blood cholesterol checked by GP

and education, household income, and age in a male and female sample......... 150 Table 4.10 Chi-square test for bivariate associations between blood glucose checked by GP

and education, household income and age in a male and female sample.......... 152 Table 4.11 Logistic regression for multivariable analysis between education and BP check-

up adjusted for age in male and female groups ................................................. 154 Table 4.12 Logistic regression for multivariable analysis between household income and

blood pressure check-up adjusted for age in male and female groups .............. 155 Table 4.13 Logistic regression for multivariable analysis between education and BC check-

up adjusted for age in male and female groups ................................................. 156 Table 4.14 Logistic regression for multivariable analysis between household income and

blood cholesterol check-up, adjusted for age in male and female groups ......... 157 Table 4.15 Logistic regression for multivariable analysis between education and blood

glucose check-up, adjusted for age in male and female groups ........................ 159 Table 4.16 Logistic regression for multivariable analysis between household income and

blood glucose check-up adjusted for age in male and female groups ............... 160 Table 5.1 Comparing education groups in terms of their mean scores on the scale1 measuring

concerns about the availability of health care ................................................... 165 Table 5.2 Comparing income groups in terms of their mean scores on the scale1 measuring

concerns about the availability of health care ................................................... 166 Table 5.3 Comparing education groups in terms of their mean scores on the scale1 measuring

concerns about accessibility to health care........................................................ 167 Table 5.4 Comparing income groups in terms of their mean scores on the scale1 measuring

concerns about accessibility to health care........................................................ 168 Table 5.5 Comparing education groups in terms of their mean score on the scale1 measuring

concerns about transport and travel time to health care .................................... 168 Table 5.6 Comparing income groups in terms of their mean score on the scale1 measuring

concerns about transport and travel time to health care .................................... 169 Table 5.7 Comparing education groups in terms of their mean score on the scale1 measuring

perception of interpersonal care from general practitioners .............................. 170

xiv

Table 5.8 Comparing income groups in terms of their mean score on the scale1 measuring

perception of interpersonal care from general practitioners .............................. 170 Table 5.9 Comparing education groups in terms of their mean score on the scale1 measuring

value of general practitioners ............................................................................ 171 Table 5.10 Comparing income groups in terms of their mean score on the scale1 measuring

perception of value of general practitioners ...................................................... 171 Table 5.11 Comparing education groups in terms of their mean score on the scale1

measuring attitudes towards health care............................................................ 172 Table 5.12 Comparing income groups in terms of their mean score on the scale1 measuring

attitudes towards health care ............................................................................. 172 Table 5.13 Comparing education groups in terms of their mean score on the scale1

measuring value of good health......................................................................... 173 Table 5.14 Comparing income groups in terms of their mean score on the scale1 measuring

value of good health .......................................................................................... 173 Table 5.15 Comparing education groups in terms of their mean score on the scale1

measuring negative aspect of social support ..................................................... 174 Table 5.16 Comparing income groups in terms of their mean score on the scale1 measuring

negative aspect of social support ....................................................................... 175 Table 5.17 Comparing education groups in terms of their mean score on the scale1

measuring positive aspect of social support ...................................................... 175 Table 5.18 Comparing income groups in terms of their mean score on the scale1 measuring

positive aspect of social support........................................................................ 176 Table 5.19 Comparing education and income group in terms of their odds ratios and 95%

confidence intervals on the question measuring a regular place of care ........... 177 Table 5.20 Comparing education and income groups in terms of their odds ratios and 95%

confidence intervals on the question measuring a regular care provider........... 177 Table 5.21 Results of logistic regression modelling between people’s concerns about

availability of health care and the use of preventive health services................. 180 Table 5.22 Results of logistic regression modelling between people’s concerns about

accessibility to health care and the use of preventive health services ............... 181 Table 5.23 Results of logistic regression modelling between people’s concerns about

transport and travel time to health care and the use of preventive health services

........................................................................................................................... 181 Table 5.24 Results of logistic regression modelling between the scale of perceived

interpersonal care and the use of preventive health services............................. 182 Table 5.25 Results of logistic regression modelling between people’s perceptions on value of

general practitioners and the use of preventive health services......................... 183

xv

Table 5.26 Results of logistic regression modelling between people’s attitude towards health

care and the use of preventive health services................................................... 183 Table 5.27 Results of logistic regression modelling between people’s perception of value of

good health and the use of preventive health services....................................... 184 Table 5.28 Results of logistic regression modelling between negative aspect of social support

and the use of preventive health services .......................................................... 184 Table 5.29 Results of logistic regression modelling between positive aspect of social support

and the use of preventive health services .......................................................... 185 Table 5.30 Results of logistic regression modelling between a regular place of care and the

use of preventive health services ....................................................................... 185 Table 5.31 Results of logistic regression modelling between a regular care provider and the

use of preventive health services ....................................................................... 186 Table 5.32 Results of the final multivariable analysis examining the influence by the

intervening factors on the relationship between education and blood pressure

check-up by GPs among males (n = 155).......................................................... 190 Table 5.33 Results of the final multivariable analysis examining the influence by the

intervening factors on the relationship between education and blood cholesterol


intervening factors on the relationship between education and blood glucose


intervening factors on the relationship between education and blood pressure

check-up by GPs among females (n = 226)....................................................... 193 Table 5.36 Results of the final multivariable analysis examining the influence by the

intervening factors on the relationship between education and blood cholesterol


intervening factors on the relationship between education and blood glucose


intervening factors on the relationship between family income and blood pressure


intervening factors on the relationship between family income and blood

cholesterol check-up by GPs among males (n = 155) ...................................... 198

xvi

Table 5.40 Results of the final multivariable analysis examining the influence by the

intervening factors on the relationship between family income and blood glucose


intervening factors on the relationship between family income and blood pressure


intervening factors on the relationship between family income and blood

cholesterol check-up by GPs among females (n = 226) .................................... 201 Table 5.43 Results of the final multivariable analysis examining the influence by the

intervening factors on the relationship between family income and blood glucose

check-up by GPs among females (n = 226)....................................................... 202

xvii

LIST OF FIGURES Figure 1.1 Age-standardised mortality rates for all causes among male groups, comparing the

least to the most socioeconomically disadvantaged areas of Australia, 1998–2000

............................................................................................................................... 5 Figure 1.2 Age-standardised mortality rates for all causes among female groups, comparing

the least to the most socioeconomically disadvantaged areas of Australia, 1998–

2000....................................................................................................................... 6 Figure 1.3 Death rates for cardiovascular diseases by age, 2002............................................. 8 Figure 1.4 Age-standardised mortality rates for CVD for the least and most

socioeconomically disadvantaged areas of Australia, 1985–1987 and 1998–2000,

among males aged 25–64 years ........................................................................... 11 Figure 1.5 Age-standardised mortality rates for CVD for the least and most

socioeconomically disadvantaged areas of Australia, 1985–1987 and 1998–2000,

among females aged 25–64 years........................................................................ 11 Figure 2.1 Trends in the age-standardised prevalence (%) of hypertension, 1980–2000 ..... 26 Figure 2.2: Primary, secondary and tertiary prevention......................................................... 29 Figure 2.3 Relationship between income and last dental check-up in the previous 12 months

among dentate persons aged 15 years or more, 2002 .......................................... 36 Figure 2.4 Percentage of blood pressure screening among different education groups aged 18

and over between 1991 and 1999, United States................................................. 39 Figure 2.5 Determinants of preventive health services utilisation ......................................... 50 Figure 2.6 Major elements of the Health Belief Model ......................................................... 68 Figure 2.7 Behavioural Model of Health Services Utilisation............................................... 70 Figure 2.8 Andersen and Newman utilisation framework (1973).......................................... 72 Figure 2.9 Aday and Andersen access framework (1975) ..................................................... 74 Figure 2.10 An emerging Andersen Behaviour model .......................................................... 76 Figure 2.11 Conceptual model of the factors affecting health care utilisation ...................... 77 Figure 2.12 PhD theoretical framework for understanding the utilisation of preventive health

services in general practice by different SEP groups .......................................... 80 Figure 3.1 Summary of study methods .................................................................................. 85 Figure 3.2 Data collection phase for self-administered mailed survey ................................ 108 Figure 3.3 Qualification distributions of the respondents from the mailed survey.............. 116 Figure 3.4 Household income distribution of the respondents from the mailed survey ...... 117 Figure 3.5 Family income groups for the data analysis ....................................................... 118

xviii

Figure 3.6 Age distribution of total study sample (n = 381)................................................ 119 Figure 3.7 Analytic approaches for the multivariable analysis............................................ 133 Figure 4.1 Trend of level of education between the study sample and the 2001 census (LGA

of Brisbane) ....................................................................................................... 144 Figure 4.2 Modelling associations between education and blood pressure check-up.......... 154 Figure 4.3 Modelling associations between household income and blood pressure check-up

........................................................................................................................... 155 Figure 4.4 Modelling associations between education and blood cholesterol check-up...... 157 Figure 4.5 Modelling associations between household’s income and blood cholesterol check-

up....................................................................................................................... 158 Figure 4.6 Modelling associations between education and blood glucose check-up.......... 159 Figure 4.7 Modelling associations between household’s income and blood glucose check-up

........................................................................................................................... 161 Figure 5.1 Relationships between a regular place of care and the use of preventive health

services among males and females.................................................................... 186 Figure 5.2 Relationships between a regular care provider and the use of preventive health

services among males and females.................................................................... 187

xix

LIST OF CONFERENCE PRESENTATIONS BY CANDIDATE ON MATTERS RELEVENT TO THESIS

Zhang J, Turrell G and Oldenburg B. Understanding the relationship between

socioeconomic status and utilisation of preventive health services related to chronic

disease prevention. Poster presentation for International Society of Behavioural

Medicine 26th Annual Meting and Scientific Sessions, Boston, USA, April 2005.

Zhang J, Oldenburg B, Turrell G and Britt H. The relationship between

socioeconomic status and preventive health services related to cardiovascular disease

provided by GPs. Oral presentation at 3rd Scientific Conference of the Australasian

Society for Behavioural Health and Medicine. Melbourne, Australia, 10–12 February

2005.

xx

LIST OF ABBREVIATIONS

ABS Australian Bureau of Statistics

BC Blood Cholesterol

BEACH Bettering the Evaluation and Care of Health

BP Blood Pressure

BRFSS Behavioural Risk Factor Surveillance System

BG Blood Glucose

CVD Cardiovascular Diseases

GP General Practitioner

GPAS General Practice Assessment Survey

HBM Health Belief Model

HILDA Household, Income and Labour Dynamics in Australian Survey

IRSD Index of Relative Socioeconomic Disadvantage

LGA Local Government Area

NHS National Health Survey

NPCRDC National Primary Care Research and Development Centre

NHMRC National Health and Medical Research Council

PCA Principal Components Analysis

QUT Queensland University of Technology

SEIFA Socio-Economic Indexes for Areas

SEP Socioeconomic Position

SES Socioeconomic Status

SPSS Statistical Package for Social Science

WHA Women’s Health Australia

xxi

STATEMENT OF ORIGINAL AUTHORSHIP

The work contained in this thesis has not been previously submitted to meet

requirements for an award at this or any other higher education institution. To the

best of my knowledge and belief, the thesis contains no material previously

published or written by another person except where due reference is made.

Signature: Date:

xxii

ACKNOWLEDGEMENTS There are many people to whom thanks are due when undertaking a project of this

size. It would be impossible to name everyone, but all the help I have received has

been greatly appreciated.

First of all, I would like to express my gratitude to my primary supervisor, Professor

Brian Oldenburg, for all of his constructive advice and assistance. He helped steer

me toward an interesting research area, and without his encouragement, interest and

enthusiasm, I could not have completed this thesis.

Sincere thanks are also due to my Associate Supervisor, Dr Gavin Turrell, for his

support and for the guidance he has provided throughout my research, as well as for

sharing his expertise in the social determinants of health and analytical research

skills. He has always been encouraging and positive in his response to my research.

Again, I could not have completed this thesis without his assistance.

My sincere thanks and appreciation also go to all who participated in the mailed

survey, for giving up their valuable time to be part of this project, and to the experts

who contributed to the development of my questionnaire.

The data were obtained from the Bettering the Evaluation and Care of Health

(BEACH) Project, which was a sub-study of my research, through the Family

Medicine Research Centre, Department of General Practice, The University of

Sydney. I would like to thank those involved in the project for their technical advice

and support, in particular, Associate Professor Helena Britt and Christopher Harrison.

I would like to express my thanks as well for the scholarship provided by the

National Health and Medical Research Council (NHMRC), which helped me to

achieve my goal.

xxiii

I would also like to express my gratitude to all of the staff from the School of Public

Health, QUT, especially Dr Diana Battistutta, for her statistical advice, Dr Elizabeth

Eakin for her initial supervision and support, and Dr Jing Sun, Dr Gary Day, Dr Scott

Wearing and Dr Monika Janda, for their professional comments.

In addition, I would also like to thank my shared office mates, Dominique Bird,

Rebecca Watson, Julie Macaranas and my fellow PhD student Elisabeth Winkler, for

their support, friendship and understanding. I also thank Genny Carter, Carmel

Stafford, Jill Nalder, Ha Le and Tuyet Le for the administrative support, as well as

Stefan Rucinski for his kindness and understanding on many occasions.

Heartfelt thanks go to my husband Keqin Jia, my son Jack, my parents and my

friends for their patience and understanding. Without their help, I could not have

completed this thesis.

Finally, I am so grateful to QUT Counselling Services for their language support,

especially Mr Martin Reese for the thesis proofreading and editing, QUT librarian

Jon Peak for solving my End Note and formatting queries, and QUT Campus

Services for the night shuttle bus, especially Mr John Rogers for his driving.

1

1 INTRODUCTION

2

1.1 Background to research

The first section of this chapter provides an overview of socioeconomic health

inequalities, socioeconomic position (SEP) and health, as well as of health service

utilisation, in relation to cardiovascular disease and diabetes, two of Australia’s

major health challenges. In addition, a conceptual framework is proposed as a way of

understanding the relationship between socioeconomic position and health.

1.1.1 Understanding socioeconomic health inequalities

There has been substantial overall improvement in the health of the Australian

population in the past two decades; however, different population subgroups have

not experienced these improvements equally. Health inequalities are systematic

differences in health status among different groups in the population, including

factors such as socioeconomic position, ethnicity, age and gender (Australian

Institute of Health and Welfare, 1995; Mathers, 1994; Turrell & Mathers, 2000;

Wilson, Oldenburg, & Lopez, 2003). A recent analysis of causes of death in

Australia indicates that there has been a significant increase in socioeconomic health

inequalities for certain age groups and conditions since the mid-1980s (Turrell &

Mathers, 2000). Despite the level of potentially avoidable mortality decreasing

overall, and across all socioeconomic groups during the 20 years covered by the

research, the rate of reduction has been higher in the highest socioeconomic group

compared to both the lowest socioeconomic group and the rest of the population

(Wilson et al., 2003). Tackling socioeconomic health inequalities remains one of

Australia’s most challenging public health issues (Oldenburg, McGuffog, & Turrell,

2000).

Turrell, Oldenburg, McGuffog and Dent (1999) have developed a conceptual

framework for understanding socioeconomic health inequalities consisting of three

closely interrelated levels that can influence health outcomes: the ‘upstream’,

‘midstream’ and ‘downstream’. At the upstream level, the relevant factors include

educational background, employment, occupation and working conditions, income,

housing and neighbourhood living conditions. These determinants are affected by

even more ‘upstream’ factors, including government policies related to income

3

distribution and welfare spending, as well as global economic and political forces.

These determinants then indirectly impact on health via psychosocial processes and

health behaviours at the ‘midstream’ level. At this level, the health care system also

plays an important role, affecting the extent of socioeconomic health inequalities

within a society. The framework also identifies the fact that cultural influences,

which in turn influence beliefs and values, have an impact on all levels.

Consequently, some reactions, such as illnesses and diseases, result from changes or

disruptions to the functioning of various physiological systems at a ‘downstream’

level. The poorer health profile of low socioeconomic groups is likely be a result of

‘more sustained or longer-term adverse changes to physiological and biological

functioning due to less favourable psychosocial influences and more harmful health-

related behaviours’ (Turrell & Mathers, 2000; Turrell et al., 1999). This framework

has evolved out of the work of McKinlay and Marceau (McKinlay & Marceau, 1999;

McKinlay & Marceau, 2004), who pointed out that ‘upstream’ healthy public policy

is an essential part of effective population health and wellbeing.

There have been a number of alternative theories to explain socioeconomic health

inequalities. Three main aetiological theories have been proposed: the so-called

‘materialist’, ‘behavioural or lifestyle’ and ‘psychosocial’ explanations (Dutton,

Turrell, & Oldenburg, 2005). The materialist explanation refers to the fact that those

with higher incomes are able to purchase better food and housing, live in safer

environments and have better access to health care. The behavioural or lifestyle

explanation refers to those health behaviours including smoking, diet and use of

health services, which are in turn influenced by many other factors, such as cognition,

attitudes and beliefs. The psychosocial explanation emphasises the importance of

factors such as the control, empowerment and social integration that may result from

low social standing and low autonomy in important areas of life, such as work. These

three foundation explanations of socioeconomic health inequalities are not mutually

exclusive; but rather can be seen as interrelated (Turrell et al., 1999).

4

1.1.2 Socioeconomic position and health

The links between poor health and socioeconomic disadvantage have been well

described in international and Australian research over the past 40 years. There are

large gradients in life expectancy by income level, education and occupation, which

have been identified repeatedly in many countries. This very large body of research

clearly demonstrates that higher socioeconomic position (SEP) is associated with

better health, and that lower SEP is associated with relatively poorer health (Harris,

Sainsbury, & Nutbeam, 1999; Hertzman, Frank, & Evans, 1994; Kaplan & Keil,

1993; Lenthe et al., 2004; Marmot, 2001, 2002; Reidpath, 2004).

A number of overseas studies have reported on the inverse effects of socioeconomic

factors on mortality. Lower levels of education, occupation and income are

associated with higher levels of mortality (Kaplan & Keil, 1993; Martijn, Anton, &

Mackenbach, 2003; Rodriguez, Paymundo, Monsalve, & Gervas, 1996; Wilkinson,

2001). Kaplan and Keil (1993) have reviewed a literature of socioeconomic factors

and diseases from the United States (US), Canada, the United Kingdom (UK),

Germany and the Nordic countries. They have concluded that there is a large body of

evidence suggesting an inverse relation between many of the indicators of SEP such

as education, income and occupation, and all-cause mortality. The socioeconomic

gradient for many different health outcomes has also been identified in a US national

sample (Adler et al., 1994; Centers for Disease Control and Prevention, 2005). The

study outcomes show that lower levels of both income and education are associated

with higher levels of mortality (Shi, Starfield, & Kawachi, 1999; Wen, Browning, &

Cagney, 2003; Williams & Collins, 1995). For example, in the United States, an

approximate doubling of risk of coronary heart disease was observed in people in the

poorest socioeconomic groups compared to those in the richest groups (Centers for

Disease Control and Prevention, 2002; Smeeth & Health, 1999). Grundy and

Sloggett (2003) also analysed the data from the Health Survey for England, which

showed that socioeconomic indicators, particularly receipt of income support, were

most consistently associated with raised odds of poor health outcomes (Van der

Heyden, Demarest, Tafforeau, & Van Oyen, 2003). In addition, in the United

Kingdom, Martijn et al. (2003) verified that death rates at all ages are two to three

times higher among people in lower social classes than among those in the higher

5

classes. Furthermore, a study of trends in socioeconomic inequalities in self-assessed

health in 10 European countries (Finland, Sweden, Norway, Denmark, England, the

Netherlands, West Germany, Austria, Italy and Spain) provided evidence that the

odds comparing extreme income quintiles increased for all countries together (Kunst

et al., 2005). Mackenbach et al.’s study (2005) also supported this finding and

showed that a higher household income is associated with better self-assessed health

in some European countries, including Belgium, Denmark, England, Finland, France,

the Netherlands and Norway.

Australian research has also demonstrated that socioeconomically disadvantaged

groups experience significantly higher mortality and morbidity rates (Draper, Turrell,

& Oldenburg, 2004; Furler et al., 2002; Hyndman & Holman, 2001; Korten et al.,

1998; Rosenman & Mackinnon, 1992). Draper et al. (2004) have documented that

low-SEP individuals had higher rates of mortality for almost all causes of death for

the years 1998–2000 (see Figures 1.1 and 1.2). The all-cause mortality rate for males

aged 25–64 years in the most disadvantaged population quintile (fifth) was 75 per

cent higher than the rate for males in the least disadvantaged quintile (377 compared

to 215 deaths per 100 000 population). For females, the all-cause mortality rate in the

most disadvantaged quintile was 52 per cent higher (204 compared to 135 deaths per

100 000 population).

0

100

200

300

400

500

600

700

800

less than 1 0-14 15-24 25-64 65-74 75 and over

Male age goups

Deat

hs p

er 1

00,0

00 p

erso

ns

Least disadvantaged Most disadvantaged

Figure 1.1 Age-standardised mortality rates for all causes among male groups, comparing the least to the most socioeconomically disadvantaged areas of Australia,

1998–2000

6

0

100

200

300

400

500

600

700

less than 1 0-14 15-24 25-64 65-74 75 and over

Female age groups

Dea

ths

per 1

00,0

00 p

erso

ns


Figure 1.2 Age-standardised mortality rates for all causes among female groups, comparing the least to the most socioeconomically disadvantaged areas of Australia,

1998–2000

Life expectancy at birth during the years 1998–2000 varied from 79.2 years for

males in the least disadvantaged quintile to 75.3 years for males in the most

disadvantaged quintile. For females, the corresponding figures were 83.6 and

81.6 years respectively. The Australian Bureau of Statistics (ABS) found that of the

total years of potential life lost due to premature death in 1997–1999, 24 per cent

came from the most disadvantaged quintile of the population, whereas only 15 per

cent came from the least disadvantaged quintile (Australian Institute of Health and

Welfare, 2004).

In addition, a number of socioeconomic indicators have been associated with

lifestyle factors that affect health. For example, smoking has been linked to lower

socioeconomic position and to many health outcomes, while some studies have

linked unemployment to alcohol consumption (Jerrett, Eyles, & Cole, 1998).

Australian studies have also found that socioeconomically disadvantaged groups are

more likely to engage in health-damaging behaviours and experience poorer

psychosocial health (Mathers, 1994; Turrell & Mathers, 2000). For example, adults

of low socioeconomic position are less likely to be physically active, more likely to

consume moderate to heavy amounts of alcohol, more likely to smoke cigarettes,

have higher daily cigarette consumption, and less likely to purchase or consume

7

foods that are consistent with dietary guideline recommendations (Harris et al., 1999;

Marmot et al., 1991; Turrell, 1998; Turrell & Mathers, 2000; Turrell & Mathers,

2001; Turrell et al., 1999; Wiggers & Sanson-Fisher, 1997; Winkleby, Kraemer, Ahn,

& Varady, 1998). People living in the most disadvantaged areas are more likely to

engage in unhealthy behaviours, including smoking, high alcohol intake, low

exercise participation, be overweight and have low or no regular intake of fruit

(Australian Institute of Health and Welfare, 2004).

The results from the 2001 Australian National Health Survey indicated that

socioeconomically disadvantaged people report more often with long-term health

conditions such as diabetes, diseases of the circulatory system, arthritis and diseases

of the ear, compared to the least disadvantaged groups (AIHW, 2004). There is also

evidence that low-SEP groups are more likely to engage in the key health behaviours,

which are linked to cardiovascular disease (CVD) and diabetes (Turrell & Mathers

2000). CVD and diabetes are chronic diseases with large differentials in disease

burden between higher and lower SEP groups in Australia (Mathers, Vos, &

Stevenson, 1999).

1.1.2.1 Burden of cardiovascular disease and socioeconomic health inequalities

Cardiovascular disease (CVD) is still Australia's greatest single health problem

(Australian Institute of Health and Welfare, 2004, 2005; Mathers, Vos, Stevenson, &

Begg, 2000; Mathers et al., 1999). It causes more deaths than any other disease

(accounting for 50 292 deaths in 2004), and in 2002 affected approximately

3.67 million Australians, accounting for 38 per cent of all deaths (Access Economics,

2005; Australian Institute of Health and Welfare, 2005). CVD is also one of the

leading causes of disability, with around 1.1 million Australians affected by one or

more disabling conditions attributed to CVD (AIHW, 2004). CVD contributes to

significant illness, disability, poor quality of life and premature death, and is the most

expensive disease group in Australia in terms of health expenditure. CVD is

estimated to be responsible for 602 558 years of healthy Australian life lost (DALYs)

in 2004—22 per cent of the total burden from all illness and injury in Australia. Of

these years, 81 per cent (490 711) were lost due to the premature death of people

with CVD, and the remaining 19 per cent of healthy years were lost due to disability

8

(Access Economics, 2005). Figure 1.3 shows the death rates for cardiovascular

diseases by age in 2002 (AIHW, 2005). As can be seen, the highest death rates for

CVD are among the oldest age group (people over 85 years); the second highest is

those aged 75–84 years, and then those aged 65–74 years. Males are more likely to

die from CVD than females in all age groups.

0

1000

2000

3000

4000

5000

6000

7000

8000

<25 25-34 35-44 45-54 55-64 65-74 75-84 >85

Age groups

Num

ber p

er 1

00,0

00 p

opul

atio

n

Male Female

Figure 1.3 Death rates for cardiovascular diseases by age, 2002

The current situation notwithstanding, in recent years there has been much progress

in improving the cardiovascular health of Australians. Death rates have fallen

significantly, some risk factors have been reduced and there have been major

advances in treatment and care.

To prevent CVD effectively, it is necessary to understand the nature of the problem

and the relationship of risk factors to the development of diseases. ‘Cardiovascular

disease’ refers to all conditions and diseases involving the heart and blood vessels.

There are many forms and causes of this diverse group of diseases. However, in

developed countries such as Australia, the main underlying problem is

atherosclerosis (AIHW, 2005). This is a condition that leads to abnormal build-ups of

fat, cholesterol and other substances, called plaque, in the inner lining of the arteries.

Plaque can occur in the arteries supplying the brain, the legs, the kidneys and, in the

case of the heart itself, the coronary arteries. The process leading to atherosclerosis is

slow and complex, often starting in childhood, and progressing with age. The major

cardiovascular diseases include coronary heart disease, stroke, heart failure,

9

peripheral vascular disease, rheumatic fever and rheumatic heart disease, and

congenital heart disease. Of the various forms of CVD, heart disease is the most

prevalent in terms of individual and community mortality rates and places the

greatest burden on health care resources. Coronary heart disease (CHD) is the most

common form of heart disease in Australia, claiming 25 439 lives in 2003. In 1998,

an estimated 224 400 Australians had disabling conditions associated with coronary

heart disease. It is the leading cause of premature death and disability. However,

since the late 1960s, death rates from coronary heart disease have fallen by around

70 per cent (AIHW, 2005). Its major clinical forms are heart attack and angina. A

heart attack occurs when a coronary plaque suddenly breaks open. This brings on a

blood clot that completely blocks blood flow to the heart muscle downstream.

Angina occurs when a plaque has markedly narrowed a coronary artery to the point

where, although the blood flow can usually meet most daily demands, it cannot

increase to meet extra demands incurred by physical activity or strong emotion,

resulting in temporary chest pain.

The prevalence of cardiovascular conditions is associated with various risk factors,

including high blood pressure, high blood cholesterol, cigarette smoking, family

history of heart disease, obesity, physical inactivity, excessive alcohol intake and

diabetes (Australian Institute of Health and Welfare, 1995). According to the AIHW

publication, Australian Facts (2004), more than 10 million adult Australians had at

least one of the following cardiovascular risk factors in 1995: tobacco smoking,

physical inactivity, high blood pressure or excess weight. About four in five men and

three in four women had at least one of these risk factors. In the period 1999–2000,

almost three million Australians (aged 25 and over) had high blood pressure or were

on medication for that condition. High blood pressure increases the risk of CVD by

two to four times. Six million people had blood cholesterol levels higher than the

upper limit recommended by the National Heart Foundation of Australia.

Table 1.1 presents the prevalence of high blood pressure, high blood cholesterol,

excess weight and diabetes by age among Australian males and females in

1999-2000. As can be seen, the males are more likely than the females to suffer from

high blood pressure, high blood cholesterol, obesity and diabetes. The older age

10

groups (55–74; 75+ years) are more likely to have these risk factors than the younger

groups (25–54 years).

Table 1.1 Prevalence of high blood pressure, high blood cholesterol, overweight and diabetes by age, 1999/2000, Australia (AIHW, 2004)

( .

People in lower socioeconomic groups are more likely to die from cardiovascular

disease (Draper et al., 2004; Moon & Waters, 2006; Turrell & Mathers, 2001). In

2000–2002, death rates from CVD in the most disadvantaged areas were 21.4 per

cent higher than in the least disadvantaged (AIHW, 2004). Higher mortality rates

from coronary heart disease and stroke were also observed among lower

socioeconomic groups (Moon & Waters, 2006; Wilson et al., 2003). People aged

25-64 years, living in the most disadvantaged areas, die from cardiovascular disease

at around twice the rate of those living in the least disadvantaged areas (Australian

Institute of Health and Welfare, 2001; Moon & Waters, 2006).

Figures 1.4 and 1.5 indicate that age-standardised mortality rates decreased

dramatically from the period 1985–1987 to the period 1998–2002 for both the least

and most socioeconomically disadvantaged areas of Australia among males and

females aged 25–64 years. Nevertheless, the mortality rates for CVD among both

males and females aged 25–64 years among the most socioeconomically

halla

This table is not available online. Please consult the hardcopy thesis available from the QUT Library

11

disadvantaged areas are much greater than those of the least socioeconomically

disadvantaged areas for both periods (Draper et al., 2004).

0

50

100

150

200

250

1985-1987 1998-2000

Males aged 25-64 years

Dea

ths

per

100,

000

pers

ons


Figure 1.4 Age-standardised mortality rates for CVD for the least and most socioeconomically disadvantaged areas of Australia, 1985–1987 and 1998–2000, among

males aged 25–64 years

0

10

20

30

40

50

60

70

80

90

1985-1987 1998-2000

Females aged 25-64 years

Deat

hs p

er 1

00,0

00 p

erso

ns


Figure 1.5 Age-standardised mortality rates for CVD for the least and most socioeconomically disadvantaged areas of Australia, 1985–1987 and 1998–2000, among

females aged 25–64 years

12

1.1.2.2 Burden of Type 2 Diabetes and socioeconomic health inequalities Diabetes, a major risk factor for heart disease, is also a serious and costly health

problem accounting for 5.8 per cent of the overall disease burden in 2003 in

Australia; it affects 7.5 per cent of the Australian population aged 25 years or older

(Australian Institute of Health and Welfare, 2002, 2006). Diabetes was the

underlying cause of 3599 deaths in 2004 (2.7 per cent of all deaths). An estimated

one million Australians have diabetes, yet half of these individuals are unaware of it

(Australian Institute of Health and Welfare, 2006).

There are two main types of diabetes—type 1 diabetes and type 2 diabetes—and they

have different causal mechanisms. Type 1 diabetes often occurs early in life and is

marked by a total or near-total lack of insulin. It results from the body destroying its

own insulin-producing cells in the pancreas. People with this form of diabetes require

daily insulin therapy to survive. Type 1 accounts for around 10–15 per cent of all

people with diabetes. Based on self-reported information from the 2001 Australian

National Health Survey, it is estimated that around 95 000 Australians (0.5 per cent

of the population) have type 1 diabetes (Australian Institute of Health and Welfare,

2005). Type 2 diabetes is marked by reduced levels of insulin (insulin deficiency)

and/or the inability of the body to use insulin properly (insulin resistance). The

disease is most common among people aged 40 years and over and accounts for

85-90 per cent of all people with diabetes. Many people with this form of diabetes

eventually need insulin therapy to control their blood glucose levels. Information

from the 1999–2000 Australian Diabetes, Obesity and Lifestyle (AusDiab) study,

which included measuring blood glucose levels, suggests that more than 850 000

Australians aged 25 years and over (7.2 per cent of that population) have type 2

diabetes. Around half of the people in the AusDiab study who tested positive for

diabetes were not aware they had the disease.

Individuals with type 2 diabetes are at increased risk of developing coronary heart

disease, stroke and peripheral vascular disease. Lifestyle factors such as obesity,

physical inactivity, high blood pressure and poor diet are major modifiable risk

factors for development of the disease or its complications. In 1999–2002 over

7 million Australians aged 25 or over were overweight or obese. These people were

13

at increased risk of developing type 2 diabetes. In 2000, 44 per cent of Australians

aged 18–75 years (around 5.8 million people) did not undertake physical activity at

high enough levels to receive health benefits. Almost 15 per cent of people did not

undertake any physical activity at all in their leisure time, increasing their risk of

developing type 2 diabetes. During 1997 and 1998, deaths with diabetes as the

underlying cause were 44 per cent higher among the lowest socioeconomic group of

the Australian population than among the highest socioeconomic group (Australian

Institute of Health and Welfare, 2001; Australian Institute of Health and Welfare and

National Heart Foundation, 2004; O'Brien, Thow, & Ofei, 2006).

Nevertheless, much of the burden caused by cardiovascular disease and type 2

diabetes is potentially preventable through primary prevention, secondary prevention,

tertiary prevention and rehabilitation. This is related to early protection from risk

factors of CVD and diabetes such as smoking and lack of physical activity, screening

for and early treatment of disease, and curing disease or extending life through

treatment (Ockene, 1992; Orchard, 1992). Due to low-SEP individuals being more

involved in unhealthy behaviours, health promotion and preventive activities need to

target the low-SEP groups to achieve health behaviour changes.

1.1.3 Socioeconomic position and health service utilisation

Evidence from both international and Australian studies that have investigated the

relationship between socioeconomic status and use of health services generally

indicates that there are higher levels of hospital admissions and medical consultations

by more disadvantaged groups. Conversely, individuals who are more disadvantaged

typically make less use of preventive care and screening services (Australian Institute

of Health and Welfare, 2004; Dunlop, Coyte, & McIsaac, 2000; Furler et al., 2002;

Goddard & Smith, 2001; Halpin, 1995; Harris et al., 1999; Hulka & Wheat, 1985;

Langlie, 1977; Marmot & Bell, 2006; Mathers, 1994; Mathers & Schofield, 1998;

Riessman, 1974; Turrell, Patterson, Oldenburg, Gould, & Roy, 2003; Wiggers,

Sanson-Fisher, & Halpin, 1995).

Dunlop et al. (2000) analysed the results of the Canadian National Population Health

Survey and demonstrated a positive relationship between health needs and the use of

14

primary care services under a universal publicly funded health care system. Those

with lower incomes were more likely to be frequent users of primary care. This was

true even after adjusting for the greater utilisation of primary care services by those

lower socioeconomic groups and, therefore, their higher exposure to the risk of

referral. In contrast, the utilisation of specialist visits was greater for those in higher

socioeconomic groups. Goddard and Smith (2001) concluded that higher rates of

general practitioner (GP) consultation are typically associated with greater

deprivation and with belonging to a lower socioeconomic group; low utilisation of

preventive services is linked to deprivation at an area level, and to poor

socioeconomic circumstances at an individual level. US studies also provide

evidence that the low-SEP groups are more likely to consult medical services due to

their increased morbidity; however, they are less likely to use preventive health

services such as screening and dental care, as well as less likely to see some

specialists (Campbell & Roland, 1996).

It was found from the 2001 Australian National Health Survey that the

socioeconomically disadvantaged report more use of medical services, but less use of

preventive health services, such as breast cancer screening and Pap-smear testing—

see Table 1.2 (AIHW, 2004). Harris et al. (1999) also suggest from Australian data

that people who are disadvantaged make poor use of preventive health services such

as screening services, even though they are more likely to suffer a higher burden of

disease.

15

Table 1.2 Proportion with selected health risk factors, long-term conditions and health actions by socioeconomic position, 2001 (%) (AIHW, 2004)

The fact that lower SEP groups are more likely to involve in unhealthy behaviour

and less likely to access and utilise preventive health services probably contributes,

at least in part, to their poorer health outcomes. To be less likely to access and utilise

preventive health services can delay diagnosis of diseases that will typically occur

later than for higher SEP patients (Campbell & Roland, 1996). As we have already

seen, there is clear evidence that many preventive activities can and do alter the

natural history of disease, and reduce the morbidity and mortality associated with a

number of diseases (Burke & Fair, 2003; Okoro, Strine, Young, Balluz, & Mokdad,

2005; Royal Australian College of General Practitioners, 1998). Timely and effective

preventive care and early disease management may reduce the risk of hospitalisation

by preventing an illness, controlling an acute episodic illness, or managing a chronic

condition (Wilson et al., 2003). Utilisation of preventive health services is very

important for the early detection of chronic diseases (Kaplan, 2000; Oldenburg &

Burton, 2004; Rose, 1992; Royal Australian College of General Practitioners, 1998).

For example, early recognition and management of hypertension, hyperlipidaema,

halla


16

anti-smoking strategies and changes in lifestyle have contributed to the decline in the

morbidity and mortality associated with ischaemic heart disease. This is a potential

pathway to the reduction of socioeconomic health inequalities. However, it has been

argued that preventive health services have a limited impact on the existence of

socioeconomic health differentials (Furler et al., 2002; Wiggers, Sanson-Fisher, &

Halpin, 1995). Nevertheless, there is evidence that improving the accessibility of

preventive health services has the potential not only to reduce the effects of

disadvantage on health, but also to reduce existing differentials in health status

(Wiggers et al., 1995). US data suggests that morbidity and mortality can be reduced

by participation in health-promoting behaviours and by timely use of preventive

health care services (Centres for Disease Control and Prevention, 2004). Furthermore,

an ecologic study using the US states as the units of analysis provides further

evidence that primary care services have a strong and significant direct influence on

life expectancy and total mortality (Shi et al., 1999). Prevention is further discussed

in Chapter 2, Section 2.3.

There is an increasing amount of evidence that primary health care has an important

role in addressing and reducing health inequalities, and that strengthening primary

health care should be a key intervention strategy for reducing health inequalities (The

Health Inequalities Research Collaboration (HIRC), 2004). To achieve this, it is

necessary to address the differences in access or utilisation of preventive health

services; in particular, targeting utilisation inequalities among socioeconomically

disadvantaged groups is essential.

In order to develop strategies that improve the use of preventive health services by

low-SEP groups, it is first necessary to examine the determinants of utilisation.

Although some international evidence for understanding why people use health

services is generally available (this is discussed in Chapter 2), the picture is less clear

in Australia.

17

1.2 Rationale and objectives for research program

While previous research has established that there is a strong relationship between

SEP and utilisation of preventive health services, the reasons for this relationship are

still uncertain (Bullough, 1972; Turrell & Mathers, 2000). This PhD research

program focuses on describing the relationship between socioeconomic position

(SEP) and utilisation of preventive health services related to cardiovascular disease

and diabetes. Such research findings can then be used to improve our understanding

of the potential factors involved in the uptake and utilisation of preventive health

services by different socioeconomic groups. Based on the existing health services

research models examined in Chapter 2, a model has been adapted to the Australian

situation in order to identify the specific factors and variables to be tested more

formally in this thesis.

This thesis considers the use of preventive health services by different

socioeconomic groups in relation to the prevention of CVD and diabetes in the

general practice setting.

The specific objectives of the research are to:

1. Identify existing relevant measures concerning the use of, and differences in

the utilisation of preventive health services and understand measurement

issues in this field of research;

2. Develop a new questionnaire concerning the use of preventive health services

in relation to CVD and diabetes;

3. Evaluate theoretical models that have been used to understand and predict the

utilisation of preventive health services by different population subgroups;

4. Describe and compare the differential uptake and use of preventive health

services in general practice settings by different socioeconomic groups;

5. Examine the relative factors impacting upon the relationship and improve our

knowledge and understanding of the determinants of uptake and utilisation of

preventive health services by different socioeconomic groups.

18

1.3 Structure of the thesis

The thesis consists of the following six chapters:

Chapter 1—Introduction: This chapter provides background information in

relation to socioeconomic position and health, and an overview of the PhD research

program aim and objectives.

Chapter 2—Literature Review: Chapter 2 presents a comprehensive and critical

literature review in relation to socioeconomic position and use of preventive health

services. The chapter also considers the literature relevant to the research setting and

the determinants and models of health service utilisation. Finally, the PhD theoretical

framework and research questions are discussed.

Chapter 3—Study Methods: This chapter discusses the study design, study

methods, and study sample, overviews the development of a new questionnaire and

presents the results of scale validation and reliability testing. A standard data

collection procedure for conducting a mail survey is outlined. The chapter also

presents measurement of main variables, an analytic plan and statistical approaches

for data analysis.

Chapter 4—Study Results 1: This chapter reports the survey participation and

representativeness of respondents of the self-administered mailed survey, and

examines the relationship between socioeconomic position and utilisation of

preventative health services based on the data collected from the survey.

Chapter 5—Study Results 2: This chapter explains the determinants of the

association between SEP and use of preventive health services based on the data

collected from the self-administered mailed survey.

Chapter 6—Discussion and Conclusions: This chapter discusses the present study

findings, strengths and limitations, study implications and recommendations for

policy and future research.

19

2 SOCIOECONOMIC POSITION AND UTILISATION OF PREVENTIVE HEALTH SERVICES IN GENERAL PRACTICE: A REVIEW OF THE LITERATURE

20

2.1 Introduction

To understand more about the relationship between SEP and the utilisation of

preventive health services in Australia, and the complex pathways involved, it is

important to consider the following key questions:

• How are Australia’s primary health care system and general practice

organised in relation to preventive health services? (Section 2.2.)

• What is the evidence base for prevention of diseases such as CVD and

diabetes? (Section 2.3.)

• What is the evidence from current studies about the use of preventive health

services by different SEP groups? (Section 2.4.)

• What factors might influence utilisation of preventive health services by

different SEP groups? (Section 2.5.)

• What are some of the existing models for understanding health service

utilisation? (Section 2.6.)

A comprehensive literature review was conducted for relevant published articles

on specific topics as discussed in this literature. The search was conducted

separately based on each sub-topic. The key search terms used were, [‘primary

health care’ or ‘general practice’],[‘prevention’ or ‘preventive activities’],

[‘preventive health services or preventive health care and socioeconomic status

or position’], [‘socioeconomic status and health’] and [determinants of health

services utilisation’. The search was conducted via the EBSCO HOST search

engine, including the Medline, PsychINFO, Primary Search and PsycARTICLES

databases. The specific search for the guidelines for preventive care was also

conducted nationally and internationally. Consequently, relevant studies and

articles from 1972 to 2004 were examined as discussed below .

2.2 Primary health care and general practice in Australia

Primary health care should be the major function and activity of a country’s health

system, and it is the principal vehicle by which to deliver the majority of health care

21

(World Health Organization, 2003b). Generally speaking, the primary health care

system is most people’s first contact point with the health care system and it also

fulfils a ‘gatekeeping’ function (Starfield, 1992). A number of international studies

have demonstrated that there is an association in developed countries between

publicly funded and well-organised primary care, and improved access and reduced

health inequalities compared to more privatised systems (Harris & Furler, 2002).

Since the Declaration of Alma-Ata in 1978 (World Health Organization, 2003a), the

health situation in many countries has changed considerably. Over the past 25 years

there have been major changes in the patterns of disease, in demographic profiles, in

exposure to major risks and in the socioeconomic environment. Governments have

also substantially changed their roles and responsibilities in relation to public health

and the organisation and delivery of health care, thereby changing the context for

framing and implementing health policy.

The primary health care system in Australia consists of a complex mix of public and

private services, and of federal and state government-funded programs. The three

major components of the system are:

1. General medical practice, which is typically provided by general

practitioners (GPs) and other doctors within a fee-for-service framework.

2. Community health services, which are primarily staffed by non-medical

health professionals and are funded and managed by state governments

and/or an area regional health service and some non-government agencies.

3. Community care programs, which exist primarily to support aged and

disabled people. They are mainly funded by the federal government’s Home

and Community Care Program. They are delivered via state-based human

services/community services and/or some health departments, and are often

operated by either local government or non-government agencies.

The system generally provides universal access to medical services and to subsidised

pharmaceuticals; as well as financial assistance to public hospitals, residential care

facilities, hostels, and home and community care. It is also the major source of

funding for health research, and provides financial assistance and support for the

22

training of health professionals and tertiary students. The range of services provided

in the community includes screening, episodic or acute services, health promotion

services and services for chronic diseases to improve the individual’s health status

and quality of life (Duckett, 2000).

A universal system of health insurance, Medicare, was introduced into Australia in

1984. Medicare's main function is to cover a large proportion of the cost of services

by medical practitioners and of public hospital care (Duckett, 2000). Under Medicare,

all permanent Australian residents are entitled to free public hospital care when

admitted to hospital as public patients. Medicare also meets the bulk of costs for all

out-of-hospital medical services, such as general practitioner visits and specialist

consultations. The Medicare contribution or benefit is 85 per cent of standard fees set

by the Australian government for each type of service. However, the vast majority of

doctors charge more than the standard fee and patients are then required to pay the

difference. State and territory governments provide public hospital services and work

closely with the federal government and professional bodies to ensure that quality of

care and appropriate standards are maintained. Private hospitals provide about a

quarter of all hospital beds in Australia. Private medical practitioners provide most

non-bed medical services and perform a large proportion of hospital services

alongside salaried doctors. Private practitioners provide most dental services.

Australians may elect either to be treated as private patients in public hospitals or to

use private hospitals (Duckett, 2000). In the private sector, patients can choose to pay

directly for medical costs or use private health insurance. About one-third of all

Australians take out private health insurance. This covers treatment as a private

patient in both public and private hospitals, as well as a range of non-medical

services such as physiotherapy, dental treatment and the purchase of spectacles

(Department of Health and Ageing, 2005).

General practice is the cornerstone of Australia’s health system, and is a medical

service that aims to offer primary, continuing and comprehensive care for individuals,

families and communities. General practice provides locally-based first contact care

and health promotion to communities, and includes GPs, pharmacists and community

health practitioners (Duckett, 2000). Around 90 per cent of Australians see a GP each

year and, through those visits, gain access to a range of diagnostic, pharmaceutical

23

and specialist services (Australian Institute of Health and Welfare, 2000, 2004). The

fact that the vast majority of people will see a GP at least once a year means that GPs

are a key resource in the delivery of preventive care, which is an important part of

the GP’s role (Harris & Mercer, 2001; Royal Australian College of General

Practitioners, 1998). General practitioners in Australia provide a great deal of both

primary and secondary preventive care, including screening for diseases and risk

factors, vaccination and preventing complications of chronic disease (Harris &

Mercer, 2001).

It is noticeable that the Australian health system is different from that of many other

countries, for example, the United Kingdom (UK) and the United States (US). A key

element of the Australian health care system is consumer choice, where Australians

have almost no restrictions on choice of a primary care provider or general practice,

both of which are usually heavily subsidised through Medicare, the universal health

insurance. This undoubtedly contributes to the historically high level of acceptability.

In the UK, consumers must sign up to a general practitioner’s list and there are

formal procedures to observe when transferring from one list to another. Similarly, in

the US most managed care plans limit consumer choice to a designated panel of

providers from whom consumers are allowed to seek (reimbursed) care (Duckett,

2000). Some challenges facing Australian primary health care systems are related to

improving equity of access, which includes providing high-quality care for socially

disadvantaged groups, ensuring access to primary health care appropriate to needs,

and developing universal and targeted policies that reduce inequalities in access to

primary health care (Harris, Harris, & Roland, 2004). In recent years there has been a

significant change in Australia towards the implementation and delivery of primary

health care. For example, an effective collaborative working relationship has been

developed between general practice and other health professionals (Delaney & Fuller,

2004). It is likely that teams of GPs, allied health providers and others will

increasingly work together to achieve the common goal of improved community

health and wellbeing outcomes through a growing emphasis on consumer

participation (Harvey, 2001).

24

2.3 Overview of prevention and preventive activities 2.3.1 Understanding of prevention Prevention is very important for the early detection of chronic diseases and offers a

valuable contribution to improving health outcomes, especially in relation to

preventable illness (Harvey, 2001; Rose, 1992; US Preventive Services Task Force,

1996). It has also been demonstrated to be more cost-effective than services with a

more curative focus, because the cost of health services is reduced following a

decrease in the overall incidence rate (Kaplan, 2000; Oldenburg & Burton, 2004;

Rose, 1992; Royal Australian College of General Practitioners, 1998). Prevention of

CVD involves reducing morbidity and mortality in people with and without

previously diagnosed disease, which can be achieved through early detection,

identification and reduction of the common major risks of CVD (Access Economics,

2005; World Health Organization, 2002). This is related to promoting healthy eating

and regular physical activity, reducing salt and saturated fat intake, giving up

smoking, maintaining a healthy weight, and reducing high blood pressure and

cholesterol levels. It is especially clear that lifestyle or behavioural interventions

offer excellent potential for the prevention of CVD, in addition to the effects of

media communications and other community-wide efforts and strategies (McMichael,

1989).

In 2002, the World Health Organization (WHO) concluded that at least five more

years of healthy life per person can be gained through cost-effective interventions by

developed countries such as Australia. In 2002, the WHO estimated that reducing

risk factors by 25 per cent can lessen a large amount of the burden of disease from

current trends; for example, reducing systolic blood pressure on average by

5-10 mmHg, or reducing cholesterol on average by 0.3–0.6 mmol/L would save

42 per cent of the current DALYs lost from these risk factors (World Health

Organization, 2002). A wide variety of public health initiatives have been undertaken

to reduce premature mortality from cardiovascular disease, which remains the

leading cause of death in Australia. For example, the most recent ‘National Strategy

for Heart, Stroke and Vascular Health in Australia’ was proposed by Commonwealth

Department of Health and Ageing in 2004 (National Heart Stroke and Vascular

Health Strategies Group, 2004). Two broad intervention approaches to reduce the

25

risks associated with CVD have been utilised: population-based strategies, such as

reduction in tobacco consumption and salt intake through legislation or voluntary

agreements, and individual targeted strategies such as identification and screening for

such factors as high blood pressure and cholesterol levels. The cost-effectiveness of

both of these has been demonstrated by a number of large randomised control trials

(Access Economics, 2005). Death rates from cardiovascular disease are falling,

coinciding with: a 21 per cent and 16 per cent fall in smoking rates for males and

females respectively over the past decade; a 50 per cent fall in the prevalence of high

blood pressure since the 1980s; a 25 per cent fall in the incidence of coronary event

over the past decade; falls in the number of coronary heart disease deaths in hospitals,

suggesting better survival of those with the disease; and rapid increases between

1997 and 2000 in the community of use of prescription drugs to lower blood pressure

and cholesterol, and antiplatelet drugs (Australian Institute of Health and Welfare,

2004, 2005; Australian Institute of Health and Welfare and National Heart

Foundation, 2004; Mathers et al., 2000; Mathers et al., 1999; National Health and

Medical Research Council, 1996b).

The benefits of incorporating prevention into medical practice have become

increasingly apparent over the past 40 to 50 years. ‘Preventive services’, which

normally refers to screening tests, counselling interventions, and immunisation and

chemoprophylaxis for the early detection of diseases, have also been associated with

substantial reductions in morbidity and mortality (US Preventive Services Task

Force, 1996). Screening asymptomatic adults for hypertension has benefits for those

patients found to have the condition, and who are treated successfully and safely to

avoid the incidence of disease, death or other complications. For example, US data

suggests that age-adjusted mortality from stroke has decreased by more than 50 per

cent since 1972, a trend attributed in part to earlier detection and treatment of

hypertension. In addition, meta-analyses by MacMahon and others (Collins et al.,

1990; MacMahon, Neal, & Rodgers, 1995) have shown a reduction of about 14 per

cent in the incidence of coronary heart disease and 42 per cent in the incidence of

stroke following treatment for hypertension. Moreover, Littenberg, Garber and Sox

(1990) have conducted a cost-effectiveness analysis of screening for hypertension,

and reported that the cost per quality-adjusted life-year (QALY) saved by screening

is between $76 and $491 on average among age and gender groups. The Australian

26

Diabetes, Obesity and Lifestyle Study (AusDiab) was conducted to determine the

prevalence of diabetes, obesity and other cardiovascular disease risk factors,

including hypertension and abnormal serum lipid profiles. The study consisted of a

baseline survey in the period 1999–2000 and a 5-year follow-up survey in 2004–05,

and was the first national study of diabetes prevalence in Australia (Barr et al., 2006;

Dunstan et al., 2001). The AusDiab study demonstrated a consistent and progressive

fall in the prevalence of hypertension (systolic blood pressure ≥ 140 mmHg, or

diastolic blood pressure ≥ 90 mmHg) in both male and female participants aged 25–

64 years after taking medications for blood pressure, and the results from the three

National Heart Foundation surveys using age standardised to the 1991 Australian

population (see Figure 2.1).

Figure 2.1 Trends in the age-standardised prevalence (%) of hypertension, 1980–2000 (AusDiab, 2001)

There have also been similar results for blood cholesterol screening. The primary

evidence to support cholesterol screening is the ability of cholesterol-lowering

interventions to reduce the risk of coronary heart disease in patients who have been

screened and diagnosed with high cholesterol. Numerous randomised trials have

provided the evidence in reduction of morbidity and mortality of CVD. For instance,

in a long-term trial of cholesterol-lowering treatment over 5.4 years, reductions in

coronary mortality (42 per cent) and all-cause mortality (30 per cent) were made

(Pedersen, Kjekshus, Berg, & Haghfelt, 1994). A meta-analysis of three major trials

showed a 20 per cent decrease in coronary events (Yusuf & Cutler, 1987) and a up-

halla

This figure is not available online. Please consult the hardcopy thesis available from the QUT Library

27

to-date meta analysis of all randomised trials from a systematic review showed that

each 10% reduction in low-density lipoprotein-cholesterol was estimated to reduce

the risk of all strokes by 16% (Amarenco, Labreuche, Lavallee, & Touboul, 2004).

People with Type 2 Diabetes have a life expectancy that can be shortened by as much

as 15 years. The best evidence for prevention of diabetes is for interventions that

target individuals at highest risk and screening for diabetes have a significant impact

on both health and life expectancy (Cohn & Sernyak, 2006; Davies, Tringham,

Troughton, & Khunti, 2004). The early detection of type 2 diabetes through

screening might also provide an opportunity to reduce the progression of

microvascular or macrovascular disease due to asymptomatic hyperglycemia through

changes to diet and physical activity (US Preventive Services Task Force, 1996). The

US Diabetes Prevention Program (DPP) demonstrated that, over 2.8 years, the

intensive lifestyle intervention reduced the incidence of type 2 diabetes by 58 per

cent and the metformin intervention reduced the incidence of type 2 diabetes by 31

per cent, compared with the placebo intervention (The Diabetes Prevention Program

Research Group, 2003).

Three types of prevention have been proposed in relation to chronic disease: primary

prevention, secondary prevention and tertiary prevention (Barros & Martinez-Giralt,

2002; Cohen & Henderson, 1988; Kenkel, 2000; Royal Australian College of

General Practitioners, 1998; Wilson et al., 2003).

1. Primary prevention is usually described as including ‘all efforts to reduce the

probability, severity and duration of future illness’ — in other words, any

protection of health by personal and community-wide effects for the purpose of

preventing disease in an asymptomatic stage (Barros & Martinez-Giralt, 2002;

Last, 1995; Wilson et al., 2003); for instance, lifestyle change such as regular

exercising, eating healthy food, giving up smoking and regular preventive

health check-ups.

2. Secondary prevention is typically considered to include ‘the prevention of

avoidable ill health or unwanted outcomes through detection of early signs of

disease (through screening for specific disease or risk factors such as high

28

blood pressure) and subsequent remedial actions or treatment’ (McKinlay &

Marceau, 2004; Wilson et al., 2003).

3. Tertiary prevention is seen as typically including the ‘control of more advanced

disease to minimise the detrimental impact this has on health’, such as

rehabilitation for heart diseases (McKinlay & Marceau, 2004; Wilson et al.,

2003).

Australian evidence suggests that about half the potentially avoidable deaths are

preventable through primary prevention, a quarter though secondary prevention, and

a quarter through tertiary prevention and rehabilitation (Wilson et al., 2003).

However, although the conceptual distinction between these three types of

prevention might be clear, for any individual prevention this distinction is usually

much less clear. Figure 2.2 shows that the three prevention phases are interrelated.

When preventive activities are referred to in the remainder of this review, such

references mainly relate to screenings for risk factors of CVD and diabetes under the

absence of illness; for example, people visiting their GPs to have their blood pressure

checked when they feel well.

halla


29

Figure 2.2: Primary, secondary and tertiary prevention (RACGP, 1996)

2.3.2 Guidelines for preventive care for general practice

The Ottawa Charter for Health Promotion (World Health Organization, 1986)

emphasised the importance of applying key preventive activities to the population in

general, and to high-risk groups in particular. Preventive activities can include a

variety of counselling interventions, screening tests and immunisation for the

prevention of more than 80 target conditions via preventive health services delivered

in the primary health care setting (Royal Australian College of General Practitioners,

2002, 2005; US Preventive Services Task Force, 1996). Screening is a means of

identifying early onset of disease, with the objective of earlier and more effective

treatment, such as screening risk factors for CVD and diabetes (Rose, 1992).

Therefore, prevention should become a greater part of a GP’s routine practice,

although prevention is under the current fee-for-service arrangements in general

practice (Duckett, 2000). For example, linking preventive activities with patients’

consultations will increase the likelihood of prevention of diseases such as CVD and

diabetes. One way of doing this is to draw up guidelines for preventive activities to

guide general practitioners. Such guidelines typically outline the kinds of preventive

activities that are appropriate for specific diseases and for different target populations.

The United States Preventive Services Task Force (USPSTF) was initially convened

by the US Public Health Service to rigorously evaluate clinical research in order to

assess the merits of preventive measures, including screening tests, counselling,

immunisations and chemoprophylaxis (US Preventive Services Task Force, 1996).

This task force eventually developed the Guide to Clinical Preventive Services,

which includes preventive evidence-based recommendations for a range of

conditions and diseases for specified target populations. It grades its

recommendations according to one of five classifications (A, B, C, D, I), reflecting

the strength of evidence and magnitude of net benefit (benefits minus harms) (US

Preventive Services Task Force, 2000-2003). For example, to prevent heart and

vascular disease, the Preventive Services Task Force strongly recommends that

clinicians screen adults aged 18 and over for high blood pressure. Their rating is an A

recommendation, which has a strong evidence. They found considerable evidence

30

that blood pressure measurement can identify adults at increased risk for

cardiovascular disease, and that treatment of high blood pressure substantially

decreases the incidence of cardiovascular disease. They concluded, therefore, that the

benefits of screening for, and treating high blood pressure in adults substantially

outweigh the harms.

The Canadian Task Force on Preventive Health Care—originally known as the

Canadian Task Force on Periodic Health Examination—has proposed clinical

guidelines for use in Canada and other countries (Stachenko, 1994). The proposed

guidelines relating to the practice of periodic health examinations target the

prevention, detection and control of specific conditions or risk factors for different

age, sex and high-risk groups (Feig, Palda, & Lipscombe, 2005).

In Australia, the Royal Australian College of General Practitioners (RACGP) has

developed and published, and then revised its Guidelines for Preventive Activities in

General Practice (Royal Australian College of General Practitioners, 2002, 2005).

The recommendations in these guidelines are derived from current evidence-based

guidelines for preventive activities. Precedence has been given to those that are most

relevant to Australian general practice. Usually this means that the recommendations

are based on Australian guidelines such as the Guidelines for Preventive

Interventions in Primary Health Care, which have been endorsed by the National

Health and Medical Research Council (National Health and Medical Research

Council, 1996a). In cases where these are neither available nor recent, other

Australian sources have been used, such as the National Heart Foundation of

Australia or overseas sources such as Canadian or US preventive guidelines.

The information in the Australian guidelines is organised at two levels. The first

level is a Lifecycle Chart, which highlights when preventive activities should be

performed and the optimum frequency for each activity. The second level is more

detailed and presents a summary of recommendations. Table 2.1 summarises those

preventive activities and recommendations in relation to the prevention of CVD and

diabetes that are most relevant to the proposed study (Royal Australian College of

General Practitioners, 2002, 2005). The RACGP regularly revises its Guidelines for

Preventive Activities in General Practice. Following advocacy by the RACGP,

screening activities within this framework are rebatable under Medicare.

31

Table 2.1 Preventive activity for cardiovascular diseases and diabetes in Australia

(RACGP, 2002)

halla


32

2.3.3 The role of the general practitioner in relation to prevention Given that guideline implementation and other preventive activities can change the

natural history of disease, health professionals have a particularly critical role to play

(Royal Australian College of General Practitioners, 1998). General practice is central

to the organisation of primary health care in Australia and provides an ideal setting in

which to reach disadvantaged groups within the community, given that over 80 per

cent of the population visit a GP annually. General practitioners, therefore, are likely

to be seeing a considerable number of disadvantaged patients, which provides an

opportunity to establish continuity of care, which has been shown to result in better

outcomes, by extension reducing health inequalities in Australia (Furler et al., 1998;

Harris et al., 2000). For example, general checks or examinations by general

practitioners (GPs) are very important processes for detecting risk factors for CVD

and diabetes; for instance, blood pressure, blood cholesterol and blood glucose

check-ups. GPs may be the only source of preventive care for some disadvantaged

groups, who may have little alternative access to preventive care, advice and

information (Furler et al., 1998; Harris et al., 2000). There is compelling evidence

supporting the importance of primary health care professionals routinely assessing

the smoking status of their patients (Oldenburg & Owen, 1995). Other evidence

demonstrates that primary health care professionals have an important role to play

towards disease prevention by addressing the particular needs of low-SEP and

disadvantaged patients (Oldenburg et al., 2000).

In addition to the Guidelines, the RACGP (1998) has also published Putting

Prevention into Practice—Guidelines for the Implementation of Prevention in

General Practice. This publication provides a realistic framework for prevention and

a range of effective strategies to improve preventive activities. As the key providers

of primary health care, general practitioners (GPs) have a central role in providing

preventive care that reaches the majority of the population, thus preventing chronic

disease.

To be effective in this role, the RACGP (2002) recommended that GPs need to be:

• Opportunistic in offering preventive care when patients present with other

problems or concerns;

33

• Anticipatory in routinely assessing the preventive care needs of their patients;

• Proactive in targeting preventive care, most intensively to high-risk

individuals, and reaching all of their patients, especially those who are least

likely to seek out assistance.

2.4 Socioeconomic position and use of preventive health services From the discussion so far, it is clear that implementing preventive activities in

general practice and encouraging people to access and utilise preventive health

services more often is very important for improving the overall health and health

status of the population. It is very important that such prevention approaches should

be targeted towards socioeconomically disadvantaged groups because of their poorer

level of health and lower health status (Harris et al., 2000). However, an examination

of the relationship between socioeconomic position (SEP) and utilisation of

preventive health services, both internationally and within Australia, generally

indicates that, although the use of several different types of preventive health

services is increasing, disparities related to SEP remain (Breen & Kessler, 1994;

Giles et al., 1993; Lorant, Boland, Humblet, & Deliege, 2002; Makuc, Freid, &

Kleinman, 1989; National Center for Health Statistics, 2004). The research on this

subject is very limited (Wiggers & Sanson-Fisher, 1997). Nevertheless, there is a

body of research that has examined such a relationship in relation to other health

issues such as immunisation, dental care, breast cancer screening and cervical cancer

screening, all of which are relevant to use of preventive health services by

disadvantaged groups in a broader range of health services. In addition, a few studies

have examined CVD/diabetes risk-factor screening and preventive check-ups. The

disadvantage groups in those studies are generally assessed using indicators such as

SEP group, race or ethnic background and rural/remote residence—people who may

have difficulties accessing and using preventive health services (Lee, Harris, &

Traynor, 1999). Much of the data are based on patients’ self-reported measures and

relate to the use of care over extended periods of time. These studies are discussed in

the following sections according to each preventive activity.

34

2.4.1 Immunisation

Immunisation refers to protecting children (and adults) against harmful infections by

using the body’s natural defence mechanisms to build resistance to specific

infections before they come in contact with them in the community (Last, 1995). The

National Health and Medical Research Council recommends a range of vaccinations

for all children, older persons and others who are medically at higher risk of

contracting vaccine-preventable diseases (Australian Institute of Health and Welfare,

2006). Some of the diseases that are preventable by routine childhood immunisation

are: pertussis (whooping cough), diphtheria, tetanus, poliomyelitis, Haemophilus

influenzae type b (Hib), measles, mumps, rubella (German measles) and hepatitis A

or B. Adults and older people can also be immunised against flu.

Socioeconomic disparities in immunisation rates among both poor and non-poor

children are well documented in the literature (Niederhauser & Stark, 2005).

Disparities continue to exist among socioeconomic groups, and ethnic and racial sub-

groups. This body of research suggests that there is a lower use of immunisation by

low-SEP families or groups. For example, Delamonica, Minujin and Gulais (2005)

analysed trends in coverage of three doses of diphtheria-pertussis-tetanus vaccine

(DPT3) in selected countries, and examined the depth of disparities by wealth. They

concluded that children from poorer households consistently lag behind children

from relatively better-off households, and the poorest groups receive fewer services.

Although this relative gap seems to be declining in many countries, such declines are

small. Lorant et al (2002) have conducted a cross-sectional Household Health

Interview Survey from a multi-stage stratified sample of 7378 people aged 25 years

and over in Belgium, and claimed that inequity appeared among SEP quintiles in

immunising flu vaccinations carried out in a general practice (GP) setting. The lower

SEP group was less likely to be immunised against influenza compared to the higher

SEP group. Some studies also confirmed that the uptake of vaccination is influenced

by the socioeconomic status of patients (Hoover, Sambamoorthi, & Crystal, 2004;

While, George, & Murgatroyd, 2005).

In Australia, Carmichael and Williams (1983) conducted an epidemiological

longitudinal study and found that recent immigrants with fewer years of schooling

35

and limited or no English utilised preventive health services, including maternal and

child health services and immunisation, significantly less than other groups. In the

same way, Jones et al.’s survey (Jones et al., 1992) indicated that families possessing

a social security ‘Health Care Card’, where the father was unemployed, who spoke

poor English or who had lived in Australia for 5 years or less were significantly less

likely to be adequately immunised. Moreover, evidence from a cluster-sample

household survey (Herceg, Daley, Schubert, Hall, & Longbottom, 1995) shows that

principal caregivers born outside Australia and with more than one child in the

household are associated with incomplete immunisation coverage for two-year-old

children. In summary, on the basis of this discussion, it seems that SEP is positively

related to the use of immunisation.

2.4.2 Dental care

Regular dental check-ups are recommended both for people with natural teeth and

people with dentures, as such check-ups are likely to prevent or correct any problems

before they become acute. Despite the Guidelines (Royal Australian College of

General Practitioners, 2002) advising that people should visit the dentist regularly,

research shows that those between the ages of 20 and 45 typically do not visit the

dentist unless a condition such as a toothache develops (Australian Unity, 2005).

Wiggers et al. (1995) examined the associations between prevalence and frequency

of health services use, and occupational prestige and educational attainment in

Australia. They observed that those respondents who were disadvantaged had a

lower prevalence of dental services use. The result is similar to that of the study that

indicated that Thai women in Australia had lower use of dental care than other

Australians (Jirojwong & Manderson, 2002). In addition, the data from Australia’s

Health 2004 (Australian Institute of Health and Welfare, 2004) reported that the

proportion of people last visiting for a dental check-up decreased with income from

35 per cent for people in households with an annual income of $60 000 or more, to

20 per cent for people in households with an annual income of less than $12 000 (see

Figure 2.3). Furthermore, the data showed that the frequency of check-up visits in the

previous few months ranged from 14 per cent of government concession card holders,

who received publicly-funded care, to 34 per cent of non-card holders. This indicated

differential use of services and the likelihood of receiving ongoing preventive care.

36

Figure 2.3 Relationship between income and last dental check-up in the previous

12 months among dentate persons aged 15 years or more, 2002 (AIHW, 2004)

2.4.3 Screening for cancer prevention

‘Screening’ refers to tests to identify risk of disease in people without symptoms; for

example, breast cancer screening every two years for women aged 50–69 years by

mammogram; and cervical cancer screening (Pap smears) for all women aged

18-70 years who have had vaginal intercourse are recommended by the Australian

Guidelines (Royal Australian College of General Practitioners, 2002, 2005). Breast

cancer is the most common cancer in women and the most common cause of cancer

death in Australian women (Zorbas, 2003). Cervical cancer is the eighth most

common cancer in Australian women. Factors affecting mortality from breast cancer

or cervical cancer include: earlier detection through screening, earlier presentation

with symptoms and improved treatments. Mammograph screening for breast cancer

offers the opportunity to change the prognosis through detection of the disease at a

preclinical and localised phase. A Pap smear for cervical cancer is a simple

procedure in which a number of cells are collected from the cervix and sent to a

laboratory, where they are tested for abnormalities. In addition, colorectal cancer

screening is recommended for the well population, from 50 years of age, every two

years. All of these cancers are preventable and curable (The Australian Government

Department of Health and Ageing, 2001). Numerous studies have indicated

socioeconomic differences in breast cancer and cervical cancer screening. These

studies are discussed below.

halla


37

A number of overseas studies suggest that socioeconomically disadvantaged people

are less likely to report an appropriate risk-screening status. For example, in Spain,

Borrell et al. (Borrell et al., 2004) investigated the relationship between class

inequalities in health status and use of both curative and preventive services, using

the Barcelona Health Interview Survey, with a sample of 5004 individuals. They

indicated that women in the higher social classes were more likely to have

gynaecological preventive procedures such as cervical smears and mammography

tests compared to women from lower social classes. They found that 60.7 per cent of

women aged over 29 in high SEP groups had periodic cervical smears, but only 32

per cent of those in low SEP groups; the corresponding figures for mammography

were 32.8 and 11.3 per cent, respectively. A number of studies have also found

similar results in Belgium and the US (Breen & Kessler, 1994; Lorant et al., 2002;

Suarez, 1994; Urban, Anderson, & Peacock, 1994). Some evidence suggested that

there was a consistent downward trend for smear screening of cervical cytological

examinations from the higher SEP groups to the lower (Makuc et al., 1989).

Similarly, women of low SEP are less likely to have attended health services for a

Pap smear, although women living in low-SEP areas have a higher incidence of

cervical cancer (Lorant et al., 2002; Makuc et al., 1989). Katz and Hofer (1994)

compared the association of income and education with breast and cervical cancer

screening in Ontario, Canada, and the United States. They suggested that women

who had higher education and income levels were more likely to receive Pap smear

and mammography screening in both countries. This finding was further confirmed

by the US Centers for Disease Control and Prevention (CDC) (2005), which

examined these relationships in 35 metropolitan areas during 2000–2002.

Australian studies have demonstrated that low-SEP individuals are less likely to

undergo cancer screening compared to high-SEP individuals. Armstrong et al. (1986)

investigated the records of 16 069 women who had undergone a Pap smear over an

eight-week period in Western Australia, and found that there was a consistent

downward trend for smear screening of cervical cytological examinations from the

higher SEP groups to the lower SEP groups. Similarly, Shelley and her colleagues

(1994) studied the records of 155 281 women aged 25–69 in New South Wales, and

demonstrated that the screening rates were lowest among women living in areas with

the most non-English speakers and the lowest socioeconomic position. Moreover, a

38

random sample of 2266 women aged 50–69 years who were examined for attendance

of mammography screening programs showed that there was higher attendance in

areas of higher SEP compared to areas of lower SEP (Hurley, Huggins, Jolley, &

Reading, 1994). Evidence reviewed by Harris and Furler (2002) also showed that

people who are socioeconomically disadvantaged are more likely to need, but are

less likely to use, preventive health services such as cancer screening tests. In

addition, Jirojwong and Manderson (2002) examined the socio-demographic factors

of users and non-users of preventive health services, including users of Pap smears

among Thai migrant women in Brisbane. They indicated that such women have a

low participation in preventive health programs and delay seeking medical treatment,

a decision which is generally influenced by cultural beliefs from their home country.

Furthermore, the 2001 Australian National Health Survey supported the findings that

the socioeconomically disadvantaged report less use of preventive health services

such as breast cancer screening and Pap smear testing (Australian Institute of Health

and Welfare, 2004).

In short, on the basis of this evidence, it seems likely that women from lower socially

disadvantaged groups are less likely than their high-SEP counterparts to undergo

cancer screening, including screening for breast cancer and cervical cancer.

2.4.4 Screening for risk factors for heart disease and diabetes

According to the Guidelines for Preventive Activities in General Practice (Royal

Australian College of General Practitioners, 2002, 2005), discussed previously in

Section 2.3.2, there is now a substantial body of research that indicates the

importance of screening regularly for the following risk factors in relation to heart

disease and diabetes: elevated blood pressure, elevated blood lipids and high blood

glucose, weight, smoking status, nutrition, physical activity and alcohol consumption.

Socioeconomic differentials in the reported prevalence of screening for

cardiovascular disease risk have been identified in the US and a number of other

developed countries. For instance, Giles et al. (1993) and Makuc et al. (1989) have

found that socioeconomically disadvantaged people are less likely to have been

screened for cardiovascular disease risks such as elevated blood cholesterol and

39

blood pressure. Hueston and Hubbard (2000) also compared the use of preventive

health services by African–Americans, in both urban and rural settings, to determine

if race and rural residence were additive risks for not accessing or using preventive

health services. They found that, in both rural and urban practices, those with lower

incomes were less likely to undergo preventive tests such as cholesterol screening.

Additionally, patients in the urban setting were more likely to receive counselling

regarding exercise and smoking than those in rural practices.

Data from the US Behavioural Risk Factor Surveillance System (Centers for Disease

Control and Prevention, 2002), carried out in 1991 and 1999 among the non-

institutionalised US population aged 18 years and over, demonstrated that prevalence

of blood pressure screening during the preceding two years was lower among

persons with less education (see Figure 2.4) and some ethnic groups, compared,

respectively, with more highly educated, non-Hispanic blacks, who were more likely

to be disadvantaged. Not surprisingly, those people with a lower prevalence of blood

pressure screening were at risk of higher blood pressure. Similarly, data from

Belgium showed that cholesterol screening favoured the higher socioeconomic

groups more than the lower SEP groups (Lorant et al., 2002).

Figure 2.4 Percentage of blood pressure screening among different education groups

aged 18 and over between 1991 and 1999, United States (Centers for Disease Control and Prevention, 2002)

halla


40

There have been very few studies carried out in Australia that examine the

relationship between SEP and the use of preventive health services related to CVD

and diabetes. Only data analysed from the 1989/90 National Health Survey (NHS)

indicated that patients of low SEP, measured by income level, among those aged

between 15 and 64 years were less likely to have a diagnostic test ordered (e.g. a

blood test), and were more likely to receive a prescription compared to patients of

high SEP (Scott, Shiell, & King, 1996). In addition, self-reported information on

lifestyle risk factors for cardiovascular disease and rates of previous screening were

provided by means of a self-administered patient questionnaire, including screening

for blood pressure, blood cholesterol, smoking, weight and alcohol intake (Heywood,

Ring, Sanson-Fisher, & Mudge, 1994). Over 90 per cent of patients reported prior

blood pressure measures in the previous two years for patients who were less than

50 years old or in the previous two years for patients who were more than 50 years

old. Fifty-one per cent of patients had undergone blood cholesterol screening in the

previous five years. However, the study did not examine socioeconomic differences

in the use and uptake of the preventive activities.

In short, although some international studies have examined the use of preventive

health services such as BP and BC screening among different SEP groups, relevant

studies of the relationship between SEP and use of preventive health services in

relation to CVD and diabetes are lacking. The picture of this use by different SEP

groups is less clear in Australia.

2.4.5 Preventive checkups and medical examination

The term ‘preventive check-ups’ or ‘medical examination’ is often used to refer to

patients who visit a GP to be examined, checked or screened under the absence of

diseases (Pattison, 2000). A thorough physical examination is then carried out, and

relevant investigations are arranged; for instance, checking blood pressure,

performing routine tests like Pap smears and checking of cholesterol levels. It is also

an ideal time to discuss lifestyle factors like smoking, excess weight and stress. The

main aim of a check-up is to detect illness at an early stage or, better still, prevent

illness occurring in the first place. A number of potentially serious health conditions,

41

such as diabetes, breast lumps, elevated blood pressure and skin cancers, are picked

up on routine medical check-ups (Pattison, 2000).

Unfortunately, there is only limited international and Australian data available that

describes the relationship between such consultations and socioeconomic factors.

Dutton’s early research (1978) empirically assessed the explanations of low use of

discretionary health care services by the poor in the US. The study used the number

of preventive health check-ups reported by respondents as an indicator of the use of

preventive health services in a household interview survey that included 1623 adults

and 1435 children, and found that there was a strong and significant positive

association between income and discretionary use when age, sex and illness levels

were taken into account. In the same way, Shankar (2000) examined the utilisation of

routine health services, using a questionnaire interview, among 1549 immigrants

aged 18–92 from El Salvador in Central America. The study found that SEP

(education and household income) was inversely associated with the person having

had a routine check-up in the previous three years. Lower socioeconomic groups

were less likely to have a routine check-up compared to higher socioeconomic

groups.

In Australia, the BEACH (Bettering the Evaluation And Care of Health) program—a

continuous study of general practice activity—reported that the need for a check-up

related to prevention was the most common reason for encounter (RFE) among the

20 most commonly recorded reasons during the 2002–2003 period (Britt et al., 2003).

However, information on the use of preventive check-up by different SEP groups

was lacking. More studies need to be undertaken. A secondary data analysis of the

BEACH program during 2002–2003 was partly covered by the research for this

thesis. The details of this study can be seen in Appendix 2.1.

In conclusion, this section has discussed all relevant uses of preventive health

services by different SEP groups, including immunisation, cancer screening, dental

care, screening for risk factors for heart disease and diabetes, and preventive check-

ups or medical examination. The results of these findings from Australian and

international studies are summarised in Tables 2.2 and 2.3 respectively. On the basis

of this evidence, it is suggested generally that the higher SEP groups were more

42

likely than the lower SEP groups to comply with the Guidelines for Preventive

Activities in General Practice: to receive vaccination, cancer screening, dental care,

screening for risk factors for heart disease and diabetes, and preventive check-ups or

medical examination. Some studies on this relationship were well documented such

as immunisation, breast cancer screening and cervical cancer screening; however,

studies on screening heart diseases and diabetes were less documented. In particular,

in Australia this data is limited.

43

Table 2.2 List of international studies that have examined the relationship between SEP and use of preventive health services

Author(s) Study Sample Study Methods Study Results

CDC (2005) Borrell et al. (2003) Lorant et al. (2002) CDC (2002) Goddard & Smith (2001) Hueston and Hubbard (2000) Shankar (2000) Urban et al. (1994) Breen and Kessler (1994) Suarez (1994) Katz and Hofer (1994) Giles et al.(1993) Makuc et al (1989) Dutton (1978) Andersen (1975)

Non-institutionalised US population aged 18 and over Non-institutionalised resident people in Spain-Barcelona A multistage stratified sample of non-institutionalised resident people (7378 aged 25+) in Belgium Non-institutionalised US population aged 18 and over Population of England. People aged 65-84. African–Americans aged 18 to 70 rural and urban US immigrants from El Salvador living WDC, 1549 individuals aged18-92. Women aged 50–75 in four suburbans Washington State Women aged over 40 450 randomly selected Mexican-American women age 40 and older Multistage random sample of women 18 years and older The Behavioral Risk factor Surveillance System US adults National Health Interview Survey aged 20-79 women 1623 adults and 1435 children National Study of Health services utilisation in 1971 including 3880 families

Behavioural Risk Factor Surveillance System (BRFSS) Telephone survey Barcelona Health Interview Survey Cross-sectional household Health Interview Survey BRFSS telephone survey Secondary data analysis— cross-section study Investigator administrated survey Questionnaire interview Survey data National Interview Survey Personal interviews National Health Interview Survey Telephone interview Interview Survey Household interview survey Household interview survey

Women with low education level and lower income had lower mammography rates than those who had higher education and higher income. Women aged over 29 in higher social class were twice more likely to have periodic cervical smears compared to those in lower social class. Inequity was high and favoured the rich for mammography and cervical screening; inequity was lower for flu immunisation and cholesterol screening but still favoured the higher socioeconomic groups. Age-adjusted prevalence of recent BP screening was lower among less-educated group and some ethnic groups. Higher rates of GP consultation are associated with greater deprivation and with lower socioeconomic group. Low utilisation of preventive services is linked to deprivation at an area level, and to poor socioeconomic circumstances at an individual level. People with lower incomes were less likely to receive the preventive health services ( Papanicolaou smear in the last 3 years, cholesterol screening in the last 5 years and tetanus shot in last 10 years among different age groups). The SEP (education and household income) inversely associated with having had a routine check-up in the past year. Low-SEP groups were less likely to have a routine check-up in then past year. The greater use of screening mammography was associated with more education and high income. Use of mammography screening was higher among women with higher incomes compared with lower income. Women with lower education and income levels were screened less often for cervical and breast cancer. Women with lower education and income level were less likely to receive breast and cervical cancer screening. People who had seen a physician within the last 2 years for preventive care reposted never being screened for High Blood Cholesterol were more common among Blacks, Hispanics and persons with less education. The poor (lower income) remained less likely than the non-poor to have recent preventive care including breast examination, Pap tests and blood pressure testing (except blood pressure testing among older women). There was a strong and significant positive relationship between income and preventive use (number of checkups and frequency of children’s check-ups). Lower education and social class along with being non-white all contribute to a lower probability of seeing the dentist. Also, non adolescents, non-middle aged adults and lower income persons to be less likely to see a dentist.

44

Table 2.3 List of Australian studies that have examined the relationship between SEP and use of preventive health services

Author(s) Study Sample Study Methods Study Results Jirojwong and Manderson (2002) Hyndman & Holman (2001) Young et al. (2001) Scott et al. (1996) Wiggers and Sanson-Fisher (1995) Herceg et al. (1995) Hurley et al. (1994) Shelley et al. (1994) Jones et al. (1992) Redman et al. (1990) Armstrong et al. (1986) Carmichael and Williams (1983)

139 Thai women in Brisbane 459 GP surgeries in metropolitan Perth 4452 women aged 45–75 59 000 people interviewed 2623 household members aged 15+ 187 two-year-old children and families Random sample of 2266 women aged 50–69 years 155281 Women aged 25–69 in NSW 520 Children attending a large inner-city paediatric hospital during 10 days (171 families) 1454 women (40+) randomly selected 16 069 Women who had smears in Western Australia 304 infants born consecutively during a 24-week period to families

Structured questionnaire survey and in-depth interview GP survey—interview Secondary data analysis from the Australian longitudinal study Secondary data analysis from 1989/90 National Health Survey Community survey Cluster-sample household survey Australian mammography screening program Attending 3 years Pap smear program Questionnaire interview Breast self-examination by a health care provider Data obtained from the records during an 8-week period. An epidemiological longitudinal study

Thai women have a low participation in preventive health programs and unhealthy life style and behaviour. Population living in the most disadvantaged areas were less likely to be able to see the local GP at short notice, to have access to a local female GP or a local service in the evenings. Women living in more remote areas had higher out-of-pocket costs and poorer access to services (including propensity to seek care). Patients of low SEP are less likely to be tested and more likely to receive a prescription compared with patients of high SEP. Women are more likely to be tested and to receive a prescription than men. The disadvantaged respondents in terms of their occupational prestige and educational attainment were lower prevalence of dentals services use. The principal caregivers being born outside Australia and having more than one child in the household are associated with incomplete immunisation coverage for two-year-old children. Attendance of mammography screening program was higher in area of higher socioeconomic position. Screening rates were lowest among women living in areas with the most non-English-speakers and the lowest socioeconomic position. Families possessing a ‘Health Care Card’, whose father was unemployed, who spoke poor English or who had lived In Australia for 5 years or less were significantly more likely to be inadequately immunised. Women were more likely to have received a breast examination in the past three years if they had reached a higher educational level. There was a consistent downward trend for smear screening of cervical cytological examinations from the higher SEP groups to the lower SEP groups. The small group whose family were the most recent immigrants, had fewer years schooling and had limited or no English, utilised preventive services (maternal and child, immunisation) significant less compared to other groups.

45

2.5 Determinants of health services utilisation

As discussed in Section 2.4, a body of research has now identified a relationship

between SEP and utilisation of the various preventive health services; however, the

reasons for this relationship are still far from clear (Bullough, 1972; Turrell &

Mathers, 2000). It is important to understand that determinants that are related to use

of preventive health services are associated with those behaviours; in particular GP-

based use of preventive health services. Granted the lack of research in preventive

health services in a GP setting, the studies in preventive health services in various

settings are discussed in the following section. This section also considers and

reviews some of the theories that have been used to inform our understanding of the

ways in which different population sub-groups use preventive health services.

2.5.1 Understanding of determinants and multiple levels of influence

Last (1995) in his Dictionary of Epidemiology defined a ‘determinant’ as ‘any factor,

whether event, characteristic, or other definable entity, that brings about change in a

health condition or other defined characteristic’. While epidemiology is typically

concerned with establishing relationships between different variables—that is, the

relationship between the causes and the effects they produce (Last, 1995)—the social

and behavioural sciences often focus more on trying to understand the ways in which

factors or variables are related to one another. For example, as discussed in Chapter 1,

Turrell et al’s (1999) framework for understanding the relationship between

socioeconomic determinants and health has illustrated clearly the complexity of such

pathways by considering the interrelationships between the three levels—the

upstream, midstream and downstream. Each of these levels contains many different

factors that are linked and related to the other levels, so there are multiple levels of

influences on health. This framework also suggests that a range of psychosocial

factors such as perceptions, networks and social support, directly influence people’s

health behaviours; for example, they can impact on the way in which people access

preventive health care services. These multiple levels of influence can be

conceptualised as operating as individual-level factors, interpersonal-level factors,

environment-level factors, system-level factors and SEP factors (Glanz & Rimer,

46

1994; Glanz, Rimer, & Lewis, 2002; Winett, King, & Altman, 1989). These are

described as follows:

• Individual-level factors refers to individual-level characteristics such as

knowledge, attitudes and beliefs, and personality traits. These factors can

include demographic variables, such as age, sex and marital status.

• Interpersonal-level factors refers to those factors influenced by family,

friends and other social groups.

• Environmental-level factors include organisational factors such as

organisational norms, standards and structures, and environmental factors

such as transport and public facilities.

• System-level factors refers to those factors influenced by societal resources,

priorities and policies.

• SEP factors refers to social structural variables, such as education, occupation

and social class, which influence health behaviours.

For example, a woman might delay obtaining a recommended cancer screening test

(a mammogram or Pap smear) because she is afraid of finding out she has cancer.

This is related to the individual level of factors. However, her decision might be

influenced by her doctor, who did not recommend a mammography or did not inform

her about the availability of mammogram services, and by her family and friends; it

might also be influenced by worry about the fee and transport (she may come from a

low-income family or may be unemployed), and distance to the clinic. These

interpersonal, environmental, societal and SEP factors also impact upon her inaction.

Similarly, a person can also delay having blood pressure, blood cholesterol and blood

glucose checked for CVD and diabetes because he or she is worried about having

CVD or a diabetes problem. This action is also impacted by his or her beliefs and

attitudes towards these preventive health check-ups, and also his or her GP, and

friends, and family.

47

2.5.2 Understanding the factors that influence the utilisation of health services

Over the past five decades, new knowledge has been generated concerning people’s

reasons for seeking different types of medical care. In general practice, for example,

the patient is self-referred and the starting point of the consultation is to find out why

the patient has attended, what health issues or problems should be addressed, and

how this should be done. Taking a decision to visit a doctor is not an automatic

response to a symptom experience (Campbell & Roland, 1996; Thorsen, Witt,

Hollnagel, & Malterud, 2001). Different disciplines have been discussed in some

studies addressing this issue, including sociology, anthropology and psychology.

How to understand this complex process and the set of steps that occurs between the

time an individual feels ‘unwell’ and a visit to the doctor or other health professional,

is influenced by many different factors in people’s decision-making processes and in

relation to their use or non-use of health services (Bowling, 2002). As discussed in

Chapter 1, three aetiological theories have been used to explain the genesis of

socioeconomic health inequalities: the ‘materialist’, ‘behavioural or lifestyle’ and

‘psychosocial’ explanations. These three explanations can also be used to explain the

relationship between SEP and use of preventive health services (Glanz et al., 2002).

To understand how people make decisions to use preventive health services it is

important to distinguish some behavioural concepts.

Researchers have discussed three sets of behaviours that are linked significantly to

people’s use of health services (Barros & Martinez-Giralt, 2002; Glanz et al., 2002;

Rosenstock, 1966). First, preventive health care behaviour has been defined as ‘any

activity undertaken by a person for the purpose of preventing disease in an

asymptomatic stage’; for instance, a medical check-up or examination is a preventive

health care behaviour. Conversely, illness behaviour is ‘any activity undertaken by a

person who feels ill, for the purpose of defining the state of his health and of

discovering a suitable remedy’. Finally, sick-role behaviour is ‘the activity

undertaken by those who consider themselves ill for the purpose of getting well’. The

present thesis emphasises the first of these in relation to preventive actions such as

early detection of illness.

48

Building on this understanding of the complex range of factors that influence

people’s use of health services, researchers in the US and Europe have tried to

answer questions such as: ‘Why do people use health services? What factors

influence people to use health services? How do poor people use health services?’ A

number of the review studies for this work are summarised in Table 2.4. Collectively,

they identify some of the most important determinants of people’s behaviour towards

the use of preventive health services; for example, socio-demographic factors, social

factors and psychological factors. These findings and determinants of health

behaviour are very important in helping to understand behaviour in relation to use of

health services.

49

Table 2.4 International Literature Review on ‘why do people use health services?’

Author(s) Type of Article Results of literature review

Thorsen et al. (2001) Denmark/Norway Campbell and Roland (1996) UK Kaplan (1993) US Hulka and Wheat (1985) US Riessman (1974) US Zubkoff and Dunlop (1974) US Schweitzer (1974) US McKinlay (1972) US Rosenstock (1966) US

Literature review Literature review Literature review Literature review Literature review Literature review Literature review Selective systematic review data in US/Europe Literature review

A theoretical clarification of the concept of the patient’s purpose of a consultation by presenting patient-centred definition, applicable for clinical work and research in general practice. While poor health status and social disadvantage increase both ‘objective’ medical need and, in turn, consultation rates, a range of other social and psychological factors influence consulting behaviour. Often, people go through ‘a search process’ in which they attempt to determine whether or not to consult a doctor. Social comparison theory can assist to understand this process. The determinants of preventive utilisation from the patient’s perspective bring forth some consistent findings including perceived health status, a regular source of care, SEP, health insurance. Two major explanations of the lower utilisation of preventive health services by the poor are ‘cultural of poverty’ and ‘social–structural influences’ on health behaviour. Determinants of use preventive health care are involved in population at risk, perceived need for care, predisposing factors, enabling factors, systemic characteristics and utilisation of model. In the examination of the literature on the use of preventive health services, several important factors have been identified that affect related consumer behaviour including socioeconomic factor such as income, changing health beliefs through health education program, significant informational and motivational role played by non-professionals and paraprofessionals in delivering preventive health services to potential consumers, and health insurance program. Six analytically distinct approaches or research strategies have been isolated including the economic, socio-demographic, geographic, social psychological, socio-cultural and organisational approaches concerning utilisation behaviour. A specific model to explain why people use health services suggested that a decision to obtain a preventive activities will not be made unless the three conditions are satisfied: psychologically readiness, beliefs feasibility of preventive activities, and capability of change through education.

50

2.5.3 Reviewing empirical studies of determinants of preventive health service utilisation

As discussed earlier, the decision to use preventive health services is influenced by

many different factors on a number of levels. SEP also has a profound influence on

the various steps in this decision-making process. There are few empirical studies

that focus on why such an association exists, although a number of researchers have

reviewed ‘why people use preventive health services’ or ‘what factors determine the

use’, as discussed in the previous section. This thesis, therefore, focuses on what

factors mediate the relationship between SEP and preventive health services (see

Figure 2.5). Empirical studies in relation to the determinants of preventive health

services utilisation are examined generally and some empirical findings in relation to

the SEP factors are discussed also. Some sub-social groupings such as ethnicity, age

and gender also appear in this discussion to provide a broad general understanding of

utilisation of preventive health services by some sub-groups that are usually more

socioeconomically disadvantaged. The possible factors influencing the association

between SEP and the use of preventive health services are summarised and

hypothesised for the purposes of this research. They are discussed under the

following five headings:

Figure 2.5 Determinants of preventive health services utilisation

Use of preventive health services

SEP - Individual-level factors: Age, gender, individual

characteristics - Interpersonal-level factors:

Family resources, social support - Environmental-level factors:

Structural factors and professional characteristics

- System-level factors: Health policies, health care system reform

51

2.5.3.1 Individual-level factors ‘Individual factors’ refers to personal characteristics primarily influencing health

behaviour. They are classified into two groups that include demographic variables

such as age and gender; and individual characteristics such as people’s knowledge,

attitudes and beliefs towards use of preventive health services.

Demographic variables

A number of overseas studies have demonstrated that there is an age and gender

difference in the use of preventive health services. Hulka and Wheat (1985) reviewed

US data and found that those younger than 15 years of age or older than 54 were

most likely to have had a routine physical examination. However, Urban et al. (1994)

showed that age was negatively related to the use of mammography screening

services in the US. Older women remained less likely to have had a recent breast

exam or Pap test, despite the fact that the incidence of these diseases rises sharply

with age (Borrell et al., 1999; Makuc et al., 1989). Women were more than three

times as likely as men to have had a routine medical check-up (Shankar, 2000).

Campbell and Roland (1996) also suggested that, in the UK, women are more likely

to be tested and to receive a prescription than men.

Australian findings regarding age prediction on use of Pap smears are consistent with

some US studies; that is, rates of Pap smear decreased with increasing age among

women aged 25–69 years (Shelley et al., 1994). Korten et al. (1998) also suggest that

there are important gender differences in the determinants of GP services use by the

elderly. In relation to health service use, men are less likely to visit a GP, less likely

to seek preventive assistance and more likely to consider waiting for appointments a

waste of time (Henning, 2001), whereas women are more likely to visit GPs for

testing (Scott et al., 1996).

Individual characteristics

Individual characteristics that influence the use of preventive health services are

related to people’s knowledge, attitudes and beliefs. A number of studies both

overseas and in Australia have provided evidence of these factors, as discussed

below.

52

A lack of knowledge of available services is a factor that influences accessing of

health care (Bentley, 2003). For example, when Breen and Kessler (1994)

investigated the reasons for women more than 40 years of age not having

mammograms in the previous three years in the US, they found that a lack of

knowledge that screening mammography is for asymptomatic women was a key

factor in women not undergoing the screening. They are more likely to believe that

the procedure is unnecessary in the absence of symptoms than women who are

screened.

Patients’ attitudes and beliefs are related to their use of preventive health services.

For example, belief in regular physical check-ups was positively associated with self-

reported utilisation of outpatient medical care (Cleary & Jette, 1984). Breen and

Kessler (1994) suggested that women reported that they did not have a mammogram

because they felt fear or discomfort. Bentley (2003), in her recent qualitative study in

the UK, also indicated that perceived fear of the diagnosis was thought to lead to

non-accessing of health care. She also examined perceptions of consumer power and

perceptions of doctors. It was evident that informants perceived that they had less

power than doctors in decisions relating to health matters, that GPs were always busy,

and that this could have undesirable consequences for patients. For example, the

participants perceived ‘being able to see the doctor of choice often involved a longer

wait, yet this was accepted as inevitable’.

In addition, such attitudes about preventive care are most closely connected to some

subgroups such as blacks and the aged (Kravits, 1975). Hochbaum (1956), from his

earlier study in the US, found that SEP (education and income) and a combination of

beliefs about susceptibility and the benefits of preventive care were independently

associated with having voluntary chest X-rays taken in the absence of symptoms.

With each SEP category, however, those who scored high on the combination of

beliefs were much more likely to have the X-ray taken than those scoring medium or

low. Dutton (1978) also supported the idea that people’s attitudes play a very

important part in explaining the income trends in use of preventive health services.

Kravits (1975) claimed, from the National Study of Health Services Utilisation

family interview survey, that blacks and low-income individuals have more negative

53

attitudes toward preventive care, and the value of medical care and doctors,

compared to the remainder of the population. For example, the answer to one of the

attitude questions, ‘if you wait long enough, you can get over most any disease

without getting medical aid’, indicated that blacks are more apt than whites to

believe this statement to be true, and low-income persons, regardless of race, are

more apt to believe this than people with higher incomes.

Furthermore, the study by Campbell and Roland (1996) in the UK showed that

minority ethnic groups had higher consultation rates than people from majority

groups, which may reflect their increased mortality, especially from CVD and

diabetes. This is because minority ethnic groups may experience particular barriers to

access to primary care services, including language barriers and cultural differences

in health perception, because ‘culture’ has been defined as acquired knowledge that

people use to interpret experience and generate behaviour—especially shared

behaviour such as ideas and patterns of thought (Bentley, 2003). Bullough (1972)

identified three types of alienation—powerlessness, hopelessness and social

isolation—as barriers to the utilisation of preventive health services. He claimed that

the direct consequences of poverty and a low educational level are most likely

reinforced by a culture of poverty, including feelings of powerlessness, hopelessness

and social isolation. Poverty is related to feelings of alienation, and the attitudes are,

in turn, related to a low level of utilisation of preventive health care.

Lorant et al. (2002) also discussed why socioeconomic differences affect the use of

preventive services, and suggested that this may be due to differences in beliefs,

help-seeking and information-seeking processes. For cancer screening, it has been

suggested that people of very low SEP may not perceive the usefulness of

asymptomatic screening. The low-SEP individuals seek information only when it is

needed, and rely on their own knowledge first, and then assess information on how it

helps and not on its credibility. Moreover, Borrell et al. (1999) added that another

possible explanation is that women from higher social classes may have a better

understanding of the potential benefits of screening tests and thus be more motivated

to undergo them. Analysing the findings from the Centers for Disease Control and

Prevention (CDC), Morbidity and Mortality Weekly Report (MMWR) (Centers for

Disease Control and Prevention, 2002) on self-reported BP screening in the US

54

found that the lowest levels of blood pressure screening were among men, Hispanics,

persons with less education and younger adults, although BP screening prevalence

was high in all states and demographic groups. One of these factors might be

associated with perception of the risk of heart disease and stroke, and the benefits of

lowering blood pressure.

In Australia, an interesting study investigating factors that predict attendance at a free

Australian mammography screening program indicated that there are cultural factors

that mean that women from non-English speaking backgrounds were less motivated

to attend, and that the promotional material may have been less accessible (Hurley et

al., 1994). Again, Shelley et al. (1994) found that there were lower Pap-smear rates

among women who spoke a language other than English at home and among women

of lower socioeconomic position.

2.5.3.2 Interpersonal-level factors

‘Interpersonal factors’ refers to health behaviour influenced by family, friends and

other social groups. Such factors are classified as variables belonging to two groups,

including family resources such as health insurance coverage and accessibility of a

regular source of health care, and social support from family, friends and other social

groups.

Family resources

Urban et al. (1994) concluded that cost and lack of insurance coverage are major

deterrents to the use of mammography screening services in the US. Suarez (1994)

also found that women with no health insurance were least likely to be screened for

mammography. It has been reported that women covered only by the public system

received less preventive care than those with mixed coverage (public and private)

(Borrell et al., 1999). Low use of preventive health services by socioeconomically

disadvantaged groups reflects cost constraints in the US. The low-SEP groups cannot

afford to purchase the services they need because their incomes are low, and

adequate insurance against health care costs is lacking. Not surprisingly, they appear

to be more sensitive than the affluent to comparable costs, which consume a larger

proportion of a poor family’s income (Dutton, 1986). Several US studies described

55

predictive factors for not using preventive services (Centers for Disease Control and

Prevention, 2002). For mammography, the suggested factors are: low income level,

Hispanic or other ethnic group and low educational level.

Having a regular source of care indicates that primary health care has become

available to patients. Hulka and Wheat (1985) reassessed the finding that Andersen

and Aday’s US nationwide survey showed a positive association between a regular

source of care and physician visits. Persons having a specific doctor as a regular

source of care were more likely to see a doctor (Andersen, 1995). Shi and Starfield

(2000) provided evidence that a source of primary care was the most notable and

significant factor impacting the health outcomes. This evidence implied that patients

with a source of primary care were much more likely to use health services than

those without a source of primary care. Dunlop et al. (2000), in their study,

Socioeconomic status and the utilisation of physicians’ services, indicated that

Canadians lacking a regular medical doctor were less likely to receive primary and

specialist care. What is more, it was also evidenced by Shankar (2000) that people

with a regular source of care and medical insurance were about two and half times as

likely as those without a regular source of care or medical insurance to have had a

routine medical check-up.

In Australia, Medicare provides financial protection against GP consultation fees for

all Australian residents. This results in significantly increased use of health services

(Duckett, 2000). However, the absence of financial barriers does not guarantee

equity. For example, GPs are more likely to have short consultation times with

socioeconomically disadvantaged individuals than with high-SEP patients. In

addition, there might be financial barriers to accessing GP services due to co-

payment issues or some other cost (transport fee) when consumers access GP

services (Young, Dobson, & Byles, 2000, 2001).

A number of Australian studies also report that patients who regularly visit only one

general practice also have more preventive care benefits from their GPs (Carmichael

& Williams, 1983; Steven, Thomas, Eckerman, Browning, & Dickens, 1999).

56

Family and social support

Family and social support networks are important predictors of use of preventive

health services. Some US studies showed that social network variables were

important in accounting for differences in use of preventive health services such as

medical check-ups, dental care and immunisations (Langlie, 1977). It has been

argued that close social support is a strong independent predictor of good health, in

addition to enabling people to cope better when they are ill (Campbell & Roland,

1996). For example, as suggested by Bullough (1972), social isolation is a barrier to

the use of preventive health services.

Australian studies have also confirmed that family and social support are important

influences on people’s use of preventive health services. For example, men with

lower social support were less likely to contact a GP (Korten et al., 1998). A recent

study by Jirojwong and Manderson (2002) investigated the use of preventive health

services, including use of Pap smears in the previous two years and dental care

within the previous 12 months, among 19–65-year-old Thai women in Brisbane. It

was found that the number of women having Pap smears was considerably lower

than the target set by the Australian government. Language was identified as one of

the barriers affecting use of health services. A language barrier limits the Thai

women’s use and understanding of health education materials provided by care

providers both verbally and in other media forms. To use preventive health services,

the women have to rely on their supporters, including spouses, older children, friends

and neighbours for information about clinic locations and types of services. Some

needed assistance to complete forms and to communicate with services.

2.5.3.3 Environmental-level factors

Behaviour in the use of health services is influenced by organisational and

environmental factors, including structural factors and physician characteristics, as

discussed below.

Structural factors

Structural factors are related to medical cost, time and organisational issues. A

number of overseas studies have investigated these influences on use of preventive

57

health services. For example, Dutton (1986) evidenced that a low-income practice

clientele and high charges were the most significant deterrents to the use of Medicaid

in the US, followed by structural barriers such as the impact of organisation and

practice patterns on the use of health services by patients. Lorant et al. (2002) also

discussed factors that might explain the inequitable use of preventive health care,

such as cancer screenings in Belgium, and that financial barriers may limit access to

cancer screening because cost sharing has a significant negative effect on the use of

mammography and is higher in the specialty sector where most cancer screening

occurs. Access to a car, distance and travel time to health services, and waiting time in health

services are positive determinants of accessibility (Andersen, Kravits, & Anderson,

1975; Bentley, 2003; Dutton, 1978). For example, it was suggested that people who

responded ‘yes’ to the question about having to wait a long time for an appointment

to see a doctor were less likely to have had a routine medical check-up than those

who responded ‘no’ (Shankar, 2000). The rural and remote location of some sub-

groups is the most important barrier in areas where physician accessibility for

preventive health care is poor (Kravits, 1975). When Hueston and Hubbard (2000)

compared the use of preventive health services by rural and urban African–American

adults, they also investigated the barriers that might have kept them from obtaining

any preventive health services. Hueston and Hubbard suggested that increasing

access to care may be an important factor if preventive services for minority and

rural populations are to increase to recommended levels.

Primary care physicians are the main source of health promotion for mammography

screening in the US. However, in the study, there was evidence that the lack of a

physician’s recommendation for mammography screening was a key reason for

women not having mammograms (Breen and Kessler, 1994). The findings from the

CDC Morbidity and Mortality Weekly Report (MMWR) (Centers for Disease Control

and Prevention, 2002) on self-reported BP screening in the US found that the lowest

levels of BP screening were among men, Hispanics, persons with low education

levels and younger adults, although the prevalence of BP screening was high in all

states and demographic groups. These factors might also be associated with limited

access to quality health care and limited socioeconomic resources.

58

System barriers focus on the various levels of availability of the health care systems

typically used by the disadvantaged. For example, in most reported data of Dutton’s

study in the US (1978), the majority of families at or below the poverty level

reported that either a hospital outpatient department or emergency room, or a public

clinic were their usual sources of primary health care, while most upper income

families used fee-for-services providers or prepaid group practices. The use rates are

low among the socioeconomically disadvantaged; in part, because of the limited

health care systems they use. This finding was also supported by Strickland and

Strickland (1996), with 281 household interviewed, 51 community leader and six

focus group interviews in five rural US counties. They pointed out that the health

providers in both the public and private sectors that serve low-income individuals

were insufficient and underfunded. There were significantly fewer individuals

eligible to receive Medicaid in the US.

In Australia, there are a number of factors in relation to structural barriers affecting

people’s use of preventive health services delivered by GPs. These include co-

payment on top of Medicare, waiting times and geographic availability of services

(Harris & Furler, 2002). Cost is a very important factor in relation to the use of

preventive health services. For example, Young et al. (2001) found that lower

socioeconomic position was associated with lower out-of-pocket costs per

consultation and was inversely related to use of preventive health services. An

interesting study to investigate factors that predict attendance at a free Australian

mammography screening program indicated a decrease in attendance with increasing

distance from the screening site (Hurley et al., 1994). In addition, the frequency of

consultation is a strong factor associating with Pap smear rates (Shelley et al., 1994).

The study indicated that women who visited a GP at least four times a year were at

least twice as likely to have had a recent Pap smear as those who averaged less than

one visit per year.

The availability of health care resources is an essential ingredient in their utilisation.

Hyndman and Holman (2001) claimed that the population living in the most

disadvantaged areas was less likely to be able to see the local GP at short notice, or

to have access to a local female GP or a local service in the evenings. Young and

59

colleagues (2000) indicated that the opportunity to participate in preventive care

programs was more limited for women living in non-urban areas because of the

shortage of medical specialists in such areas. As discussed previously, many studies

indicated that people with low SEP were more likely to access multiple-care services

(no regular health care), which gave little opportunity for continuity of care of the

family by one practitioner, thus seriously limiting the role of that practitioner in

preventive health care and education (Carmichael & Williams, 1983; Duckett, 2000;

Steven et al., 1999).

Characteristics of the health professional

Professional characteristics such as gender, training, experience and attitudes towards

preventive practices and patients impact on the use of health services. A physician’s

gender influences the use of cancer screening. In Britain, a large proportion of

women would prefer to have their smears taken by another woman (Cullum &

Savory, 1983) (Standing & Mercer, 1984). Dutton (1986) discussed the idea that

specialised training may encourage more intensive evaluation of patients, such as

comprehensive diagnostic testing. In addition, physician attitudes towards preventive

care should correlate with preventive practices, while general satisfaction might be

either a cause or a consequence of higher use rates. For example, low levels of

screening for breast and cervical cancer were due to a lack of emphasis on cancer

prevention for women who have had contact with the health care system (Makuc et

al., 1989).

In Australia, characteristics of GPs such as gender are also related to use of

preventive health care. For screening Pap smears, female GPs took twice as many

smears as did male GPs. This is possibly because female GPs are more likely to

encourage their patients to be screened, or women preferentially seek female GPs

when they wish to be screened (Armstrong et al., 1986). However, the shortage of

female GPs in rural areas was reflected by the responses from women in a sub-study

of the Australian Longitudinal Study on Women’s Health (Young et al., 2000), rating

their access to a female provider as either fair or poor. Thus, women in rural areas

have difficulty seeing the GP of their choice.

60

The impact of general practitioners’ attitudes towards low-income patients on the use

of health services has been observed. Despite higher rates of chronic disease and

lower rates of preventive care uptake, socioeconomically disadvantaged groups

receive fewer long GP consultations than SEP-advantaged groups (Furler et al.,

2002). However, when Wiggers and Sanson-Fisher (1997) assessed whether

practitioner provision of preventive care in general practice consultations is

associated with patients’ educational and occupational status, their study showed that

practitioners were less likely to discuss preventive care topics (smoking, alcohol

consumption, exercise, cervical cancer screening, breast cancer screening and blood

pressure measurement) with patients of high educational or occupational status. The

pattern of results suggested that previous findings of low-SEP groups having a

poorer preventive care status may not be attributable to differentials in practitioners’

provision of preventive care.

The decisions made by GPs can have a substantial effect on the use of health care

services that have more beneficial uses, such as diagnostic testing. However, this is

not only a concern in the use of health services. Differences in patient’s SEP and

professional practice patterns may be factors influencing people’s use of health

services. Scott et al. (1996) explored this relationship between SEP and GPs’

decision making, analysing sub data from the Australian National Health Survey.

Respondents aged 15 or over who reported visiting a GP for a ‘check up or

examinations’ were chosen. The results showed that a diagnostic test was less likely

to be ordered or undertaken for respondents with annual incomes of less than

$10 000 compared to those with incomes of $30 000 or more.

2.5.3.4 System-level factors

Health policies

‘Institutional factors’ refers to rules, regulations and policies in formal structures,

which influence health behaviours. Different countries have different health policies.

There are only two comparisons between the US and Australian health system, as

discussed below, because of the study model driving from the US.

61

In the US, the health care system is a very large national share of privately funded

insurance by managed care organisations (Duckett, 2000). Medicaid is a program

that pays for medical assistance for certain individuals and families with low

incomes and resources. Medicare is the national health insurance program for people

aged 65 or older, and some people under age 65 who have disabilities. However,

17 per cent of the US population remains uninsured. This factor strongly influences

the way people use health services. For example, Urban et al. (1994) claimed that

cost and lack of insurance coverage are major deterrents on use of mammography

screening services in the US. They suggested that state legislation mandating

coverage for screening mammography by private insurers, and federal legislation

mandating such coverage by Medicaid, may have an important impact on use by

creating a ‘standard of care’ recognised by all physicians.

On the other hand, the Australian health insurance scheme, Medicare, is a universal

care scheme to ensure that all Australians have equal access to care in a public

system. Under this care system, people have free choice of medical care, especially

in ambulatory care. Despite this, there are equity, efficiency and acceptability issues

challenging the Australian health care system (Duckett, 2000). Although access to

general practice services is now available without significant financial barriers, as

demonstrated by the generally high level of bulk billing—for example, the bulk-

billing rate increased from 45.2 per cent in 1984–85 to a high of 72.3 per cent in

1999–2000, decreased to 67.5 per cent in 2003–04, and reached to 70.2 per cent in

2004–05 (Australian Institute of Health and Welfare, 2006)—(although levels of

bulk billing are lower in rural areas), the absence of financial barriers does not

guarantee equity. For example, as discussed previously, consultation times with

people from low-SEP groups are shorter than those for high-SEP groups, and there

are still financial barriers to accessing specialist services. Although direct costs can

be met by Medicare, a co-payment by the user may be required if the provider does

not directly bill the service to Medicare. Bulk billing and special services under

Medicare influence the way patients use health services. For example, the choice of

bulk billing for patients depends on a decision made by their GP. Doctors vary their

charges according to the patient’s ability to pay (Young et al. 2001), and they tend to

use bulk billing more frequently for low-SEP groups and frequent users of health

services (Young et al. 2000).

62

Health care system changes

Changes to the health care system are also required to improve use of health services.

In the US, a number of studies have found that, compared with services that are less

well-integrated, there is an association between the provision of more ‘holistic’ and

proactive community-based health care services, and improved health outcomes at

lower cost (Harris & Furler, 2002). The US managed care organisations are designed

to reduce costs to the insurer, with the aim of providing lower premiums to

consumers, and only contract with a limited number of providers. Importantly, this

can influence the treatment behaviour of the selected providers (Duckett, 2000). The

increasing formation of integrated health care systems and managed care

organisations may also create new opportunities for pursuing better preventive

practices (US Preventive Services Task Force, 1996).

In Australia, the most common source of primary care—general practice—is not

subject to direct government control. GPs may locate their services essentially

wherever they please, and can provide whatever hours and types of services they

wish. A study by Hyndman and Holman (2001) indicated that there was less out-of-

hours service and less provision of some day services in disadvantaged areas, in spite

of a seemingly adequate supply of physical resources. A health care system is

therefore needed that is oriented to the needs of populations and communities, and in

which the various elements of primary care, such as general practice and community

health, work more effectively together by establishing integrative structures at the

local level (Harris & Furler, 2002). Additionally, to reduce access barriers to health

care services, the Divisions of General Practice should be addressing health

inequalities within their communities and providing need assessments and

partnerships (Harris & Mercer, 2001).

2.5.3.5 A summary of the potential factors influencing SEP and the use of preventive health services

From this review, two firm conclusions can be drawn. The first of these is that SEP

and a range of other factors and influences are strong predictors of access to and use

of preventive health services. The second is that this relationship is strongly

influenced by a number of different factors, many of which are beyond the individual

63

level of influence and which occur at the interpersonal, environmental and system

levels.

A few studies have considered how these different factors interact with the

overarching relationships between SEP and use of preventive health services.

Riessman (1974) discussed the reason for lower utilisation of preventive health

services by the low-SEP groups in the US and suggested two major explanations.

One was the ‘culture of poverty’ influence on use of health services, meaning

‘culture’ comprising a body of interrelated social, economic and psychological traits

that are transmitted from generation to generation. For example, low-SEP groups

have used preventive health services less and are also less knowledgeable about

appropriate health behaviour because they possess a culture which does not place a

high value on health. Another explanation was related to economic and

sociostructural influences. Economic influence has been shown to have a great

influence on the use of preventive health services in the US, as discussed previously.

These factors include the price of services, the presence or absence of health

insurance, and family income. ‘Sociostructural influence’ refers to professional and

organisational factors that can lead to inadequate use by low-SEP groups, such as

long waiting times and professionals’ attitudes towards patients.

Dutton (1978) also conducted an empirical study to explain the low use of

discretionary health services (the number of preventive health examinations reported

by respondents) by the socioeconomically disadvantaged in the US. Dutton used

similar explanations to those of Riessman (1974), including cost constraints, the

culture of poverty and system barriers. The results from this study suggested that

neither financial access nor health education, without accompanying improvements

in delivery systems, will eliminate income differentials in use. In addition, Strickland

and Strickland (1996) concluded that the different levels of factors interacted to deter

lower income groups from using preventive services in several ways, including

community, programmatic and individual levels.

Australian evidence has been reviewed by Harris and Furler (2002), who concluded

that utilisation of health services by different socioeconomic groups was mediated

and moderated by the following factors:

64

• geographic availability of services, especially in rural and outer urban

areas;

• the cost of health care services, for example, ‘co-payments’ on top of

Medicare;

• waiting times for publicly funded health services;

• a range of specific barriers to disadvantaged groups, including cultural

and language barriers.

It is unclear what the most salient factors are, which factors determine the

socioeconomically disadvantaged groups’ use of preventive health services, and

whether these factors mediate the relationship between SEP and use of preventive

health services in relation to CVD and diabetes in Australia. More studies are needed

to improve the understanding of determinants of the relationship between SEP and

use of preventive health services in Australia.

Table 2.5 provides a summary of the five levels of determinants of utilisation of

preventive health services and relevant evidence in this field from the literature

review.

65

Table 2.5 Determinants of preventive health services utilisation from the literature review

Determinants Definition Classifications Summary of Study Findings Individual-level factors Interpersonal-level factors Environmental-level factors System-level factors SEP factor

- Health behaviour primarily a result of personal characteristics. -Health behaviour is influenced by family, friends, and other social groups. -Health behaviour is influenced by organisational/environmental factors. - Rules, regulations and policies in formal structures, which influence health behaviours. -Social structural variables, such as ethnicity, education and occupation and social class, which influence health behaviours.

Demographic variables: age, gender Individual characteristics: knowledge, attitudes and beliefs Family resources: health insurance coverage, accessibility of a regular source of health care Social Support: from family, friends and other social group Structural factors: Financial, time, organisational and practice patterns Physicians’ practice patterns Professional characteristics: professional gender, training and personal characteristics Health policies: Medicare (bulk billing) (AUS) and Medicaid (US) Structural reform of health care system Attitudes and beliefs towards use of preventive health services Regular source of care Social support Availability and accessibility to health care Internal-care Health system—bulk billing

Most overseas and Australian data showed that older women remained less likely to have had a recent breast exam or Pap test. Women were more likely to have had a routine medical check-up compared to men. People’s knowledge, attitudes and beliefs are related to their use of preventive health services. Most overseas and Australia data showed that accessibility of a regular source of health care, health insurance and social support positively impact on use of preventive health services. Both overseas and Australian studies evidenced that cost (co-payment), distance, limited hours and patient-sharing by doctors; a lack of emphasis on prevention by professionals, professional gender and attitudes towards patients all deterred/associated the use of preventive health services. Cost and lack of insurance coverage were major deterrents on use of preventive health services in the US. In Australian context, bulk-billing, special services under Medicare influence the behaviour of patients to use of health services. Overseas studies showed that SEP groups were more sensitive to cost. Individual attitudes towards health services and professionals were more likely to be negative. Culture and language were barriers to use of preventive health services by the poor. Practitioners’ attitudes towards low-income groups also impacted on the use. Australian studies also confirmed the length of consultation by GPs for low-SEP groups was less than that for higher SEP groups. Cultural and language barriers to disadvantaged groups applied when to access health services.

66

From the previous review of determinants of health service utilisation and

understanding of socioeconomic influences on the use of preventive health services,

the following potential key factors have been identified, and might explain the

relationship between SEP and the use of preventive health services in relation to

CVD and diabetes in Australia.

• Bulk billing: This is a system-level factor. It is related to the bulk-billing

policy applied in Australian general practice. This is a factor that influences

use of preventive health services, given that the rates of bulk billing are

fluctuating and patients have to contribute a co-payment in some

circumstances. As Australian evidence has indicated, low-SEP individuals

have lower out-of-pocket costs compared to high-SEP individuals. Numerous

international studies also indicate that cost is a major constraint in the use of

health services and preventive health services.

• Availability and accessibility to health care: These are environmental-level

factors. ‘Availability’ means the health services should be available for

people to use. These factors are related to the availability of doctors,

including male and female doctors for people to choose, consultation times,

after-hours services and health sources. ‘Accessibility’ means the health

services must be accessible. This is related to distance and travel times to

health services, access to cars, public transport and waiting times. Both

Australian and international studies show that availability of and accessibility

to health services are positively affecting people’s use of preventive health

services.

• Interpersonal care: This is an environmental (organisational)-level factor

and is related to the professional’s attitudes towards patients. Australian

evidence indicates that low-SEP individuals are more likely to receive a short

GP consultation compared to high-SEP individuals. US data also shows that

health professionals’ service for low-income individuals is insufficient in

terms of consultation time. This factor may also contribute to the SEP

differences in use of preventive health services.

67

• Regular source of care: These factors are at an interpersonal level. Having

regular health care refers to people who have a specific doctor or medical

centre ready to access or use whenever possible. A number of international

studies have examined this relationship and have indicated that there is a

positive association between having a regular source of care and use of health

services. A few Australian studies have also showed this positive association.

• Social support: This is also an interpersonal level factor. Social support

refers to the social network of family, friends and other social groups. Both

Australian and US studies indicate that people with less social support are

less likely to contact their doctors. Social support is an important variable in

accounting for differences in use of preventive health services.

• Attitudes and belief towards preventive health care and general

practitioners: These are individual-level factors. Some of these examples

include people’s attitudes towards preventive care: screening; value of health

services and doctors, and belief in the value of preventive health check-ups.

The international evidence from previous discussions generally indicates that

low-income individuals have more negative attitudes and beliefs towards

preventive care and health services compared to high-income individuals.

One Australian study has also shown that disadvantaged groups are less

motivated to attend screening programs. Attitudes and beliefs of low-SEP

groups, therefore, play a prominent part in explaining socioeconomic

differentials in utilisation of preventive health services.

In summary, these six areas are hypothesised to be the most likely to impact on the

use of preventive health services by different SEP groups. They will be examined

further in this thesis. As noted from the previous discussion, many studies have been

conducted in the US and other countries. Most of the findings were based on US

conceptual models. Do the same identified factors similarly influence the use of

preventive health services in an Australian context? How can the US model be

transferred to an Australian model? The most relevant models of health service

utilisation are reviewed in the next section.

68

2.6 Models of health services utilisation

Most commonly used theoretical models for understanding individuals’ use of health

services in this literature have included the Health Belief Model and health services

utilisation models developed by Andersen and his colleagues. In addition, another

model of health service utilisation developed by Dutton is also presented. These

models are critically reviewed in this section in order to understand determinants of

preventive health services more broadly.

2.6.1 Health Belief Model

The Health Belief Model (HBM) was one of the first, and has been one of the most

commonly used theoretical frameworks to explain and predict health-related

behaviours (Bowling, 2002; Glanz & Rimer, 1994; Glanz et al., 2002; Gochman,

1997; Nutbeam & Harris, 1999). It was originally developed by US researchers in the

1950s to explain why people failed to participate in programs to prevent and detect

diseases, for example health checks and immunisation programs, and it has been

substantially developed and modified for application to other types of health

behaviours (Janz & Becker, 1984; Nutbeam & Harris, 1999). The major elements of

the HBM are summarised in Figure 2.6.

Figure 2.6 Major elements of the Health Belief Model

(Nutbeam & Harris, 1999)

halla


69

The model predicts that individuals will take action to protect or promote health if

they perceive themselves to be susceptible to a condition or problem, and if they

believe it will have potentially serious consequences; that is, the perceived threat is

high. The model also predicts that individuals believe a course of action is available

that will reduce their susceptibility, or minimise the consequences; that is, the

benefits of taking action outweigh the costs or barriers. For example, to apply this

model to the prevention of CVD and diabetes, individuals need to:

• believe that they are at risk of CVD or diabetes;

• believe that the consequences of the diseases are serious;

• believe the supportive information for action, which may initiate a response

such as media or GP;

• believe preventive actions, such as healthy lifestyle and regular health checks,

will greatly reduce the risk of diseases;

• identify the potential barriers to taking action that will underweigh the

benefits of action to reduce risk, such as costs, transport to the services and

pressure from family, friends and community;

• believe in their ability to take effective action, for example, exercising, eating

healthy food and undergoing regular health checks.

The model has been refined substantially by including factors such as additional

personal characteristics, social circumstances and the concept of self-efficacy;

collectively, these have improved the strength of the model in predicting behaviour

change (Nutbeam & Harris, 1999).

As mentioned above, the HBM has been most useful when applied to behaviours for

which it was developed originally. On the other hand, it has turned out to be less

useful in predicting more long-term, complex and socially determined behaviours

(Nutbeam & Harris, 1999). A critical review of the model by Janz and Becker (1984)

pointed to the limitations of the HBM in predicting and explaining health behaviour:

‘the Health Belief Model is a psychosocial model; as such, it is limited to accounting

for as much of the variance in an individual’s health behaviour as can be explained

70

by their attitudes and beliefs. It is clear that other forces influence health actions as

well.’ These ‘other forces’ include social, economic and environmental conditions,

which significantly shape the barriers to action that are fundamental to the model.

For example, limited access to health care services and health resources can greatly

impede effective health actions such as preventive check-ups, and will in turn impact

on the individual’s perceptions of barriers and benefits that are integral to the model

(Nutbeam & Harris, 1999).

2.6.2 Health services utilisation models

Evolving out of the Health Belief Model, Andersen’s Behavioural Model of Health

Services Utilisation was proposed in the late 1960s (Andersen, 1968) (see Figure 2.7).

Figure 2.7 Behavioural Model of Health Services Utilisation

(Andersen, 1968)

This model was initially proposed and empirically tested in a series of studies to

assist in understanding why families use health services; to define and measure

halla


71

equitable access to health care; and to assist in developing policies to promote

equitable access. The Andersen model has made the following contributions to health

services research. First, it has helped organise and integrate an array of correlates of

health and health care behaviour from diverse literature sources in sociology,

psychology, economics and medicine into predisposing, enabling and need predictors

of families’ use of physician, hospital and dental services. Secondly, it has

systematically characterised the array of predictors of health services utilisation,

including independent variables such as demographic variables (age, sex and family

size), social structure (employment, social class and occupation) and health beliefs as

predisposing factors, family and community resources as enabling factors, and health

status as need factors. Thirdly, it delineated the indicators of health services

utilisation (dependent variables) according to the type of service such as hospital,

physician and dentist, and the reason for use (discretionary or nondiscretionary).

Fourthly, it also specified the hypothesised relationships between the predictors and

indicators of utilisation. Finally, it provided an integrated theoretical and empirical

approach that has been widely applicable to diverse populations at international,

national and local levels, and important health policy problems such as access and

equity (Gochman, 1997).

Subsequently, Andersen’s model has been modified, expanded and applied to

examine the predictors of health service utilisation in relation to an array of health

problems and issues. Approximately ten years after Andersen’s original model,

Andersen and Newman (Gochman, 1997) formulated an expanded model which

added to and elaborated the components in order to be more responsive to societal

and policy changes affecting health care, as well as to more fully reflect the

increasing complexity of health care service delivery. The framework hypothesises

that societal determinants affect individual determinants, both directly and indirectly,

through the health services system. Individual determinants have the most immediate

influence on people’s decisions about the use of services (see Figure 2.8).

72

Figure 2.8 Andersen and Newman utilisation framework (1973) (Gochman, 1997)

Individuals, rather than families, were the units of analysis in the revised framework.

The previous three components: predisposing, enabling and need predictors were

incorporated as the principal individual determinants of health service utilisation, and

the framework breaks out societal determinants and the health services system as

important aggregate determinants of individuals’ health care-seeking behaviour.

Importantly, the revised framework emphasised a multifaceted concept of utilisation.

As well as the type of utilisation acknowledged in Andersen’s original model, which

refers to the category of services rendered, it also includes the purpose and the unit of

analysis under the health services utilisation. The former refers to the reason care

was sought: for health maintenance in the absence of no or minor symptoms (primary

care), for the diagnosis or treatment of illness (secondary care) and for rehabilitation

in the case of a long-term health problem (tertiary care). (The latter refers to whether

the services were received during a particular time period, how many units of

services were received during that period, as well as patterns of providers, such as

referrals and continuity of care.)

halla


73

The expanded model was mainly used to compare medical care use in Sweden and

the US (Aday & Awe, 1997). It can be generalised to measure the equity of the

distribution of health services and evaluate the intervention potential of the various

predictors, as well as to empirically document the strengths of the casual connections

and interactions among the hypothesised predictors of use. However, Andersen and

Newman acknowledged that the postulated causal links between the societal factors

and resulting utilisation behaviour can only be assumed, since the data, methods and

theory that control their influence are not well developed. For example, the

influences of the health services system on an individual’s determinants poses a

range of questions regarding the impact of health policy changes on an individual’s

use of health services.

A few years later, the Aday and Andersen access framework was derived from the

Andersen and Newman framework, which was developed to guide the first national

survey of access to medical care in the US. It provided a systematic basis for

assessing the performance of major governmental and private programs in order to

improve access and use of medical care in the US (Aday & Awe, 1997). The

framework hypothesised the relationships among five components, which are

indicated by the arrows (see Figure 2.9).

74

Figure 2.9 Aday and Andersen access framework (1975) (Aday & Awe, 1997)

Health policy may affect the characteristics of health delivery systems directly and

may change the characteristics of the population at risk, either directly or indirectly,

through the health delivery system. For example, health policies such as increasing

the supply of doctors in an area can enhance the availability of health delivery

systems. The delivery system may directly affect utilisation patterns and consumer

satisfaction. The characteristics of the population may also directly affect use and

satisfaction.

halla


75

As a starting point, health policy replaced societal norms in the Andersen and

Newman model for considering the predictors of the utilisation and satisfaction with

medical care. The model specified the characteristics of the health services system

with more detailed dimensions; for example, availability, entry and structure, and the

characteristics of population at risk such as age, gender, race and place of

residence—termed ‘immutable’. These predictors may be seen as structural and

process predictors to the access to health services. In this expanded behaviour model,

the principal indicators of realised access were objective measures of health services

utilisation, and subjective assessments by consumers of their actual utilisation

experience which essentially reflected the extent to which the system and population

characteristics actually predict whether or not or how much care is sought and how

satisfied consumers are with the health system.

Indeed, both of the expanded models have been applied extensively in guiding the

design and implementation of large-scale international, national and local surveys.

Andersen and his colleagues conducted related program evaluations during the1960s,

1970s and 1980s. Their models have also been widely applied by other investigators

to explain and predict a variety of health care utilisation behaviours.

However, there have been a number of criticisms of Andersen’s original and

expanded models. As explained by Gochman (1997), the criticisms are broadly

related to: (1) the definition and measurement of the major predictors, for example,

the validity of the study concepts concerning how the predisposing, enabling and

need factors are measured, and other variables need to be added in to adequately

capture relevant predictors; (2) the definition and measurement of indicators of

health services utilisation, for example, special consideration on the appropriate type,

site, purpose and time interval of utilisation in measuring a given type of service such

as mammography screening, rehabilitation services or mental health services; (3)

there is a need to test the array of causal relationships implied in the models more

fully; and (4) finally, major criticisms of the empirical research based on the model

include the fact that it tends to explain a very small percentage of the overall

variation in utilisation.

76

In the mid-1990s, Andersen (1995) reviewed the development and application of his

model over a 25-year period. He finally revised his model (see Figure 2.10).

Figure 2.10 An emerging Andersen Behaviour model (Andersen, 1995)

This model portrays ‘the multiple influences on health services’ use and,

subsequently, on health outcomes such as perceived health status. The multiple

influences refer to health care system (policy and health reform, resources and

organisation) and external environment (physical, political and economic

components), and impact on population health behaviour such as use of health

services through population characteristics (predisposing factors and enabling

resources and need factors). Shankar (2000) used Andersen’s final model to

determine the factors associated with utilisation of routine medical check-ups in a

sample of immigrants from El Salvador now living in the US. This study focused on

population characteristics, including predisposing factors such as age, gender,

education, marital status and beliefs about cancer; enabling variables such as income,

transportation, source of medical care, insurance coverage and years in the US. It

found that those predisposing and enabling factors that were inversely associated

halla


77

with having had a routine check-up in the past year were education, household

income, having to wait for an appointment and the number of years resident in the

US. For instance, those who responded ‘yes’ to the question about long waits for

appointments were less likely to have had a routine medical check-up than those who

responded ‘no’. The people with higher SEP (education and income) were more

likely to undergo a medial check-up.

2.6.3 Other models of health services utilisation

Building on Andersen’s earlier model, Dutton (1986) explored some additional

factors affecting health care utilisation. In her study, a survey of 3058 family

members and their usual providers in Washington, D.C., utilisation was viewed as

the product of patient characteristics plus provider and system attributes. The study

incorporated Andersen’s categorical terms of factors such as predisposing, enabling

and need factors in the patient characteristics in this model. The dependent variables

include patient-controlled use by children and adults, and use controlled primarily by

doctors. For each measurement of use, an equation containing variables representing

provider and system features plus various patient characteristics was estimated (see

Figure 2.11).

Figure 2.11 Conceptual model of the factors affecting health care utilisation (Solid arrows denote expected major relationships, while dashed arrows denote less important

relationships) (Dutton, 1986)

halla


78

Dutton’s conceptual model of the factors affecting health care utilisation has paid

more attention to the role of providers and system features, and has focused on

multiple determinants of utilisation rather than a selected set of factors compared to

most population-based studies. Additionally, many previous studies have been

limited to one or two types of setting, but Dutton’s model has considered widely

varying types of health services settings such as solo physicians, fee-for-services

groups, prepaid group practices, public clinics and hospital clinics. Moreover, this

model has been applied particularly to the study sample of low-SEP and minority

groups. Based on this conceptual model, Dutton found that low-income groups and

high charges were the most significant deterrents to use, followed by absence of

Medicaid, distance to health care, limited hours and patient-sharing by physicians.

Charges and distance had a disproportionate impact on the low-SEP groups.

2.6.4 Summary of the reviewed models

Based on the foregoing discussion, it can be concluded that all models are

interrelated. Each has its merits and can contribute to health services research in

different ways. Put simply, HBM emphasises the role of perceptions, attitudes and

beliefs in explaining individual behaviour. It is more useful for studies focused on

health-related behaviours, but it is not so useful in predicting more socially or

environmentally determined behaviours. Importantly, the expanded Health Services

Utilisation models by Andersen et al., as well as the model by Dutton, identify the

multi-faceted and multi-level factors influencing different types of health services

utilisation. Thus, these latter models more fully reflect the increasing complexity of

health services utilisation. However, it is not certain that current Health Services

Utilisation models are appropriate for studying the mediators of the relationship

between SEP and use of preventive health services.

Reviewing the literature examined in the previous section, there are three issues to be

considered. First, the Health Services Utilisation models examine how these

individual factors influence, directly or indirectly, the different levels of different

types of use of health services (including preventive health services). Socioeconomic

factors are only recognised as predisposing and enabling factors in the Health

79

Services Utilisation models. Second, in this thesis, it is necessary to understand why

different SEP groups vary in their use of preventive health services and what

individual factors at the different levels mediate the relationship between SEP and

the use of preventive health services. Third, as discussed in the previous section, six

areas of potential factors at four different levels, which possibly explain the

relationship, are hypothesised. The current models, therefore, are insufficient to be

applied in this research study.

2.7 PhD theoretical framework and research questions

Taking all of the above issues into consideration, it is important to construct a

theoretical framework for the study of the relationship between SEP and use of

preventive health services in relation to CVD and diabetes in Australia. The

theoretical framework that has been developed is based on an amalgamation of the

models previously discussed, considering the multi-faceted factors influencing the

socioeconomic differences in use of preventive health services. Figure 2.12 presents

the PhD theoretical framework of the relationship between SEP and preventive

health services in a general practice setting.

80

System-level Factors (Health Policy and Health care system) - bulk-billing in general practice

Environmental-level Factors

- Accessibility to health services - Availability of health services - Perceived interpersonal care

Interpersonal-level Factors

- A regular source of health care Social/family relationships - Social support

Individual-level Factors

- Attitudes towards preventive health services

- Value of general practitioners - Belief in preventive care

- Value of good health

Figure 2.12 PhD theoretical framework for understanding the utilisation of preventive health services in general practice by different SEP groups

Socioeconomic Position (SEP)

- Education - Income (Family

income) Demographic Characteristics - Age (25-64 years old) - Sex

Utilisation of preventive health services –GP

- Blood pressure screening - Cholesterol screening - Blood glucose screening

Risk factors of CVD diseases and diabetes (type-2)

81

The framework illustrates the primary relationship between SEP and preventive

health services in relation to screening for and detection of risk factors for CVD and

diabetes in general practice. There are four groups of factors identified. The key

components in this framework are described in the following paragraphs.

The first component is Socioeconomic Position (SEP); this is the main predictive

factor as an independent variable. It is measured by individual education level and

family income. It has already been evident in many studies, as discussed previously,

that different age and gender groups use preventive health services differently. For

this study, the 25–64 age group has been chosen, and the reason for this is discussed

in Chapter 3. The classified SEP groups, therefore, will be stratified by gender and

adjusted for age.

The second component is Utilisation of Preventive Health Services in general

practice as a main outcome variable; this is related to primary prevention, including

blood pressure, blood cholesterol and blood glucose check-ups for risk factors of

cardiovascular diseases and diabetes.

The third component is the middle triangle, encompassing four levels of factors; that

is, system, environmental, interpersonal and individual, as discussed in Section 2.5.

From the discussion of empirical studies six core areas of factors at these four levels

were hypothesised to be the most likely intervening variables to impact the

relationship between SEP and use of preventive health services. It is worth noticing

that these hypothesised factors are not particularly salient for CVD and diabetes, and

they can predict many aspects of preventive health services use by SEP groups. For

the purpose of this research program, the identified factors are used to predict the

utilisation of preventive health services with regard to prevention of CVD and

diabetes by different SEP groups. The following factors are included: (1) societal

factors that refer to health policy and the health care system, such as bulk billing; (2)

environmental factors that refer to structural barriers and GP’s attitudes, including

availability of and accessibility to health services, for example transport, accessible

services at a cost, appointment times, travel and waiting time, and perceived

interpersonal care from GPs; (3) interpersonal factors that refer to family resources,

82

including a regular source of health care and social support; and (4) individual

factors that refer to personal health beliefs and attitude, including the value of

preventive health services, the value of general practitioners, and the value of good

health. More detailed scales related to the factors of each proposed area will be

specified in Chapter 3.

Therefore, utilising this framework, this PhD research project aims to examine and

explain the relationship between SEP and the utilisation of preventive health services

in relation to CVD and diabetes in Australia. It addresses two key research

questions:

1. Do different SEP groups in Australia show different patterns of use of

preventive health services related to CVD and diabetes in general practice?

2. What are the key factors that influence this relationship in general practice in

an Australian context?

2.8 Summary

Socioeconomically disadvantaged groups experience significantly higher mortality

and morbidity rates than socioeconomically advantaged groups. People in lower

socioeconomic groups are more likely to die from cardiovascular disease and

diabetes than people in higher socioeconomic groups. It is evident that prevention

can reduce socioeconomic health inequalities, and that the use of preventive health

services is related to the reduction of morbidity and mortality. However, an

increasing number of studies have examined the relationship between SEP and use of

preventive health services, and indicated that low-SEP groups are less likely to use

preventive health services, including immunisation, cancer screenings, dental care,

screening for risk factors for heart diseases and diabetes, and preventive check-ups.

GPs play an important role in promoting preventive activities in the primary health

care setting in Australia, and low-SEP groups should be targeted.

There are many factors that influence the relationship between SEP and utilisation of

preventive health services. The following potential factors arising from the literature

83

review are proposed, that SEP differences in the use of preventive health services are

likely to be due to the following common factors:

• Health system bulk billing;

• Perceived availability of and accessibility to health care;

• Perceived interpersonal care from general practitioners;

• Regular source of care;

• Social support; and,

• Attitudes and beliefs towards preventive health care and general practitioners.

To date, most studies of preventive health service utilisation in relation to CVD and

diabetes have been conducted in North America, and relatively little such research

has been conducted in Australia. A theoretical framework derived from existing

models of health service utilisation is used to guide the PhD research in order to

explore the relationship between SEP and use of preventive health services in

relation to CVD and diabetes in Australia. The methods used in this study are

described in the next chapter.

84

3 STUDY METHODS

85

3.1 Introduction This chapter describes the study design, study population and sample selection, and

also outlines the research instrument for a survey concerning the use of preventive

health services in the general adult population. It also provides a detailed account of

the data collection procedure for the mailed survey and describes the study

measurements, statistical methods and analytical plan for data analysis. Figure 3.1

provides an outline of the study methods used.

Figure 3.1 Summary of study methods

Study design and study population

Sample selection

Survey questionnaire

Measurement

Sample calculation

Sampling frame

Sampling procedure

Identify questions and scales

Development of questionnaire questions and scales

Analytical approach and data analysis

Mailed survey — Five contacts

Data entry

SEP variables

Demographic variables

Intervening variables

Dependent variables

Identify domains for questionnaire

Pre-testing questionnaire

Expert review of draft questionnaire

Data collection

Data management

Sample exclusion Exclusion of respondents with CVD and diabetes

Management of missing data

Data coding

Data verification

Data cleaning

Final modification of questionnaire

86

3.2 Overview of study design and study population

This study is a cross-sectional study, and a self-administered mailed survey in the

form of a questionnaire has been employed for data collection. In order to describe

and characterise the utilisation of preventive health services in general practice in

Australia, a cross-sectional study design was employed because as this design can

provide a good overview or ‘snapshot’ of thoughts, feelings or behaviours in relation

to utilisation of health services by a large number of people. In addition, the cross-

sectional study design is economical in relation to time and resources, as large

numbers of people can be surveyed more quickly than in a longitudinal study

(Bowling, 2002).

Similarly, a mailed survey was employed in the study because this method of data

collection is relatively quick and more cost-effective than personal or telephone

interviews, due to the time constraints of the PhD research program and the scarcity

of funding. In addition, this allows people to complete a questionnaire in the privacy

of their own home, thus minimising social desirability and interview bias (Bowling,

2002; De Leeuw, 2001; Dillman, 2000; Turrell, 2000; Van Campen, Sixma, Kerssens,

& Peters, 1998). Nevertheless, both the cross-sectional design and mailed survey

method have their advantages and disadvantages. These are discussed in Chapter 6.

Individuals were recruited from the Brisbane Local Government Area in the State of

Queensland, Australia. The study population included working adults aged

25–64 years at the time of the data collection period, September to December 2004.

Following the method used by the Australian Bureau of Statistics (Australian Bureau

of Statistics, 1995), the working age population was divided into three groups:

youths (15–24 years), prime working age (25–44 years) and older working age

(45–64 years). The latter two age groups were chosen for this study because there are

evidence-based guidelines for both CVD and diabetes prevention that are relevant for

those between 25–64 years of age.

87

3.3 Sample selection

3.3.1 Sample calculation

The required sample size was based on an estimated proportion of 20–25 per cent of

the general population who had attended a GP for a check-up/examination between

February 1995 and January 1996 (Lorant et al., 2002). From the literature review

(Chapter 2), it had already been established that low socioeconomic groups are less

likely to use preventive health services than more advantaged groups. It was assumed

that this figure (20–25 per cent) applied to the medium-SEP group, and that an

absolute difference of 15 per cent or greater would be meaningful to report as

statistically significant. A total of 97 respondents per group was required in order to

detect a 10 per cent rate of preventive health service use among the low-SEP group

compared to a rate of 25 per cent among the medium-SEP group and, similarly, to

detect a 25 per cent as opposed to a 40 per cent rate of preventive health service

usage among medium- and high-SEP groups. This assumes a 5 per cent type I error

(two-tailed) and 80 per cent power. Hence, to cover all comparisons, a total of 97

respondents per group was needed to add a 40 per cent allowance for non-response, a

15 per cent allowance for multivariable modelling to adjust for confounding, a 20 per

cent measurement error and a 20–30 per cent assumption to allow investigation of

people having CVD and diabetes conditions that were separated from healthy

populations. Finally, a total of 800 subjects needed to be sampled. The sample

calculation procedure is presented according to Kirkwood’s formula (2003).

3.3.2 Sample calculation procedure Number of participants per group = 2.82 × [P1× (1 – P1) + P2 × (1– P2)] / (P1– P2)2 (P1 = 0.1, P2 = 0.25, 2.82: an index of 80% of power) N = 2.82 × [0.1 × 0.9 + 0.25 × 0.75] / 0.152 = 7.84 × 0.2775 / 0.0225

= 97 + statistical adjustment (15%) = 15 + measurement error (20%) = 20 Sample per group = 132 × 3 groups = 396

88

+ 60% response rate = 555 (or 50% response rate) = 595 (600) + 20–30% (CVD an diabetes conditions) = 800 3.3.3 Sampling frame

The study sample was drawn from the Australian Electoral Roll, which is maintained

and regularly updated by the Australian Electoral Commission (AEC). The electoral

roll includes all eligible citizens 18 years of age and over who are registered to vote

in Australian elections. A formal letter requested extraction of a list of all people

living in the Brisbane area in the age range 25–64 (see Appendix 3.1).

A formal agreement was made between the AEC and the Queensland University of

Technology (QUT) in order to receive the data. Data extractions from the AEC have

their own policy requirements. The data are confidential and may only be used for

purposes permitted by the Commonwealth. The Electoral Act 1918, ‘Agreement for

the Safeguard of Electoral Information used in Medical research and Public Health

Screening Programs’, sets out the conditions for appropriate use. The data were then

extracted by the AEC, based on five-year age ranges across the eight age groups: 25–

29, 30–34, 35–39, 40–44, 45–49, 50–54, 55–59 and 60–64 years. Electors’ ages were

calculated using 7 September 2004 as the reference point. In order to obtain the

electorates that encompassed all Brisbane suburbs, the 12 federal divisions covering

all Brisbane electorates were identified. The information obtained for each electorate

included elector ID, surname, given names, gender, enrolled address and postal

address.

3.3.4 Sampling procedure

Once obtained from the AEC, elector data were imported into an SPSS file. The files

obtained for each of the eight age groups were combined into one file, which

included 752 054 records of 25–64-year-old electors for the 12 federal divisions. The

data set was then reduced by selecting the Brisbane local government area, based on

the enrolled address (residency address), down to 438 702 records. The final sample

of 800 participants was then randomly selected using the Data Selection function in

the Statistical Package for Social Science (SPSS)12.01(SPSS, 2004). The final

89

selected sample comprised 393 (49.12 per cent) males and 407 (50.88 per cent)

females.

3.4 Survey questionnaire

A number of steps are critically important in the development of a satisfactory

questionnaire (Bowling, 2002; Boyce, 2002; Dillman, 2000; Oppenheim, 1992;

Windsor, Baranowski, Clark, & Cutter, 1994). The key steps used to establish a

satisfactory questionnaire for this study included:

1. identification of salient and relevant domains;

2. identification of existing adequate scales, as well as of those domains that

required additional scales to be developed;

3. development of relevant questions and scales, and construction of the

questionnaire;

4. expert review of the draft questionnaire; and

5. pre-testing and refining of the questionnaire.

3.4.1 Domains for questionnaire Six key domains were developed from the research framework, as outlined in

Chapter 2. These domains—the key variables, definitions and descriptions that were

developed for the initial questionnaire—are summarised in Table 3.1.

90

Table 3.1 Domains of the initial questionnaire and their variables and definitions

3.4.2 Identifying questions and scales

A search for relevant published articles was conducted in order to identify relevant

questions and scales consistent with the domains listed in Table 3.1. The key search

terms used to identify potentially relevant questionnaires, questions and /or scales

were, ‘preventive health services or preventive health care and beliefs or attitudes or

Domains

Variables Definitions

Utilisation of preventive health services — GP Self-reported preventive

care activities Blood pressure measurement within the past 2 years; Cholesterol screening within the past 5 years; Blood glucose screening within the past 5 years.

Socio-demographic variables Income Actual total income of the family to which the individual belongs

in thousand of dollars. Education The individual level education by post-school qualification

categories. Age The actual age in years of the individual. Gender Male and female groups. Perceived system-level factor Health care system Bulk billing — if GPs bulk bill patients for the consultation fee.

Perceived environmental level factors Transport

Transport to GP clinic: public / own car to access to health services.

Travel time Travel time to the regular source of medical care.

Accessibility of preventive health services

The cost of seeing a GP Out-of-pocket cost from patients related to the consultation fee. A number of doctors to choose from

Available doctors for patients to choose e.g. female doctors.

Able to see preferred doctor

A doctor who patients would like to see.

Appointment time Elapsed time between initial request and the date of the appointment for medical care.

Waiting time Waiting time in GP’s office or clinic.

Availability of preventive health services

Consultation time How long GP consults with the patient. GP’s attitudes towards patients

Inter-personal care scale GP’s perceived attitude towards patients.

Interpersonal level Family sources Regular source of care Regularly visit GPs at one practice or one doctor.

Social support Social support Support from family, friends and neighbours for use of preventive health services.

Individual level Value of GPs An index based on Andersen, patients’ perceptions towards

doctors. Value of health care An index based on Andersen, patients’ perceptions towards

health care.

Attitudes and beliefs towards health care

Value of good health An index based on Andersen, patients’ attitudes towards doctor’s advice about improving their health.

91

behaviours and instrument or scales’. The search was conducted via the EBSCO

HOST search engine, including the Medline, PsychINFO, Primary Search and

PsycARTICLES databases. The search was also conducted through international and

Australian instrument web sites, including the Australian Centre on Quality of Life

Instrument Database at http://acqol.deakin.edu.au/instruments/index.htm, and

BUROS Institute of Mental Measurements at www.unl.edu/buros. Another type of

search was conducted through specific health service researchers; for example,

searching for articles by overseas authors Andersen and Dutton, and studies by

Australian authors Mark Harris and Sanson-Fisher. Consequently, relevant studies

from 1972 to 2004 were examined.

Potentially relevant scales and questions relating to each domain were then extracted

from a review of all the existing questionnaires and formative studies. Six key

questionnaires and studies were identified.

1. A behavioural model of families’ use of health services (Andersen et al.,

1975). In this study, Andersen initially proposed the model and tested it

empirically to assist in understanding why families use health services; to

define and measure equitable access to health care; and to assist in

developing policies to promote equitable access. The study developed some

relevant variables to measure families’ attitudes towards, and beliefs about,

health service use.

2. General Practice Assessment Survey (GPAS) (Ramsay, Campbell, Schroter,

Green, & Roland, 2000). This is a questionnaire developed at the National

Primary Care Research and Development Centre (NPCRDC) in Manchester.

GPAS focuses on questions about access, interpersonal aspects of care and

continuity of care. This questionnaire is also used as a quality control tool.

GPAS has two versions: one to be sent by post and the other given to patients

after medical consultation. The questions in GPAS have been tested

extensively and validated in general practice in the UK.

3. 2001 National Health Survey (Australian Bureau of Statistics, 2000). This

survey was conducted throughout Australia from February to November 2001

(General Practice Statistics and Classification Unit, 2003). It was the fifth in

the series of health surveys conducted by the ABS. The survey collected

information about the health status of the population, including long-term

92

medical conditions and recent injuries; use of health services and visits to

hospital, and other actions people had recently taken for their health; health-

related aspects of people’s lifestyles, such as smoking, diet, exercise and

alcohol consumption; and demographic and socioeconomic characteristics.

4. The Household, Income and Labour Dynamics in Australian (HILDA)

survey project (Butterworth & Crosier, 2004; Wooden & Watson, 2000).

This is a household-based panel study which began in 2001. It collects

information about economic and subjective wellbeing, labour market

dynamics and family dynamics. Interviews are conducted annually with all

adult members of each household.

5. The Australian Longitudinal Study on Women's Health—widely known

as Women's Health Australia (WHA) (Loxton, Schofield, & Hussain, 2004;

WHA, 2004). This is a longitudinal population-based survey that examines

the health of over 40 000 Australian women over a 20-year period. It was first

funded in 1995, and designed to explore factors that influence health among

women who are broadly representative of the entire Australian population.

The study goes beyond a narrow perspective that equates women's health

with reproductive and sexual health, and takes a comprehensive view of all

aspects of health throughout a woman's lifespan. The project assesses:

• physical and emotional health (including wellbeing, major diagnoses,

symptoms)

• use of health services (GP, specialist and other visits, access, satisfaction)

• health behaviours and risk factors (diet, exercise, smoking, alcohol, other

drugs)

• time use (including paid and unpaid work, family roles and leisure)

• socio-demographic factors (location, education, employment, family

composition)

• life stages and key events (such as childbirth, divorce, widowhood).

6. Brisbane (Australia) Food Study. In this study, Turrell et al. (Turrell,

Patterson, Oldenburg, Gould, & Roy, 2003) collected area-and individual-

level information to examine the relative contributions of environmental and

individual factors to socioeconomic variability in food purchasing behaviour

in 2000. The data were collected by home-based face-to-face interviews of

93

1003 households from 50 small areas. The data consisted of a range of

socioeconomic and demographic measurements.

The authors of each of these studies were contacted, and permission was sought to

use and/or adapt the previously developed measures. Additional psychometric

information and any other relevant information for each of these scales was also

requested. The criteria for suitability of use for the present research included: the

scale had been tested psychometrically and validated; the content of the scale was

suitable for the new research purpose; and the questionnaire items and permission

could be obtained from the author. Subsequently, six scales and one dichotomous

question with two items were selected according to these criteria. These scales

covered the following areas: measuring people’s attitudes and perceptions towards

preventive health care and health services at individual, interpersonal, environmental

and system levels. They included: concerns about availability and accessibility to

health care, which were adapted from the WHA survey; perceived interpersonal

care, which was adapted from British general practice assessment survey; attitude

towards health care, value of GPs and value of good health, which were adapted

from Andersen’s research questionnaire; and social support, which was adapted from

the HILDA survey; and a regular source of care, including a regular care provider

and place of care, which was adapted from the WHA survey. The remaining

background questions were also adapted from the 2001 NHS and developed by the

research team. Questions relating to self-reported preventive services use were

developed based on the Guidelines for Preventive Activities in General Practice

(RACGP, 2002). SEP and social demographic variables were employed from BFS.

Table 3.2 provides the details of questions and scales used in developing of the

questionnaire.

Based on this review of existing relevant scales and measures, a number of important

issues arise. First, although Health Services Utilisation Models have been applied for

many years in some countries, little progress has been made in developing valid and

reliable scales for measuring their concepts and variables. For example, to measure

health-related attitudes, Dutton (Dutton, 1978) used a health belief scale based on

three questions ranked from 1 = low to 4 = high, and Shankar (Shankar, 2000) used

agree and disagree, while Andersen used a 5-point-format response, including

94

strongly agree, tend to agree, tend to disagree, strongly disagree and non applicable.

Second, there is no standard measurement, or some measurements have not been

used widely to measure specific determinants of preventive health services utilisation.

Third, there is little evidence as to whether these measurements have already been

formally tested with regard to their validity and reliability from this review. Fourth,

most of the studies have been done in the United States (US). The health care

systems in Australia and the US are different, as discussed in Chapter 2. For example,

Australia has the Medicare system, which gives people free access to public hospitals.

Some general practices bulk bill patients. The private insurance policy and

organisational system of health services are all different when compared to other

countries. Finally, there are not many specific measurements to measure use of

preventive health services in relation to CVD and diabetes.

Since some of the scales chosen for the questionnaire were developed by overseas

researchers such as those in America, some of the questions and items had to be

revised in order to fit the Australian context. This was necessary because of

differences of language and culture. In addition, some adapted items from the

Australian studies have been reworded. Therefore, reliability and validity testing of

adapted and developed scales was conducted and the scales were then employed in a

population-based survey to explain the influence of SEP on the utilisation of

preventive health services. In order to measure those determinants of preventive

health service utilisation that were identified from the literature early on, a number of

questions and measurement tools needed to be adapted and developed. However, to

increase the face validity of the scales, the original phrasing of items is preserved as

far as possible. These are discussed in the following section.

95

Table 3.2 Details of questions and scales used in questionnaire development

No Description Sources Psychometric information

Citations

Health status and lifestyle 1 Health status 2001 NHS Q202 n/a ABS (2000) 2001 National Health Survey: Survey Questionnaires—The

adult questionnaire. .abs.gov.au/Websitedbs/D3110122.NSF/4a255eef008309e44a255eef00061e

57/2656b8f177b47ec4ca256c5d0002bb94/$FILE/typ2026adult.pdf. 2 Health problem Developed n/a 3 Condition of CVD

Condition of diabetes, information on Medication Adapted from 2001 NHS Q450 & Q506 n/a ABS (2000) 2001 National Health Survey: Survey Questionnaires—The adult

questionnaire. 4 Family history Developed n/a 5 Smoking status 2001 NHS Q220-222 n/a ABS (2000) 2001 National Health Survey: Survey Questionnaires—The adult

questionnaire. Utilisation of GP health services: 6 All consultations or visits to a GP

Health check up — outcome variable Adapted from 2001 NHS Q 726 n/a same as above.

7 Self-reported preventive care activities: screenings and information required related to CVD and diabetes

Developed from the Guidelines — red book

n/a Guidelines for Preventive Activities in General Practice, RACGP, 2002.

Access to GP health services: 8 Government concession card information 2001NHS Q 759

n/a ABS (2000) 2001 National Health Survey: Survey Questionnaires—The adult

questionnaire. 9 Private health insurance information 2001 NHS Q751,Q753 n/a The same above. 10 A regular care provider and a regular place of care Women’s Health Survey Q14

( two singe items) n/a Women’s Health Australia (2004), Fourth survey for mid-age women.

http://www.alswh.org.au/Surveys_data/Surveys/Mid4Survey.pdf

Attitudes and perceptions towards GPs and health services :

11 Concerns about availability and accessibility of health care

Adapted from WHS Q27 No information Women’s Health Australia (2004), Fourth survey for mid-age women. http://www.alswh.org.au/Surveys_data/Surveys/Mid4Survey.pdf

12 Personal beliefs and attitude can be affected by health care — inter-personal care

GPAS questionnaire (3 items)

Reliability = 0.937

Ramsay et al. (2000) The general practice assessment survey (GPAS): tests of data quality and measurement properties, Family Practice, 17 (5): 372–9.

13 Personal beliefs and attitude can be affected health care - value of GPs

Adapted from Andersen’s scale (5 items) Reproducibility = 0.90

R Andersen (1974) A behavioural model of families’ use of health services. Centre for Health Administration Studies, University of Chicago

14 Attitude towards GP health services Adapted from Andersen’s scale (4 items) Reproducibility = 0.89

As above.

Personal beliefs on health and social support 15 Value of good health Adapted and developed from Andersen’s

scale (5 items) Reproducibility = 0.91

As above.

16 Salience of health Developed (single item) n/a n/a 17 Personal beliefs and attitude can be affected by health

care- Social / family support HILDA survey project (10 items) No information Wooden and Watson (2000), The household, income and labour dynamics in

Australia (HILDA) survey: An introduction to the proposed survey design and plan. HILDA project discussion paper series NO. 1/00.

18-29

SEP and demographic information Brisbane Food Study (BFS) (12 questions)

n/a Turrell et al. (2003), Measuring socioeconomic position in dietary research: is choice of socioeconomic indicator important? Public Health Nutrition, 6(2), 191–200.

96

3.4.3 Development of questionnaire questions and scales 3.4.3.1 Formatting questions and scales

Most of the questions were developed as pre-coded (closed) questions that were

followed by a list of possible answers. Respondents could tick whichever pre-coded

answer applied. Additionally, a few questions were filter questions that respondents

could skip if they did not apply to them. Questions were numbered and items were

also clearly labelled. Some filter questions clearly provided directions in order to

help respondents understand which question to go to next.

Three types of question were employed in the questionnaire: single-item questions,

battery questions and scales. The single-item measure uses a single question to

measure the concept of interest; for example, a question with regard to health status.

Batteries of questions are a series of single items, each relating to the same variable

of interest (Bowling, 2002), but each item is analysed and presented individually.

Scales involve a series of items about a specific domain that can be summed to yield

a score with different levels of meaningful degrees (Bowling, 2002).

The response formats for questions in the questionnaire included a dichotomous

response format (Yes–No), a four-point, five-point and a six-point response format.

The latter format requires writing a collection of statements, each of which is clearly

positive or clearly negative with respect to the construct of interest. For example, the

respondent reads each statement and selects a response from a six-point continuum:

strongly agree, moderately agree, agree, disagree, moderately disagree, strongly

disagree. Neutral statements were not included. The survey contained a series of

‘opinion’ statements about an issue; that is, the extent to which the respondent agrees

or disagrees with each statement.

3.4.3.2 Structuring of questionnaire

Questions were then grouped together by subject and topic. This method was felt to

be more professional, structured and easy for respondents to follow. Equally

important, sensitive questions related to income were placed later in the

questionnaire (Boyce, 2002). Based on this principle, six topics of questions were

97

determined in the questionnaire based on previously identified domains (see

Section 3.4.1) and each of the topics was placed together. There were questions

related to:

• health status and lifestyle;

• use of preventive health services;

• access to GP health services;

• attitudes towards GP health services;

• personal beliefs on health and social support; and

• SEP and demographic information (see also Table 3.2).

Subsequently, a draft questionnaire, A survey about the use of health services

(Health Service Utilisation Questionnaire), was constructed, with 29 questions

(including six scales), using the six topic areas and divided into two parts.

The first part of the Health Service Utilisation Questionnaire (Part A) contained 17

closed questions that addressed issues related to respondents’ own health and use of

health services and included the first five topics as follows.

Health status and lifestyle

This topic included perceived health status, health problems, smoking status, CVD

condition and diabetes, and family history of CVD and diabetes. These variables

were added because it was useful to obtain a profile of the health status of the

population and also to check the quality of data; in particular, to identify any health

problems, conditions of CVD and diabetes from the general population.

• Health status: This item was derived from the 2001 National Health

Survey (Australian Bureau of Statistics, 2000), which is conducted by the

Australian Bureau of Statistics every five years. This was a single-item

question with five dimensions of response: excellent, very good, good,

fair, poor. The purpose of this measure was to assess the perceived health

status from the sample population and to compare it with the national data.

• Health problem: This item was developed as a single dichotomous

question designed to identify whether the participants had illness

conditions. The people with these conditions were excluded when

98

analysing the use of preventive health services because the purpose of this

research was to investigate people’s preventive health behaviours in the

absence of illness.

• Condition of CVD and diabetes: This question was also adapted from

the 2001 NHS. It was a two-item question with a dichotomous response.

The purpose of creating this question was the same as the health problem

question. Additionally, it provided a profile of CVD and diabetes

conditions in the sample population.

• Family history of CVD or diabetes: This question was developed also

as a two-item question with a dichotomous response. As is well known, a

family history of CVD or diabetes is a very important factor for the

development of CVD and diabetes. This information provided baseline

data for the sample population.

• Smoking status: This question was employed from the 2001 NHS and a

included a filter question asking about current and past smoking status.

The question also provided a profile of the sample population.

Utilisation of GP Health Services

• Consultations or visits to a GP: This was a developed measure based on

the purpose of this research and was an open question. The question asked

participants about all of their consultations or visits to a GP, including

health examinations or check-ups that they had undergone in the previous

12 months.

• Self-reported preventive activities: This question was developed based

on the Guidelines for Preventive Activities in General Practice (RACGP,

2002). The question contained eight preventive activities, with six

response choices about accessing GP health services including, in the last

12 months, in the last 2 years, 2–5 years ago, more than 5 years ago,

never and don’t know. The question measured self-reported preventive

activities related to CVD and diabetes, including blood pressure screening,

blood cholesterol screening, blood glucose screening and weight

assessment; and consultation on diet, physical activity, smoking and

alcohol.

99

Access to GP health services

• Information on government concession card: This question was

derived from the 2001 NHS and was a dichotomous question with four

items. The question asked whether people had a health care card,

pensioner concession card, Commonwealth senior’s health care or

veterans’ affairs card that helped them to obtain bulk billing and a

pharmaceutical benefit, or other particular concessions. This could affect

people’s use of a health service.

• Private health insurance information: This question was derived from

the 2001 NHS. The question asked respondents whether they had any of

the different types of private health insurance, as this would affect their

access to certain health services.

• Regular source of care from GP: This question was derived from the

2004 WHA. The question had two items with a four-response format:

always, most of the time, sometimes, rarely or never. The question asked

whether people had a regular source of care; for instance, going to the

same GP or seeing the same doctor.

Attitudes and perceptions towards GPs and health services

• Concerns about availability and accessibility of health care: This scale

was adapted from the 2004 WHS mid-age questionnaire. The scale

included nine items that measure people’s concerns about availability and

accessibility of GP health care, particularly being able to find a GP who

bulk bills; choice of GP; length of time to make appointment; waiting

time; cost of seeing a GP; transport to see a GP and travelling time to a

GP. The response format was: very important, fairly important, not very

important, not at all important, I don’t think about it, not applicable.

• Professional attitudes towards patients—perceived interpersonal care:

This was derived from 2004 GPAS questionnaire, including three items.

The scale measured personal beliefs about and attitudes towards health

care received from a GP, involving GPs spending time with patients,

showing patience, showing caring and concern. It had a five-point

response format: excellent, very good, good, fair, poor.

100

• Value of GPs: This scale was adapted from Andersen’s scale, including

five items. It measured how people value GPs and health care. It had a

six-point response format: strongly agree, moderately agree, agree,

disagree, moderately disagree, strongly disagree.

• Attitude towards health care: This scale was also adapted from

Andersen’s scale, including four items. It measured people’s attitudes

towards health care received from GPs. It had six-point response format:

strongly agree, moderately agree, agree, disagree, moderately disagree,

strongly disagree.

Personal beliefs on health and social support

• Value of good health: This scale was adapted from Andersen’s scale,

including five items. The intention was to measure how people’s values

of good health can affect their use of health services. A six-point response

format was used: strongly agree, moderately agree, agree, disagree,

moderately disagree, strongly disagree.

• Salience of health: This was a single item and was based on Dutton’s

study (Dutton, 1978). It was intended to measure how frequently people

discuss health issues with family, friends and colleagues, reflecting their

beliefs and attitudes towards health care and influencing their use of

health services. The response format included: a lot of the time, some of

the time, almost never, never.

• Social support: This was adapted from the HILDA survey and measured

social and family support for an individual’s health care. The question

was a 10-item scale with a six-point response format: strongly agree,

moderately agree, agree, disagree, moderately disagree, strongly disagree.

The second part of the Health Service Utilisation Questionnaire (Part B) included 12

social demographic questions. Part B sought information on respondents’ SEP and

demographic characteristics. All questions in relation to SEP were derived from the

BFS (Turrell et al., 2003). These were measurements that had been used in many

Australian studies, and they are summarised as follows.

101

Socio-demographic measurement questions provide useful information on

geographic location, living situation, employment situation, and occupation and

socioeconomic situation, and are described as follows:

• Country of birth, including two response choices: Australia and other

• Age in years

• Gender: male or female

• living arrangements, with six response choices, including—

– Single and living by yourself

– Single and living with friends or relatives

– Single parent living with one or more children

– Couple (married or defacto) living with no children

– Couple (married or defacto) living with one or more children

– Other (specify).

• Length of residence in current address, in years or months or days

• SEP questions included education or highest qualification: education

information was obtained by asking respondents two questions. The first

question asked respondents whether they had any post-school educational

qualification. If the answer was ‘yes’, they were directed to the next

qualification question, which included nine choices. The choices were:

trade or business certificate, apprenticeship, associate diploma,

undergraduate diploma, bachelor degree, post graduate diploma, master’s

degree, doctorate and other.

• Employment situation: this question gave respondents ten response

choices, including: full-time, part-time, casual, work without pay, home

duties, unemployed, retired, permanently unable to work, student and

other employment situation.

• Current job information: If respondents were employed, they then needed

to answer whether they were employed by someone else, or were self-

employed in own business with employees or with no employees.

• Occupation: Occupation information was obtained by asking respondents

two questions. The first question asked respondents to write their current

occupations; the second question asked respondents to provide what main

tasks they dealt with. By doing this, a precise occupation information

102

from respondents was obtained. This occupation question was not used to

measure SEP; the reasons are discussed in Chapter 7.

• Family income: Family income information was obtained by asking

respondents to estimate their household’s total annual, fortnightly or

weekly income, including pensions, allowances and investments. The

question used a single measure consisting of 14 narrow-ranged income

categories and one choice for respondents who did not wish to respond.

3.4.4 Expert review of draft questionnaire

It is important for experts and knowledgeable peers to review and comment on a

questionnaire (Bowling, 2002; Dillman, 2000). The expert review conducted as part

of this research focused on two main issues: first, whether the proposed scales and

questions provided sufficient coverage of the required domain areas; and, secondly,

whether the expert was aware of other existing measures or questionnaires.

Eight national and international experts were identified as having expertise relevant

to this field of research (see Appendix 3.2 for details). Six experts finally agreed to

review the questionnaire.

A review package was sent to the experts by email, and contained the following four

documents:

1. Reference table of the questions and scales (see Table 3.2)

2. Conceptual framework identifying the key variables (see Figure 2.11)

3. One-page summary of the intended purpose of the questionnaire, which

helped conceptualise the development of this questionnaire and its intended

purpose (not shown), and

4. A draft questionnaire.

Comments and suggestions were provided by six experts through regular mail and

email. The general comments were related to the reading level of the questionnaire,

formatting of the questionnaire, and the wording and sequence of questions. For

example, the reading level could have been too high and better considered for

12-year-old students. Question formatting may have been needed to be left justified

103

instead of right. Some questions needed to be worded consistently and organised in a

logical sequence. The experts also provided specific comments on each question or

item of scales; for example, the order of certain items and wording of questions. For

the question about health services use, the experts asked: ‘Are a general health

examination and a health check-up the same thing?’ The time frame for check-up

would be for a longer period. How many people would have a general check-up more

often than once every year or so? What should a general examination include? They

also suggested that there needed to be a question about whether people would see

another doctor sometimes for specific things such as screenings. For example, when

a person sees a doctor other than their usual doctor, what are the reasons for this? Do

people see the doctor for certain things such as screening for cancer, women’s and

men’s health problems? The experts suggested considering other relevant

questionnaires: SNAP (S: quit Smoking, N: better Nutrition, A: moderate Alcohol

and P: more Physical activity), a survey on behavioural risk factors in general

practice, and the Quality of Life Questionnaire.

3.4.5 Pre-testing of questionnaire

Once the experts’ feedback had been incorporated into the questionnaire, it was then

pre-tested. The purpose of questionnaire pre-testing is to test the proposed

administration procedures, to establish whether people understand the questions, and

to identify any other issues that might arise, for example the length of the

questionnaire. (Dillman, 2000).

A detailed checklist for pre-testing was developed, and contained 10 questions (see

Table 3.3). In addition, a letter of invitation was prepared to inform participants

about the purpose of the pre-testing, the objective of the survey and to provide

instructions on how to complete the questionnaire (See Appendix 3.3).

Ethical approval was obtained from the University Human Research Ethics

Committee, Queensland University of Technology, before conducting the pre-testing

(See Appendix 3.4).

104

Table 3.3 Pilot survey checklist

PLEASE READ THIS BEFORE COMPLETING THE QUESTIONNAIRE PLEASE MAKE COMMENTS IN THE MARGINS WHILE FILLING OUT THE SURVEY 1. How long did it take you to complete the survey? __________ minutes 2. Did you complete it all in one go or over 2 or more sessions?

All in one go Over 2 or more sessions

3. How did you find the length of the survey?

Too long Long but bearable Fine Too Short

4. How did you find the size of the print?

Readable Difficult to read at times (please mark these questions in the margins of the survey)

5. Did you find the tick boxes (e.g. ) impractically close or too small?

Yes No

6. Please mark these questions which you found too technical or difficult to understand?

Yes (please mark these words in the margins of the survey) No

7. How did you find the instructions? e.g. please go to Question 30, Please tick one only

Understandable/unambiguous Some were hard to understand (please mark these instructions in the margins of the

survey) 8. Could you please mark those questions which are repetitive?

Yes (please mark these questions in the margins of the survey) No

9. Did you find any of the questions offensive or too sensitive?

Yes (please mark these questions in the margins of the survey) No

10. Do you have any other comments about the survey? __________________________________________________________________ __________________________________________________________________ __________________________________________________________________

105

The questionnaires for pre-testing were distributed to each participant, with an

invitation letter and a checklist. Each participant was required to read the checklist

prior to filling out the questionnaire. Feedback and comments on the questionnaire

were obtained and the questionnaire was revised accordingly.

Working individuals aged from 25 to 64 years, and from different SEP backgrounds

were selected from blue-collar workers, general staff at Kelvin Grove campus of the

Queensland University of Technology (QUT) and from a local church congregation.

Both QUT and the local church congregation are located in the inner city of Brisbane,

in the state of Queensland, Australia. More people from low-SEP backgrounds were

selected because of greater concern for low-SEP groups in relation to readability of

the questionnaire and their comprehension ability. There were 22 surveys given out

for pre-testing. Twenty-two questionnaires were returned. The response rate for the

pre-testing was 100 per cent. The results of the pre-testing were as follows.

The average age of participants for the pre-testing was 37. Twelve males and ten

females participated in this pre-testing. The average length of time for filling out the

questionnaire was 22 minutes (completion times ranged from 10 to 45 minutes). The

majority of participants filled out the questionnaire all in one go, and agreed that it

was readable and understandable, despite being long. There were 11 participants who

did not give any comments, while 11 participants wrote specific comments. General

comments were given by those participants, and related to clarity of instruction of

question, repetition of some items, irrelevancy of items, length of questions,

similarity of items and sensitivity of questions such as income.

3.4.6 Final modification of questionnaire

Following completion of the above steps, the final Health Service Utilisation

Questionnaire consisted of a 10-page A4 booklet (See the questionnaire in

Appendix 3.5). The questionnaire included 79 items that consisted of: 12 socio-

demographic items; 10 items assessing health status, disease conditions and smoking

status; 20 items assessing use of health services; and 37 items assessing the factors

that might affect use of health services utilisation, which were presented in the

following order:

106

1. Health status and lifestyle: covering perceived health status, health

problems, smoking status, CVD condition and diabetes, and family history of

CVD and diabetes. This included five questions with a total of 10 items.

2. Use of preventive health services: covering consultations or visits to a GP

and self-reported preventive activities. This included a total of 10 items.

3. Access to GP health services: covering information on government

concession cards, private health insurance and regular sources of care from

GPs. This included a total of 10 items.

4. Attitudes towards GP health services: covering information with regard to

concerns about availability of and accessibility to health care, professional

attitudes towards patients—perceived interpersonal care, value of GPs, and

attitudes towards health care. This included a total of 21 items.

5. Personal beliefs about health and social support: including two scales and

one question related to the value of good health, salience of health and social

support. This included a total of 16 items.

6. SEP and demographic information: including 12 items with one choice of

answer on income, education, age and gender variables.

3.5 Data collection

3.5.1 Preparation phase The following steps were carried out prior to survey administration. These included

budgeting for the survey, designing and writing cover letters, thank-you and

reminder postcards, printing the mailing package and other related administrative

tasks. The costs included the printing of the questionnaire, postage and other costs

(data entry, sample purchase and instant lottery tickets). The cover letters were

prepared and edited, then hand-signed individually by the principal supervisor and

PhD candidate—it has been reported that personalising mailings for general public

samples can increase the response rate by 5–11 per cent (Dillman, 2000). A thank-

you and reminder postcard was also prepared. Fifteen hundred copies of the survey,

820 copies of the postcards and 1500 copies of C5 reply-paid envelopes were printed.

107

3.5.2 Mailing contacts Each selected respondent was contacted up to four times during the period

18 October to 26 November 2004.

Contact 1: Eight hundred surveys were mailed-out, with each package containing a

personalised cover letter, a survey, an instant lottery scratch-it, as well as a pre-

addressed and prepaid reply envelope. The cover letter explained the importance of

the research and the reason a response to the survey was important (see

Appendix 3.6). In order to increase the response rate, a gratuity (lottery scratch-it) for

each potential participant (800) was sent with the first mailing package. In this regard,

research has shown consistently that the inclusion of a small token financial gratuity

in advance can improve response rates significantly (Dillman, 2000).

Contact 2: Eight hundred thank-you and reminder postcards were sent a week after

the survey (see Appendix 3.7). They expressed appreciation for responding if this

had already occurred, and indicated that if the completed questionnaires had not been

mailed, it was hoped that they would be returned soon.

Contact 3: There were 377 replacement questionnaires with cover letters (see

Appendix 3.8) that were sent to non-respondents four weeks after the first

questionnaire mail-out.

Contact 4: There were 276 reminder letters sent to those people who had still not

returned their survey 6 weeks after the first mail-out. The final contact letter is

presented in Appendix 3.9. Overall, a total of 1400 surveys were sent off. Figure 3.2

shows a summary of contacts with survey respondents.

108

Week 1 Week 2 Week 3 Week 4 Week 5 Week 6

First mailing Thank-you First replacement Final contact (n = 800) Postcard (n = 377) Second replacement

(n = 800) (n = 276)

Figure 3.2 Data collection phase for self-administered mailed survey

3.5.3 Data recording and storage A respondent spreadsheet was developed, including each respondent’s ID, name,

gender, mail address and date of receipt of survey, to record the survey procedure

and to follow up the survey process to check whether the survey was returned or not.

Additionally, a survey diary was used to record the mailed survey procedure, the

number of copies of the questionnaire sent off and other administrative matters; for

example, to record the information, if any, of respondents calling the researchers or

ethics committee.

Each survey was checked for missing data and ineligible writing as soon as it was

returned. Any returned blank surveys were noted as unwilling and excluded when the

data was entered.

Every effort was made to ensure and maintain the confidentiality of survey

respondents. The mailing list of respondents was kept confidential. Each respondent

was assigned a unique identification number (ID). There were no names of

respondents appearing on the survey. The returned surveys were also recorded and

stored confidentially in A4 printing paper storage boxes with covers. The storage

boxes were then safely stored in a locked filing cabinet.

3.6 Data management

The process of preparing data for analysis commenced after the data collection phase

was completed and included the following four steps.

109

3.6.1 Data coding

Coding is a method of conceptualising research data and classifying it into

meaningful and relevant categories for the respondents in the study (Bowling, 2002).

The coding frame was set up after the data were collected. The basic rules for the

development of the coding frame were that the codes were mutually exclusive,

coding formats for each item were comprehensive and the codes were applied

consistently (Bowling, 2002). The number of digits assigned for each code depended

on whether the variable was a single-, double- or triple-column variable. For example,

since the maximum number of participants was 800, the code for ID was three digits.

It is customary in most research to use the codes 9 or 99 or 999 for inadequate

responses. This was also applied in this research. Codes for double-ticked responses

were 8 or 88, and skipped answers were left blank. A numerical code for each item

response from the survey respondents was also allocated. A master copy of the

questionnaire was prepared as a code book prior to data entry for data analysis

purposes. The full ranges of valid codes were written in the code book against each

question, as well as missing values and double-ticked values.

3.6.2 Data entry

Data entry was conducted by an experienced data entry professional using the

Notepad program. The data from each survey were entered into the dataset against

each ID number. The original entered data was then transferred into the Statistical

Package for the Social Science (SPSS) system file.12.01 (SPSS, 2004) for data

analysis. Each variable referred to a single item of the scales and needed to be named

or labelled before the data could be analysed. Each code for each item remained the

same as the code when the data was entered; for example, male = 1 and female = 2.

A new code was given when a new variable was created.

3.6.3 Data verification

To verify data and detect any entry mistakes, 51 subjects (10 per cent of the original

sample size) were randomly selected from the total sample of 519. The data

verification was processed in SPSS Data Entry Builder 2.0 by the PhD candidate.

There was 0.17 per cent of disagreement between the professional and verifying data

110

entry. The data were checked for any discrepancies between the two sets of coding,

errors were corrected and agreement was reached over any genuine differences in

interpretation between the data entry professional and PhD research candidate.

3.6.4 Data cleaning

To eliminate the errors that occurred during the data collection, coding and input

stages, data cleaning commenced once the data entry was completed. A standard

procedure for data cleaning was followed, using two types of checks. The first was a

check for outliers and wild codes by inspecting frequency distributions. The second

data-cleaning procedure involved consistency checks, which focused on internal data

consistency; for example, checking the filter questions for impossible combinations

and inconsistent values and meanings.

3.7 Sample exclusion

3.7.1 Management of missing data

Three types of missing values required attention in this study. The first one involved

item non-response, where there was no response to a particular question or item. The

second occurred when the answer was marked in a way that was unclear, or fell

outside the range of permissible responses for each of item of scale. In order to

minimise the amount of missing data for each item, a ‘best estimation’ method was

used. This used an estimated average value of the respondents who had a similar age,

gender, income and education to the respondents who missed the item (Bowling,

2002; De Leeuw, 2001; Dengler, Roberts, & Rushton, 1997). For example, if a

respondent who was male and 35 years old, and who had obtained a bachelor’s

degree in a middle-income group, missed one item of the social support scale, the

estimation was then made based on an average of score of the item of social support

that was generated from those who were male respondents and 35 years old, with the

equivalent level of education and income as close as possible. The third missing

value was for respondents who did not wish answer a specific question or item, or

missed out the question; for example, questions about family income and education.

Criteria for the management of this type of missing data were set up. All cases of

111

missed data were retained in the dataset in order to keep a maximum number of

participants for the analysis. The SPSS program can recognise those missed values as

‘missing’ in each individual variable. As a result, these ‘missing’ values were

excluded in each single data-analysis procedure.

3.7.2 Exclusion of respondents with CVD and diabetes conditions 3.7.2.1 CVD and diabetes problems and use of preventive health services

Among the usable sample of the mailed survey (N = 519), 22.2 per cent (n = 115) of

the respondents had CVD conditions and 7.5 per cent (n = 39) of the respondents had

diabetes. Importantly, it is noticeable from a preliminary data analysis that

respondents with CVD and diabetes conditions were more likely to go for a blood

pressure, blood cholesterol and blood glucose check-up than the rest of the

respondents. As shown in Table 3.4, the respondents with CVD conditions were

absolutely 20 per cent more likely to go for a BP check-up than the respondents

without a CVD condition, and this relationship was statistically significant (P =

0.000); the respondents with diabetes were absolutely 25 per cent more likely to go

for a BP check-up than the rest of respondents, and the relationship was also

statistically significant (P = 0.001). Similarly, the respondents with CVD conditions

were also nearly 30 per cent more likely to go for a blood cholesterol check-up and

21 per cent more for a blood glucose check-up than the rest of the respondents, and

these relationships were statistically significant (P = 0.000). Furthermore, the

respondents with diabetes were 33 per cent more likely to go for a blood cholesterol

check-up and 39.2 per cent more likely to go for a blood glucose check-up than the

rest of the respondents. These relationships were statistically significant (P = 0.000).

112

Table 3.4 Chi-square test for bivariate associations between chronic conditions and BP, BC and BG checked by GP in a total sample of 519 respondents

Chronic

conditions

No* BP check-up P -value

N = 499 (%)

No* BC check-up P -value

N = 492 (%)

No* BG check-up P -value

N = 487 (%)

CVD Yes No Missing

Diabetes Yes No Missing

113 90.3 0.000a 368 70.1 18

39 97.4 0.001d 432 72.2 28

111 87.4 0.000 b 362 58.0 19 38 94.7 0.000e 426 61.7 28

109 78.0 0.000 c 359 56.8 19 39 97.4 0.000f 421 58.2 27

Notes: *Missing values for each variable were excluded.

a. X2 = 17.60 df = 1; b.X2 = 30.91 df = 1; c. X2 = 14.96 df = 1; d. X2 = 10.628 df = 1; e. X 2= 15.17 df = 1; f. X2 = 21.59 df = 1.

Table 3.5 shows the results of logistical regression of multivariable analysis between

CVD and diabetes conditions, and the use of preventive health services, adjusted by

age and sex among the total sample (N = 519). As can be seen, respondents who had

CVD and diabetes conditions were 3.6 times more likely than those who did not have

the conditions to go for BP and BC check-ups and 2.6 times more likely to go for a

BG check-up. The two former relationships were statistically significant.

Considering this evidence, and to achieve a ‘healthy population’ for measuring

primary preventive health service utilisation, the final sample for the data analysis

excluded the people who had a CVD or diabetes condition.

Table 3.5 The logistic regression for multivariable analysis between BP, BC and BG Check-up and CVD and diabetes conditions adjusted for age and sex

Exploratory variables BP check-up OR 95%CI

BC check-up OR 95%CI

BG check-up OR 95%CI

CVD and diabetes problems No Yes

1.000 3.586 1.860–6.914

1.000 3.641 2.016–6.578

1.000 2.556 0.595–1.754

Age

1.034 1.013–1.057

1.070 1.047–1.092

1.048 1.029–1.077

Sex Male Female

1.000 1.688 1.096–2.600

1.000 0.718 0.471–1.095

1.000 1.051 0.752–1.860

113

3.7.2.2 CVD and diabetes problems and SEP

Table 3.6 also presents the Chi-square results for socioeconomic position among

respondents who had CVD and diabetes conditions in the original sample of 519

respondents. The table shows that respondents who were in both low-education and

low-income groups were more likely than their counterparts to have CVD and

diabetes conditions. The relationships among the four relationships were statistically

significant (P < 0.05).

Table 3.6 Chi-square test for socioeconomic profile among respondents who had CVD and diabetes conditions in a total sample of 519 respondents

Chi-square test CVD

N (%) Case (%) X2 DF P-value

Diabetes

N (%) Case (%) X2 DF P-value

Total sample Education Bachelor degree and higher Diploma and vocational Non post-school

qualification Missing

486* 112 (23.0) 173 15.0 10.25 2 0.006 148 25.7 165 29.1 (33)

477* 39 (8.2) 173 2.3 12.49 2 0.002 144 11.1 160 11.9 (42)

Total sample Family income Aus $52 000 or more $31 200–51 999 < $31 199 Don’t wish to answer Missing

491* 114 (23.2) 256 18.0 18.36 3 0.000 72 23.6 68 42.6 95 23.2 (28)

481* 38 (7.9) 256 5.1 11.55 3 0.009 68 4.4 66 13.6 93 14.0 (38)


These relationships were further tested by logistical modelling adjusted for age and

sex among the total sample (n = 519) (see Table 3.7). Respondents from low-

education backgrounds and low-income families were twice as likely to have CVD

and diabetes conditions. These relationships were statistically significant (for

education 95 per cent CI: 1.106–3.393; for income 95 per cent CI: 2.310 1.260–

4.235). It was concluded that the more disadvantaged groups were more likely to

have CVD and diabetes conditions. This result implied that a number of respondents

who were in low socioeconomic positions were excluded from the final study sample.

114

This caused an even more disproportionably under-representation of low-SEP groups

in the final data analysis. This is discussed further in Chapter 6.

Table 3.7 The logistic regression for multivariable analysis between CVD and diabetes problems and SEP variables adjusted for age and sex

Exploratory variables CVD and diabetes (total sample = 519) OR 95% CI OR 95%CI

Education Bachelor degree and higher Diploma and vocational Non post-school qualification

1.000 2.107 1.207–3.680 1.937 1.106–3.393

–

Income Aus$52 000 or more $31 200–51 999 <$31 199 Don’t wish to answer

–

1.000 1.121 0.585–2.147 2.310 1.260–4.235 1.514 0.860–2.666

Age 1.066 1.043–1.089

1.071 1.049–1.095

Sex Male Female

1.000 0.715 0.460–1.111

1.000 0.681 0.440–1.055

In summary, the total usable sample from the self-administered questionnaire survey

was 519. The study sample size (n = 381) for the final data analysis excluded those

respondents who had a CVD or diabetes condition (n = 138). The rationale for this

exclusion was that the purpose of the study was to investigate the utilisation of

preventive health services by different SEP individuals under absence of illness; that

is, for the purpose of primary prevention. Excluding the people with CVD and

diabetes conditions ensured that the sample was a substantially ‘healthy’ population.

In addition, the patterns of the check-ups between these two populations were

different from a preliminary bivariate data analysis. Respondents who had CVD or

diabetes conditions might have been required by their GPs to have regular check-ups

and they were more like to have these check-ups compared to the ‘healthy’

population, as discussed previously. Furthermore, respondents who had CVD or

diabetes conditions were more socioeconomically disadvantaged, as discussed

previously. In order to have a clear picture of utilisation of preventive health services

by a general population, the study sample only employed 381 respondents without

those chronic conditions.

115

3.8 Study measurement 3.8.1 Socioeconomic position Socioeconomic position (SEP) refers to ‘the social and economic factors that

influence what position(s) individuals and groups hold within the structure of society,

for example, what social and economic factors are the best indicators of location in

the social structure that may have influence on health’ (Lynch & Kaplan, 2000).

Three levels of SEP measurement have been commonly utilised in health research:

the individual level, the household level and the area-level (Dutton et al., 2005;

Krieger, Williams, & Moss, 1997; Lynch & Kaplan, 2000; Turrell et al., 1999).

Individual-level SEP measures are typically based on personal characteristics such as

personal education and income, from data that is usually easily obtained. In contrast,

the household level of SEP-measurement can provide combined information from all

members of a household, for example, ‘family income’. Area-level SEP indicators,

often called ‘neighbourhood level’ measures, can provide information about

‘contextual conditions’, for example, the level of neighbourhood educational

attainment.

Socioeconomic position was measured on the basis of the respondents’ education

levels and their family income. These two measures were chosen on the basis of

consideration of a multi-level of measures of SEP in case a misleading picture of

socioeconomic differences in the utilisation of preventive health services occurred

because it was only based on one measure of SEP. These multilevel approaches to

data analysis can strengthen our understanding of the links between SEP and health

services use. Education and family income measures are more easily interpreted and

used for data analysis and are also common measures for all other national data,

which is easy to obtain for the study representativeness comparison. However, family

income also has problems of missing data. This is further discussed in Chapter 6.

3.8.1.1 Education

Education is an individual level of SEP indicators. Previous studies have suggested

that education is the most stable indicator of SEP as it captures processes that occur

116

early in life and that tend not to change over time (Galobardes, Morabia, & Bernstein,

2001).

Figure 3.3 shows the education distribution for the respondents to the mailed survey.

About 67 per cent (n = 244) of respondents had a qualification and 30 per cent (n =

110) of respondents had no post-school qualification. The majority of respondents

had obtained a bachelor’s degree and an equal number of respondents had an

associate diploma and postgraduate diploma.

2.4

30.4

3.9

1.8

5

7.1

25.7

3.7

6.8

3.1

10

0 5 10 15 20 25 30 35

Missing

Non post-school qualification

Others

Doctorate

Masters degree

Post graduate diploma

Bachelor degree

Undergraduate diploma

Associate diploma

Apprenticeship

Trade or business certificate

Qua

lific

atio

ns

Percentage

Figure 3.3 Qualification distributions of the respondents from the mailed survey

These educational categories were then combined into three categories for data

analysis. The ‘other’ group was reclassified and assigned to a qualification group:

bachelor degree or higher, diploma and vocational and non post-school qualification,

according to a classification of the BFS study (Turrell, 2000).

3.8.1.2 Family income

It should be noted, however, that family income has its disadvantages, such as

missing data. In this study, nearly 20 per cent of the respondents were not willing to

answer this question or just missed it out, so missing data was excluded from the data

117

analysis. Thus, the sample size for each of the income groups was diminished and the

power of the data was limited.

The income distribution for the respondents is presented in Figure 3.4. As can be

seen, the majority of respondents’ family incomes were $52 000 and above, and a

small number of respondents’ family incomes were below $31 200. Just above 18 per

cent of the respondents did not wish to answer this question.

0

20

40

60

80

100

120

140

$832

0-103

99

$104

00-15

559

$156

00-20

799

$208

00-25

999

$260

00-31

199

$312

00-36

399

$364

00-41

599

$416

00-51

999

$520

00-77

999

$780

00 or

more

Don't w

ish to

answ

er

Missing

Family Income Range

Freq

uenc

y

Figure 3.4 Household income distribution of the respondents from the mailed survey

The original measure was then collapsed into three income groups according to their

frequency distribution and a classification from BFS (Turrell, 2000) (Figure 3.5).

The three income groups were classified as A$52 000 or more, accounting for 54 per

cent (n = 206); $36 400–51 999, accounting for 15.7 per cent (n = 60); and less than

$36 399 for 10.5 per cent (n = 40). It should be noted that respondents who ‘don’t

wish to answer’ the income question (almost 20 per cent) were excluded from this

data analysis. Nevertheless, it is important that the choice, I don’t wish to answer this

question, was given in the income question, because questions about people’s

incomes can be sensitive.

118

54.1

15.7

10.5

18.4

1.30

10

20

30

40

50

60

Aus $52,000 ormore

$36,400-51,999 $36,399 or less Don't wish toanswer

Missing

Family income groups

Perc

enta

ge

Figure 3.5 Family income groups for the data analysis

3.8.2 Demographic variables 3.8.2.1 Age

Age was a continuous interval measure ranging in yearly increments from 25 to 64.

Figure 3.6 shows the frequency of age in the total study sample (n = 381). For

bivariate analysis, age was categorised into four groups: 25–34, 35–44, 45–54 and

55–64, but for the multivariable analysis, where age was treated as a potential

confounder, age was used as a continuous variable. From a preliminary analysis of

this present study, it was found that age was significantly related to the use of

preventive health services. Therefore, age was adjusted for all modelling analysis.

119

604020

Age

40

30

20

10

0

Freq

uenc

y

3

998

17

5

19

13

18

6

26

12

19

24

9

2122

15

32

27

7

25

15

108

2

Figure 3.6 Age distribution of total study sample (n = 381)

3.8.2.2 Gender

There were 155 (40.7 per cent) males and 226 (59.3 per cent) females in the study

sample. The data were analysed separately by gender, as is discussed in Section

3.9.1.1. A comparison of gender profile between the study sample and 2001 census

data is made in Chapter 4.

3.8.3 The potential factors of socioeconomic differences in utilisation of

preventive health services Initially, six relevant scales were adapted and/or further developed from existing

questionnaires and formative studies. The six scales were: concerns about

availability and accessibility to health care, perceived interpersonal care, attitude

towards health care, value of general practitioners, value of good health and social

support. The reliability testing for each of the adapted or developed scales was based

on data collected from the self-administered mailed survey of a general population.

The following steps were used to generate a total score for each scale:

120

1. Each item of the scale was assigned a number from high to low, based on

each scale format. For example, for the scale response format very likely,

likely, unlikely and very likely, numbering was 4, 3, 2 and 1 for each item. If

the scale response format was strongly agree, moderately agree, agree,

disagree, moderately disagree and strongly disagree, then the numbers were

assigned as 6, 5, 4, 3, 2 and 1. The number was entered into the database

when the data entry was conducted.

2. Negatively worded items were reversed before a total score was calculated

for the scales in order to check reliability.

3. Treating each of the scales as an interval scale, the scores from each of the

items were added up to give an overall score for each scale. Subsequently,

the overall score for each scale was generated from the items that made up

each scale using ‘computing’ techniques in SPSS12.01(SPSS, 2004).

4. Finally, each scale was checked for normality to perform further analysis.

The psychometric properties of those adapted scales were evaluated before the data

were analysed. Face and content validity had been evaluated previously during the

development of the scales and the questionnaire. Face validity was assessed in terms

of the presentation and relevance of the questionnaire by the research team and the

expert review panels. Content validity was also assessed, to check that the

content/domain was relevant to the original purpose of this research project.

Principal Components Analyses (PCA) with a varimax rotation were conducted so as

to extract the maximum amount of variance of the loadings within components

across each of the five variables (Tabachnick & Fidell, 2001). Because of the

different response formats, separate PCAs were conducted for each of the six

variables. Items were excluded if the loading of coefficients was less than 0.5, to

allow for a moderate level (20–30 per cent) of overlapping variance (Tabachnick &

Fidell, 2001). Each PCA was then repeated until a stable solution was achieved. To

identify and label each component, further investigation of the highest loading items

was required.

Internal consistency was used to assess reliability. Reliability coefficients were

calculated for each of these scales prior to any subsequent analysis. This is often

121

called ‘Cronbach’s Alpha’ on the items in the scale. It is the most commonly used

method of assessing reliability. Mean scores and standard deviations for each

factorial scale were used to assess the variation in response within the sample

population. Ideally, the Cronbach’s Alpha coefficient of a scale should be above 0.7

(Nunnally & Bernstein, 1994). A score slightly lower than 0.7 indicates a marginally

unacceptable level of internal reliability and that scale should be used with caution.

A score that is markedly lower than 0.7 indicates an unacceptable level of internal

reliability.

Using the exclusion criteria of factor loadings of Principal Component Analysis less

than 0.5, three components were extracted from the scale (concerns about availability

and accessibility to health care). In addition, two components were extracted from

the scale (social support). Finally, nine scales and two dichotomous variables were

derived from the 37 items, following Principal Component Analysis and reliability

testing. The final component loading for each scale was presented in Table 3.8 and

the results of the internal reliability test for each scale, including mean scores,

standard deviations and Cronbach’s Alpha reliability coefficients were presented in

Table 3.9.

122

Table 3.8 Results of final Principal Component Analysis for the scales

Notes: 1. The score for this item has been reversely converted from original score, which gave a negative coefficient. 2. Final retained loadings 3. All missing values were excluded from the analysis

Varimax Rotation Component 1 Component 2 Component 3

Concerns about availability and accessibility to health care3 i The amount of time I get to spend with a doctor d The amount of time it takes to get an appointment c Being able to see my preferred doctor every time h The amount of time I have to wait in the waiting room a Being able to find a doctor who bulk bills e The cost of seeing a doctor b Having a number of doctors to choose from g The amount of time to travel to see a doctor f Transport to see a doctor or medical centre % of variance explained

.722 .721 .694 .555 -.058 .164 .388 .180 .157 38.72%

.045 .287 –.033 .151 .897 .863 .519 .148 .200 15.37%

.303 .121 –.032 .348 .105 .202 .134 .909 .871 11.92%

Professional attitudes towards patients—inter-personal care3 a The amount of time the doctor spends with you? b The doctor’s patience with your questions or worries? c The doctor’s caring and concern for you? % of variance explained

.954 .953 .923 88.973%

Value of GPs3 a The care I have received from doctors in the last few years has been good b Doctors belong to a very high status profession c A person understands their own health better than most doctors do d1 It is very important to choose your doctor carefully in order to get good medical care e Many doctors are more interested in their incomes than in making sure everyone receives adequate medical care % of variance explained

.839 .771 .010 -.259 .545 32.78%

–.030 –.157 .719 .734 .563 28.44%

Attitudes towards health care3 a If you wait long enough, you can get over almost any disease without seeing a doctor b I avoid seeing a doctor whenever possible c I only go to a doctor if there is no other option d1 Even if a person is feeling okay, they should get a general examination or check up every year or so % of variance explained

.679 (.694) 2 .884 (.884) 2 .855 (.859) 2 .229 50.67% (66.67%)

Value of good health 3 a Get more rest and sleep b Get more exercise d Cut down on the amount of work you do c Stop eating some favourite foods e Spend more time doing things with family and friends % of variance explained

.767 .725 .723 .739 .731 54.34%

Social support 3 a People don’t visit me as often b Need help but can’t get it d Don’t have anyone to talk e No one to lean on g Feel very lonely c 1 Have a lot of friends f 1 Someone can cheer me up h1 Enjoy the time spending with people i1 Talking someone feel better j1 Need help can find someone % of variance explained

.743 (.704)2 .797 (.793)2 .773 (.857)2 .756 (.845)2 .828 (.828)2 .456 .371 .068 -.015 .336 35.06% (65.17%)

–.015 .067 .314 .345 .116 .279 (.503)2 .655 (.763)2 .762 (.713)2 .852 (.784)2 .776 (.852)2 26.53% (53.68%)

123

Table 3.9 Mean scores, standard deviations and Cronbach’s Alpha reliability coefficients for the factorially derived scales

Scales Cronbach’s

Alpha Number of Items Number of Cases

Concerns about availability & accessibility to health care (1) Availability (2) Accessibility (3) Transport and travel time to access health care

0.6891 0.723 0.856

4 3 2

3713 3483 3443

Perceived interpersonal care 0.938 3 381

Value of GPs 0.6171 0.4122

2 (a + b) 3(c, d*,e)

380 380

Attitude towards healthcare

0.746 0.636

3 (excluding item d*) 4

381 381

Value of good health

0.788 5 381

Social support Negative aspect Positive aspect

0.865 0.773

5 5

381 381

Notes: 1. Score slightly lower than 0.7 indicating marginally unacceptable level of internal reliability and that scale should be used with caution.

2. Score markedly lower than 0.7 indicating unacceptable level of internal reliability. 3. ‘Not applicable’ was excluded from the scales.

As a requirement of multivariable analysis, each scale must be distributed normally

when it is used as a continuous variable. The normality of each scale, therefore, was

checked based on a normality checking criteria developed by Battistutta (2003). The

criteria are derived following descriptive statistics for data: mean, median, sd,

minimum, maximum, skewness, kurtosis and histogram. They include the following

aspects:

• Is the mean within ± 10% of the median value?

• Do mean ± 3 sd approximate the minimum and maximum values in the

dataset?

• Is the skewness coefficient within ± 3?

• Is the kurtosis coefficient within ± 3?

• Does the histogram look bell-shaped?

124

The histograms for each scale were created and statistical descriptions were produced

and analysed (not shown). These scales are discussed individually, based on

Tables 3.8 and 3.9.

After the validity and reliability of the scales were tested, nine scales were

determined from a final extraction and exclusion of items of the six original scales.

3.8.3.1 Concerns about availability of health care

This scale was extracted from the original scale of Concerns about availability and

accessibility to health care and included the following four items. Reponses to these

four items were summed to range from 4–20 to provide a single measure, with a

higher score reflecting greater concerns:

1. The amount of time I get to spend with a doctor. 2. The amount of time it takes to get an appointment to see a doctor. 3. Being able to see my preferred doctor every time. 4. The amount of time I have to wait in the waiting room to see a doctor.

After factor loadings of Principal Component Analysis, this scale was extracted from

the original scale of Concerns about availability and accessibility to health care,

which consisted of four items and explained 38.72 per cent of the variance. The scale

has an internal consistency with Cronbach Alpha coefficients reported of 0.69

(n = 371). This scale can be used, but with caution. The scale was assessed separately

for the male and female groups. Based on the normality checking criteria, it

suggested that the data were normally distributed for the male group (maximum or

minimum = Mean 16.16 ± 3 × SD 2.78) and the female group (Mean 16.84 ± 3

× SD 2.69).

3.8.3.2 Concerns about accessibility to health care

This is the second component from the scale of Concerns about availability and

accessibility to health care and included the following three items. Reponses to these

three items were summed to range from 3–15 to give a single measure, with a higher

score reflecting greater concerns:

125

1. Being able to find a doctor who bulk bills. 2. The cost of seeing a doctor. 3. Having a number of doctors to choose from in the one medical

practice/centre.

This scale was extracted from the scale of Concerns about availability and

accessibility to health care after factor loadings of Principal Component Analysis. It

included three items that explained 15.37 per cent of the variance. The scale has

good internal consistency with Cronbach Alpha coefficients reported of 0.72

(n = 348). The data were also normally distributed for the male (Mean 10.41 ± 3 ×

SD 2.96) and female groups (Mean 10.96 ±3 × SD 2.77) based on the normality

checking criteria.

3.8.3.3 Concerns about transport and travel time to health care

This scale included the following two items, which were derived from the scale of

Concerns about availability and accessibility to health care. Reponses to these two

items were summed to range from 2–10 to give a single measure, with a higher score

reflecting greater concerns:

1. The amount of time to travel to see a doctor. 2. Transport to see a doctor or medical centre.

This scale was extracted from the scale of Concerns about availability and

accessibility to health care after factor loadings of Principal Component Analysis,

which included two items that explained 11.92 per cent of the variance. The scale has

a good internal consistency with Cronbach’s Alpha coefficients reported of 0.856

(n = 344). The data were normally distributed for both the male (Mean 5.78 ± 3 × SD

2.25) and female groups (Mean 6.38 ± 3 × SD 2.28) according to the normality

checking criteria.

126

3.8.3.4 Perceived interpersonal care

The perceived interpersonal care scale included the following three items, where the

original items for the perceived interpersonal care scale still remained. Reponses to

these three items were summed to range from 3–15 to give a single measure, with a

higher score reflecting better perceived care:

1. The amount of time the doctor spends with you. 2. The doctor’s patience with your questions or worries. 3. The doctor’s caring and concern for you.

This scale remained the initial three items that explained 88.97 per cent of the

variance after factor loadings of Principal Component Analysis. The scale has a good

internal consistency, with Cronbach Alpha coefficients reported of 0.94. The data

were also normally distributed for males (Mean 11.09 ± 3 × SD 2.84) and females

(Mean 11.85 ± 3 × SD 2.65) according to the normality checking criteria.

3.8.3.5 Value of general practitioners This scale includes the following two items:

1. The care I have received from doctors in the last few years has been good. 2. Doctors belong to a very high status profession.

Reponses to these two items were summed to range from 2–12 to give a single

measure, with a higher score representing greater value.

Using the factor loadings of Principal Component Analysis, the scale produced two

factors that explained 32.78 per cent and 28.44 per cent of the variance, respectively.

The first scale included two items and the second scale included three items.

However, the two scales did not have good internal consistencies, with Cronbach

Alpha coefficients of 0.62 and 0.41, respectively. Therefore, according to Nunnally

& Bernstein (Nunnally & Bernstein, 1994), the scale derived from the first two items

needed to be used with caution and the scale derived from the three items was

unacceptable. The data for the former scale were normally distributed for males

(Mean 10.03 ± 3 × SD 1.68) and females (Mean 9.81 ± 3 × SD 1.79) according to the

normality checking criteria.

127

3.8.3.6 Attitudes towards health care

The final scale included three items:

1. If you wait long enough, you can get over almost any disease without seeing a doctor.

2. I avoid seeing a doctor whenever possible. 3. I only go to a doctor if there is no other option.

Reponses to these three items were summed to range from 3–18 to give a single

measure, with a higher score reflecting positive attitudes.

Using the exclusion criteria of factor loadings less than 0.5, one item for the scale

was dropped from the analyses. The final scale included three items that explained

50.67 per cent of the variance. The scale had a good internal consistency, with a

Cronbach Alpha coefficient of 0.75 (See Table 3.5). The data were also normally

distributed for males (Mean 11.85 ± 3 × SD 3.40) and females (Mean 11.59 ± 3 × SD

3.14) according to the normality checking criteria.

3.8.3.7 Value of good health

The value scale of good health included the following five items, where the original

items for the scale still remained. Reponses to these five items were summed to range

from 5–20 to give a single measure, with a higher score reflecting greater value:

1. Get more rest and sleep. 2. Get more exercise. 3. Cut down on the amount of work you do. 4. Stop eating some favourite foods. 5. Spend more time doing things with family and friends.

This scale remained the initial five items after the factor loadings of Principal

Component Analysis that explained 54.34 per cent of the variance. The scale has a

good internal consistency, with Cronbach Alpha coefficients reported of 0.79 (see

Table 3.6). The data were also normally distributed for males (Mean 14.94 ± 3 × SD

2.30) and females (Mean 15.25 ± 3 × SD 2.64) according to the normality checking

criteria.

128

3.8.3.8 Social support

One negative and one positive aspect of the social support scale was derived from the

original social support scale. The negative aspect of social support scale included the

following five items. Reponses to these five items were summed to range from 5–30

to give a single measure, with a least score reflecting a lower level of social support:

1. People don’t visit me as often as I would like. 2. I often need help from other people but can’t get it. 3. I don’t have anyone that I can really talk to. 4. I have no one to lean on in times of trouble. 5. I can feel very lonely.

One negative and one positive social support scale was derived from the original

social support scale. The negative social support scale was extracted from the initial

social support scale, including five items that explained 35.06 per cent of the

variance.

The scale has a good internal consistency, with Cronbach Alpha coefficients reported

of 0.87(n = 381). The data were also normally distributed for males (Mean 21.32 ± 3

× SD 4.63) and females (Mean 21.46 ± 3 × SD 5.05).

The positive aspect of the social support scale included the following five items.

Reponses to these five items were summed to range from 5–30 to give a single

measure, with a higher score reflecting a higher level of social support:

1. I seem to have a lot of friends. 2. There is someone who can always cheer me up. 3. I enjoy the time spend with the people who are important to me. 4. When something is on my mind, just talking with the people I know makes me

feel better. 5. When I need someone to help me, I can usually find someone.

The positive social support scale was extracted from the initial social support scale

down to five items that explained 26.53 per cent of the variance. The final scale had

a good internal consistency, with Cronbach Alpha coefficients reported of

0.77(n = 381). The data were also normally distributed for males (Mean 22.45 ± 3 ×

SD 4.09) and females (Mean 24.01 ± 3 × SD 3.73) according to the normality

checking criteria.

129

3.8.4 Other intervening variables: a regular source of care A ‘regular source of care’ refers to individuals regularly visiting one general practice

or medical centre, or seeing a regular doctor when they need. Two variables were

derived from the survey, including a regular place of care and a regular care provider

as potential mediators. These two variables were dichotomised into (i) always and

most of the time; (ii) sometimes and rarely or never. Respondents who responded

always and most of the time to the item about visiting the same general practice or

medical centre were recognised as having a regular place of care; otherwise they did

not have a regular place of care; that is, they sometimes and rarely or never visited

the same GP or medical centre. This principle was also applied in explaining a

‘regular care provider’, which recognised respondents who always and most of the

time saw their GPs or the same doctor as a regular care provider; adversely, those

who sometimes and rarely or never saw the same care provider were seen as not

having a regular care provider.

3.8.5 Outcome variables

As discussed in Chapter 2, identification and screening risk factors of CVD and

diabetes are important for a reduction of morbidity and mortality of these two

diseases. This study measured what percentage of people had their blood pressure,

blood cholesterol and blood glucose checked by their GPs in the preceding two years

and five years of the point of time of data collection. These outcome measures were

defined as preventive health checkups. The preventive health checkups were

involved two possible ways in this study: (1) the people presented for a health

checkup of their own volition and (2) GPs conducted opportunistic preventive health

checkups. Importantly, it should be noted that this study cannot identify which ways

the people had their preventive health checkups at general practice service. These

three dependent variables were defined according to the Guidelines for Preventive

Activities in General Practice developed by the Royal Australian College of General

Practitioners (2002).

3.8.5.1 Blood pressure check-up

130

This was used as a dichotomous variable. The Guidelines suggest that blood pressure

levels should be measured in all adults from age 18 years at least every two years at

non risk. Therefore, the Yes group included people who had their blood pressure

checked according to the Guidelines within the previous two years, and included two

categories: in the last 12 months and in the last 2 years. The rest of the categories

became the No group, who did not comply with the guidelines.

3.8.5.2 Blood cholesterol check-up

Blood cholesterol was also a dichotomous variable. In the same way as for blood

pressure screening, the Guidelines suggest that screening healthy people without risk

factors is recommended every five years. In this case, the Yes group included the

respondents who went to their GP for a blood cholesterol screening in the last 12

months, in the last 2 years, and 2–5 years ago. The No group included those

respondents who went to see their GPs for more than 5 years ago and never.

3.8.5.3 Blood glucose check-up

Blood glucose was examined as a dichotomous variable. For blood glucose screening,

the Guidelines suggest that all people should be screened every three years.

Therefore, the Yes group included those respondents who had a blood glucose check

in the last 12 months, in the last 2 years, 2–5 years ago, and the No group included

those who had a blood glucose check more than 5 years ago and never. Table 3.10

presents the classification for the three dependent variables.

Table 3.10 Classification category for three dependent variables

‘Yes’ — checked ‘No’ — not checked Blood pressure check-up ‘in the last 12 months’ + ‘in the

last 2 years’ ‘2–5 years ago’ + ‘more than 5 years ago’ + ‘never’

Blood cholesterol check-up ‘in the last 12 months’ + ‘in the last 2 years’ + ‘2–5 years ago’

‘more than 5 years ago’+ ‘never’

Blood glucose check-up ‘in the last 12 months’ + ‘in the last 2 years’ + ‘2–5 years ago’

‘more than 5 years ago’+ ‘never’

131

3.9 Analytical approach and statistical analysis 3.9.1 Gender separation Data were stratified by gender, as the literature review had already demonstrated that

gender was a factor related to differential use of preventive health services. In

addition, a preliminary analysis of this present study showed that the patterns of the

use of preventive health services by males and females were different. In order to

reflect a correct pattern of utilisation of preventive health services by working-age

adults in Australia, data were then analysed separately by gender.

3.9.2 Analytical hypothesis Education and use of preventive health services The relationship between education and use of preventive health services is

presumed to reflect a person’s ability to access and interpret health-related

information. Someone with a high level of education may have broader knowledge of

preventive care and health, and may be more likely to take action to prevent diseases

or improve their health status. Thus, socioeconomic position, on the basis of

respondents’ education level, is appropriate in this study. The following hypotheses

for this study were made with respect to education and modifiable factors, and

utilisation of preventive health services:

1. Participants with lower levels of education will exhibit a utilisation pattern

that is least consistent with preventive guideline recommendations;

2. Participants with lower levels of education will exhibit higher levels of

concern about the availability of and accessibility to health care, and

transport and travel time; lower levels of perceived interpersonal care, the

value of general practitioners, the value of good health and social support;

will have more negative attitudes towards health care, and are less likely to

have a regular source of care; and

3. Education differences in perceptions of the concerns, values, attitudes,

social support and having a regular source of care will make a significant

contribution to education variations in the utilisation of preventive health

services.

132

Family income and use of preventive health services

Family income measures a household level of SEP indicators and is more likely to

have the greatest direct relevance to the use of preventive health services, because it

affects availability of economic resources, which then affects ability to access health

care services, and affordability of higher levels of use of preventive health services

(Dutton et al., 2005). Therefore, for this study, the following hypotheses were made

with respect to family income and modifiable factors, and utilisation of preventive

health services:

1. Participants from a low-income family will exhibit a utilisation pattern that is

least consistent with preventive guideline recommendations;

2. Participants from a low-income family will exhibit higher levels of concern

about availability of and accessibility to health care and transport and travel

time; lower levels of perceived interpersonal care, the value of general

practitioners, the value of good health and social support; will have more

negative attitudes towards health care, and are less likely to have a regular

source of care; and

3. Differences in perceptions of the concerns, values, attitudes, social support

and having regular source of care between low- and high-income families

will partly account for variations in their utilisation of preventive health

services.

3.9.3 Data analysis 3.9.3.1 Statistical analysis SPSS 12.01-14.0 was used for all statistical analyses. The data analysis included

survey response rate, representativeness of the sample and socio-demographic

characteristics of respondents, using statistical description techniques, and bivariate

analyses and multivariable analysis.

Bivariate analyses (Chi-square tests) were performed between SEP variables, age

groups and each outcome variable, before the multivariable analysis was performed.

P-values were used to estimate the relationships. Multivariable analysis, including

the Logistic Regression Model (LRM) and General Linear Model (GLM), was used

to examine the relationship between SEP and the use of preventive health services

133

since the outcome variables were dichotomous, and the relationship between SEP

and those intervening variables were continuous outcomes. All intervening variables

were used as intermediaries when adjusting in the modelling. All assumptions were

tested before the modelling analysis was performed, including linearity between the

dependent variable and intervening variables (for LRM), or the covariate (age) for all

SEP groups, and equality of variance of the dependent variables across the SEP

groups (for GLM).

3.9.3.2 Analytical steps In order to examine whether an intervening variable mediates the relationship

between SEP (education or family income) and the use of preventive health services,

an ‘elaboration’ data analysis technique was applied (Turrell, 1997). This approach

states that a significant association between an independent and a dependent variable

(in this case SEP and use of preventive health services) is partially explained if the

introduction of a third variable changes this association. It is important that this third

variable can only explain an association if it is significantly related to both the

independent and dependent variables. Therefore, this involved four steps to examine

the multivariable relationships (see Figure 3.7):

STEP 1

STEP2 STEP 3

STEP 4

Figure 3.7 Analytic approaches for the multivariable analysis

Preventive health service utilisation - Blood pressure check-up - Blood cholesterol check-up - Blood glucose check-up

SEP - Income - Education

Intervening variables - Nine scales - Two dichotomous variables

134

STEP 1: In order to assess whether there were relationships between SEP and use of

preventive health services in relation to CVD and diabetes, LRM was applied,

adjusting for age. These results are reported in Chapter 4.

STEP 2: To investigate why different SEP groups differ in their use of preventive

health services, it is important to look at the relationship between SEP and the

intervening variables. Thus, GLM was conducted to look at associations between

SEP and the scales, adjusting for age. To assess the difference between

socioeconomic groups, pairwise comparisons were performed. In addition, the

Logistic Regression Model was employed to look at the associations between SEP

and the variables of a regular place of care and a regular care provider, as these

variables were categorical. These results are reported in Chapter 5.

STEP 3: A logistic regression model was also undertaken to examine the

relationship between those intervening variables, and the use of preventive health

and age was also adjusted in the model. Those nine scales are employed in the

models as categorical variables (including low, middle and high groups) to

understand the relationship better across the three different scores. These results are

reported in Chapter 5.

STEP 4: To assess whether the potential factors mediate the relationship between

SEP and the use of preventive health services, multivariable logistic regression

models were employed, adjusting for age, and then adjusting for each of the

intervening variables to identify the contribution from each of the intervening

variables, and finally, adjusting all intervening variables in the model, considering

the interrelationships among those variables. Only those intervening variables were

adjusted in this model that had relationships with both SEP and the use of preventive

health services. These results are reported in Chapter 5.

3.9.3.3 Decision-rules for interpretation of study findings Owing to the limited sample size in this study and a power issue due to stratifying of

the data by gender, the study made decision rules for interpretation of the study

findings as follows:

135

• Adjusted odds ratios (OR) with 95 per cent confidence intervals (CI) were

used to estimate the logistic modelling relationships. A relative difference of

10 per cent in the odds ratios between SEP groups in terms of the use of

preventive health services was deemed meaningful.

• P-values less than 0.05 and 0.10 were considered a statistical significance for

GLM modelling, given the limitation of the sample size for the male and

female groups.

• Any relative changes greater than 10 per cent in odds ratio after adding an

intervening variable and all intervening variables together in the final model

would suggest that the modifiable factor(s) accounted for some of the original

socioeconomic variability in the use of preventive health services.

3.10 Summary

This chapter has described the methods and procedures for conducting a population-

based survey. Using simple random selection, details of 800 individuals residing in

the Brisbane metropolitan area were extracted from the Australian Electoral Roll to

receive a mailed survey. This chapter has described the development of a new

questionnaire for measuring the social determinants of utilisation of preventive

health services in relation to CVD and diabetes. Following an extensive review of

existing relevant scales and questionnaires, a new questionnaire was developed

according to the following five steps:

1. identification of the salient and relevant domains for the questionnaire;

2. identification of existing adequate scales for the questionnaire, as well as of

those domains that required additional scales to be developed;

3. development of relevant questions and scales and construction of the

questionnaire;

4. expert review of the draft questionnaire; and

5. pre-testing and refining of the questionnaire.

136

The final Health Service Utilisation Questionnaire, including a total of 79 items, was

established for a self-administered mailed survey to examine purposely the

relationship between socioeconomic position and the use of preventive health

services in relation to CVD and diabetes. Six key domains were developed from the

research framework, and the relevant scales and questions relating to each domain

were then identified from six relevant questionnaires and studies. The relevant scales

and questions were reviewed and discussed in terms of their suitability for the study.

Consequently, six scales and one single-item question were selected according to the

selection criteria, the authors for the specific questionnaire and questions and scales

were contacted, and permissions were requested to use and adapt any of the

previously developed measures.

The chapter has described a mailed survey data collection process, which

implemented Dillman’s mailed survey strategies and stretched over a six-week

period, with up to four contacts carried out by mail with each participant. The

chapter has addressed the issues with the study sample of exclusion and gender

separation for data analysis and discussed the analytical approach and statistical

methods for data analysis.

In addition, this chapter has summarised the measurement of main variables for the

study. The study measurement was developed and the scales were validated and

tested for reliability. The six available scales were tested for validity and reliability.

Subsequently, nine relevant scales were determined based on the exclusion criteria of

factor loadings of Principal Component Analysis less than 0.5 and Cronbach’s Alpha

reliability coefficients greater than or close to 0.7. The measurement for the study

included SEP variables, 11 intervening variables and three outcome variables: blood

pressure, blood cholesterol and blood glucose check-up. SEP was measured by

individual education level and family income. The measurement will be used to

examine purposely the relationship between socioeconomic position and the use of

preventive health services in relation to CVD and diabetes. The results of this

examination are presented in the next chapter.

137

4 RESULTS: DESCRIPTION OF STUDY SAMPLE AND EXAMINING THE RELATIONSHIP BETWEEN SOCIOECONOMIC POSITIONS AND THE USE OF PREVENTIVE HEALTH SERVICES IN GENERAL PRACTICE

138

4.1 Introduction This chapter examines the first PhD research question, as outlined in Chapter 2: to

explore the relationship between SEP and the utilisation of preventive health services

in relation to CVD and diabetes in Australia. The chapter begins with a discussion of

the survey response rate from the self-administered questionnaire and then describes

the socio-demographic profile of the respondents. It also examines the

representativeness of the sample and compares each of the SEP groups using defined

socioeconomic indicators, including education and income, in terms of their reported

use of preventive health services. Finally, it undertakes a multivariable analysis to

examine the relationship between SEP and the use of preventive health services by

adjusting for age as a potential confounding variable.

4.2 Survey response rate Table 4.1 examines the survey response rate for each of the three mailings. There

were 800 questionnaires sent in the first mailing. Ten surveys were sent back as

‘RTS’ (return to sender), which indicated that the respondents no longer lived at the

identified address. On the basis of the first mailing, 408 useable questionnaires out of

790 respondents were returned, which represented a within mailing response rate of

51.6 per cent (408/790) and 78.6 per cent (408/519) of the final sample.

Following the mail-out of 377 replacement questionnaires three weeks after the

initial mailing, 19.8 per cent (n = 73) of these were returned. Eight surveys were

‘return-to-sender’. Subsequently, the second replacement questionnaires were sent

out to 279 respondents and only 14 per cent (n = 38) were returned, including seven

‘RTS’. It is noted that the majority of returns came through the initial mailing.

Notwithstanding this, the follow-ups were essential to increase the response rate and

minimise the bias often associated with non-response. After three mailings, a final

useable response rate of 65.6 per cent was achieved, which included a sample of 519

respondents.

139

Table 4.1 Response rates and useable returns for different mailings

Mailings No. of

questionnaires sent

No of RTSa

No. of useable returns

Within mailing response

rateb

Percent of final sample

(n = 519)

Overall response

rate for each mailing

First mailing First replacement Second replacement

800

377

276

10

8

7

408

73

38

51.6

19.8

14.1

78.6

14.1

7.3

51.6

9.2

4.8

Total 1453 25 519 – 100.0 65.6 Note: a: RTS: return-to-sender. b. Equals the number of useable returns divided by the number of questionnaire sent excluding number of RTS.

4.3 Socio-demographic profile and health status characteristics of respondents

This section focuses on three aspects: a profile of CVD and diabetes problems, socio-

demographic characteristics and self-reported preventive health services utilisation

by respondents to the mailed survey. 4.3.1 Socio-demographic profile

Table 4.2 summarises the socio-demographic profile of the respondents (N = 381) in

terms of their age, gender, education, household income and self-assessed health

status. There were a total of 381 respondents, including 155 males (40.7 per cent)

and 226 females (59.3 per cent). The female sample (n = 226) was 18 per cent larger

than the male sample (n = 155). The average age of the sample was 42. The majority

of the respondents were aged between aged 30–39 years, while the minority were

aged 60–64 years.

140

Table 4.2 Socio-demographic characteristics of the study sample (n = 381)

Study variables No of cases Percentage

Gender Male Female Age groups

25–29 30–34 35–39 40–44 45–49 50–54 55–59 60–64

Education Bachelor degree and higher Diploma Vocational Non post-school qualification Missing Household income Aus $52 000 or more $31 200–51 999 < $31 199 Don’t wish to answer Missing Self-assessed health status Excellent Very good Good Fair Poor Missing

155 40.7

226 59.3

46 12.1 65 17.1 61 16.0 54 14.2 54 14.2

40 10.5 38 10.0 23 6.0

151 39.6 40 10.5

65 17.1 116 30.4 9 2.4 206 54.1 60 15.7 40 10.5 70 18.4 5 1.3

76 19.9 163 42.8 111 29.1 22 5.8 8 2.1 1 0.3

Most of the respondents perceived themselves as being in very good health. Overall,

34.5 per cent of respondents reported that they had a higher education qualification,

including a bachelor degree and higher; however, almost the same proportion (33.3

per cent) of the respondents reported that they had no post-high school qualification.

A relative small proportion of respondents (10.0 per cent) reported that they had a

diploma qualification and 20 per cent reported that they had a vocational

qualification.

Over half of the respondents were in the highest income category (household income

above $52 000), while just over 10 per cent of the respondents were in the lowest

income category (household income below $31 199). The middle household-income

141

group accounted for about 16 per cent of the respondents. However, over 18 per cent

of the respondents did not wish to answer the income question and 1.3 per cent had

data missing.

4.3.2 Self-reported preventive health service utilisation

Table 4.3 presents the self-reported preventive health service utilisation among the

study sample. Of the total sample of 381 respondents, 66.4 per cent (n = 253) had

visited a GP to have their blood pressure checked in the previous two years and 95

per cent confidence interval is 0.60–0.72; 54 per cent (n = 205) of the respondents

had their blood cholesterol checked in the previous five years and 95 per cent

confidence interval is 0.47–0.61; and 52 per cent (n = 198) of the respondents had

their blood glucose checked in the previous five years and 95 per cent confidence

interval is 0.45–0.59.

Table 4.3 Self-reported preventive health service utilisation of the study sample

No of cases BP check-up

% (95% CI) No of cases BC check-up % (95% CI)

No of cases BG check-up % (95% CI)

Yes

No

Missing

253 66.4 (0.60–0.72)

110 28.9 (0.21–0.37)

18 4.7 (–0.05–0.14)

205 53.8 (0.47–0.61)

153 40.2 (0.32–0.48)

23 6.0 (–0.04–0.16)

198 52.0 (0.45–0.59)

157 41.2 (0.33–0.49)

26 6.8 (–0.03–0.16)

Total 381 100.0 381 100.0 381 100.0

Notes: a. Blood pressure checked in the last 2 years. b. Blood cholesterol checked in the last 5 years.

c. Blood glucose checked in the last 5 years.

4.4 Representativeness of sample

The following comparisons are made between the study sample (n = 381) and the

2001 census for the local government area (LGA) of Brisbane (n = 850 111) among

25–64-year olds and other national data sets in terms of age, gender, education and

health service utilisation.

142

4.4.1 Age profile

Table 4.4 compares the age distribution of the study sample with the 2001 census for

the LGA of Brisbane. Overall, the study sample differed in age from the population

by only 1.5 per cent for males and 1.1 per cent for females, which was an average

difference between the two samples. However, it can be seen from examining each

age group separately that some age groups are better represented than others. The

least represented age group in the study sample is people aged 50–54 years by 2.5

per cent on average, and the second least-represented age group is the youngest age

group (25–29-year olds) by 2.2 per cent on average. The male age group (30–34-year

olds) in the study sample is overrepresented by 5.5 per cent. However, overall, the

distribution of age group in the study sample is not very divergent from the

population sample.

Table 4.4 Comparison of the age profile of the study sample with the 2001 census data (LGA of Brisbane) among 25–64-year olds

Age groups Male

Study sample 2001 census % (n = 318) % (Brisbane)

Female Study sample 2001 census

% (n = 318) % (Brisbane)

25–29

30–34

35–39

40–44

45–49

50–54

55–59

60–64

12.3 14.4

20.0 14.5

14.8 14.1

13.5 13.9

13.5 13.2

10.3 12.8

9.7 9.8

5.8 7.3

11.9 14.2

15.0 14.7

16.8 14.5

14.6 14.4

14.6 13.3

10.6 12.6

10.2 9.3

6.2 7.0

Total (No.) 155 416 318 226 433 793

4.4.2 Gender profile

Table 4.5 compares the gender profile of the study sample with the 2001 census data

for the LGA of Brisbane. Interestingly, there is an 8.2 per cent difference for both

males and females in the study sample compared to the 2001 Brisbane census data.

However, males are underrepresented in the study sample and females are

143

overrepresented. Overall, both census data and study sample have more females than

males, in particular in the study sample.

Table 4.5 Comparison of the gender profile of the study sample with 2001 census data (LGA of Brisbane) among 25–64-year olds

Gender N Study sample (%) N 2001 census (Brisbane) (%) Male

155 40.7

415 704 48.9

Female

226 59.3

434 407 51.1

Total 318 100.0 850 111 100.0

4.4.3 Education profile

Table 4.6 compares the study sample and the 2001 census for the LGA of Brisbane

data in terms of levels of education in 25–64-year olds. There are two noticeable

differences between these two samples. Half of the respondents in the study sample

have a higher education level, while only one-quarter have the same education level

in the Brisbane 2001 census data. The respondents with a lower level of education

such as a certificate or no qualification are underrepresented in the study sample. The

most overrepresented education group is respondents with a bachelor’s degree, by

12.7 per cent; on the other hand, respondents who have no qualifications are

underrepresented by 18.9 per cent in the study sample when compared to the census

data. On the basis of this evidence, it is concluded that the respondents to the study

sample have a higher level of education than those in the census data. This result is

similar to other comparable Australian studies. For example, Turrell and Najman

(1995) conducted a study on people’s food behaviours using a mailed survey, and

selected their samples from the electoral roll in Brisbane. They claimed that the low-

education groups were disproportionately underrepresented on electoral rolls. In

addition, Burton et al.’s study (2003) on physical activity, which selected samples

from Brisbane electoral rolls and obtained a study sample through a mailed survey,

also reported that low level of education and low-income individuals were

underrepresented.

144

Table 4.6 Comparison of the individual education profile of the study sample with 2001 census data (LGA of Brisbane) among 25–64-year olds

Level of education Study sample Cumulative ( %) (%)

2001 census Cumulative (%) (%)

1. Doctoral degree 2. Master degree 3. Graduate diploma 4. Bachelor degree 5. Undergraduate diploma and Associate diploma 6. Certificate 7. No post-school qualification 1 8. Not stated 2

1.8 6.8 5.0 13.9 7.1 39.6 25.7 50.1 10.5 63.2 13.1 93.6 30.4 100.0 6.4

0.73 2.74 2.01 4.71 1.97 17.75 13.04 25.11 7.36 42.73 17.62 92.00 49.27 100.00 8.00

Total 100.0 100.00

Notes: 1.Includes persons who do not have a qualification and persons who have a qualification out of scope of the Australian Standard Classification of Education. 2. Includes 'Inadequately described.

At the same time, the distribution of education level of the study sample follows a

similar trend to the census sample. To illustrate this, Figure 4.1 depicts the education

distribution for both samples. For example, the majority of respondents have a

bachelor’s degree, and also many respondents have a certificate in both data sets

compared to the rest of the education groups. The two study samples show a similar

distribution. This is important evidence that the data from the mailed survey is

representative, in spite of the overrepresentation of higher levels of education.

0

10

20

30

40

50

60

Doc

tora

lD

egre

e

Mas

ter D

egre

e

Gra

duat

eD

iplo

ma

Bac

helo

rD

egre

e

Und

ergr

adua

teD

iplo

ma

and

Ass

ocia

teD

iplo

ma

Cer

tific

ate

Not

app

licab

le

Not

sta

ted

Perc

enta

ge

Study sample 2001 Brisbane Census

Figure 4.1 Trend of level of education between the study sample and the 2001 census

(LGA of Brisbane)

145

4.4.4 Household income profile

The study also intended to make a comparison for household income data between

the study sample and the 2001 Census data. The Census collected each individual’s

income information per household, while the present study collected family income.

There is no comparable family income data in the specific age group (25–64 years)

available. Nevertheless, from the distribution of the individual education levels, it is

assumed that the study sample also has a higher proportion of high-income

respondents compared to the 2001 Census for LGA of Brisbane. This is evident in

the later description of the socio-demographic characteristics of the study sample,

with 54 per cent of the respondents in the highest income category.

4.4.5 Health service utilisation

Table 4.7 compares the study sample with available national data and studies in

terms of health service utilisation. The percentage of people visiting their GP

(85.0 per cent) in the previous 12 months in the study sample is almost the same as

that reported as the annual average GP consultation rate (86.0 per cent) in Australia

from the Department of Health and Aged Care (2000). In addition, comparisons for

blood pressure and blood cholesterol screening are made. An Australian study

(Heywood et al., 1994) that examined prevalence of screening rates for risk factors

for CVD among people aged 18–75 years in Brisbane and Toowoomba suggested

that over 90 per cent of respondents self-reported having their blood pressure

checked in the previous two years and 51 per cent reported having their blood

cholesterol checked in the previous five years. Compared with these data, the

percentage of respondents having a blood pressure screening in the study sample is

underrepresented, but there is a similar rate of blood cholesterol screening. Given the

two studies did not have a similar population background, this comparison was not

entirely analogous. Nevertheless, this is the best available data for this comparison. It

is useful to help understand the general distribution of the use of preventive health

services. Furthermore, there are no available data on blood glucose screening for this

comparison.

146

Table 4.7 Comparison of respondents in the study sample and other data in terms of health service utilisation

Health service utilisation Study sample (%)

Department of Health & Aged Care (2000)

GP consultation in the last 12months

Blood pressure checked in the last 2 years

Blood cholesterol checked in the last 5 years

Blood glucose checked in the last 5 years

85.0

66.4

53.8

52.0

86.0%

90% (Heywood et al. 1994)1

51% (Heywood et al. 1994) 1

Not available

Notes: 1. The study was conducted in Queensland between August 1989 and April 1990, including two cities: Brisbane and Toowoomba among 18–75-year olds.

4.5 Bivariate associations between SEP, age and the use of preventive health services

To understand the relationship between SEP and the use of preventive health services,

bivariate associations between education, family income, age and the use of

preventive health services were explored before the multivariable associations were

examined. The relationship between age and the use of preventive health services

was examined to determine whether age is a potential confounder for the relationship

between SEP and the use of preventive health services. All data were analysed

separately by gender and discussed by each outcome variable individually.

4.5.1 Blood pressure check-up

Table 4.8 presents blood pressure check-up by education, household income and age

group among males and females. Just above 60 per cent (95 per cent

CI: 0.52–0.68) of all male respondents (n = 155) had a BP check-up from a GP in the

preceding two years, while above 70 per cent (95 per cent CI: 0.64–0.76) of all

female respondents (n = 226) had their BP checked. The bivariate relationships

comparing male and female groups are discussed below:

4.5.1.1 Education

As can be seen from Table 4.8, there was a very similar proportion of BP check-up

from high- (63.3 per cent), middle- (64.0 per cent) and low-education groups among

males (66.7 per cent). There were no statistical differences (P = 0.951). On the other

hand, for females, there were different proportions across the education groups in

147

terms of BP checked, with a 11.3 per cent difference between high- and low-

education groups, and 21.3 per cent between high- and middle-education groups.

However, there were no significant associations between having BP checked and

education (P = 0.195).


Of male income groups, the proportion of respondents from the middle-income

group who went for a BP check-up was different from the high-income group by

11.16 per cent. This bivariate association was not significant (P = 0.799). Of female

income groups, almost 80 per cent of the respondents in both the low- and the

middle-income groups had a BP check-up. There were 17.7 per cent of differences

for a BP check-up between the high-income group and the middle- or low-income

groups; these associations, however, were not statistically significant (P = 0.236)

(see Table 4.8).

4.5.1.3 Age

Among the male age groups, there was an increasing trend for BP check-up from the

younger age group to the older age group. Age was strongly related to the BP check-

up (P = 0.005). For the female age groups, the old-age group was more likely to go

for a BP check-up than the other groups. The relationship was statistically significant

(P = 0.068) (see Table 4.8).

148

Table 4.8 Chi-square test for bivariate associations between blood pressure checked by GP and education, household income and age in a male and female sample

Male BP checked by GP Chi-Square Test N BP Checked

(%)

X2 DF P-value

Total sample Education Bachelor degree and higher Diploma and vocational Non post-school qualification Missing* Family Income Aus $52 000 or more $31 200–51 999 < $31 199 Don’t wish to answer** Missing* Age groups 25–34 35–44 45–54 55–64

155

60 50 30

15 89 19

14 21

12

47 40 35

23

60.6

63.3 64.0 66.7

66.3 73.7 64.3

44.7 75.0 71.4 78.3

0.100

0.450

12.671

2

2

3

0.951

0.799

0.005

Female BP checked by GP Chi-Square Test N BP Check-up

(%)

X2 DF P-value

Total sample Education Bachelor degree and higher Diploma and vocational Non post-school qualification Missing* Household Income Aus $52 000 or more $31 200–51 999 < $31 199 Don’t wish to answer** Missing* Age groups 25–34 35–44 45–54 55–64

226

83 51 80 12

110 39 24 43 10

60 69 54 35

70.4

66.3 80.4 73.8

67.3 79.5 79.2

71.7 73.9 63.0 88.6

3.269

2.886

7.138

2

2

3

0.195

0.236

0.068

Notes: *Missing values for each variable were excluded. ** ‘Don’t wish to answer’ was excluded from the analysis.

149

4.5.2 Blood cholesterol check-up

Table 4.9 shows the bivariate relationships between blood cholesterol check-up and

SEP variables, and age groups among males and females. Fifty-nine per cent (95 per

cent CI: 0.51–0.67) of all male respondents (n = 155) had their BC checked by their

GP in the previous two years while over 50 per cent (95 per cent CI: 0.43–0.57) of all

female respondents (n = 226) had their BC checked. The bivariate relationships are

discussed below by comparing male and female groups.

4.5.2.1 Education

As Table 4.9 shows, among males, the higher education group were nearly 20 per

cent more likely to go to a GP for a BC check-up than the low-education group.

However, there were no statistical associations among these groups (P = 0.192). Of

female education groups, there was nearly the same proportion of BC check-up

across the education groups. There were no differences among these groups

(P = 0.655).


Among the male groups, the higher income groups were almost 10 per cent more

likely to have a BC check-up than the low-income group, but there was no statistical

significance (P = 0.634); while, among the three female income groups, there was

nearly a 20 per cent difference between the high-income group and the low-income

group in terms of BC check-up. However, no bivariate statistical relationship existed

(P = 0.714) (Table 4.9).

4.5.2.3 Age

Among both male and female age groups, there was an increasing trend for BC

check-up from the younger to the older age group. Age has a positive relationship

with BC check-up (P = 0.000) (Table 4.9).

150

Table 4.9 Chi-square test for bivariate associations between blood cholesterol checked by GP and education, household income, and age in a male and female sample

Male BC checked by GP Chi-Square Test N BC Checked

(%)

X2 DF P-value


155 61

49 32

13

91 17 15 21

10 46 41 37

23

58.7

65.6 63.3 46.9

62.6 52.9 53.3

34.8 73.2 70.3 82.6

3.301

0.894

21.834

2

2

3

0.192

0.639

0.000

Female BC checked by GP Chi-Square Test N BC Check-up

(%)

X2 DF P-value


226

79 51 79 17 107 37 23 42

17 53 69 53

36

50.4

50.6 58.8 53.2 54.2 62.2 43.5

30.2 53.6 64.2 75.0

0.845

2.006 20.693

2

2 3

0.655

0.367

0.000

Notes: *Missing values for each variable were excluded. ** ‘Don’t wish to answer’ was excluded from the analysis.

151

4.5.3 Blood glucose check-up

Table 4.10 shows the bivariate relationships between blood glucose check-ups and

SEP variables, and age groups among males and females. Forty-nine per cent (95 per

cent CI: 0.41–0.57) of all male respondents (n = 155) had their BG checked by their

GP in the preceding two years, while 54 per cent (95 per cent CI: 0.48–0.60) of all

female respondents (n = 226) had their BG checked. The bivariate relationships are

discussed below by comparing male and female groups.

4.5.3.1 Education

As shown in Table 4.10, among males, there was an 11.3 per cent difference between

the high-education group and the low-education group in terms of a BG check-up.

Among females, those in the high-education group were more likely to go for a

check-up than their low counterparts; while the middle group were slightly more

likely to go to a GP for a BG check-up than the low-education group. However, these

relationships were not statistically significant (male P = 0.817; female P = 0.579).


There was about a 10 per cent difference in terms of BG check-ups between the

middle-income group and the low-income group among males; and a 20 per cent

difference between the middle-income group and the low-income group among

females. Statistically, however, there were no bivariate relationships between income

and BG check-up among either males or females (male: P = 0.864; female: P = 0.179)

(Table 4.10).

4.5.3.3 Age

Among both the male and female age groups, there was an increasing trend for BG

check-up from the younger age group to the older age group. Age has a positive

relationship with BG check-up as well (male: P = 0.003; female: P = 0.032)

(Table 4.10).

152

Table 4.10 Chi-square test for bivariate associations between blood glucose checked by GP and education, household income and age in a male and female sample

Male BG checked by GP Chi-Square Test N BG Checked

(%)

X2 DF P-value


155

57 47 31

20

87 16 13

21 18

44 38 34 23

49.0

54.4 55.3 48.4

51.7 56.3 46.2 34.1 63.2 55.9 78.3

0.405

0.293 13.810

2

2

3

0.817 0.864 0.003

Female BG checked by GP Chi-Square Test N BG Check-up

(%)

X2 DF P-value


226

77 53 82 14

108 38 23 45

12

57 68 59

36

54.0

54.5 62.3 53.7

52.8 68.4

47.8

43.9 55.9 58.2 75.0

1.092

3.442

8.792

2

2

3

0.579

0.179

0.032


** ‘Don’t wish to answer’ was excluded from the analysis.

On the basis of this bivariate analysis, it is concluded that there were no statistical

differences between education, income and BP, and BC and BG check-ups, although

some differences among these relationships occurred. Age was found to have

153

positive associations with BP, BC and BG check-ups. Therefore, age was adjusted

when multivariable analysis was applied.

4.6 Multivariable relationship between SEP and the use of preventive health services

To explore the relationship between SEP (education and income) and the use of

preventive health services (BP, BC and BG check-up) at multivariable levels, logistic

regression was performed by adjusting for age as a continuous variable, for both

male and female groups. When comparing the three education levels, the high level

of education was used as a reference group. When comparing the three income

groups, the high-income group was used as a reference group. The results are

discussed below.

4.6.1 Blood pressure check-up

4.6.1.1 Education

Table 4.11 compares each of the education groups in terms of adjusted odds ratio and

95 per cent confidence interval on blood pressure check-up. After adjusting for age,

the results of the modelling show that the middle- and low- education groups were

slightly less likely to go to a GP for a BP check-up compared to the high-education

groups. The odds ratio was 0.92 (95 per cent CI 0.41–2.05) for the middle-education

group and 0.92 (95 per cent CI 0.35–2.41) for the low-education group.

However, this relationship in the female groups differed from that of the male groups.

As can be seen in the table, both the low- and middle-education groups were more

likely to visit a GP for a BP check-up than the high-education group.

Notwithstanding, the low-education group (OR 1.32 95 per cent CI 0.66–2.66) was

less likely to go for a BP check-up than the middle group (OR 1.99 95 per cent

CI 0.86–4.58).

154

Table 4.11 Logistic regression for multivariable analysis between education and BP check-up adjusted for age in male and female groups

Education

Males

OR 95%CI Females

OR 95%CI

Bachelor degree and higher Diploma and Vocational Non post–school

Age

1.00 0.92 0.41–2.05 0.92 0.35–2.41 1.05 1.01–1.09

1.00 1.99 0.86–4.58 1.32 0.66–2.66 1.01 0.98–1.05

Furthermore, Figure 4.2 also illustrates these relationships. It can be seen that there

are no obvious trends across the education groups for males or females.

0

0.5

1

1.5

2

2.5

3

3.5

4

4.5

5

Bachelordegree and

higher

Diplomaand Vocational

Non post-school

Females Bachelordegree and

higher

Diplomaand Vocational

Non post-school

Males Females

Adju

sted

odd

s ra

tios

Figure 4.2 Modelling associations between education and blood pressure check-up

4.6.1.2 Income

Table 4.12 compares each of the income groups in terms of adjusted odds ratio and

95 per cent confidence intervals on blood pressure check-up.

155

Table 4.12 Logistic regression for multivariable analysis between household income and blood pressure check-up adjusted for age in male and female groups

Household income Males

OR 95%CI Females

OR 95%CI A$52 000 or more $31 200–51 999 <$31 199 Age

1.00 1.27 0.40–4.01 0.68 0.19–2.36 1.05 1.01–1.09

1.00 1.86 0.77–4.46 1.79 0.61–5.19 1.01 0.98–1.05

After adjusting for age, among males, the low-income group was one and a half

times less likely to visit a GP for a BP check-up compared to the high-income group

(OR 0.68 95 per cent CI 0.19–2.36). Among females, both the low-income group and

middle-income group were more likely than the high-income group to have a BP

check-up (OR: 1.86 95 per cent CI: 0.77-4.46; 1.79 95 per cent CI: 0.61–5.19). The

low-income groups were slightly less likely to have their BP checked than the middle

groups. Figure 4.3 also portrays these relationships.

0

0.5

1

1.5

2

2.5

3

3.5

4

4.5

5

5.5

Aus$52000or more

$31200-51999

<$31199 Females Aus$52000or more

$31200-51999

<$31199

Males Females

Adju

sted

odd

s ra

tio

Figure 4.3 Modelling associations between household income and blood pressure check-up

156

4.6.2 Blood cholesterol check-up

4.6.2.1 Education


95 per cent confidence intervals on blood cholesterol check-up.

Table 4.13 Logistic regression for multivariable analysis between education and BC check-up adjusted for age in male and female groups

Education

Males OR 95%CI

Females OR 95%CI

Bachelor degree and higher Diploma and vocational Non post-school

Age

1.00 0.69 0.29–1.62 0.26 0.09–0.71 1.08 1.04–1.13

1.00 1.15 0.54–2.44 0.76 0.39–1.50 1.06 1.03–1.09

After adjusting for age, the results from the logistic regression modelling showed that,

among males, there was a strong decreasing trend from the high-education group

across to the low-education group. The low-education group was nearly four-fold

less likely to visit a GP for a BC check-up than the high-education group (OR 0.26,

95 per cent CI 0.09–0.71). The middle-education group was one and half times less

likely to have a BC check-up than the high-education group (OR 0.69, 95 per cent CI

0.29–1.62).

Among the female group, after adjusting for age, the low-education group was one

and a quarter times less likely to have a BC check-up than the high-education group

OR 0.76, 95 per cent CI 0.39–1.50). These relationships can also be observed in

Figure 4.4. It is obvious that the higher the education level, the higher the likelihood

of a BC check-up.

157

0

0.5

1

1.5

2

2.5

3

Bachelordegree and

higher

Diplomaand

Vocational

Nonpost-school

Females Bachelordegree and

higher

Diplomaand

Vocational

Nonpost-school

Males Females

Adj

uste

d od

ds ra

tio

Figure 4.4 Modelling associations between education and blood cholesterol check-up

4.6.2.2 Income


95 per cent confidence intervals on blood cholesterol check-up. After adjusting for

age, the results from the logistic regression analysis showed that there was an

increasing trend in term of BC check-up across the three income groups, from low to

high, among males. The low-income group was two and half times less likely than

the high-income group to have a BC check-up OR 0.39, 95 per cent CI 0.11–1.39).

The middle-income group was 2.2 times less likely than the high-income group to

have a BC check-up (OR 0.45, 95 per cent CI 0.14–1.43). However, there was only

slightly difference between the middle-income group and the low-income group with

the two confidence intervals overlapped.

Table 4.14 Logistic regression for multivariable analysis between household income and blood cholesterol check-up, adjusted for age in male and female groups


OR 95%CI Females

OR 95%CI A$52 000 or more $31 200–51 999 <$31 199 Age

1.00 0.45 0.14–1.43 0.39 0.11–1.39 1.08 1.04–1.13

1.00 1.29 0.57–2.94 0.45 0.17–1.19 1.07 1.04–1.11

158

Among females, the low-income group was nearly two-fold less likely to have a

BC check-up than the high-income group (OR 0.45, 95 per cent CI 0.17–1.19). These

relationships can also be depicted diagrammatically, as in Figure 4.5. Obviously, the

low-income groups in both males and females were less likely to go for a blood

cholesterol check-up compared to the higher income groups. Additionally, there is a

positive correlation between income and blood cholesterol check-up among the male

groups. The more affluent people were more likely to go for a BC check-up than the

less affluent.

0

0.5

1

1.5

2

2.5

3

3.5

Aus$52000or more

$31200-51999

<$31199 Aus$52000or more

$31200-51999

<$31199

Males Females

Adj

uste

d od

ds r

atio

Figure 4.5 Modelling associations between household’s income and blood cholesterol

check-up 4.6.3 Blood glucose check-up

4.6.3.1 Education


95 per cent confidence interval on blood glucose check-up. After adjusting for age,

there was also a decreased trend in BG check-up from the high-education group to

the low-education group. Among males, the low-education group was 1.8-fold less

likely to visit a GP for a BG check-up than the high-education group (OR 0.55, 95

per cent CI 0.22–1.43). The results also showed that the middle-education group was

slightly less likely than the high-education group to have a BG check-up (OR 0.89 95

per cent CI 0.39–1.99). Among the female groups, after adjusting for age, the low-

education group was 1.4-fold less likely to have a BG check-up than the high

159

education group (OR 0.73, 95 per cent CI 0.38–1.42). The middle-education group

was slightly more likely to have a BG check-up than the high-education group (OR

1.17, 95 per cent CI 0.56–2.44).

Table 4.15 Logistic regression for multivariable analysis between education and blood glucose check-up, adjusted for age in male and female groups

Education

Males

OR 95%CI Females

OR 95%CI

Bachelor degree and higher Diploma and vocational Non post-school

Age

1.00 0.89 0.39–1.99 0.55 0.22–1.43 1.06 1.02–1.09

1.00 1.17 0.56–2.44 0.73 0.38–1.42 1.05 1.02–1.08

Figure 4.6 also pictures these relationships. It can be seen that the lower education

groups were less likely than the higher education groups to go for a blood glucose

check-up in both male and female groups. In addition, there was also a positive

correlation between education and blood glucose check-up among males. The more

educated people were, the higher the likelihood of a blood glucose check-up.

0

0.5

1

1.5

2

2.5

3

Bachelordegree and

higher

Diplomaand

Vocational

Nonpost-school

Bachelordegree and

higher

Diplomaand

Vocational

Nonpost-school

Males Females

Adj

uste

d od

ds ra

tio

Figure 4.6 Modelling associations between education and blood glucose

check-up

160

4.6.3.2 Income


95 per cent confidence intervals on blood glucose check-up. After adjusting for age,

there was a decreased trend for BG check-up across the three income groups, from

high to low, among males. The low-income group was 1.7-fold less likely than the

higher income groups to have a BG check-up (OR 0.58, 95 per cent CI 0.16–2.03).

The middle-income group was slightly less likely to have a BG check-up compared

to the high-income group (OR 0.89, 95 per cent CI 0.29–2.80).

Among females, the low-income group was 1.4-fold less likely to have a BG check-

up compared to the high-income group (OR 0.71, 95 per cent CI 0.28–1.79). The

middle-income group was nearly two times more likely to have a BG check-up than

the high-income group (OR 1.89, 0.85–4.22).

Table 4.16 Logistic regression for multivariable analysis between household income and blood glucose check-up adjusted for age in male and female groups


OR 95%CI Females

OR 95%CI Aus$52 000 or more $31 200–51 999 <$31 199 Age

1.00 0.89 0.29–2.80 0.58 0.16–2.03 1.06 1.02–1.10

1.00 1.89 0.85–4.22 0.71 0.28–1.79 1.06 1.01–1.08

Figure 4.7 also illustrates these associations. A positive correlation is seen across the

male income groups and indicates that the low-income groups were less likely to

have their blood glucose checked compared to the high-income group.

161

0

0.5

1

1.5

2

2.5

3

3.5

4

4.5

Aus$52000or more

$31200-51999

<$31199 Aus$52000or more

$31200-51999

<$31199

Males Females

Adju

sted

odd

s ra

tio

Figure 4.7 Modelling associations between household’s income and blood glucose

check-up

4.7 Summary This chapter has reported that an overall 65.6 per cent survey response rate was

achieved. Comparisons were made between the study sample and the 2001 Census

for the local government area of Brisbane in terms of age group, gender and level of

education among 25–64-year olds. It was suggested that the distribution of age and

gender was not very divergent from the 2001 census data. In addition, the

comparison also suggested that the respondents of the study sample had a higher

level of qualification compared to the census data. In addition, comparisons were

made in terms of utilisation of health services with other national data and studies. It

showed there was a similar GP consultation rate in the study sample compared with

the Australian average GP consultation rate in the previous 12 months.

Each of the SEP groups was examined in terms of the key dependent variables of

blood pressure, blood cholesterol and blood glucose check-up. Multivariable

analyses were undertaken to examine the relationship between SEP and use of

preventive health services, adjusted for age as a confounding variable. In general,

most results from the multivariable analyse between SEP and the use of preventive

health services indicated that the sub-groups with lower levels of education and

lower incomes were less likely than the higher SEP groups to present to the GP for a

162

check up of their blood pressure, blood cholesterol and blood glucose. However,

there was only one statistically significant result from the 12 examinations of the

multivariable analysis. Therefore, the results need to be interpreted with caution. A

few results also indicated that the low-education or income groups were more likely

to have their BP checked compared to the high-education or -income groups among

females. The possible reasons are explored in Chapter 6. Nevertheless, the general

trend from most of the findings indicated that the low-SEP groups were less likely

than the high-SEP groups to use preventive health services. Why do these

relationships exist? Chapter 5 will examine this question.

163

5 RESULTS: UNDERSTANDING THE RELATIONSHIP BETWEEN SOCIOECONOMIC POSITION AND USE OF PREVENTIVE HEALTH SERVICES

164

5.1 Introduction The question addressed in this chapter is how to explain the findings in Chapter 4

concerning the relationship between SEP and use of preventive health services. Put

simply, why are socioeconomically disadvantaged individuals less likely to visit a

GP for a blood pressure, blood cholesterol and/or blood glucose check-up, all of

which are recommended preventive activities?

As argued earlier in the literature review, socioeconomic differences in the use of

preventive health services are likely, in part, to be due to the fact that these groups

are differentially predisposed towards utilisation of GP-based preventive activities

because of their different attitudes, beliefs and concerns at the individual level. They

are also differentially exposed to interpersonal-, environmental- and system-level

influences that are linked into other social and economic constraints, and

psychological factors. Four sets of relationships need to be investigated in order to

establish this causal pathway. The first step is to examine the relationship between

the measures of SEP and the use of preventive health services, which has been

examined earlier in Chapter 4. The other three steps are examined in this chapter.

These are: utilisation of multivariable approaches to test the relationship between the

measures of SEP and the intervening variables, to explore the relationship between

the intervening variables and the use of preventive health services and, finally, to

examine how these intervening factors are associated with the relationship between

SEP and the use of preventive health services.

5.2 Modelling associations between SEP and the intervening factors The multivariable associations between SEP variables and 11 intervening variables,

adjusted for age, are examined in this section. Adjusted means of those scales among

each education and income group, and 95 per cent confidence intervals are presented,

and odds ratio and 95 per cent confidence intervals are represented for the two

categorical variables. The results of paired-mean comparisons among education and

income groups for both males and females are also represented, along with F-ratios

and P-values; they are discussed below.

165

5.2.1 Concerns about availability of health care Table 5.1 compares education groups in terms of their mean score on the scale

measuring concerns about the availability of health care. Based on the statistical

evidence in the table, there are no differences of the mean scores across the education

groups for either males (P = 0.193) or females (P = 0.322) with regard to concerns

about the availability of doctors, consultation times, the amount of time to obtain an

appointment, waiting times and being able to see a preferred doctor.

Table 5.1 Comparing education groups in terms of their mean scores on the scale1 measuring concerns about the availability of health care

Education groups Males N2 Mean 95% CI

Females N2 Mean 95% CI

1. High 2. Medium 3. Low

63 15.7 15.0–16.4 49 16.7 15.9–17.5 30 16.1 15.0–17.1

84 16.5 15.9–17.1 53 17.0 16.3–17.8 83 17.1 16.5–17.7

F = 1.666 df = 2 P = 0.193 F = 1.141 df = 2 P = 0.322 Adjusted paired–mean comparisons (P-values)

1. 2. 3.

1 2 0.072 0.622 0.311

1 2 0.255 0.162 0.902

Notes: 1. The scale measuring concerns about the availability of health care was scored to range from 4–20, with a higher score reflecting greater concerns. 2. The response for ‘Not applicable’ was excluded from the scale.

Table 5.2 compares income groups in terms of their mean score on the scale

measuring concerns about the availability of health care. Based on the evidence in

the table, there are no differences across income groups in terms of the scale score

among males (P > 0.10). However, there are some differences among female income

groups (P < 0.05). As can be seen from the paired-mean comparison, a significant

difference appears between the low- and the middle-income groups (P = 0.01), and

between the low- and high-income groups (P < 0.10). Comparatively, a lower level

of concern about availability of health care is experienced among a low-income

female group, while the level of concern is higher among a middle-income group.

There is no statistical difference between the middle- and high-income groups

(P > 0.10). In short, the low-income female group has a lower level of concern about

the availability of health care compared to the higher income groups.

166

Table 5.2 Comparing income groups in terms of their mean scores on the scale1 measuring concerns about the availability of health care

Income groups Males N2 Mean 95% CI



92 16.1 15.5–16.7 19 16.1 14.9–17.4 13 16.1 14.6–17.6

112 16.6 16.1–17.1 39 17.3 16.4–18.2 25 15.4 14.3–16.5


1. 2. 3.

1 2 0.966 0.989 0.986

1 2 0.187 0.060 0.010

Notes: 1. The scale measuring concerns about the availability of health care was scored to range from 4–20, with a higher score reflecting greater concerns. 2. The response for ‘Not applicable’ was excluded from the scale. 5.2.2 Concerns about accessibility to health care Table 5.3 compares education groups in terms of their mean score on the scale

measuring concerns about accessibility to health care. Some significant associations

between education and this scale appear in both male and female groups (P < 0.05).

As can be seen from the paired-mean comparison, the middle-education group is

more likely to be concerned about accessibility to health care, whether the health

service bulk billed patients, the cost of the consultation and the number of doctors

able to be chosen from, compared to the high-education group among both males

(P < 0.01) and females (P < 0.05); while only females from the low-education groups

are more likely to be concerned about their accessibility to health care compared to

the high-income group (P = 0.01). There are no differences between the low- and

middle-income groups, for either males or females (P > 0.10). It is suggested that the

lower the education level, the higher the concerns about accessibility to health care

among females, and the middle-education group are more likely to be concerned than

the high-education group among males.

167

Table 5.3 Comparing education groups in terms of their mean scores on the scale1 measuring concerns about accessibility to health care




62 9.7 8.9–10.4 51 11.3 10.5–12.1 28 10.5 9.5–11.6

76 10.2 9.6–10.8 48 11.4 10.6–12.1

76 11.4 10.8–12.0 F = 4.379 df = 2 P = 0.014 F = 4.065 df = 2 P = 0.019 Adjusted paired–mean comparisons (P-values)

1. 2. 3.

1 2 0.004 0.189 0.275

1 2 0.026 0.010 0.943

Notes: 1. The scale measuring concerns about the accessibility to health care was scored to range from 3–15, with a higher score reflecting greater concerns. 2. The response for ‘Not applicable’ was excluded from the scale. Table 5.4 compares income groups in terms of their mean scores on the scale

measuring concerns about the accessibility to health care. As can be seen from the

results of paired comparison, there are no differences among male income groups

(P > 0.10) and significant differences appear in female income groups (P < 0.05). It

shows that the female respondents with a low income are more likely than those with

a high income to be concerned about their accessibility to health care (P < 0.05). In

addition, there is a borderline significance between the middle- and high-income

groups (P < 0.10). This indicates that concerns about the accessibility to health care

are low among female respondents with a high income compared to those with a

middle income. There are no statistical significances among the low- and middle-

income groups. A similar conclusion to that drawn for the education groups—that the

lower the income levels, the higher the concerns about accessibility to health care

among females but not males.

168

Table 5.4 Comparing income groups in terms of their mean scores on the scale1 measuring concerns about accessibility to health care




90 9.9 9.4–10.6 20 11.2 9.9–12.5 12 11.3 9.6–12.9

103 10.4 9.9–10.9 37 11.4 10.5–12.3 21 11.9 10.8–13.1


1. 2. 3.

1 2 0.091 0.146 0.930

1 2 0.057 0.019 0.465

Notes: 1.The scale measuring concerns about the accessibility to health care was scored to range from 3–15, with a higher score reflecting greater concerns. 2. The response for ‘Not applicable’ was excluded from the scale. 5.2.3 Concerns about transport and travel time to health care Table 5.5 compares education groups in terms of their mean scores on the scale

measuring concerns about transport and travel time to health care. As can be seen

from the paired-mean comparisons, there are no statistical differences across the

three male education groups (P > 0.10). On the other hand, there is close to

borderline significance across the female education groups (P < 0.10). A statistical

significance occurs between the low- and the high-education groups (P < 0.10) and

the low- and the middle-education groups (P < 0.05) respectively. These results

indicate that the female respondents with a low level of education are more likely to

be concerned about transport and travel time to health care compared to their higher

education counterparts. However, the result needs to be interpreted with caution.

There is no significant difference between the middle- and the high-education groups.

Table 5.5 Comparing education groups in terms of their mean score on the scale1 measuring concerns about transport and travel time to health care




59 5.53 4.94–6.11 48 5.82 5.18–6.47 26 5.81 4.93–6.69

77 6.2 5.7–6.7 51 5.9 5.3–6.6

75 6.8 6.3–7.4 F = 0.275 df = 2 P = 0.760 F = 2.591 df = 2 P = 0.077 Adjusted paired–mean comparisons (P-values)

1. 2. 3.

1 2 0.498 0.596 0.979

1 2 0.567 0.092 0.035

Notes: 1.The scale measuring concerns about transport and travel time to health care was scored to range from 2–10, with a higher score reflecting greater concerns. 2. The response for ‘Not applicable’ was excluded from the scale.

169

Table 5.6 compares income groups in terms of their mean scores on the scale

measuring concerns about accessibility to health care. Based on the evidence of the

paired-mean comparisons, statistical significances appear in male groups (P < 0.05)

and no significant results appear among the female groups (P > 0.10). It indicates

that the males with a low income are more likely than those with a high income to be

concerned about transport and travel time (P < 0.01). Similarly, the males with a

middle income are more likely than those with a high income to be concerned

transport and travel time (P < 0.01). There is no difference between the low- and the

middle-income groups (P > 0.10). It can be summarised from this evidence that male

respondents with a lower income are more concerned about transport and travel time

to access health services than those with a high income.

Table 5.6 Comparing income groups in terms of their mean score on the scale1 measuring concerns about transport and travel time to health care




85 5.4 4.9–5.8 19 6.3 5.4–7.3 12 7.3 6.1–8.5

104 5.9 5.5–6.3 38 6.3 5.6–7.0 23 6.9 5.9–7.8


1. 2. 3.

1 2 0.076 0.004 0.228

1 2 0.339 0.054 0.314

Notes: 1. The scale measuring concerns about transport and travel time to health care was scored to range from 2–10, with a higher score reflecting greater concerns. 2. The response for ‘Not applicable’ was excluded from the scale. 5.2.4 Perceived interpersonal care from general practitioners Table 5.7 compares education groups in terms of their mean scores on the scale

measuring perceived interpersonal care from general practitioners. On the basis of

the results, there are no significant associations between education and this scale

among both males and females (P > 0.10).

170

Table 5.7 Comparing education groups in terms of their mean score on the scale1 measuring perception of interpersonal care from general practitioners

Education groups Males N Means 95% CI

Females N Means 95% CI


65 10.6 9.9–11.3 52 11.2 10.5–11.9 33 11.6 10.7–12.6

86 11.9 11.4–12.5 53 11.9 11.2–12.7 83 11.7 11.1–12.3


1. 2. 3.

1 2 0.225 0.095 0.501

1 2 0.989 0.558 0.607

Note: 1 The scale measuring perception of the interpersonal care was scored to range from 3–15, with a higher score reflecting better perceived care. Table 5.8 compares income groups in terms of their mean scores on the scale

measuring perceived interpersonal care from general practitioners. As can be seen

from the table, for both males and females, there are no statistically significant

associations across income groups in regard to receiving interpersonal care from

doctors, including the amount of time the doctor spends with them, the doctor’s

patience with questions or worries, and the doctor’s caring and concern (P > 0.10).

Table 5.8 Comparing income groups in terms of their mean score on the scale1 measuring perception of interpersonal care from general practitioners

Income groups Males N Mean 95% CI

Females N Mean 95% CI


93 11.2 10.7–11.8 21 11.0 9.8–12.2 15 10.1 8.6–11.5

113 11.9 11.4–12.4 39 11.8 10.9–12.6 25 10.8 9.7–11.9


1. 2. 3.

1 2 0.771 0.140 0.303

1 2 0.821 0.077 0.169

Note: 1. The scale measuring perception of the interpersonal care was scored to range from 3–15, with a higher score reflecting better perceived care. 5.2.5 Value of general practitioners Table 5.9 compares education groups in terms of their mean scores on the scale

measuring value of general practitioners. Based on the evidence from the table, no

statistically significant associations exist among these education groups for either

males or females with regard to their perception on the value of general practitioners,

(P > 0.10).

171

Table 5.9 Comparing education groups in terms of their mean score on the scale1 measuring value of general practitioners

Education groups Males N Mean 95% CI



64 10.1 9.7–10.5 52 9.9 9.5–10.4

33 10.0 9.5–10.6

86 9.6 9.2–9.9 53 9.7 9.2–10.2 83 10.1 9.7–10.5


1. 2. 3.

1 2 0.661 0.833 0.870

1 2 0.687 0.055 0.184

Note: 1. The scale measuring value of general practitioners was scored to range from 2–12, with a higher score reflecting greater value. Table 5.10 compares income groups in terms of their mean scores on the scale

measuring value of general practitioners. Similarly, based on the evidence, no

associations exist across income groups for either males or females (P > 0.10) with

regard to the perception of the value of general practitioners; that is, whether the care

received from doctors in the past has been good, and whether doctors belong to a

very high-status profession.

Table 5.10 Comparing income groups in terms of their mean score on the scale1 measuring perception of value of general practitioners




93 10.09 9.75–10.43 21 10.33 9.62–11.04

14 9.57 8.69–10.44

113 9.7 9.4–10.1 39 9.7 9.2–10.3 25 9.1 8.4–9.8


1. 2. 3.

1 2 0.558 0.272 0.182

1 2 0.998 0.112 0.167

Note: 1. The scale measuring value of general practitioners was scored to range from 2–12, with a higher score reflecting greater value. 5.2.6 Attitudes towards health care Table 5.11 compares education groups in terms of their mean score on the scale

measuring attitudes towards health care. Some significant associations appear across

the education groups among both males and females (P < 0.05). The table shows that

respondents with a middle level of education have less positive attitudes towards

health care than those with a higher level of education—for males (P < 0.01) and

females (P < 0.05). There are no associations between the low- and the high-

172

education groups. In addition, there is a borderline significant difference between the

low- and the middle-education groups among females (P < 0.10), but no such

association among males (P > 0.10). The lower education group had more positive

attitudes than the middle-education group. The result needs to be interpreted with

caution.

Table 5.11 Comparing education groups in terms of their mean score on the scale1 measuring attitudes towards health care




65 12.4 11.6–13.3 52 10.8 9. 9–11.7 33 12.0 10.9–13.6

86 12.1 11.4–12.7 53 10.7 9.9–11.6 83 11.7 11.0–12.4


1. 2. 3.

1 2 0.009 0.551 0.107

1 2 0.016 0.469 0.075

Note: 1. The scale measuring attitudes towards health care was scored to range from 3–18, with a higher score reflecting positive attitudes. Table 5.12 compares income groups in terms of their mean scores on the scale

measuring attitudes towards health care. Based on the evidence in the table, no

statistical significances appear across either male or female income groups (P > 0.10).

Table 5.12 Comparing income groups in terms of their mean score on the scale1 measuring attitudes towards health care




93 11.90 11.21–12.59 21 11.29 9.84–12.74 15 12.01 10.27–13.74

113 11.9 11.3–12.5 39 11.6 10.7–12.6 25 12.0 10.8–13.2


1. 2. 3.

1 2 0.454 0.914 0.529

1 2 0.642 0.840 0.609

Note: 1. The scale measuring attitudes towards health care was scored to range from 3–18, with a higher score reflecting positive attitudes. 5.2.7 Value of good health Table 5.13 compares education groups in terms of their mean scores on the scale

measuring value of good health. Based on this evidence, there are no significant

173

associations between education and value of good health scores in either male or

female groups (P > 0.10).

Table 5.13 Comparing education groups in terms of their mean score on the scale1 measuring value of good health




65 14.9 14.3–15.4 52 14.9 14.3–15.6 33 15.0 14.2–15.9

86 15.1 14.6–15.7 53 15.2 14.5–15.9

83 15.3 14.7–15.9


1. 2. 3.

1 2 0.869 0.720 0.833

1 2 0.882 0.670 0.814

Note: 1. The scale measuring value of good health was scored to range from 5–20, with a higher score reflecting greater value.

Table 5.14 compares income groups in terms of their mean scores on the scale

measuring value of good health. Two significant associations across female income

groups are found (P < 0.05), while no associations exist in the male group (P > 0.10).

It is indicated that the female respondents with a low income are more likely to

follow a GP’s suggestions to improve their health than those who had a middle

income (P = 0.015) and a high income (P = 0.031). There is no significant

association between the middle- and the high-income groups (P > 0.10).

Table 5.14 Comparing income groups in terms of their mean score on the scale1 measuring value of good health




93 14.8 14.3–15.2 21 15.3 14.3–16.3 15 15.3 14.1–16.5

113 15.2 14.7–15.7 39 14.8 13.9–15.6 25 16.4 15.4–17.5

F = 0.737 df = 2 P = 0.481 F = 3.205 df = 2 P = 0.043 Adjusted paired-mean comparisons (P-values)

1. 2. 3.

1 2 0.323 0.387 0.990

1 2 0.436 0.031 0.015

Note: 1. The scale measuring value of good health was scored to range from 5–20 with a higher score reflecting greater value. 5.2.8 Social support Table 5.15 compares education groups in terms of their mean scores on the scale

measuring negative aspects of social support. There are significant differences across

male (P < 0.05) and female education groups (P < 0.10). The table shows that the

174

respondents with a low level of education are more likely than those with a high

education level to have a low level of social support (P < 0.05). In addition, the male

middle-education group also experienced a lower level of social support than the

high-education group (P < 0.05), while the female low-education group has a lower

level of social support than their middle counterparts (P < 0.10). Overall, it can be

concluded that the respondents with lower levels of education are more likely to have

a lower level of social support than their high counterparts.

Table 5.15 Comparing education groups in terms of their mean score on the scale1 measuring negative aspect of social support




65 22.5 21.3–23.6 52 20.7 19.5–21.9 33 20.1 18.5–21.7

86 22.3 21.2–23.4 53 22.0 20.7–23.4 83 20.5 19.4–21.6


1. 2. 3.

1 2 0.045 0.018 0.528

1 2 0.783 0.028 0.089

Note: 1. The scale measuring social support was scored to range from 5–30, with a lower score reflecting a lower level of social

support.

Table 5.16 compares income groups in terms of their mean score on the scale

measuring negative aspect of social support. Evidence shows that there are

significant differences across male (P < 0.10) and female income groups (P < 0.05).

Male respondents from the middle-income group have a lower level of social support

than their high counterparts (P < 0.05). However, no associations appear between the

low- and high-income groups, and between the low- and the middle-income groups

(P > 0.10). Female respondents with a low or middle income are more likely than

those with a high income to have a lower level of social support (P < 0.05), while no

statistical association exists between the low and the middle income groups among

females. In short, these results indicate that there is a positive association between

income and social support scale, with the higher income level and the higher level of

social support among females, while the only association found is that the middle

group has a lower level of social support than their high counterparts.

175

Table 5.16 Comparing income groups in terms of their mean score on the scale1 measuring negative aspect of social support




93 21.9 21.1–22.9 21 19.6 17.6–21.6 15 20.9 18.6–23.3

113 22.4 21.5–23.3 39 20.0 18.5–21.6 25 20.2 18.2–22.1


1. 2. 3.

1 2 0.032 0.428 0.373

1 2 0.011 0.042 0.923

Note: 1. The scale measuring social support was scored to range from 5–30, with a lower score reflecting a lower level of social support. Table 5.17 compares education groups in terms of their mean scores on the scale

measuring positive aspect of social support. Based on this evidence, there are no

significant differences across the education groups for both males and females

(P > 0.10).

Table 5.17 Comparing education groups in terms of their mean score on the scale1 measuring positive aspect of social support




65 22.9 21.9–23.9 52 22.3 21.2–23.4 33 21.7 20.3–23.2

86 24.1 23.3–24.9 53 24.7 23.7–25.8 83 23.4 22.5–24.2


1. 2. 3.

1 2 0.393 0.174 0.542

1 2 0.344 0.197 0.036

Note: 1. The scale measuring social support was scored to range from 5–30, with a higher score reflecting a higher level of social support. Table 5.18 compares income groups in terms of their mean scores on the scale

measuring positive aspect of social support. As can be seen from the table, no

statistical significances are found in either male or female groups (P > 0.10).

176

Table 5.18 Comparing income groups in terms of their mean score on the scale1 measuring positive aspect of social support

Income groups Males N Means 95% CI

Females N Means 95% CI


93 22.9 22.1–23.7 21 21.9 20.2–23.7 15 21.9 19.8–23.9

113 24.1 23.4–24.8 39 24.5 23.3–25.7 25 22.9 21.4–24.4

F = 0.771 df = 2 P = 0.465 F = 1.340 df = 2 P = 0.265

Adjusted paired–mean comparisons (P-values)

1. 2. 3.

1 2 0.323 0.363 0.960

1 2 0.578 0.170 0.112

Note: 1. The scale measuring social support was scored to range from 5–30, with a higher score reflecting a higher level of social support. 5.2.9 A regular source of care A regular place of care

Table 5.19 compares education and income groups in terms of their odds ratios and

95 per cent confidence intervals on the measure of a regular place of care. On the

basis of this evidence, male respondents with a low level of education are 1.2 times

less likely than those who had a high level of education to have a regular place of

care (OR: 0.87 95 per cent CI: 0.26–2.94), while the respondents with a middle level

of education are more likely than their high counterparts to have a regular place of

care (OR: 1.74 95 per cent CI: 0.54–5.60). In contrast, there is a reverse pattern of

this association among female education groups. It should be noted that the odds

ratio is quite large and the 95 per cent CI is much wider in the middle female

education group. This may be due to the limited sample size.

177

Table 5.19 Comparing education and income group in terms of their odds ratios and 95% confidence intervals on the question measuring a regular place of care

SEP variables Males N OR 95% CI

Females N OR 95% CI

Education groups 1. High 2. Medium 3. Low

60 1.00 52 1.74 0.54–5.60 32 0.87 0.26–2.94

83 1.00 50 5.74 0.70–46.81 81 1.31 0.43–3.98

Income groups 1. High 2. Medium 3. Low

89 1.0 20 0.44 0.12–1.71 15 0.32 0.07–1.55

108 1.0

39 1.33 0.35–5.08 24 1.10 0.22–5.43

Note: The question measuring a regular place of care was categorised as Yes and No.

When comparing the relationship between income and the measure of a regular place

of care, it has been found that the low-income male respondents are three times (95

per cent CI: 0.12–1.71) less likely to visit the same medical centre than those with a

high income; and, male respondents with a middle income are two times (95 per cent

CI: 0.07–1.55) less likely to visit the same medical centre than those with a high

income. However, these associations are reversed among females. The females with

a low or middle income are more likely to have a regular place of care compared to

those in the high-income group.

A regular health provider

Table 5.20 compares education and income groups in terms of their odds ratios and

95 per cent confidence intervals on the measure of a regular care provider.

Table 5.20 Comparing education and income groups in terms of their odds ratios and 95% confidence intervals on the question measuring a regular care provider

SEP variables Males N OR 95% CI

Females N OR 95% CI

Education groups 1. High 2. Medium 3. Low

56 1.00 51 1.21 0.48–3.02

33 1.08 0.37–3.14

78 1.00 50 0.91 0.35–2.34

80 1.03 0.43–2.43 Income groups


85 1.0 21 1.16 0.33–4.11

14 0.51 0.13–2.07

104 1.00

37 1.33 0.46–3.89 23 0.74 0.24–2.28

Note: The question measuring a regular care provider was categorised as Yes and No.

As can be seen, there are no differences between the low- and the high-education

groups with regard to a regular care provider for either male or female groups.

Respondents with a middle level of education are slightly less likely than those with

178

a high level of education to have a regular care provider in the female group (OR:

0.91 95 per cent CI: 0.35–2.34), while this association is reversed in the male group

(OR: 1.21 95 per cent CI: 0.48–3.02). On the other hand, both male and female

respondents in a low-income group are consistently less likely to have a regular care

provider compared to their high-income counterparts. Their odds ratios are 0.51 (95

per cent CI: 0.13–2.07) and 0.74 (95 per cent CI: 0.24–2.28) respectively. On the

contrary, both male and female respondents in a middle-income group are more

likely to see the same doctor than those with a high income.

In summary, from the 44 preceding examinations on the relations between SEP

(education and income) and the 11 intervening variables for both male and female

groups, the main results can be highlighted based on their significant differences as

follows.

• The low-income female group has a lower level of concern about the

availability of health care compared to their higher income (both high and

middle groups) counterparts.

• The middle-education group is more likely to be concerned about

accessibility to health care compared to the high-education group, among

both males and females, while only females from low-education groups are

more likely to be concerned about their accessibility to health care compared

to the high-education group. Additionally, female respondents from both low-

and middle-income groups are more likely to be concerned about

accessibility to health care than those from the high-income group.

• The female respondents with a low level of education are more likely to be

concerned about transport and travel time to health care compared to their

higher education counterparts (including both middle and high). In addition,

male respondents with a lower income (middle and low groups) are

concerned more about transport and travel time to access health services than

those with a high income.

• There are no significant associations between education, income and the

scales measuring perceived interpersonal care from general practitioners, and

the value of general practitioners among both males and females.

179

• Respondents with a middle level of education have less positive attitudes

towards health care than those with a higher level of education, for both

males and females. In addition, there is a borderline significant difference

between the low- and the middle-education groups among females. It

indicates that the lower education group has a more positive attitude than the

middle-education group. However, the result needs to be interpreted with

caution because of the imprecision of the estimate due to limited sample size.

• Female respondents with a low income are more likely to follow a GP’s

suggestions to improve their health than those with a middle or high income.

• Male respondents with lower level of education (including low- and middle-

education groups) are more likely to have a lower level of social support than

their high counterparts, while female respondents with a low level of

education have experienced a lower level of social support compared to their

high counterparts (both middle- and high-education groups). Besides, male

respondents from the middle-income group have a lower level of social

support than their high counterparts. Female respondents with a low- or

middle-income level are more likely than those with a high income to have a

lower level of social support.

• There are no significant differences between SEP and the measures of a

regular place of care and a regular car provider. However, some trends are

found based on the difference in odds ratios. They are:

– Male respondents in the low-education group are less likely than those

in high-education group to have a regular place of care. In addition, for

males, the lower the income the less likelihood there is of having a

regular place of care. In contrast, there is a reverse pattern of this

association among female education and income groups.

– There are no differences between the low- and the high-education

groups (either male or female) with regard to their regular care provider.

In contrast, both male and female respondents in a low-income group

are consistently less likely to have a regular care provider than their

high-income counterparts.

180

5.3 Modelling associations between the intervening variables and the use of preventive health services

This section examines the modelling associations between each of the intervening

variables and each of the three dependent variables, including blood pressure (BP),

blood cholesterol (BC) and blood glucose (BG) check-ups, adjusted for age, among

the male and female groups respectively. Logistic regression modellings are applied.

As mentioned in Chapter 3, those nine scales are employed in the models as

categorical variables (including low, middle and high groups). Odds ratios and 95 per

cent confidence intervals (CIs) are presented. They are discussed as follows.

5.3.1 Concerns about availability to health care Table 5.21 presents the results from the modelling associations between people’s

concerns about the availability of health care and the use of preventive health

services among male and female groups. As can be seen in the table, the male

respondents with the high score of concerns are more than two times more likely to

have BP, BC and BG check-ups in comparison to those with a low score of concerns.

On the basis of this evidence, it is indicated that the more the male respondents are

concerned about the availability of health care (with higher score), the more they are

likely to visit their GP for a BP, BC and BG check-up. However, the pattern is

reversed among females.

Table 5.21 Results of logistic regression modelling between people’s concerns about availability of health care and the use of preventive health services

Male Female Concerns about availability of health care2

BP

OR1 95%CI

BC

OR 95%CI

BG

OR 95%CI

BP

OR 95%CI

BC

OR 95%CI

BG

OR 95%CI

Low Middle High

1.000 1.45 0.61–3.44 2.03 0.74–5.55

1.000 1.57 0.67–3.67 2.13 0.78–5.83

1.000 1.10 0.48–2.55 2.33 0.86–6.29

1.000 0.96 0.47–1.95 0.80 0.37–1.74

1.000 1.32 0.68–2.55 1.01 0.48–2.09

1.000 1.12 0.59–2.13 0.79 0.39–1.63

Notes: 1. OR: age adjusted odds ratio. 2. The scale measuring concerns about the availability of health care was scored to range from 1–20, with a higher score reflecting greater concerns. 5.3.2 Concerns about accessibility to health care Table 5.22 presents the results from the modelling associations between the scale of

concerns about the accessibility to health care and the three dependent variables. As

can be seen in the table, among both males and females, the respondents with the

181

high score of concerns are one to one and half folds less likely to have BP, BC and

BG check-ups in comparison to those with a low score of concerns. This indicates

that the people who are more concerned about accessibility to health care are less

likely to use preventive health services.

Table 5.22 Results of logistic regression modelling between people’s concerns about accessibility to health care and the use of preventive health services

Male Female Concerns about accessibility to health care2

BP

OR1 95%CI

BC

OR 95%CI

BG

OR 95%CI

BP

OR 95%CI

BC

OR 95%CI

BG

OR 95%CI

Low Middle High

1.000 0.74 0.31–1.78 0.87 0.36–2.09

1.000 1.57 0.63–3.92 0.67 0.27–1.65

1.000 1.26 0.52–3.08 0.90 0.38–2.17

1.000 1.23 0.56–2.69 0.89 0.42–1.91

1.000 0.80 0.39–1.64 0.75 0.36–1.57

1.000 0.79 0.39–1.58 0.75 0.37–1.51

Notes: 1. OR: age adjusted odds ratio. 2. The scale measuring concerns about the accessibility to health care was scored to range from 1–15, with a higher score reflecting greater concerns. 5.3.3 Concerns about transport and travel time to health care

Table 5.23 presents the results from logistic regression modelling between people’s

concerns about transport and travel time to health care and use of preventive health

services among males and females. Except for BP check-ups among females, most of

the results indicate that respondents who are more concerned about transport and

travel time to access health services are about one to one and half times less likely to

visit a GP for preventive health check-ups.

Table 5.23 Results of logistic regression modelling between people’s concerns about transport and travel time to health care and the use of preventive health services

Male Female Concerns about transport and travel time to health care2

BP

OR1 95%CI

BC

OR 95%CI

BG

OR 95%CI

BP

OR 95%CI

BC

OR 95%CI

BG

OR 95%CI

Low Middle High

1.000 0.76 0.32–1.78 0.76 0.29–1.97

1.000 1.03 0.43–2.46 0.66 0.26–1.71

1.000 1.14 0.49–2.64 0.60 0.23–1.55

1.000 1.46 0.68–3.17 1.58 0.73–3.41

1.000 0.67 0.32–1.41 0.65 0.31–1.34

1.000 0.75 0.37–1.55 0.94 0.47–1.92

Notes: 1. OR: age adjusted odds ratio. 2. The scale measuring concerns about transport and travel time to health care was scored to range from 1–10, with a higher score reflecting greater concerns.

5.3.4 Perceived interpersonal care Table 5.24 presents the results of logistic regression modelling between the scale of

perceived interpersonal care and the use of preventive health services among the

182

male and female groups. The results from the logistic regression modelling for both

males and females suggest that perceptions towards interpersonal care are related to

the use of preventive health services. For example, among males, respondents who

had a middle score are one and half times more likely than those with a low score to

have a BG check-up. This relationship is statistically significant (OR 2.32 95 per cent

CI 1.02–5.28). Some borderline significance also occurs between the high score of

the perceived care and BG check-up among males (OR: 2.27; 95 per cent CI: 0.93–

5.57), and between the middle score and BP check-up among females (OR: 2.14;

95 per cent CI: 0.99–4.60). On the basis of this evidence, it is indicated that

respondents who thought that doctors spending time with them, doctors’ patience and

caring and concern for them was important are more likely to use preventive health

services.

Table 5.24 Results of logistic regression modelling between the scale of perceived interpersonal care and the use of preventive health services

Male Female Perceived interpersonal care2

BP

OR1 95%CI

BC

OR 95%CI

BG

OR 95%CI

BP

OR 95%CI

BC

OR 95%CI

BG

OR 95%CI Low Middle High

1.000 0.90 0.39–2.06 1.19 0.49–2.92

1.000 1.52 0.66–3.51 1.42 0.59–3.42

1.000 2.32 1.02–5.28 2.27 0.93–5.57

1.000 2.14 0.99–4.60 1.63 0.79–3.36

1.000 1.29 0.63–2.67 1.56 0.77–3.17

1.000 1.78 0.88–3.59 1.48 0.75–2.93

Notes: 1. OR: age adjusted odds ratio. 2. The scale measuring perception of the interpersonal care was scored to range from 1–15, with a higher score reflecting better perceived care.

5.3.5 Value of general practitioners

Table 5.25 presents the results of logistic regression modelling between the scale of

value of general practitioners and the use of preventive health services among the

male and female groups. Similar results to those for interpersonal care appear in the

table. These also suggest that those respondents who hold positive perceptions on the

value of general practitioners are more likely to use preventive health services. For

example, the respondents who had positive perceptions on the value of GPs are

nearly two and half times (95 per cent CI: 0.95–6.19) more likely to have their BP

checked than those who had lower value perceptions among males. However, this

association is of borderline significance and the result needs to be interpreted with

caution.

183

Table 5.25 Results of logistic regression modelling between people’s perceptions on value of general practitioners and the use of preventive health services

Male Female Value of general practitioner2

BP

OR1 95%CI

BC

OR 95%CI

BG

OR 95%CI

BP

OR 95%CI

BC

OR 95%CI

BG


1.000 1.06 0.46–2.43 2.42 0.95–6.19

1.000 0.79 0.34–1.85 1.25 0.51–3.08

1.000 0.98 0.42–2.28 1.84 0.77–4.44

1.000 1.33 0.68–2.62 1.21 0.54–2.72

1.000 1.23 0.65–2.33 1.08 0.49–2.29

1.000 1.66 0.89–3.11 1.15 0.55–2.40

Notes: 1. OR: age adjusted odds ratio. 2. The scale measuring value of general practitioners was scored to range from 1–12, with a higher score reflecting greater value.

5.3.6 Attitude towards health care

Table 5.26 presents the results of logistic regression modelling between people’s

attitude towards health care and the use of preventive health services among the male

and female groups. Most of the results in the table indicate that the respondents with

a positive attitude towards health care are more likely to use the preventive health

services compared to the respondents with negative attitudes. In particular, the

female respondents with a positive attitude towards health care are about two and

half times more likely to have BC and BG check-ups than those with negative

attitudes (BC check-up: OR 2.54 95 per cent CI 1.18–5.44; BG check-up: OR 2.40

95 per cent CI 1.14–5.06).

Table 5.26 Results of logistic regression modelling between people’s attitude towards health care and the use of preventive health services

Male Female Attitudes

towards health care2

BP

OR1 95%CI

BC

OR 95%CI

BG

OR 95%CI

BP

OR 95%CI

BC

OR 95%CI

BG


1.000 1.36 0.59–3.12 1.13 0.46–2.74

1.000 0.85 0.38–1.92 1.26 0.49–3.21

1.000 0.77 0.34–1.73 0.84 0.34–2.08

1.000 1.56 0.78–3.12 1.53 0.71–3.32

1.000 1.74 0.89–3.37 2.54 1.18–5.44

1.000 1.79 0.95–3.38 2.40 1.14–5.06

Notes: 1. OR: age adjusted odds ratio. 2. The scale measuring attitudes towards health care was scored to range from 1–18, with a higher score reflecting positive attitudes.

5.3.7 Value of good health Table 5.27 presents the results of logistic regression modelling between people’s

perceptions of the value of good health and use of preventive health services among

the male and female groups.

184

Table 5.27 Results of logistic regression modelling between people’s perception of value of good health and the use of preventive health services

Male Female Value of good health2

BP

OR1 95%CI

BC

OR 95%CI

BG

OR 95%CI

BP

OR 95%CI

BC

OR 95%CI

BG

OR 95%CI Low Middle

High

1.000 1.28 0.58–2.82 0.73 0.29–1.85

1.000 0.75 0.34–1.65 0.83 0.32–2.18

1.000 0.74 0.34–1.61 0.99 0.38–2.56

1.000 1.19 0.59–2.40 1.22 0.57–2.61

1.000 1.55 0.79–3.01 1.61 0.78–3.34

1.000 1.11 0.58–2.12 1.39 0.69–2.81

Notes: 1. OR: age adjusted odds ratio. 2. The scale measuring value of good health was scored to range from 1–20, with a higher score reflecting greater value.

On the basis of this evidence, it can be inferred that the male respondents who were

willing to follow a doctors’ advice to improve their health (high score) are slightly

less likely to use health services, whereas these scores are the opposite for females.

For example, female respondents who are willing to follow a doctor’s advice to

improve their health (high score) are more likely to use health services than those

who dislike (low score) having their BC checked.

5.3.8 Social support

Table 5.28 presents the results of logistic regression modelling between social

support scale and use of preventive health services among the male and female

groups. As can be seen, the results of relationships between negative aspect of social

support and use of preventive health services are not consistent. For BP check-up

among males, the respondents with a low level of social support are more likely to go

for a BP check-up compared to those respondents with a high social support score.

However, for BC check-ups among males, the respondents with a higher score of

social support are more likely to have a BC check-up. Of females, for all check-ups,

the respondents with less social support are more likely to have check-ups.

Table 5.28 Results of logistic regression modelling between negative aspect of social support and the use of preventive health services

Male Female Negative aspect of Social support2

BP

OR1 95%CI

BC

OR 95%CI

BG

OR 95%CI

BP

OR 95%CI

BC

OR 95%CI

BG

OR 95%CI Low Middle

High

1.000 0.68 0.25–1.83 0.52 0.24–1.15

1.000 1.19 0.43–3.25 1.29 0.58–2.88

1.000 1.09 0.41–2.94 0.99 0.45–2.19

1.000 0.65 0.27–1.55 0.71 0.37–1.38

1.000 0.54 0.23–1.28 0.91 0.49–1.70

1.000 0.52 0.22–1.22 0.82 0.45–1.50

Notes: 1. OR: age adjusted odds ratio. 2. The scale measuring social support was scored to range from 1–30, with a least score reflecting a lower level of social support.

185

Table 5.29 presents the results of logistic regression modelling between positive

aspects of social support and use of preventive health services. Most of the results

between positive aspects of social support and the use of preventive services indicate

that the respondents who were more positive with their social support are more likely

to have their BP, BC and BG (except for BG check-up among females) check-ups.

Two borderline significances appear between the middle social support score and BC

check-up, and between the high social support score and BG check-up. However,

these results need to be interpreted with caution.

Table 5.29 Results of logistic regression modelling between positive aspect of social support and the use of preventive health services

Male Female Positive social support 2

BP OR1 95%CI

BC OR 95%CI

BG OR 95%CI

BP OR 95%CI

BC OR 95%CI

BG OR 95%CI

Low Middle

High

1.000 1.08 0.50–2.33 1.64 0.59–4.58

1.000 2.17 0.98–4.83 2.32 0.83–6.47

1.000 1.64 0.76–3.57 2.65 0.96–7.37

1.000 1.15 0.56–2.34 1.22 0.56–2.67

1.000 0.83 0.42–1.65 1.07 0.52–2.22

1.000 0.79 0.41–1.55 0.68 0.33–1.38

Notes: 1. OR: age adjusted odds ratio. 2. The scale measuring social support was scored to range from 1–30, with a higher score reflecting a higher level of social support.

5.3.9 A regular source of care A regular place of care

Table 5.30 presents the results of logistic regression modelling between a regular

source of care and the use of preventive health services among the male and female

groups. Evidence shows that respondents (both male and female) who often visited

the same medical centre when they went to see a GP were more likely to have BP,

BC and BG check-ups. For example, male respondents who often visited the same

medical centre are two times more likely to have their BP checked compared to those

who did not (OR 2.96, 95 per cent CI 1.047–8.35).

Table 5.30 Results of logistic regression modelling between a regular place of care and the use of preventive health services

Male Female A regular place of care2

BP

OR1 95%CI

BC

OR 95%CI

BG

OR 95%CI

BP

OR 95%CI

BC

OR 95%CI

BG

OR 95%CI

No Yes

1.000 2.96 1.05–8.35

1.000 1.31 0.47–3.68

1.000 1.82 0.60–5.48

1.000 1.29 0.46–3.69

1.000 1.17 0.40–3.39

1.000 1.19 0.43–3.30

Notes: 1. OR: age adjusted odds ratio. 2. The question measuring a regular place of care categorised as Yes and No.

186

Figure 5.1 illustrates these relationships. It is obvious that respondents (both males

and females) with a regular place of care are more likely to use preventive health

services than those who do not .

0

1

2

3

4

5

6

7

8

9

BP BC BG BP BC BG

Males Females

Adju

sted

odd

s ra

tio

Without a regular place of care With a regular place of care

Figure 5.1 Relationships between a regular place of care and the use of preventive health services among males and females

A regular care provider

Table 5.31 presents the results of logistic regression modelling between a regular

care provider and the use of preventive health services among the male and female

groups. Strong evidence shows that the respondents (both males and females) who

often saw the same doctor when they went to see a GP were more likely to have BP,

BC and BG check-ups, except in the case of BP check-ups for the male respondents.

For example, the female respondents who often saw the same doctor were one and

half times more likely to have their BG checked than those who did not (OR 2.43 95

per cent CI 1.14–5.18).

Table 5.31 Results of logistic regression modelling between a regular care provider and the use of preventive health services

Male Female A regular care provider2

BP

OR1 95%CI

BC

AOR 95%CI

BG

AOR 95%CI

BP

AOR 95%CI

BC

AOR 95%CI

BG

AOR 95%CI No

Yes

1.000 0.62 0.25–1.53

1.000 1.14 0.47–2.75

1.000 1.24 0.51–3.04

1.000 1.71 0.78–3.75

1.000 2.13 0.97–4.63

1.000 2.43 1.14–5.18

Notes: 1. OR: age adjusted odds ratio.2. The question measuring a regular place of care categorised as Yes and No.

187

Figure 5.2 also presents these relationships. Except for BP check-up among males,

respondents with a regular care provider are more likely than those without a regular

care provider to visit a GP for a check-up.

0

1

2

3

4

5

6

BP BC BG BP BC BG

Males Females

Adju

sted

odd

s ra

tio

Without a regular care provider With a regular care provider

Figure 5.2 Relationships between a regular care provider and the use of preventive

health services among males and females

In summary, from the 22 preceding examinations of the relations between those

intervening variables and the use of preventive health services for male and female

groups, the main results can be highlighted based on both trend and significance

level as follows:

• The more concern about availability of health care, the more use of

preventive health services among males, while some of the results suggest

that females who are more concerned are less likely to use such services.

• Those more concerned about accessibility to health care are generally less

likely to use the preventive health services.

• Those more concerned about transport and travel time are less likely to use

the preventive health services.

• Those perceiving interpersonal care from general practitioners as important

are more likely to use the preventive health services. The relationship

between the scale and blood glucose check-up is statistically significant.

• Those with more positive perceptions of the value of general practitioners are

more likely to use preventive health services.

188

• Those with more positive attitudes towards health care are more likely to use

preventive health services. In particular, two significant results appear

between the high score and blood cholesterol and blood glucose check-ups

among females.

• Those among the middle-score female group who place great value on good

health are slightly likely to use preventive health services, except for blood

pressure check-up. In contrast, those females placing great value on good

health are more likely to use preventive health services.

• The results between negative aspects of social support and use of preventive

health services are not consistent. Of males, the results suggest that those

having a higher level of social support are more likely to go for a blood

cholesterol check-up, but not for a blood pressure check-up. However, the

associations are not significant. On the other hand, the results between

positive social support and the use shows that the respondents who are more

positive about their social support are more likely to have blood pressure,

blood cholesterol and blood glucose check-ups (except for blood glucose

check-ups among females). Two borderline significances appear among some

groups. However, these results need to be interpreted with caution.

• The consistent results show that those having a regular place of care are more

likely to use preventive health services. A significant result appears between

having a regular place of care and blood pressure check-up.

• The majority of the results also indicate that those having a regular care

provider are more likely to use preventive health services. A significant result

appears for blood glucose check-up and a borderline significance for blood

cholesterol check-up among females. However, the results need to be

interpreted with caution.

5.4 Modelling associations between SEP and the use of preventive health services adjusting for each of the intervening variables

This section discusses the final multivariable logistic regression modelling

relationships between SEP (education and income) and GP-based use of preventive

health services (BP, BC and BG check-ups), adjusted for age by the addition of each

189

individual intervening variable (IV) and, finally, inclusion of all intervening

variables (IVs) among male and female groups. This examines whether those

potential factors mediate the relationship between SEP and the use of preventive

health services. Only those intervening variables were adjusted in the final model if

they have relationships both with SEP and the use of preventive health services, as

discussed in Sections 5.2 and 5.3. Odds ratios and 95 per cent confidence intervals

(CIs) are presented. They are discussed below under headings for each of the SEP

indicators and each of the outcome variables among males and females.

5.4.1 The associations between education and use of GP based preventive health service adjustment for intervening variables among males

Based on the preceding analyses in Sections 5.2 and 5.3, four intervening variables

were identified, which might explain the relationship between education and GP-

based blood pressure, blood cholesterol and blood glucose checkups among males.

These four intervening variables include concern about accessibility to health care in

terms of being able to find a doctor who bulk bills, the cost of seeing a doctor and

having a number of doctors to choose from in the one medical practice/centre;

attitudes towards health care; social support and a regular place of care. They are

discussed under each of the outcome variables.


Table 5.32 examines the influence by the four intervening variables on the

relationship between education and blood pressure check-up among males. Model 1

presents the original relationship between education and blood pressure check-up

without adjusting any intervening variables. As can be seen in Model 1, the lower

education groups are slight less likely than the high education group to go for a blood

pressure check-up. After the inclusion of either each of the individual intervening

variables (Model 2–5) or all intervening variables (Model 6), the relationship

between education and blood pressure check-up is still the same as the original

model.

190

Table 5.32 Results of the final multivariable analysis examining the influence by the intervening factors on the relationship between education and blood pressure check-up

by GPs among males (n = 155)

Blood Pressure Check-up

Intervening

factors Model 1

OR 95%CI Model 2

OR 95%CI Model 3

OR 95%CI Model 4

OR 95%CI Model 5

OR 95%CI Model 6

OR 95%CI Education

High Mid

Low

1.00 0.92 0.41–2.05 0.92 0.35–2.41

1.00 0.97 0.39–1.99 0.94 0.22–1.43

1.00 0.91 0.39–2.09 0.92 0.35–2.41

1.00 0.95 0.42–2.14 0.96 0.37–2.63

1.00 0.89 0.39–2.04 0.91 0.34–2.46

1.00 0.93 0.39–2.26 0.98 0.34–2.86

Concern about accessibility to health care

1.02 0.89–1.16

1.05 0.92–1.19

Attitudes towards health care

0.99 0.89–1.11

0.98 0.88–1.10

Social support

1.04 0.95–1.14

1.07 0.97–1.19

A regular place of care

2.42 0.84–7.04

2.69 0.86–8.41

Notes: 1. Unadjusted Model: only adjusted for age in the model. 2. Adjusted for the scale measuring ‘Concern about accessibility to health care’. 3. Adjusted for the scale measuring ‘Attitudes towards health care’. 4. Adjusted for the scale measuring ‘Social support’. 5. Adjusted for the scale measuring ‘A regular place of caret’. 6. Adjusted all above intervening factors.



relationship between education and blood cholesterol check-up among males. As can

be seen in Model 1, the low education groups are nearly four-fold less likely than the

high education groups to go for a blood cholesterol check-up. After the inclusion of

either each of the intervening variables (Model 2–5) or all the intervening variables

(Model 6), the relationship between education and blood cholesterol check-up has no

change greater than 10 per cent.

191

Table 5.33 Results of the final multivariable analysis examining the influence by the intervening factors on the relationship between education and blood cholesterol check-

up by GPs among males (n = 155)

Blood Cholesterol Check-up

Intervening factors Model 1

OR 95%CI Model 2

OR 95%CI Model 3

OR 95%CI Model 4

OR 95%CI Model 5

OR 95%CI Model 6

OR 95%CI Education

High Mid

Low

1.00 0.69 0.30–1.62 0.26 0.09–0.71

1.00 0.73 0.30–1.77 0.25 0.09–0.74

1.00 0.72 0.30–1.69 0.27 0.09–0.72

1.00 0.72 0.31–1.69 0.27 0.10–0.75

1.00 0.64 0.27–1.52 0.27 0.09–0.75

1.00 0.71 0.28–1.81 0.28 0.09–0.86


1.05 0.91–1.19

1.05 0.92–1.21


1.02 0.91–1.13

1.03 0.91–1.16

Social support

1.04 0.95–1.14

1.06 0.95–1.17


1.22 0.40–3.70

1.62 0.48–5.49

Notes: 1. Unadjusted Model: only adjusted for age in the model. 2. Adjusted for the scale measuring ‘Concern about accessibility to health care’. 3. Adjusted for the scale measuring ‘Attitudes towards health care’. 4. Adjusted for the scale measuring ‘Social support’. 5. Adjusted for the scale measuring ‘A regular place of caret’. 6. Adjusted all above intervening factors. 5.4.1.3 Blood glucose check-up


relationship between education and blood glucose check-up among males. As can be

seen from Model 1, the low education group is as nearly twice less likely as the high

education group to have a blood glucose check-up. The relationship is still the same

after inclusion of each of the intervening variables in Models 2–5. However, there

are some changes in odds ratios among education groups (all changes in odds ratios

with more than 10 per cent are highlighted in the table) that occur after the inclusion

of all intervening variables (Model 6). The difference of blood glucose check-up

between low and high education groups declines.

192

Table 5.34 Results of the final multivariable analysis examining the influence by the intervening factors on the relationship between education and blood glucose check-up

by GPs among males (n = 155)

Notes: 1. Unadjusted Model: only adjusted for age in the model. 2. Adjusted for the scale measuring ‘Concern about accessibility to health care’. 3. Adjusted for the scale measuring ‘Attitudes towards health care’. 4. Adjusted for the scale measuring ‘Social support’. 5. Adjusted for the scale measuring ‘A regular place of caret’. 6. Adjusted all above intervening factors. 5.4.2 The associations between education and use of GP based preventive

health service adjustment for intervening variables among females Based on the preceding analyses in Sections 5.2 and 5.3, four intervening variables

were identified, which might explain the relationship between education and GP-

based blood pressure, blood cholesterol and blood glucose check-ups among females.

These variables include concern about accessibility to health care, concern about

transport and travel time to health care, attitudes towards health care and social

support. They are discussed under each of the outcome variables.



relationship between education and blood pressure check-up among females. As can

be seen in Model 1, the lower education groups are one and half or two times more

likely to go for a blood pressure check-up compared to the high education group.

After the addition of the factor ‘concern about accessibility to health care’ (Model 2),

Blood Glucose Check-up


OR 95%CI Model 2

OR 95%CI Model 3

OR 95%CI Model 4

OR 95%CI Model 5

OR 95%CI Model 6

OR 95%CI Education

High Mid

Low

1.00 0.89 0.39–1.99 0.55 0.22–1.43

1.00 0.85 0.36–1.99 0.56 0.21–1.52

1.00 0.82 0.35–1.90 0.54 0.21–1.38

1.00 0.92 0.41–2.09 0.59 0.23–1.54

1.00 0.85 0.38–1.95 0.59 0.23–1.56

1.00 0.77 0.31–1.90 0.62 0.22–1.78


1.08 0.95–1.23

1.10 0.96–1.26


0.96 0.86–1.07

0.95 0.84–1.07

Social support

1.04 0.95–1.13

1.08 0.97–1.19


1.95 0.60–6.32

2.54 0.69–9.24

193

the differences of blood pressure check-up between the low and the high education

group or between the mid- and high-education groups decline by more than 10 per

cent. After inclusion of the factor ‘concern about transport and travel time to health

care’ (Model 3) or ‘attitudes towards health care’ (Model 4), the difference of blood

pressure check-up between the mid and high groups increases by more than 10 per

cent. No changes occur after inclusion of the factor ‘social support’ (Model 5). In

addition, there is a change in odds ratio greater than 10 per cent among low-

education groups that occurs after the inclusion of all intervening variables (Model 6).

The difference of blood pressure check-up between the low- and the high-education

groups declines.

Table 5.35 Results of the final multivariable analysis examining the influence by the intervening factors on the relationship between education and blood pressure check-up

by GPs among females (n = 226)


Intervening

factors Model 1

OR 95%CI Model 2

OR 95%CI Model 3

OR 95%CI Model 4

OR 95%CI Model 5

OR 95%CI Model 6

OR 95%CI Education

High Mid

Low

1.00 1.99 0.86–4.58 1.32 0.66–2.66

1.00 1.71 0.69–4.19 0.99 0.47–2.08

1.00 2.27 0.94–5.48 1.24 0.59–2.59

1.00 2.25 0.96–5.29 1.37 0.68–2.78

1.00 1.92 0.83–4.46 1.36 0.67–2.76

1.00 2.02 0.79–5.15 1.11 0.51–2.42


1.01 0.89–1.14

1.04 0.92–1.19

Concern about transport and travel time to health care

1.09 0.94–1.26

1.06 0.89–1.25


1.09 0.99–1.21

1.11 0.99–1.26

Social support

1.05 0.96–1.13

1.06 0.97-1.16

Notes: 1. Unadjusted Model: only adjusted for age in the model. 2. Adjusted for the scale measuring ‘Concern about accessibility to health care’; 3. Adjusted for the scale measuring ‘Concern about transport and travel time too health care’; 4. Adjusted for the scale measuring ‘Attitudes towards health care’; 5. Adjusted for the scale measuring ‘Social support’; 6. Adjusted all above intervening factors. 5.4.2.2 Blood cholesterol check-up Table 5.36 examines the influence by the four intervening variables on the

relationship between education and blood cholesterol check-up among females. As

can be seen in Model 1, the low-education group is more than one and a quarter

194

times less likely than the high-education group to have a blood cholesterol check-up,

and the mid-education group is 1.15 times more likely to have a blood cholesterol

check-up. There are some changes in odds ratios greater than 10 per cent that occur

after the inclusion of the individual factors: concern about accessibility to health care

(Model 2), concern about transport and travel time (Model 3), attitudes towards

health care (Model 4) and all intervening variables (Model 6). The differences

between the low- and the high-education groups for blood cholesterol check-up are

widened. In addition, the direction is changed between the mid- and the high-

education groups in terms of blood cholesterol check-up. The middle-education

group are about one and a quarter times less likely to go for a blood cholesterol

check-up compared to the high-education group.

Table 5.36 Results of the final multivariable analysis examining the influence by the intervening factors on the relationship between education and blood cholesterol check-

up by GPs among females (n = 226)


Intervening

factors Model 1 OR 95%CI

Model 2 OR 95%CI

Model 3 OR 95%CI

Model 4 OR 95%CI

Model 5 OR 95%CI

Model 6 OR 95%CI

Education High Mid

Low

1.00 1.15 0.54–2.44 0.76 0.39–1.50

1.00 1.05 0.47–2.37 0.67 0.33–1.41

1.00 1.04 0.48–2.67 0.68 0.33–1.39

1.00 1.29 0.60–2.79 0.76 0.39–1.54

1.00 1.14 0.54–2.42 0.77 0.39–1.52

1.00 0.89 0.38–2.11 0.64 0.29–1.37


0.97 0.87–1.08

1.02 0.89–1.17


0.95 0.83–1.09

0.89 0.76–1.05


1.11 1.01–1.23

1.12 1.00–1.25

Social support

1.02 0.94–1.09

1.01 0.92–1.10

Notes: 1. Unadjusted Model: only adjusted for age in the model. 2. Adjusted for the scale measuring ‘Concern about accessibility to health care’. 3. Adjusted for the scale measuring ‘Concern about transport and travel time too health care’. 4. Adjusted for the scale measuring ‘Attitudes towards health care’. 5. Adjusted for the scale measuring ‘Social support’. 6. Adjusted all above intervening factors.

5.4.2.3 Blood glucose check-up


relationship between education and blood glucose check-up. It can be seen in Model

1 that the low-education groups are nearly one and a half times more likely than the

195

high education group to have a blood glucose check-up. However, there are some

changes in odds ratios greater than 10 per cent that occur after the inclusion of the

individual factor: concern about transport and travel time to health care (Model 3),

attitudes towards health care (Model 4), and spontaneously adjusting all intervening

variables (Model 6). The differences between the low- or middle- and the high-

education groups in terms of blood glucose check-up are widened.

Table 5.37 Results of the final multivariable analysis examining the influence by the intervening factors on the relationship between education and blood glucose check-up

by GPs among females (n = 226)


Intervening

factors Model 1

OR 95%CI Model 2

OR 95%CI Model 3

OR 95%CI Model 4

OR 95%CI Model 5

OR 95%CI Model 6

OR 95%CI Education

High Mid

Low

1.00 1.17 0.56–2.44 0.73 0.38–1.42

1.00 1.23 0.56–2.71 0.69 0.34–1.39

1.00 1.19 0.56–2.58 0.61 0.30–1.23

1.00 1.31 0.61–2.79 0.73 0.37–1.43

1.00 1.18 0.56–2.48 0.72 0.37–1.40

1.00 1.32 0.58–3.02 0.59 0.28–1.22


0.96 0.86–1.07

0.99 0.89–1.13


1.02 0.89–1.16

0.98 0.84–1.15


1.11 1.01–1.23

1.12 1.01–1.25

Social support

0.98 0.91–1.06

0.98 0.90–1.06

Notes: 1. Unadjusted Model: only adjusted for age in the model. 2. Adjusted for the scale measuring ‘Concern about accessibility to health care’. 3. Adjusted for the scale measuring ‘Concern about transport and travel time too health care’. 4. Adjusted for the scale measuring ‘Attitudes towards health care’. 5. Adjusted for the scale measuring ‘Social support’. 6. Adjusted all above intervening factors. 5.4.3 The associations between income and use of GP based preventive health

service adjustment for intervening variables among males

Based on the preceding analyses in Sections 5.2 and 5.3, four intervening variables

were identified, including concern about transport and travel time to health care,

social support, a regular place of care and a regular care provider, which might

account for some of the variations on GP-based use of preventive health services by

196

male respondents with different levels of income. They are discussed under each of

the outcome variables.


Table 5.38 examines the influence of the four intervening variables on the

relationship between family income and blood pressure check-up among males. As

can be seen from Model 1, the people in the low-income group are one and a half

times less likely to have their blood pressure checked than the high-income group.

After inclusion of the factor, ‘Concern about transport and travel time to health care’

(Model 2), the difference between these two groups in terms of blood pressure

check-up no longer exists. In addition, the difference between these two groups after

inclusion of the factor, ‘A regular place of care’, is also narrowed (Model 4). There

are no changes in odds ratios that occur after inclusion of the factor, ‘Social support’,

(Model 3). Furthermore, changes in odds ratios greater than 10 per cent occur

spontaneously after the addition of all intervening variables (Model 6). The

relationship between education and blood pressure check-up is reversed. Both the

low- and middle-education groups are more likely than the high-education group to

have a blood pressure check-up.

197

Table 5.38 Results of the final multivariable analysis examining the influence by the intervening factors on the relationship between family income and blood pressure

check-up by GPs among males (n = 155)



OR 95%CI Model 2

OR 95%CI Model 3

OR 95%CI Model 4

OR 95%CI Model 5

OR 95%CI Model 6

OR 95%CI Family income

High Mid

Low

1.00 1.27 0.40–4.00 0.68 0.19–2.36

1.00 1.77 0.49–6.29 0.99 0.23–4.28

1.00 1.29 0.40–4.09 0.68 0.19–2.39

1.00 1.36 0.41–4.49 0.81 0.22–3.01

1.00 1.44 0.45–4.59 0.67 0.19–2.43

1.00 2.42 0.61–9.51 1.31 0.23–7.50


0.91 0.75–1.11

0.95 0.76–1.18

Social support

1.01 0.91–1.12

1.04 0.92–1.18


2.99 0.93–9.63

10.65 1.76–64.59


0.85 0.31–2.29

0.34 0.08–1.51

Notes: 1. Unadjusted Model: only adjusted for age in the model. 2. Adjusted for the scale measuring ‘Concern about transport and travel time’. 3. Adjusted for the scale measuring ‘Social support’. 4. Adjusted for the scale measuring ‘A regular place of care’. 5. Adjusted for the scale measuring ‘A regular care provider’. 6. Adjusted all above intervening factors. 5.4.3.2 Blood cholesterol check-up


relationship between family income and blood cholesterol check-ups among males.

As can be seen in Model 1, the lower the family income the less likely the person is

to have a blood cholesterol check-up without inclusion any intervening variables.

However, a number of changes in odds ratios greater than 10 per cent that are

highlighted in the table occur after addition of the factors, ‘Concern about transport

and travel time to health care (Model 2) and ‘A regular place of care’ (Model 4) and

all intervening variables (Model 6). The differences between the low- and the high-

or the middle- and the high-income groups in terms of blood cholesterol check-up are

reduced.

198

Table 5.39 Results of the final multivariable analysis examining the influence by the intervening factors on the relationship between family income and blood cholesterol



Intervening

factors Model 1

OR 95%CI Model 2

OR 95%CI Model 3

OR 95%CI Model 4

OR 95%CI Model 5

OR 95%CI Model 6


High Mid

Low

1.00 0.45 0.14–1.43 0.39 0.11–1.39

1.00 0.56 0.16–2.00 0.55 0.13–2.28

1.00 0.46 0.14–1.50 0.42 0.12–1.47

1.00 0.54 0.16–1.77 0.39 0.11–1.40

1.00 0.42 0.13–1.37 0.38 0.10–1.41

1.00 0.60 0.15–2.41 0.57 0.12–2.79

Concerned about transport and travel time to health care

0.91 0.74–1.11

0.92 0.73–1.16

Social support

1.05 0.95–1.16

1.07 0.95–1.22


1.34 0.39–4.49

1.47 0.27–8.09


1.79 0.65–4.96

2.14 0.58–7.84

Notes: 1. Unadjusted Model: only adjusted for age in the model. 2. Adjusted for the scale measuring ‘Concern about transport and travel time’. 3. Adjusted for the scale measuring ‘Social support’. 4. Adjusted for the scale measuring ‘A regular place of care’. 5. Adjusted for the scale measuring ‘A regular care provider’;. 6. Adjusted all above intervening factors. 5.4.3.3 Blood glucose check-up


relationship between family income and blood glucose check-up among males. As

can be seen in Model 1, the lower the income group the less likely to have a blood

glucose check-up. However, a number of changes occur after addition of the factors,

‘Concern about transport and travel time to health care’ (Model 2) and ‘A regular

place of care’ (Model 4). The differences among the education groups are reduced

and some relationships are reversed. There are no changes in odds ratios greater than

10 per cent that occur after adjusting the factors, ‘Social support’ (Model 3) and ‘A

regular care provider’ (Model 5). Furthermore, after inclusion of all intervening

variables spontaneously in the final model (Model 6), the relationships are reversed.

Both lower income groups are more likely to have a blood glucose check-up.

199

Table 5.40 Results of the final multivariable analysis examining the influence by the intervening factors on the relationship between family income and blood glucose



Intervening

factors Model 1

OR 95%CI Model 2

OR 95%CI Model 3

OR 95%CI Model 4

OR 95%CI Model 5

OR 95%CI Model 6


High Mid

Low

1.00 0.89 0.29–2.80 0.58 0.16–2.03

1.00 1.03 0.29–3.56 0.82 0.19–3.41

1.00 0.94 0.29–3.01 0.61 0.17–2.20

1.00 1.19 0.35–4.05 0.62 0.17–2.29

1.00 0.90 0.29–2.86 0.52 0.14–1.97

1.00 1.56 0.37–6.59 1.19 0.22–6.54


0.89 0.74–1.09

0.89 0.72–1.13

Social support

1.05 0.96–1.16

1.09 0.96–1.23


1.95 0.52–7.31

7.22 0.72–72.03


1.28 0.46–3.51 1.31 0.36–4.81

Notes: 1. Unadjusted Model: only adjusted for age in the model. 2. Adjusted for the scale measuring ‘Concern about transport and travel time’. 3. Adjusted for the scale measuring ‘Social support’. 4. Adjusted for the scale measuring ‘A regular place of care’. 5. Adjusted for the scale measuring ‘A regular care provider’. 6. Adjusted all above intervening factors. 5.4.4 The associations between income and use of GP-based preventive health

service adjustment for intervening variables among females

Based on the preceding analyses in Sections 5.2 and 5.3, five intervening variables

were identified, including concern about availability of health care, concern about

accessibility to health care, value of good health, social support, and a regular care

provider, which might explain the relationship between family income and GP-based

blood pressure, blood cholesterol and blood glucose check-ups among females. They

are discussed under each of the outcome variables.


Table 5.41 examines the influence of the five intervening variables on the

relationship between family income and blood pressure check-up among females. As

can be seen in Model 1, both the low- and middle-income groups are nearly twice as

more likely to go for a blood pressure check-up compared their high-income

counterparts. However, after inclusion of the intervening variables, ‘Concern about

accessibility to health’ (Model 3) or ‘A regular care provider’ (Model 6), a number of

200

changes in odds ratios greater than 10 per cent occur. The differences among income

groups in terms of blood pressure check-up are either reduced or increased. There are

no changes in odds ratios after adjusting for ‘Concern about availability of health

care’ (Model 2), or ‘Value of good health’ (Model 4) and or ‘social support’ (Model

5). Furthermore, after adjustment for all intervening variables combined, the middle-

income group is two and a half times more likely to have a blood pressure check-up

compared the high-income group. The difference between these two income groups

in terms of blood pressure check-up is increased.

Table 5.41 Results of the final multivariable analysis examining the influence by the intervening factors on the relationship between family income and blood pressure

check-up by GPs among females (n = 226)


Intervening factors

Model 1 OR 95%CI

Model 2 OR 95%CI

Model 3 OR 95%CI

Model 4 OR 95%CI

Model 5 OR 95%CI

Model 6 OR 95%CI

Model 7 OR 95%CI

Family income High Mid Low

1.00 1.86 0.77–4.46 1.79 0.61–5.19

1.00 1.78 0.74–4.31 1.75 0.59–5.17

1.00 2.04 0.79–5.22 1.52 0.49–4.62

1.00 1.86 0.77–4.47 1.78 0.60–5.24

1.00 1.84 0.77–4.44 1.94 0.66–5.73

1.00 2.31 0.87–6.11 2.09 0.65–6.80

1.00 2.45 0.82–7.32 1.69 0.47–6.12

Concern about availability of health care

1.01 0.89–1.15

0.91 0.77–1.09


0.98 0.85–1.12

1.04 0.88–1.22


1.00 0.88–1.14

0.98 0.84–1.14

Social support

1.05 0.96–1.15

1.09 0.97–1.22


1.76 0.73–4.27

1.92 0.71–5.18

Notes: 1. Unadjusted Model: only adjusted for age in the model. 2. Adjusted for the scale measuring ‘Concern about availability of health care’. 3. Adjusted for the scale measuring ‘Concern about accessibility to health care’. 4. Adjusted for the scale measuring ‘Value of good health. 5. Adjusted for the scale measuring ‘Social support’. 6. Adjusted for the scale measuring ‘A regular care provider’. 7. Adjusted all above intervening factors.



relationship between family income and blood cholesterol check-up among females.

As can be seen in Model 1, the low-income group is more than two-fold less likely

than the high-income group to go for a blood cholesterol check-up. However, after

201

inclusion of the individual intervening variables, ‘Concern about accessibility to

health care’ (See Model 3) or ‘Value of good health’ (Model 4) or ‘A regular care

provider’ (Model 6), the difference between these two income groups in terms of

blood cholesterol check-up is increased. Furthermore, the difference for this

relationship is further increased after adjustment for all intervening variables together

(see Model 7). The low-income group is nearly five-fold less likely than the high-

income group to have their blood cholesterol checked.

Table 5.42 Results of the final multivariable analysis examining the influence by the intervening factors on the relationship between family income and blood cholesterol

check-up by GPs among females (n = 226)


Intervening factors

Model 1 OR 95%CI

Model 2 OR 95%CI

Model 3 OR 95%CI

Model 4 OR 95%CI

Model 5 OR 95%CI

Model 6 OR 95%CI

Model 7 OR 95%CI


1.00 1.29 0.57–2.94 0.45 0.17–1.19

1.00 1.33 0.58–3.02 0.45 0.17–1.19

1.00 1.19 0.51–2.80 0.39 0.13–1.14

1.00 1.37 0.59–3.14 0.38 0.14–1.03

1.00 1.29 0.57–2.94 0.46 0.17–1.21

1.00 1.44 0.60–3.43 0.40 0.15–1.12

1.00 1.43 0.55–3.64 0.22 0.06–0.80


0.98 0.87–1.10

0.94 0.81–1.09


1.00 0.88–1.14

1.04 0.89–1.21


1.12 0.98–1.27

1.12 0.96–1.29

Social support

1.01 0.93–1.10

0.99 0.89–1.09


2.18 0.89–5.31

1.90 0.72–5.03

Notes: 1. Unadjusted Model: only adjusted for age in the model. 2. Adjusted for the scale measuring ‘Concern about availability of health care’. 3. Adjusted for the scale measuring ‘Concern about accessibility to health care’. 4. Adjusted for the scale measuring ‘Value of good health. 5. Adjusted for the scale measuring ‘Social support’. 6. Adjusted for the scale measuring ‘A regular care provider’. 7. Adjusted all above intervening factors. 5.4.4.3 Blood glucose check-up


relationship between family income and blood pressure check-up among females. As

can be seen in Model 1, the low-income group is nearly one and half times less likely

to go for a blood glucose check-up compared to the high-income group. However,

202

the difference between these two groups in terms of blood glucose check-up is

increased after adjustment for the intervening variables, ‘Value of good health’

(Model 4) or ‘A regular care provider’ (Model 6), individually. Furthermore, the

difference among the income groups in terms of blood glucose check-up is increased

again after adjustment for all the five intervening variables together (Model 7).

Table 5.43 Results of the final multivariable analysis examining the influence by the intervening factors on the relationship between family income and blood glucose check-

up by GPs among females (n = 226)

Notes: 1. Unadjusted Model: only adjusted for age in the model. 2. Adjusted for the scale measuring ‘Concern about availability of health care’. 3. Adjusted for the scale measuring ‘Concern about accessibility to health care’. 4. Adjusted for the scale measuring ‘Value of good health. 5. Adjusted for the scale measuring ‘Social support’. 6. Adjusted for the scale measuring ‘A regular care provider’. 7. Adjusted all above intervening factors.


Intervening factors

Model 1 OR 95%CI

Model 2 OR 95%CI

Model 3 OR 95%CI

Model 4 OR 95%CI

Model 5 OR 95%CI

Model 6 OR 95%CI

Model 7 OR 95%CI


1.00 1.89 0.85–4.22 0.71 0.28–1.79

1.00 1.99 0.89–4.47 0.66 0.26–1.72

1.00 1.79 0.78–4.12 0.70 0.26–1.91

1.00 1.98 0.88–4.45 0.62 0.24–1.61

1.00 1.91 0.86–4.27 0.66 0.26–1.69

1.00 2.17 0.91–5.17 0.67 0.25–1.76

1.00 2.42 0.96–6.12 0.42 0.13–1.38


0.95 0.84–1.06

0.94 0.81–1.09


0.96 0.85–1.08

0.98 0.84–1.13


1.09 0.97–1.24

1.11 0.96–1.29

Social support

0.96 0.88–1.04

0.92 0.83–1.03


2..73 1.14–6.57

2.78 1.05–7.373

203

5.5 Summary This chapter has examined the potential modifiable factors which might influence the

relationship between SEP and the use of preventive health services. In order to

examine these associations to answer the key research question, the following

examinations were undertaken. Section 5.2 examined the relationship between SEP

and each of the potential intervening variables, which indicated that some of the

potential intervening variables were to some extent statistically significant in relation

to socioeconomic position. Section 5.3 further examined the relationship between

each of the intervening variables and the dependent variables. This also indicated

some of the intervening variables were related to the use of preventive health

services. Section 5.4 examined whether the potential intervening variables, which

have relationships both with SEP and the use of preventive health services, mediate

the relationship between SEP (education and family income) and the GP-based use

of preventive health services (blood pressure, blood cholesterol and blood glucose

check-up) by comparing between an unadjusted model and an adjusted model for

each of the intervening variables, and also all intervening variables combined, among

males and females. Each of results is presented in each of the sections and the final

results will be discussed in Chapter 6.

204

6 DISCUSSION AND CONCLUSIONS

205

6.1 Introduction

It is now well established that socioeconomically disadvantaged groups experience

poorer health than the rest of the population. This relationship has been observed in

numerous studies over the past 50 years. Cardiovascular diseases and diabetes are

two of the most important health challenges facing Australia now and in the

foreseeable future. Substantial socioeconomic inequalities also exist in relation to

both CVD and diabetes. For example, CVD death rates among people from the most

disadvantaged areas of Australia were about twice as high as those for individuals

from the least disadvantaged areas in 2002 (Australian Institute of Health and

Welfare, 2002, 2006; Moon & Waters, 2006). Deaths with diabetes as the underlying

cause were 44 per cent higher in the lowest socioeconomic group of the Australian

population compared to the highest socioeconomic group (Australian Institute of

Health and Welfare, 2001; Australian Institute of Health and Welfare and National

Heart Foundation, 2004; Moon & Waters, 2006; O'Brien et al., 2006).

Prevention of CVD aims to reduce morbidity and mortality in people both with and

without previously diagnosed disease, by identifying and reducing the common

major risks of CVD (Access Economics, 2005; World Health Organization, 2002).

These efforts include the promotion of healthy eating and regular physical activity,

the reduction of salt and saturated fat intakes, giving up smoking, maintaining a

healthy weight, and reducing blood pressure and cholesterol levels. A wide variety of

public health initiatives have been undertaken to reduce premature mortality from

cardiovascular disease in Australia since the number of deaths peaked in the mid-

1960s. These initiatives have included population-based strategies involving the use

of mass media, legislation or voluntary agreements, as well as more individually

focused strategies that involve early detection and identification by screening of

elevated risk factors. People are also encouraged to see their GPs regularly for a

preventive health care assessment (Australian Institute of Health and Welfare and

National Heart Foundation, 2004).

Australians have almost no restrictions on choice of a primary care provider or

general practice, both of which are usually heavily subsidised through Medicare, the

universal health insurance’. However, SEP differences in use of heath services still

206

exist. There is limited research investigating the relationship between SEP and

preventive care in general practice in Australia. However, some research has

demonstrated that low-SEP groups are less likely to use preventive health services

than high-SEP groups. Why do low-SEP individuals use preventive health services

less? Results from an empirical review of studies on determinants in relation to the

use of preventive health services suggest that there are many factors influencing

people’s decisions to visit their GP for preventive care (Glanz & Rimer, 1994; Glanz

et al., 2002; Winett et al., 1989). These were related to individual-level factors (e.g.

attitudes and beliefs towards preventive health care), interpersonal-level factors (e.g.

regular source of care and perceived social support) and perceptions of

environmental or organisation-level and system-level factors (perceived interpersonal

care, availability of and accessibility to health care, including bulk billing).

The present study has described the relationship between socioeconomic position

(SEP) and utilisation of preventive health services in relation to cardiovascular

disease and diabetes; it has also examined the potential factors of uptake and

utilisation of GP-based preventive health services by different socioeconomic groups

among the adult working-aged population in Brisbane, Queensland, Australia.

This chapter considers the major research findings of this thesis, discusses the major

study strengths and limitations, and then considers the implications and

recommendations for policy and future research in this area.

6.2 Summary of findings 6.2.1 Examining the relationship between socioeconomic position and the use

of preventive health services Most of the results examining the relationship between SEP and the use of preventive

health services in general practice were suggestive of socioeconomically

disadvantaged people being less likely to utilise preventive health care for diabetes

and cardiovascular disease. For males in particular, the low-SEP groups recorded the

least use of preventive health services among the three education and income groups,

including blood cholesterol and blood glucose check-ups, while the high-SEP groups

207

recorded the greatest use of preventive health services. For females, while the results

suggested that low-SEP groups were less likely than the high-SEP groups to have

blood cholesterol and blood glucose check-ups, this was not the case for blood

pressure check-ups. General speaking, there was a similar pattern for education and

income and the use of preventive health services among both males and females.

6.2.2 Understanding the relationship between socioeconomic position and use of preventive health services

The findings that considered the mediating factors between SEP and the use of

preventive health services suggest that socioeconomically disadvantaged adults (as

measured by both level of education and income) are more concerned with transport

and travel time to health care, and about accessibility to health care in terms of being

able to find a GP who bulk bills, the cost of seeing a GP and having a choice of GP.

They are also less likely to have a regular place of care and social support. Such

results suggest that these factors are likely to result in their lesser use of preventive

health services than their high-SEP counterparts. In addition, the study findings

suggest that respondents with a low level of education are less likely to have positive

attitudes towards health care, and those from low-income families are less likely to

have a regular care provider than their high-SEP counterparts.

6.3 Discussion of findings 6.3.1 SEP and the use of preventive health services

A review of the results of both international and Australian studies in Chapter 2

indicated that use of preventive health services was generally correlated with SEP

(Breen & Kessler, 1994; Giles et al., 1993; Lorant et al., 2002; Makuc et al., 1989;

National Center for Health Statistics, 2004). This pattern was found in most studies,

irrespective of how SEP was measured or what types of preventive health services

were examined.

Although a number of the results from the present study are also consistent with this

finding, this relationship was not found for BP check-up. This might be due to the

208

manner in which BP screening is conducted by GPs, because it is generally easier to

perform a BP check than to test for elevated BC and BG. Moreover, they often

conduct a BP check-up are during a consultation for another ‘problem’. However, for

BC and BG check-ups, GPs have to refer a patient to a pathologist and it can take a

number of days to receive the result. Moreover, many campaigns have been

conducted over the last 30 years to encourage GPs to take patients BP more routinely

in order to prevent heart disease (Australian Institute of Health and Welfare and

National Heart Foundation, 2004).

The results also suggest that the SEP patterning of use of preventive health services

for males and females is different. Females were more likely than males to have their

BP, BC and BG checked than males. This might be the case because women are

more concerned about preventive health efforts than their male counterparts (Cleary,

Zaborski, & Ayanian, 2004; Young et al., 2001). For instance, women are more

likely to visit GPs for testing and to receive a prescription than men (Scott et al.,

1996). Men, in relation to health services use, are less likely to visit a GP, less likely

to seek preventive assistance and more likely to consider waiting for appointments a

waste of time (Henning, 2001). Given that women also take part in breast cancer

screening and Pap smear testing, this may also give them more opportunity to access

preventive health services than men (Shelley et al., 1994).

6.3.2 Factors influencing the use of preventive health services by different SEP

groups

6.3.2.1 Concern about transport and travel time to health care

The effect of individuals’ concerns about transport and travel time to health care on

the relationship between SEP and use of preventive health services shows seven

changes in odds ratio greater than 10 per cent that occurred among 24 examinations.

It is suggested that individuals’ concerns about transport and travel time to health

care explain some of the variations in the use of preventive health services by

different SEP groups.

The examination of SEP and the scale indicated that female respondents with a low

level of education were more concerned about transport and travel time to health care

209

than their higher education counterparts, and male respondents with a lower income

were concerned more about transport and travel time to access health services than

those with a high income. At the same time, most of the results from the examination

between the scale and the use of preventive health services were also evident, except

for the female respondents, who went for a BP check-up; the rest of the respondents,

who were more concerned about transport and travel time to access the health

services, were less likely to visit their GPs for preventive health check-ups. In brief,

individuals who were more concerned about transport and travel time to health care

were the most vulnerable people, and were less likely to use preventive health

services.

A number of studies have provided evidence that transport and travel time are factors

that influence an individual’s use of health services, especially preventive health

services (Andersen et al., 1975; Bentley, 2003; Dutton, 1978; Hyndman & Holman,

2001). For example, it has been found that not having a car and distance both have a

marked effect on consultation rates, and accessibility to health care seemed to be

correlated, along with SEP, with poorer service provision for disadvantaged groups.

Distance had the greatest barrier impact on the poor than on the more affluent

(Dutton, 1986). Dutton (1986) found that, among the poor, distance was significantly

associated with lower rates of children’s check-ups and weakly associated with lower

rates of initial visits by both children and adults. McCarthy (1999) also pointed out

that travel patterns are related to income in the United Kingdom; that is, people in

low-income families are more likely to use public transport or to walk or cycle, while

high-income families are more likely to drive a car. Although the distribution of

health services was generally equitable for different SEP areas in Australia

(Hyndman and Holman, 2001), this still involved distance and travel time to some

extent. The low-SEP groups were more concerned about transport and travel time to

health care, both of which lead to less use of preventive health services, and this may

be due to these groups’ lack of transportation, which affects their ability to utilise

preventive health services (Strickland & Strickland, 1996).

210

6.3.2.2 Concern about accessibility to health care

The effect of individuals’ concerns about the accessibility to health care on the

relationship between SEP (both measured by education and family income) and use

of preventive health services shows four changes in odds ratio greater than 10 per

cent that occurred among 24 examinations. The results of this study suggest that

being able to find a bulk-billing service, the cost of GP consultations and having a

choice of doctor explained some of the variations in the use of preventive health

services by different socioeconomic groups.

An examination of the relationship between SEP and the scale suggested that the

lower the education or income level, the higher the concern about accessibility of

health care among either males or females. However, the results of the examination

of the relationship between the scale and use of preventive health services indicated

that people who were more concerned about accessibility to health care were less

likely to use preventive health services. Australian studies indicated that cost is a

factor in determining whether people use preventive health services. For instance,

one of the Australian studies indicated that individuals in the low-SEP groups were

less likely to spend on out-of-pocket costs (co-payment) per consultation, leading to

a lesser use of health services (Young et al., 2001). Out-of-pocket costs may affect

how often consumers use health services, both curative and preventive. Although

Medicare provides a fixed rebate for services fees, there is no legislation restricting

the amount a doctor can charge for a service. General speaking, bulk billing is

considered to make health services delivered by GPs very accessible in the

Australian context. Some evidence suggested that older people and those living in

socioeconomically disadvantaged areas in Australia have higher bulk-billing rates

(McClelland, 1991). However, there has been a lack of data to assess the impact that

the direct cost has on use of health services. Certainly, a number of overseas studies

have demonstrated that medical costs and charges are the most significant deterrents

to use of preventive health services such as cancer screening, preventive check-ups

and initial illness visits; in particular, these deterrents have the greatest impact on the

poor (Dutton, 1986) because socioeconomically disadvantaged groups are more cost

constrained (Dutton, 1978) and they tend to respond to higher prices by cutting back

on preventive care (Dutton, 1986).

211

6.3.2.3 A regular source of care A regular place of care The results from the final modelling showed that six changes in odds ratio greater

than 10 per cent were found among 24 examinations between SEP and the use of

preventive health services, after adjustment for the variable measuring a regular

place of care. This is strongly suggestive of people having a regular place of care as a

mediating factor in explaining the variation in use of preventive health services by

different SEP groups in this study.

On the basis of the examination between SEP and a regular place of care, it is

suggested that the male respondents with a low level of education and income are

less likely than those with a high level of education and income to have a regular

place of care, while there is a reversed pattern of this association among females.

This may be due to the fact that females are more likely to use preventive health

services where they may be likely to visit the same general practice or medical centre

than men. In addition, evidence shows that both male and female respondents who

often visit the same medical centre are more likely to have BP, BC and BG check-

ups.

The evidence of this study also suggested that low-SEP groups generally are less

likely to have a regular place of care, which leads to a low use of preventive health

services. The finding of this study is consistent with other studies both overseas and

in Australia, as examined in Chapter 2, which suggested that despite being less likely

to report having a regular place of care, those socioeconomically disadvantaged

groups are less likely to use preventive health services (Carmichael & Williams,

1983; Corbie-Smith, Flagg, Doyle, & O'Brien, 2002; Hulka & Wheat, 1985; Shi et

al., 1999; Steven et al., 1999).


The results from the final modelling showed that six changes in odds ratio greater

than 10 per cent were found among 24 examinations between SEP and the use of

preventive health services, after adjustment for the variable measuring a regular care

212

provider. It is suggested that people having a regular care provider accounts for some

variations in the use of preventive health services by SEP groups.

The results of the examination between SEP and a regular care provider showed that

low-income respondents were much less likely to visit the same GP compared to

high-income respondents among both males and females. However, it seems no

trends appear among education groups in terms of regular care providers. Again,

evidence showed that, except for the male respondents who went for a BP check-up,

the respondents who often saw the same doctor were more likely to have BP, BC and

BG check-ups.

The findings of this study are evidence that low-income individuals are less likely to

have a regular care provider, and thus a low use of preventive health services.

Consistent with other studies, having a regular care provider is positively associated

with use of preventive health services (Andersen, 1995; Dunlop et al., 2000;

Shankar, 2000).

Why are low-SEP groups less likely to have a regular source of care? This could be

because of various environmental/organisational and system-level factors. For

example, one of the US studies indicated that low-income individuals were serviced

insufficiently and those services were under-funded. Australian studies also provided

evidence that people who had a low-SEP background were less likely to be able to

see the local GP at short notice or to access a local female GP in the evenings

(Hyndman & Holman, 2001; Strickland & Strickland, 1996). In addition, Furler et al.

(2002) claimed that people in disadvantaged areas visit GPs more often annually, but

are less likely to have a long consultation. A number of studies suggested that longer

consultations were associated with higher patient satisfaction (Berkestijn, Kastein,

Lodder, Melker, & Bartelink, 1999; Freeman, Horder, Howie, & Hungin, 2002;

Howie et al., 1999). Therefore, these factors undoubtedly, but to varying degrees,

influence low-SEP groups as to choice of specific health services that can be a

regular source of care.

213

6.3.2.4 Attitudes towards health care Similarly, the results from the final logistic modelling showed that four changes in

odds ratio greater than 10 per cent were found among the 24 examinations of the

relationships between SEP and the use of preventive health services, after adjustment

for the scale measuring attitudes towards health care. It is suggested that the factor of

attitudes towards health care has some contribution to the variations of the use of

preventive health services by different SEP groups.

The examination of the association between SEP and attitudes towards health care

showed that both male and female respondents with a lower (middle) level of

education were less positive towards health care than those with a higher level of

education. On the other hand, it also found that respondents with a positive attitude

towards health care were more likely to use preventive health services compared to

those respondents with negative attitudes. In particular, female respondents with a

positive attitude towards health care were about one and a half times more likely to

have blood cholesterol and blood glucose check-ups than those with negative

attitudes. As discussed in Chapter 2, socioeconomically disadvantaged groups are

more likely to have negative attitudes towards preventive health care compared to

socioeconomically advantaged groups (Kravits, 1975), which leads to a lesser use of

preventive health services (Dutton, 1978).

6.3.2.5 Social support There were six relative changes in odds ratio greater than 10 per cent that occurred

among 24 examinations, after adjustment for the social support scale in the

modelling relationship between SEP and the use of preventive health services. The

findings of this research consistently showed that, overall, the respondents with

lower levels of education and income had a lower level of social support than their

high counterparts, among both males and females. Most of the results between social

support and the use of preventive services indicate that the respondents who were

more positive about their social support were more likely to have BP, BC and BG

(except BG among females) check-ups.

214

Social relationship has increasingly been linked to health outcomes. A number of

studies show that lack of social support is linked to increased risk of various diseases,

as well as to length of life (Ryff, Singer, & Palmersheim, 2004). Social support is

good for health under all circumstances. Stansfeld (1999) reviewed a number of

studies and indicated that higher income and education were associated with larger

networks, more contact with network members and more voluntary associations, and

those adults in metropolitan Toronto with a higher socioeconomic status perceived

higher levels of social support. In addition, social support is an important positive

predictor of use of health services, as discussed in Chapter 2. An Australian study

also pointed out that the stronger the social support network, the higher the usage of

preventive health services (Jirojwong & Manderson, 2002; Korten et al., 1998). The

findings of this present study also support the literature that people who have more

positive social support are more likely to use preventive health services. It can be

concluded that social support has some contribution to the variation of use of

preventive health services by different SEP groups in this study.

6.4 Strengths and limitations

6.4.1 Study design and survey implementation Although it was appropriate to employ a cross-sectional study design in order to

describe use of preventive health services in general practice in Australia, by

different sociodemographic groups, such a study design cannot be used to establish

any causal relationships between the variables examined. More research using

prospective study designs is required.

Although the local electoral roll was used to obtain a random sample for a specific

age group (25–64 years) and detailed addresses for the mailed survey, the study

sampling frame tended to under-represent the low-SEP groups. From the results of

socio-demographic analysis in Chapter 4, it is seen that 30 per cent of the

respondents had no educational qualifications, whereas 40 per cent of the

respondents had higher educational qualifications, including bachelor degrees and

higher. The results also indicated that just above 10 per cent of the respondents were

in the lowest income category and over 50 per cent of the respondents were in the

highest income category. In addition, the results of analysing the study

representativeness in comparison with 2001 census Brisbane data showed that

215

respondents with a higher level of education were over-represented. This result is in

accordance with other Australian studies by Turrell and Najman (1995), and Burton

et al. (2004). It should be noted also that the present study excluded the respondents

who had CVD and diabetes conditions, as they were more likely to be the socially

disadvantaged (lower education and income level). This may result in a limited

power when undertaking model estimation across SEP groups. Thus, the study

results need to be interpreted with consideration of this limitation.

A mailed survey was employed in the present study, as this is a common data-

collection method used in survey research. This method of data collection is

relatively quick and more cost-effective than personal or telephone interviews.

People can complete surveys in the privacy of their own homes, thus minimising

social desirability and interview bias. Mailed surveys have been proven to be suitable

for receiving high response rates for sensitive topics such as family income (Bowling,

2002; De Leeuw, 2001; Dillman, 2000; Turrell, 2000; Van Campen et al., 1998). On

the other hand, mailed surveys are less suitable for complex issues and long

questionnaires; therefore, the questions and items had to be limited so as to achieve a

good response rate (Bowling, 2002). Further, item non-response, where data on

particular items are missing, is a common problem with mailed surveys rather than

interview studies (Van Campen et al., 1998). Nevertheless, every effort was made to

present and structure the Health Service Utilisation Questionnaire for this present

study as simply and straightforwardly as possible, to further facilitate a good

response rate, and to conduct pre-testing and pilot studies to prevent item non-

response (De Leeuw, 2001). In addition, efforts were also made through: (1) rigorous

implementation of Dillman’s mailed survey strategies (Dillman, 2000) to the data

collection process; and (2) using a best-estimation technique for item non-response to

reduce missing items.

Implementing Dillman’s mailed survey strategies (Dillman, 2000) for the data

collection achieved an acceptable response rate of 65.6 per cent. A recent mailed

survey study by Burton et al. (2005), which was also conducted in Brisbane,

achieved a lower response rate of 56.7 per cent. However, another study employing a

similar survey methodology, conducted by Turrell (1997), achieved a higher

216

response rate of 80 per cent. Overall, the whole procedure of the mailed survey was

generally satisfactory, allowing data analysis for this research program.

6.4.2 Study measurement Socioeconomic position (SEP) was measured on the basis of respondents’ education

level and family income. These two measures were chosen in order to avoid a

misleading picture of socioeconomic disparities in the utilisation of preventive health

services that can arise if only one SEP indicator is used.

As discussed in Chapter 3, socioeconomic position on the basis of education level is

appropriate in this study, because the relationship between education and use of

preventive health services is presumed to reflect a person’s ability to access and

interpret health-related information. Someone with a high level of education may

have broader knowledge about preventive care and health, and may be more likely to

take action to prevent diseases or improve their health status (Galobardes et al.,

2001). Family income is also an appropriate measure of SEP because it makes

available economic resources to allow access to health care services and it allows

affordability of higher levels of use of preventive health services (Dutton et al., 2005).

It should be noted, however, that family income has its disadvantages, such as

missing data. In this present study, nearly 20 per cent of the respondents were not

willing to answer this question or missed it out, so this excluded from the data

analysis.

Nevertheless, the patterns of use of preventive health services were generally

consistent for the two indicators of SEP. It should be noted, however, that the

independent effect of education and family income was not taken into account in this

present study owing to the limited sample size. This means that when the effect of

education on the use of preventive health services was analysed, the family income

was not adjusted. Conversely, when the effect of family income was analysed,

education was not adjusted. It is argued that the two indicators may measure different

pathways through which socioeconomic position can have an independent effect on

the use of preventive health services (Dutton et al., 2005; Galobardes et al., 2001).

217

In addition, it should be noted that the preventive health check-up (BP, BC and BG)

was captured those people who were self-motivated to visit their GP for preventive

health checkups, or who visited their GP for other reasons and then GP performed

opportunistic health checks. The latter may accounts for the majority of preventive

care in general practice. Importantly, it should be noted that this study cannot

identify which ways the people had their preventive health checkups at general

practice service. The study sample was a healthy general population after excluding

those people who had cardiovascular diseases and diabetes and some other chronic

conditions. The health check-up was not included those people who called to visit

their GPs.

Data were based on self-report which is a commonly used method of collecting data

about individuals’ health and risk-factor status (Newell, Girgis, Sanson-Fisher, &

Savolainen, 1999). Notwithstanding this, there are concerns about the validity and

reliability of such data, so it is imperative that their validity and reliability is

established. Newell et al. (1999) critically reviewed 66 studies based on self-reports

of health behaviours and risk factors relating to cancer and cardiovascular diseases in

the general population. They concluded that self-reported data consistently

underestimated the proportion of individuals considered ‘at-risk’, for example, from

smoking, being overweight, hypertension, diabetes, inadequate Pap testing,

mammographic screening and no clinical breast examination. In the present study,

respondents were required to recall GP visits in the previous 12 months, 2 years, 2–5

years and more than 5 years. Respondents may have had trouble remembering

exactly when they last had their check-ups (recall bias) or alternatively, they may

have lacked the knowledge of the screening tests to answer the question accurately

(reporting bias).

6.4.3 Data analysis

This present study addressed specific limitations in the current evidence base that

were identified in the previous literature review, and which claimed that age and

gender were found to be confounders of SEP and utilisation behaviour. This present

study also confirmed the finding that patterns of use of preventive health services

were different by age and gender from a preliminary analysis. It is indicated from the

218

preliminary results that the use of preventive health services increased significantly

when age increased. Importantly, most of the older respondents were more likely to

have a lower level of education and to be from low-income families. It also

suggested that the patterns of use between males and females are different. The data,

therefore, stratified by gender and age, was adjusted as a confounder in the data

modelling analysis.

However, given the limited funding for this research study, the sample size was

calculated based on an 80 per cent power. Additionally, gender separation was not

considered when the sample size was calculated. When the sample was separately

analysed by gender and each of the outcome variables, the statistical power was

limited. The associations examined in this study may occur by chance. In addition,

the data have excluded the people who had CVD and diabetes. The removal of those

people has removed those at highset risk (they were more likely to have CVD and

diabetes) among the low SEP groups who were more likely to visited GP for the

preventive health checkups. In this case, the study sample was biased with low SEP

groups. Therefore, the results need to be interpreted with understanding of the data

limitation.

6.4.4 Generalisability of study findings

Notwithstanding all of the above limitations, the study findings are arguably quite

generalisable to adults living in most metropolitan areas of Australia for the

following reasons:

1. This study successfully implemented a mail survey with a good response rate

(65 per cent) from a general adult population in the Brisbane metropolitan

area.

2. The comparisons in terms of age and gender between the study sample with

the 2001 census for LGA of Brisbane showed that, overall, the distribution of

age group and gender in the study sample is not very divergent from the

population sample.

219

3. The percentage of people visiting their GP (85.0 per cent) in the previous 12

months in the present study is almost the same as that reported as the annual

average GP consultation rate (86.0 per cent) in Australia.

However, there are also some limitations to the generalisability of the study findings,

because:

1. The study sample was drawn from the electoral roll in the local government

area of Brisbane, using a method which has been shown to under-represent

socially disadvantaged individuals (Burton, Turrell, & Oldenburg, 2004;

Turrell & Najman, 1995).

2. The results from the representativeness analysis indicated that the

respondents with a higher level of education in the study sample were over-

represented, whereas the respondents with a lower level of education were

underrepresented when compared to the 2001 census of the local government

area of Brisbane.

3. A comparison with two other studies in terms of utilisation of health services

indicated a slight under-representation of participants who had received a BP

check-up. This is probably due to the exclusion of utilisation of other health

services in hospitals and specialists practices.

6.5 Study implications

6.5.1 Preventive health service utilisation in general practice The findings from the present study suggest that disadvantaged individuals make less

use of GP-based preventive health services in relation to cardiovascular diseases and

diabetes. Clearly, GPs have an important role in promoting preventive activities for

cardiovascular diseases and diabetes and reducing health inequalities, given the

majority of the Australians visit their GPs at least annually. For example, Koning et

al. (2004) conducted an audit study among general practitioners, and this indicated

that improving preventive care delivery in general practice could reduce the outcome

of stroke. In addition, both the OXCHECK Study and the Family Heart Study in the

United Kingdom, which were two large, randomised, controlled trials of

cardiovascular risk screening and intervention in general practice, found significant

220

reductions in blood pressure and cholesterol levels compared with controls, and 12

per cent overall cardiovascular risk reduction was achieved (Fowler, 2001).

Therefore, GPs need to address the needs of their socioeconomically disadvantaged

patients effectively in the prevention of CVD and diabetes. It is necessary to use

different strategies for different SEP patients, for example, spending more time

discussing preventive care with low-SEP patients (Lee, 1999), in particular,

conducting a needs assessment to identify the priority areas for preventive activities

for various target groups (Royal Australian College of General Practitioners, 1998).

6.5.2 Health policy and implications for strategies Most findings of the present study showed a positive relationship between SEP and

use of preventive health services. As discussed in Chapter 1, multi-level (upstream,

midstream and downstream levels) and multifaceted strategies are needed to reduce

socioeconomic health inequalities (Mackenbach & Stronks, 2004; Oldenburg et al.,

2000). Evidence-based intervention strategies for reducing socioeconomic health

inequalities are partly associated with changing social and economic policies,

empowering individuals and strengthening their social and family networks, and

improving the equity of the health care system (Oldenburg et al., 2000). It is most

important that relevant health care services should be made available to low-SEP

individuals in ways that enhance the availability and accessibility of preventive

health services. In order to improve equity of access in providing high-quality care

for socially disadvantaged groups and ensuring access to primary health care

appropriate to needs, socially disadvantaged groups need to be targeted (Harris et al.,

2004).

The findings of the present study indicate that low-SEP groups are less likely to use

preventive health services than high-SEP groups, owing to a lack of a regular source

of care and social support. The strategies of this intervention should be to implement

community development programs in disadvantaged areas that focus on creating

supportive community networks, with funding to invest in health services and to

keep the community informed about the availability of health services.

221

The present study also found that low-SEP groups were more concerned about

accessibility to health services in terms of being able to find a doctor who bulk bills,

the cost of seeing a doctor, having a number of doctors to choose from, as well as

transport and travel time to health care—which may possibly explain why the low-

SEP groups make less use of preventive health services. Policies should be pursued

to build up national health resources such as increasing GP bulk-billing services,

providing an economically, geographically and culturally accessible health care

system through improved transport systems with regular and frequent connections to

medical centres (McCarthy, 1999; Oldenburg et al., 2000; Wilkinson & Marmot,

2003). Thus, the health policies for transport should give the highest priority to

public transport, walking and cycling, given that low-SEP groups are concerned

more about transport and travel time to health care. In addition, direct subsidies of

public transport may be an efficient way of redressing income and access inequalities

(McCarthy, 1999).

Furthermore, the findings of this study showed that people’s attitudes towards health

care also explain some of the variations in the use of preventive health services by

different SEP groups. These factors also need to be considered when policy makers

and health providers promote preventive activities to reduce health inequalities. This

is not an easy task, because it involves ‘a gradual transformation of complex social,

political and economic relations through which inequalities are constituted’ (Short,

1999). A change in health providers’ attitudes towards their socioeconomically

disadvantaged patients and the provision of more social support to

socioeconomically disadvantaged individuals and communities may be a simple

starting point.

6.5.3 Models of health service utilisation

The present study reviewed current models and theories for understanding health

services utilisation. Most emphasis was given to Andersen’s model of health services

utilisation, which identifies the multiple influences on health services use and,

subsequently, on health outcomes. This model was initially proposed and empirically

tested in a series of studies to assist in understanding why families use health

services; to define and measure equitable access to health care; and to assist in

222

developing policies to promote equitable access. The Andersen model has

systematically characterised the array of predictors of health services utilisation, and

delineated the indicators of health services utilisation according to the type of service

such as hospital, physician and dentist, and the reason for use. It also specified the

hypothesised relationships between the predictors and indicators of utilisation. and

provided an integrated theoretical and empirical approach that has been widely

applicable to diverse populations at international, national and local levels, and

important health policy problems such as access and equity (Gochman, 1997).

Subsequently, Andersen’s model has been modified, expanded and applied to

examine the predictors of health service utilisation in relation to an array of health

problems and issues. Approximately ten years after Andersen’s original model,

Andersen and Newman (Gochman, 1997) formulated an expanded model which

added to and elaborated the components in order to be more responsive to societal

and policy changes affecting health care, as well as to more fully reflect the

increasing complexity of health care service delivery. The framework hypothesises

that societal determinants affect individual determinants, both directly and indirectly,

through the health services system. Individual determinants have the most immediate

influence on people’s decisions about the use of services.

Considering the multi-faceted factors influencing the use of preventive health

services and the multilevel influences on socioeconomic health inequalities, a

theoretical framework was developed which built on Andersen’s model. The

framework has been adapted to the Australian situation and used the proposed areas

of factors as intervening variables which are tested more formally in this research. It

guided this research to understand the relationship between SEP and use of

preventive health services in relation to CVD and diabetes in Australia. It provides

an overall picture of the relationship between SEP and preventive health services in

relation to the screenings for risk factors of CVD and diabetes in general practice,

which intervene by the four hierarchies of factors from the top to the bottom of the

triangle. This framework can also be widely employed to further understand this

relationship and other health services utilisation studies. However, its

appropriateness needs to be considered in terms of different health care system and

different study setting as well as different health services (hospital, specialist) when

223

applied. In addition, the measures to assess the factors influencing the relationship

between SEP and preventive health services at environmental- and system-level

factors are derived from the people’s perception. The direct measures at these levels

are needed. More factors at the different level that might influence on the

relationship are needed to be explored. More empirical studies are also required to

test this theoretical framework.

6.5.4 Methodological implications The research findings relate primarily to GP-based utilisation of preventive health

services and not to other health services such as hospital care, specialist medical

practitioners nor health promotion or prevention provide via other services or

programs. The results of this present study, therefore, only represent a partial

utilisation of preventive health services by the general adult population. Additionally,

since the factors that influence the use of preventive health services may differ

according to the type of services used, future research in different health services

may be needed to obtain a whole picture of the utilisation of preventive health

services, and to compare different utilisations of different health services, as well as

the salient factors for each of the services.

In order to capture the full range of health services utilisation behaviours of low-SEP

groups, a multi-method study needs to be considered, if possible. For example,

qualitative and quantitative approaches to data collection can be combined with other

methods, such as personal and telephone interview surveys, because different

methods of data collection often produce different results (Dillman, 2000; Van

Campen et al., 1998).

With regard to self-reported data, due to the lack of accuracy of respondents’ self-

reports, strategies need to be considered that can maximise the accuracy of self-

reported data. For example, these strategies include ensuring respondents understand

the questions posed by pre-testing questionnaires prior to a survey, and that the

questions have ‘clear, exhaustive, mutually exclusive’ response options during a

questionnaire design; phrasing questions with socially desirable responses, and using

224

‘bounded recall’ to improve respondents’ recall of less recent events. Researchers

also need to consider other data sources where possible (Newell et al., 1999).

In addition, the scales measuring people’s perceptions, attitudes and beliefs that were

assessed as factorially valid but with low internal reliability need to be further

developed. Given this present study used people’s perceived measures of

environmental and system-level factors, more direct measures to assess the factors

influencing the relationship between SEP and preventive health services at

environmental- and system-level factors are necessary.

6.6 Conclusions and recommendations for policy and future study

Three major conclusions can be drawn from the present study. First, some important

differences were found between SEP and use of preventive health services in relation

to CVD and diabetes among the adult population. Second, socioeconomically

disadvantaged adults are more affected by concerns about transport and travel time to

health care and concerns about accessibility to health care in terms of being able to

find a GP who bulk bills, the cost of seeing a GP and having a choice of GPs. They

are less likely to have a regular care provider and regular place of care, and have less

positive attitudes towards health care and less social support. These potential factors

account for some of the variations in GP-based use of preventive health services by

different SEP groups. Third, to a lesser extent, concerns about availability of health

care and how people value good health both have a slight association with this

relationship, whereas people’s perception of interpersonal care and the value of

general practitioners are unlikely to be associated with the relationship between SEP

and the use of preventive health services.

On the basis of the preceding discussion, the following is a summary of the key

policy recommendations. First, general practitioners need to pay more attention to

socioeconomically disadvantaged groups, such as providing longer consultation

times, addressing patient needs and actively offering information on preventive care.

Secondly, implementation of health promotion and education programs is needed in

disadvantaged areas to keep the community informed about the availability of health

services, and health services should be made more accessible. Thirdly, investment

225

should be made in physical assets such as increasing GP bulk-billing services and

providing an economically, geographically and culturally accessible health care

system through improved transport systems, with regular and frequent connections to

medical centres. Next, the health policies for transport should give the highest

priority to public transport, walking and cycling, and direct subsidies of public

transport may be an efficient way of redressing income and access inequalities.

Finally, improving access to a regular source of care may be an important step in

encouraging low-SEP individuals to initiate use of preventive health services in a

general practice setting.

Future research in different health services may be needed to obtain a whole picture

of utilisation of preventive health services. More empirical studies are required to test

the theoretical framework for understanding preventive health services use, with

consideration of its appropriateness in terms of different health care systems. More

direct measures to assess the factors influencing the relationship between SEP and

preventive health services at environmental- and system-level factors are needed. In

order to capture the full range of health service utilisation behaviours of low-SEP

groups, a multi-method study needs to be considered if possible.

226

APPENDICES

227

Appendix 2.1 BEACH study: This was a manuscript that was submitted to BMC: Journal of Health Services Research

253

Appendix 3.1 Request letter to Australian Electoral Commission

255

Appendix 3.2 List of names of experts for questionnaire review

Name Organisations

Professor Ron Andersen Scool of Public health, University of California, USA

Paul Magnus Australian Institute of Health and Welfare (AIHW)

Dr Anna Hawkes National Heart Foundation

Peter Abernethy National Heart Foundation

Helena Britt Family Medicine Research Centre,

AIHW General Practice Statistics and Classification Unit,

University of Sydney;

Prof. Mark Harris University of New South Wales

256

Appendix 3.3 Cover letter for pre-testing the questionnaire

257

Appendix 3.4 Ethical clearance for questionnaire pre-testing and mailed survey

258

Appendix 3.5 Health Service Utilisation Questionnaire

269

Appendix 3.6 Cover letter for the first mailing

270

Appendix 3.7 Thank you and reminder postcard

271

Appendix 3.8 Cover letter for the first replacement mailing

272

Appendix 3.9 Cover letter for the second replacement mailing

273

REFERENCES

274

Access Economics. (2005). The shifting burden of cardiovascular disease in Australia. Retrieved May 29, 2006, from http://www.heartfoundation. com.au/media/nhfa_shifting_burden_cvd_0505.pdf.

Aday, L. A., & Awe, W. C. (1997). Health services utilisation models. In D. S. Gochman (Ed.), Handbook of health behavior research I: Personal and social determinants. New York: Plenum Press.

Adler, N. E., Boyce, T., Chesney, M. A., Cohen, S., Folkman, S., Kahn, R. L., et al. (1994). Socioeconomic status and health: The challenge of the gradient. American Psychologist, 49(1), 15–24.

Amarenco, P., Labreuche, J., Lavallee, P., & Touboul, P.-J. (2004). Statins in Stroke Prevention and Carotid Atherosclerosis: Systematic Review and Up-to-Date Meta-Analysis. Stroke, 35(12), 2902-2909.

Andersen, R. (1968). A behavioral model of families' use of health services. Chicago: University of Chicago.

Andersen, R. (1995). Revisiting the behavioral model and access to medical care: Does it matter? Journal of Health and Social Behavior, 36(1), 1–10.

Andersen, R., Kravits, J., & Anderson, O. W. (1975). Equity in health services: Empirical analyses in social policy. Cambridge: Ballinger Publishing Company.

Armstrong, B. K., Rouse, I. L., & Butler, T. L. (1986). Cervical cytology in Western Australia. The Medical Journal of Australia, 144, 239–247.

Australian Bureau of Statistics. (1995). Australian social trends 1995: Population projections: Projections of the working age population. Retrieved November 10, 2005, from http://www.abs.gov.au/Ausstats/[email protected]/94713ad445ff1425ca25682000192af2/3dc3ec0170d0e0e7ca2569ee0015d899!OpenDocument.

Australian Bureau of Statistics. (2000). 2001 National health survey: Survey questionnaires: The adult questionnaire. Retrieved May 24, 2004, from http://www.abs.gov.au/Websitedbs/D3110122.NSF/4a255eef008309e44a255eef00061e57/2656b8f177b47ec4ca256c5d0002bb94/$FILE/typ2026adult.pdf.

Australian Institute of Health and Welfare. (1995). Risk factors for cardiovascular disease: A summary of Australian data. Canberra: Author.

Australian Institute of Health and Welfare. (2000). General practice in Australia: 2000. Canberra: Author.

Australian Institute of Health and Welfare. (2001). Heart, stroke and vascular diseases — Australian facts 2001. Cardiovascular disease series No. 14. AHIW Cat. No. CVD 13.Canberra: Author.

Australian Institute of Health and Welfare. (2002). Diabetes: Australian facts 2002. AIHW Cat. No. CVD 20. Canberra: Author.

Australian Institute of Health and Welfare. (2004). Australia's Health 2004. Canberra: Author.

Australian Institute of Health and Welfare. (2005). Cardiovascular disease. Retrieved July 26, 2005, from http://www.aihw.gov.au/cvd/index.cfm.

Australian Institute of Health and Welfare. (2006). Australia's health 2006. Canberra: Author.

Australian Institute of Health and Welfare and National Heart Foundation. (2004). Heart, stroke and vascular diseases: Australian facts 2004. Cardiovascular disease series No 22. AIHW Cat. No.CVD 27. Canberra: Author.

Australian Unity. (2005). Dental Care. Retrieved August 19, 2005, from http://www.australianunity.com.au/au/hser/dental/default.asp.

275

Barr, E. L. M., Magliano, D. J., Zimmet, P. Z., Polkinghorne, K. R., Atkins, R. C.,

Dunstan, D. W., et al. (2006). AusDiab 2005: The Australian diabetes, obesity and lifestyle study — tracking the accelerating epidemic: Its causes and outcomes. Melbourne: The International Diabetes Institute.

Barros, P. P., & Martinez–Giralt, X. (2002). Preventive health care and payment systems to providers. EcoPapers. Retrieved March 18, 2006, from http://econpapers.repec.org/scripts/redir.pl?u=http%3A%2F%2Fpareto.uab. es%2Fwp%2F2002%2F50702.pdf;h=repec:aub:autbar:507.02

Battistutta, D. (2003). Assessing adequacy of normality assumption: 2006 PUN 105 Health statistics study guide. Brisbane: Queensland University of Technology.

Bentley, J. M. (2003). Barriers to accessing health care: The perspective of elderly people within a village community. International Journal of Nursing Studies, 40, 9–21.

Berkestijn, L. G. V., Kastein, M. R., Lodder, A., Melker, R. A. D., & Bartelink, M.-L. (1999). Do better quality consultations result in better health? Relationship between quality of consultations and health status of patients with non-acute abdominal complaints in general practice. Family Practice, 16(6), 566.

Borrell, C., Rohlfs, I., Ferrando, J., Pasarin, M. I., Dominguez-Berjon, F., & Plasencia, A. (1999). Social inequalities in perceived health and the use of health services in a southern European urban area. International Journal of Health Services, 29(4), 743–764.

Borrell, C., Rohlfs, I., Ferrando, J., Pasarin, M. I., Dominguez–Berjon, F., & Plasencia, A. (2004). Social inequalities in perceived health and the use of health services in a southern European urban area. In Navarro, V & M. C (Eds.), Political and economic determinants of population health and well-being: Controversies and developments. New York: Baywood.

Bowling, A. (2002). Research methods in health: Investigating health and health services (2nd ed.). Berkshire: McGraw-Hill Education.

Boyce, J. (2002). Market research in practice. Sydney: McGraw-Hill Australia. Breen, N., & Kessler, L. (1994). Changes in the use of screening mammography:

Evidence from the 1987 and 1990 national health interview surveys. American Journal of Public Health, 84(1), 62.

Britt, H., Miller, G. C., Knox, S., Charles, J., Valenti, L., & Henderson, J. (2003). General practice activity in Australia 2002–03. General Practice Series No. 14. AIHW Cat. No. GEP 14. Canberra: Author.

Bullough, B. (1972). Poverty, ethnic identity and preventive health care. Journal of Health Social Behavior, 13(4), 347–359.

Burke, L. E., & Fair, J. (2003). Promoting prevention — skill sets and attributes of health care providers who deliver behavioral interventions. Journal of Cardiovascular Nursing, 18(4), 256–266.

Burton, N. W., Turrell, G., & Oldenburg, F. B. (2004). Item non-response in a population-based mail survey of physical activity. Journal of Physical Activity and Health, 1(4), 345–361.

Butterworth, P., & Crosier, T. (2004). The validity of the SF-36 in an Australian national household survey: Demonstrating the applicability of the household income and labour dynamics in Australia (HILDA) survey to examination of health inequalities. BMC Public Health, 4( 44).

276

Campbell, S. M., & Roland, M. O. (1996). Why do people consult the doctor? Family Practice, 13(1), 75–83.

Carmichael, A., & Williams, H. E. (1983). Use of health care services for an infant population in a poor socio-economic status, multi-ethnic municipality in Melbourne. Australian Paediatric Journal, 19, 225–229.

Centers for Disease Control and Prevention. (2002). State-specific trends in self-reported blood pressure screening and high blood pressure — United States, 1991–1999. MMWR, 51(21), 456–460.

Centers for Disease Control and Prevention. (2004). Surveillance for certain health behaviours among selected local areas — United States, behavioral risk factor surveillance system, 2002.

Centers for Disease Control and Prevention. (2005a). Breast cancer screening and socioeconomic status — 35 metropolitan areas, 2000 and 2002. MMWR, 54(39), 981–985.

Centers for Disease Control and Prevention. (2005b). Health disparities experienced by Black or African Americans — United States. MMWR, 54(01), 1–3.

Cleary, P., & Jette, A. M. (1984). The validity of self-reposted physician utilisation measures. Medical Care, 22(9), 796–803.

Cleary, P. D., Zaborski, L. B., & Ayanian, J. Z. (2004). Sex differences in health over the course of midlife. In O. G. Brim, C. D. Ryff & R. C. Kessler (Eds.), How healthy are we?: A national study of well-being at midlife. Chicago: The University of Chicago.

Cohen, D. R., & Henderson, J. B. (1988). Health, prevention, and economics. Oxford & New York: Oxford University Press.

Cohn, T. A., & Sernyak, M. J. (2006). Metabolic monitoring for patients treated with antipsychotic medications. Canadian Journal Of Psychiatry. Revue Canadienne De Psychiatrie, 51(8), 492-501.

Collins, R., Peto, R., MacMahon, S., Hebert, P., Fiebach, N. H., Eberlein, K. A., et al. (1990). Blood pressure, stroke, and coronary heart disease. Part 2, Short-term reductions in blood pressure: Overview of randomised drug trials in their epidemiological context. Lancet, 335(8693), 827–838.

Corbie-Smith, G., Flagg, E. W., Doyle, J. P., & O'Brien, M. A. (2002). Influence of usual source of care on differences by race/ethnicity in receipt of preventive services. Journal General Internal Medicine, 17(June), 458–464.

Cullum, D. E., & Savory, J. N. (1983). Patient preferences for cervical cytology. British Medical Journal, 287, 329–332.

Davies, M. J., Tringham, J. R., Troughton, J., & Khunti, K. K. (2004). Prevention of Type 2 diabetes mellitus. A review of the evidence and its application in a UK setting. Diabetic Medicine: A Journal Of The British Diabetic Association, 21(5), 403-414.

De Leeuw, E. D. (2001). Reducing missing data in surveys: An overview of methods. Quality and Quantity, 35, 147–160.

Delamonica, E., Minujin, A., & Gulais, J. (2005). Monitoring equity in immunisation coverage. Bulletin of the World Health Organization, 83 (5), 384391.

Delaney, L., & Fuller, C. (2004). General practice/primary care agencies systems integration: A patient-centred approach. Australian Journal of Primary Health, 10(2), 112–120.

Dengler, R., Roberts, H., & Rushton, L. (1997). Lifestyle surveys — the complete answer? Journal of Epidemiology and Community Health, 51(1), 46–51.

277

Department of Health and Ageing. (2005). Health care in Australia. Retrieved May 5, 2005, from http://www.dfat.gov.au/facts/health_care.html.

Dillman, D. A. (2000). Mail and Internet surveys: The tailored design method (2nd Ed.). New York: John Wiley and Sons.

Draper, G., Turrell, G., & Oldenburg, B. F. (2004). Health inequalities in Australia: Mortality. Canberra: Queensland University of Technology and the Australian Institute of Health and Welfare.

Duckett, S. (2000). The Australian health care system. Melbourne: Oxford University Press.

Dunlop, S., Coyte, P. C., & McIsaac, W. (2000). Socio-economic status and the utilisation of physicians' services: Results from the Canadian National Population Health Survey. Social Science & Medicine, 51, 123–133.

Dunstan, D., Zimmet, P., Welborn, T., Sicree, R., Armstrong, T., Atkins, R., et al. (2001). Diabesity and associated disorders in Australia — 2000: The accelerating epidemic. Melbourne: The International Diabetes Institute.

Dutton, D. B. (1986). Financial, organization and professional factors affecting health care utilization. Social Science & Medicine, 23(7), 721–735.

Dutton, D. B. (1978). Explaining the low use of health services by the poor: Costs, attitudes, or delivery systems? American Sociological Review, 43 (June), 348–368.

Dutton, T., Turrell, G., & Oldenburg, B. F. (2005). Measuring socioeconomic position in population health monitoring and health research. Brisbane, Australia: Queensland University of Technology.

Feig, D. S., Palda, V. A., & Lipscombe, L. (2005). Canadian task force on preventive health care: Screening for type 2 diabetes mellitus to prevent vascular complications: Updated recommendations from the Canadian task force on preventive health care. CMAJ, 172(2), 177–180.

Fowler, G. (2001). Commentary: GP 'check-ups' still of limited value. International Journal of Epidemiology, 30, 942–943.

Freeman, G. K., Horder, J. P., Howie, J. G. R., & Hungin, A. P. (2002). Evolving general practice consultation in Britain: Issues of length and context. British Medical Journal, 324(7342), 880.

Furler, J., Mihalopoulos, C., Lee, P., Harris, L., Powell-Davies, G., Naccarella, L., et al. (1998). Action on health inequalities through general practice. [Discussion paper]. Melbourne: Integration and Access SERU and Centre for Health Equity Training , Research and Evaluation.

Furler, J. S., Harris, E., Chondros, P., Davies, P. W. P., Harris, M. F., & Young, D. Y. L. (2002). The inverse care law revisited: Impact of disadvantaged location on accessing longer GP consultation times. Medical Journal of Australia, 177, 80–83.

Galobardes, B., Morabia, A., & Bernstein, M. S. (2001). Diet and socioeconomic position: Does the use of different indicators matter? International Journal of Epidemiology, 30, 334–340.

General Practice Statistics and Classification Unit, T. U. O. S. (2003). Bettering the Evaluation and Care of Health (BEACH) 2001–2002. Australian Family Physician, Jan–Feb; 32( (1–2)), 59–63.

Giles, W. H., Anda, R. F., Jones, D. H., Serdula, M. K., Merritt, R. K., & DeStefano, F. (1993). Recent trends in the identification and treatment of high blood cholesterol by physicians: Progress and missed opportunities. JAMA, 269(9), 1133–1138.

278

Glanz, K., & Rimer, B. K. (1994). Theory at a glance: A guide for health promotion practice. Bethesda, MD.: National Cancer Institute.

Glanz, K., Rimer, B. K., & Lewis, F. M. (2002). Health behavior and health education. San Francisco: Jossey-Bass.

Gochman, D. S. (1997). Handbook of health behavior research I. New York and London: Plenum Press.

Goddard, M., & Smith, P. (2001). Equity of access to health care services: Theory and evidence from the UK. Social Science & Medicine, 53, 1149–1162.

Grundy, E., & Sloggett, A. (2003). Health inequalities in the older population: The role of personal capital, social resources and socio-economic circumstances. Social Science & Medicine, 56(5), 935–947.

Halpin, S. J. (1995). Prevalence and frequency of health service use: Associations with occupational prestige and educational attainment. Australian Journal of Public Health, 19(5), 512–519.

Harris, E., Sainsbury, P., & Nutbeam, D. (1999). Perspectives on health inequity. Sydney: The Australian Centre for Health Promotion.

Harris, E., Traynor, V., Rose, V., Furler, J., Powell Davies, P. G., Harris, M. F., et al. (2000). Action on health inequalities through general practice: The role of divisions of general practice. Sydney: Centre for Health Equity Training Research and Evaluation.

Harris, M., & Furler, J. (2002). How can primary care increase equity in health? NSW Public Health Bulletin, 13(3), 35–38.

Harris, M., & Knowlden, S. (1999). Clinical perspective: A general practitioner response to health differentials. In E. Harris, P. Sainsbury & D. Nutbeam (Eds.), Perspectives on health inequity. Springfield: McMillan.

Harris, M., Harris, E., & Roland, M. (2004). Access to primary health care: Three challenges to equity. Australian Journal of Primary Health, 10(3), 21–29.

Harris, M., & Mercer, P. J. T. (2001). Reactive or preventive: the role of general practice in achieving a healthier Australia. Medical Journal of Australia, 175, 92–93.

Harvey, P. (2001). Preventive social health programs: Are they Australia's answer to rising health care costs in rural communities? Australian Journal of Rural Health, 9, 293–296.

Henning, D. (2001). Providing health care to men. Current Therapeutics, May, 11–14.

Herceg, A., Daley, C., Schubert, P., Hall, R., & Longbottom, H. (1995). A population-based survey of immunisation coverage in two-year-old children. Australian Journal of Public Health, 19(5), 465–470.

Hertzman, C., Frank, J., & Evans, R. G. (1994). Heterogeneities in health status and the determinants of population health. In R. G. Evans, M. L. Barer & T. R. Marmor (Eds.). Why are some people healthy and others not?: The determinants of health of populations. New York: Aldine de Gruyter.

Heywood, A., Ring, I., Sanson-Fisher, R., & Mudge, P. (1994). Screening for cardiovascular disease and risk reduction counselling behaviour of general practitioners. Preventive Medicine, 23, 292–301.

Hochbaum, G. M. (1956). Why people seek diagnostic x-rays. Public Health Report, 71(4), 377–380.

Hoover, D. R., Sambamoorthi, U., & Crystal, S. (2004). Characteristics of community dwelling elderly not vaccinated for pneumococcus in 1998 and 2001. Preventive Medicine, 39, 517–527.

279

Howie, J. G., Heaney, D. J., Maxwell, M., Walker, J. J., Freeman, G. K., & H. R. (1999). Quality at general practice consultations: Cross sectional survey. BMJ [Clinical Research Ed.] [NLM – MEDLINE], 319(7212), 738.

Hueston, W. J., & Hubbard, E. (2000). Preventive services for rural and urban African American adults. American Medical Association, 9, 263–266.

Hulka, B. S., & Wheat, J. R. (1985). Patterns of utilization: The patient perspective. Medical Care, 23(5), 438–460.

Hurley, S. F., Huggins, R. M., Jolley, D. J., & Reading, D. (1994). Recruitment activities and sociodemographic factors that predict attendance at a mammographic screening program. American Journal of Public Health, 84(10), 1655.

Hyndman, J. C. G., & Holman, C. D. A. J. (2001). Accessibility and spatial distribution of general practice services in an Australian city by levels of social disadvantage. Social Science & Medicine, 53, 1599–1609.

Janz, N. K., & Becker, M. H. (1984). The health belief model: A decade later. Health Education Quarterly, 11(1), 1–47.

Jerrett, M., Eyles, J., & Cole, D. (1998). Socioeconomic and environmeantal covariates of premature mortality in Ontario. Social Science & Medicine, 47(1), 33–49.

Jirojwong, S., & Manderson, L. (2002). Physical health and preventive health behaviors among Thai women in Brisbane, Australia. Health Care for Women International, 23, 197–206.

Jones, K., Fasher, B., Hanson, R., Burgess, M., Isaacs, D., Joshi, P., et al. (1992). Immunisation status of casualty attenders: Risk factors for non-compliance and attitudes to 'on the spot' immunisation. Journal of Paediatrics and Child Health, 28, 451–454.

Kaplan, G., & Keil, J. (1993). Socioeconomic factors and cardiovascular disease: A review of the literature. Circulation, 88(4), 1973–1998.

Kaplan, R. M. (2000). Two pathways to prevention. American Psychologist, 55(4), 382–396.

Katz, S. J., & Hofer, T. P. ( 1994). Socioeconomic disparities in preventive care persist despite universal coverage: Breast and cervical cancer screening in Ontario and the United States. JAMA, 272(7), 530–534.

Kenkel, D. S. (2000). Prevention. In A. A. N. Culyer, JP (Ed.), Health economics (Vol. 1). Amsterdam: Elsevier Science.

Koning, J. S. d., Klazinga, N. S., Koudstaal, P. J., Prins, A., Dippel, D. W. J., Heeringa, J., et al. (2004). Quality of care in stroke prevention: Results of an audit study among general practitioners. Preventive Medicine, 38, 129–136.

Korten, A. E., Jacomb, P. A., Jiao, Z., Christensen, H., Jorm, A. F., Henderson, A. S., et al. (1998). Predictors of GP service use: A community survey of an elderly Australian sample. Australian and New Zealand Journal of Public Health, 22(5), 609–615.

Kravits, J. (1975). The relationship of attitudes to discretionary physician and dentist use by race and income. In R. Andersen, J. Kravits & O. W. Anderson (Eds.), Equity in health services: Empirical analyses in social policy. Cambridge: Ballinger.

Krieger, N., Williams, D. R., & Moss, N. E. (1997). Measuring social class in US public health research: Concepts, methodologies, and guidelines. Annual Review of Public Health, 18(1), 341–378.

280

Kunst, A. E., Bos, V., Lahelma, E., Bartley, M., Lissau, I., Regidor, E., et al. (2005). Trends in socioeconomic inequalities in self-assessed health in 10 European countries. International Journal of Epidemiology, 34, 295–305.

Langlie, J. K. (1977). Social network, health beliefs, and preventive health behavior. Journal of Health and Social Behavior, 18 (September), 244–260.

Last, J. M. (1995). A dictionary of epidemiology. Oxford: Oxford University Press. Lee, P., Harris, E., & Traynor, V. (1999). Addressing disadvantage in GP obstetric

shared care. Sydney: CHETRE. Lenthe, F. J. V., Schrijvers, C. T. M., Droomers, M., Joung, I. M. A., Louwman, M.

J., & Mackenbach, J. P. (2004). Investigating explanations of socioeconomic inequalities in health. European Journal of Public Health, 14, 63–70.

Littenberg, B., Garber, A. M., & Sox, H. C. (1990). Screening for hypertension. Annals of Internal Medicine, 112(3), 192–202.

Lorant, V., Boland, B., Humblet, P., & Deliege, D. (2002). Equity in prevention and health care. Journal of Epidemiology Community Health, 56(7), 510–516.

Loxton, D., Schofield, M., & Hussain, R. (2004). History of domestic violence and health service use among mid-aged Australian women. Australian and New Zealand Journal of Public Health, 28(4), 383–388.

Lynch, J., & Kaplan, G. (2000). Socioeconomic position. In L. Berkman & I. Kawachi (Eds.), Social epidemiology. New York: Oxford University Press.

Mackenbach, J. P., Martikainen, P., Looman, C. W., Dalstra, J. A., Kunst, A. E., Lahelma, E., et al. (2005). The shape of the relationship between income and self-assessed health: an international study. International Journal of Epidemiology, 34, 286–293.

Mackenbach, J. P., & Stronks, K. (2004). The development of a strategy for tackling health inequalities in the Netherlands. International Journal for Equity in Health, 3(11), 1–7.

MacMahon, S., Neal, B., & Rodgers, A. (1995). Blood pressure lowering for the primary and secondary prevention of coronary and cerebrovascular disease. Schweiz Med Wochenschr, 125(51–52), 2479–2486.

Makuc, D. M., Freid, V. M., & Kleinman, J. C. (1989). National trends in the use of preventive health care by women. American Journal of Public Health, 79(1), 21.

Marmot, M. (2001). A social view of health and disease. In T. Heller, R. Muston, M. Sidell & C. Lloyd (Eds.), Working for health. London: SAGE Publications.

Marmot, M. (2002). The influence of income on health: Views of an epidemiologist. Health Affairs, 21(2), 31–46.

Marmot, M., & Bell, R. (2006). The socieconomically disadvantaged. In B. S. Levy & V. W. Sidel (Eds.). Social injustice and public health. Oxford: Oxford University Press.

Marmot, M. G., Smith, G. D., Stansfeld, S., Patel, C., North, F., Head, J., et al. (1991). Health inequalities among British civil servants: The Whitehall II study. The Lancet, 337(8), 1387–1393.

Martijn, H., Anton, K. E., & Mackenbach, J. P. (2003). Socioeconomic inequalities in morbidity among the elderly: A European overview. Social Science & Medicine, 57, 861–873.

Mathers, C. D. (1994). Health differentials among adult Australians aged 25–64 years. Canberra: Australian Government Publishing Service.

Mathers, C. D., & Schofield, D. J. (1998). The health consequences of unemployment: The evidence. Medical Journal of Australia, 168, 178–182.

281

Mathers, C. D., Vos, E. T., Stevenson, C. E., & Begg, S. J. (2000). The Australian burden of disease study: Measuring the loss of health from diseases, injuries and risk factors. Medical Journal of Australia, 172, 592–596.

Mathers, C. D., Vos, T., & Stevenson, C. (1999). The burden of disease and injury in Australia. Canberra: AIHW.

McCarthy, M. (1999). Transport and health. In M. Marmot & R. G. Wilkinson (Eds.), Social determinants of health. Oxford: Oxford University Press.

McClelland, A. (1991). Spending on health : The distribution of direct payments for health and medical services (Background paper No 7). Canberra: National Health Strategy.

McKinlay, J. B., & Marceau, L. D. (1999). A tale of 3 tails. American Journal of Public Health, 89(3), 295–298.

McKinlay, J. B., & Marceau, L. D. (2004). Upstream health public policy: Lessons from the battle of tobacco. In Navarro V & M. C (Eds.), Political and economic determinants of population health and well-being: Controversies and developments. New York: Baywood.

McLennan, W. (1997). Australian standard classification of occupations (ASCO) (2nd ed.). Canberra: Australian Bureau of Statistics.

McMichael, A. J. (1989). Coronary heart disease: Interplay between changing concepts of aetiology, risk distribution, and social strategies for prevention. Community Health Studies, 13(1), 5–13.

Moon, L., & Waters, A.-M. (2006). Socioeconomic inequalities in cardiovascular disease in Australia: Current picture and trends since the 1990s. Canberra: AIHW.

Morris, S. (2001). Economics and equity in the distribution of scarce health care resources. In T. Heller, R. Muston, M. Sidell & C. Lloyd (Eds.), Working for health. London: SAGE Publications.

National Center for Health Statistics. (2004). Health, United States, 2004 with chartbook on trends in the health of Americans. Hyattsville: Maryland.

National Health and Medical Research Council. (1996a). Guidelines for preventive interventions in primary health care: Cardiovascular disease and cancer — report of the assessment of preventive activities in the health care system initiative. Canberra: Australian Government Publishing Services.

National Health and Medical Research Council. (1996b). Prevention and improved management of cardiovascular disease. In B. Oldenburg (Ed.), Promoting the health of Australians: Case studies of achievements in improving the health of the population. (pp. 35–44). Canberra: The Australian Government Publishing Service.

National Heart Stroke and Vascular Health Strategies Group. (2004). National strategy for heart, strokeand vascular health in Australia: The Department of Health and Ageing.

Newell, S. A., Girgis, A., Sanson-Fisher, R. W., & Savolainen, N. J. (1999). The accuracy of self-reported health behaviors and risk factors relating to cancer and cardiovascular disease in the general population: A critical review. American Journal of Preventive Medicine, 17(3), 211–229.

Niederhauser, V. P., & Stark, M. (2005). Narrowing the gap in childhood immunisation disparities. Pediatric Nursing, 31 (5), 380–386.

Nunnally, J. C., & Bernstein, I. H. (1994). Psychometric theory. New York: McGraw-Hill Book Company.

282

Nutbeam, D., & Harris, E. (1999). Theory in a nutshell: A guide to health promotion theory. Sydney: The McGraw-Hill Book Company.

O'Brien, K., Thow, A., & Ofei, S. (2006). Diabetes hospitalisations in Australia, 2003–4. (Vol. no. 47). Canberra: AIHW.

Ockene, I. S. (1992). The rationale for intervention. In I. S. Ockene & J. K. Ockene (Eds.), Prevention of coronary heart diease. Boston: Little, Brown and Company.

Okoro, C. A., Strine, T. W., Young, S. L., Balluz, L. S., & Mokdad, A. H. (2005). Access to health care among older adults and receipt of preventive services: Results from the behavioral risk factor surveillance system, 2002. Preventive Medicine, 40, 337–343.

Oldenburg, B., & Burton, N. W. (2004). Primary prevention and health psychology: An ounce of prevention is worth a pound of cure. In A. A.Kaptein & J. Weinman (Eds.), An introduction to health psychology. Oxford: Blackwell Publishers.

Oldenburg, B., & Owen, N. (1995). Preventive care in general practice in Australia: A public health perspective. Patient Education and Counseling, 25, 305–310.

Oldenburg, B., McGuffog, I. D., & Turrell, G. (2000). Socioeconomic determinants of health in Australia: Policy responses and intervention options. Medical Journal of Australia, 172, 489–492.

Oppenheim, A. N. (1992). Questionnaire design, interviewing and attitude measurement. New York: Printer Publishers.

Orchard, T. J. (1992). Intervention for the prevention of coronary heart disease in diabetes. In I. S. Ockene & J. K. Ockene (Eds.), Prevention of coronary heart disease. Boston: Little, Brown and Company.

Pattison, A. (2000). Having a checkup. Retrieved August 17, 2005, from http://www.nevdgp.org.au/Pattison/Checkup/full.htm.

Pedersen, T., Kjekshus, J., Berg, K., & Haghfelt, T. (1994). Randomised trial of cholesterol lowering in 4444 patients with coronary heart disease: The Scandinavian Simvastatin Survival Study (4S). The Lancet, 344(8934), 1383.

Ramsay, J., Campbell, J., Schroter, S., Green, J., & Roland, M. (2000). The general practice assessment survey (GPAS): Tests of data quality and measurement properties. Family Practice, 17(5), 372–379.

Reidpath, D. D. (2004). Social determinants of health. In Helen Keleher & B. Murphy (Eds.), Understanding health: A determinants approach. New York: Oxford University Press.

Riessman, C. K. (1974). The use of health services by the poor. Social Policy, May/June, 41–49.

Rodriguez, G., Paymundo, P. M., Monsalve, M. H., & Gervas, J. (1996). Socioeconomic status, chronic morbidity and health service utilisation by families. Family Practice, 13(4), 382–385.

Rose, G. (1992). The strategy of preventive medicine. Oxford: Oxford University Press.

Rosenman, S. J., & Mackinnon, A. (1992). General practitioner services under Medicare. Australian Journal of Public Health, 16(4), 419–426.

Rosenstock, I. M. (1966). Why people use health services. Milbank Memorial Fund Quarterly, 44 (July, Part 2), 94–127.

Royal Australian College of General Practitioners. (1998). Putting prevention into practice: A guide for the implementation of prevention in the general practice setting. Melbourne: RACGP.

283

Royal Australian College of General Practitioners. (2002). Guidelines for preventive activities in general practice. Melbourne: Author.

Royal Australian College of General Practitioners. (2005). Guidelines for preventive activities in general practice. Melbourne: Author.

Ryff, C. D., Singer, B. H., & Palmersheim, K. A. (2004). Social inequalities in health and well-being: The role of relational and religious protective factors. In O. G. Brim, C. D. Ryff & R. C. Kessler (Eds.), How healthy are we?: A national study of well-being at midlife. Chicago: The University of Chicago.

Scott, A., Shiell, A., & King, M. (1996). Is general practitioner decision making associated with patient socio-economic status? Social Science & Medicine, 42(1), 35–46.

Shankar, C. (2000). Utilisation of routine medical services among immigrants from El Salvador. Journal of Health Care for the Poor and Underserved, 11(2), 125–134.

Shelley, J. M., Irwig, L. M., Simpson, J. M., & Macaskill, P. (1994). Who has pap smears in New South Wales? Patterns of screening across sociodemographic groups. Australian Journal of Public Health, 18(4), 406–411.

Shi, L., & Starfield, B. (2000). Primary care, income inequality, and self-rated health in the United States: A mixed-level analysis. International Journal of Health Services, 30(3), 541–555.

Shi, L., Starfield, B., & Kawachi, I. (1999). Income inequality, primary care, and health indicators. The Journal of Family Practice, 48(4), 275–284.

Short, S. (1999). Sociological perspective: Why do disadvantaged people behave so badly? In E. Harris, P. Sainsbury & D. Nutbeam (Eds.), Perspectives on health inequity. Springfield: McMillan.

Smeeth, L., & Health, I. (1999). Tackling health inequalities in primary care. British Medical Journal, 318, 1020–1021.

SPSS, I. (2004). The Statistical Package for the Social Science (SPSS) system file (12.01 edition): SPSS for Windows. Chicago.

Stachenko, S. (1994). Preventive guidelines: Their role in clinical prevention and health promotion. In Canadian task force on the periodic health examination: The Canadian guide to clinical preventive health care. Ottawa: Health Canada.

Standing, P., & Mercer, S. (1984). Quinquennial cervical smears: Every woman's right and every general practitioner's responsibility. British Medical Journal, 289, 883–886.

Stansfeld, S. A. (1999). Social support and social cohesion. In M. G. Marmot & R. G. Wilkinson (Eds.), Social determinants of health. New York: Oxford University Press.

Starfield, B. (1992). Primary care: Concept, evaluation and policy. New York: Oxford University Press.

Steven, I., Thomas, S., Eckerman, E., Browning, C., & Dickens, E. (1999). The provision of preventive care by general practitioners measured by patient completed questionnaires. Journal of Quality in Clinical Practice, 19, 195–201.

Strickland, J., & Strickland, D. L. (1996). Barriers to preventive health services for minority households in the rural south. The Journal of Rural Health, 12(3), 206–217.

284

Suarez, L. (1994). Pap smear and mammogram screening in Mexican–American women: The effects of acculturation. American Journal of Public Health, 84(5), 742–746.

Tabachnick, B. G., & Fidell, L. S. (2001). Using multivariate statistics. Needham Heights MA: Pearson Education Company.

The Australian Government Department of Health and Ageing. (2001). Cervical cancer: The National cervical screening program. Retrieved August 19, 2005, from http://www.cervicalscreen.health.gov.au/facts/index.html.

The Diabetes Prevention Program Research Group. (2003). Costs associated with the primary prevention of type 2 diabetes mellitus in the diabetes prevention program. Diabetes Care, 26(1), 36–46.

The Health Inequalities Research Collaboration (HIRC), P. H. C. N. (2004). Action on health inequalities: A primary health care research program. Canberra: Author.

Thorsen, H., Witt, K., Hollnagel, H., & Malterud, K. (2001). The purpose of the general practice consultation from the patient's perspective — theoretical aspects. Family Practice, 18(6), 638–643.

Turrell, G. (1997). Educational differences in dietary guideline food practices: Are they associated with educational differences in food and nutrition knowledge? Australian Journal of Nutrition and Dietetics, March; 54 (1), 25–33.

Turrell, G. (1998). Socioeconomic differences in food preference and their influence on health food purchasing choices. Journal of Human Nutrition and Dietetics, 11, 135–149.

Turrell, G. (2000). Income non-reporting: Implications for health inequalities research. Journal Epidemiology and Community Health, 54, 207–214.

Turrell, G., & Mathers, C. D. (2000). Socioeconomic status and health in Australia. Medical Journal of Australia, 172, 434–438.

Turrell, G., & Mathers, C. D. (2001). Socioeconomic inequalities in all cause and specific cause mortality in Australia: 1985–87 and 1995–97. International Journal of Epidemiology, 30, 231–239.

Turrell, G., & Najman, J. M. (1995). Collecting food-related data from low socioeconomic groups: How adequate are our current research designs? Australian Journal of Public Health, 19(4), 410–416.

Turrell, G., Oldenburg, B. F., McGuffog, I. D., & Dent, R. (1999). Socioeconomic determinants of health: Towards a national research program and a policy and intervention agenda. Canberra: Ausinfo.

Turrell, G., Patterson, C., Oldenburg, B. F., Gould, T., & Roy, M.-A. (2003). The socioeconomic patterning of survey participation and non-response error in a multilevel study of food purchasing behaviour: Area — and individual-level characteristics. Public Health Nutrition, 6, 179 –187.

Urban, N., Anderson, G. L., & Peacock, S. (1994). Mammography screening: How important is cost as a barrier to use? American Journal of Public Health, 84(1), 50.

US Preventive Services Task Force. (1996). Guide to clinical preventive services (2nd ed.). Baltimore: Williams & Wilkins.

US Preventive Services Task Force. (2000–2003). US Preventive Services Task Force Ratings: Strength of recommendations and quality of evidence. Guide to clinical preventive services. Retrieved August 16, 2005, from http://www.ahrq.gov/clinic/3rduspstf/ratings.htm.

285

Van Campen, C., Sixma, H., Kerssens, J. J., & Peters, L. (1998). Comparisons of the costs and quality of patient data collection by mail versus telephone versus in-person interviews. European Journal of Public Health, 8(66–70).

Van der Heyden, J. H. A., Demarest, S., Tafforeau, J., & Van Oyen, H. (2003). Socioeconomic differences in the utilisation of health services in Belgium. Health Policy, 65, 153–165.

Wen, M., Browning, C. R., & Cagney, K. A. (2003). Poverty, affluence, and income inequality: Neighborhood economic structure and its implications for health. Social Science & Medicine, 57, 843–860.

WHA. (2004). Women's Health Australia: Fourth survey for mid age women. Retrieved August 18, 2004, from http://www.newcastle.edu.au/centre/wha/ Surveys_data/Surveys/Mid4Survey.pdf.

While, A., George, C., & Murgatroyd, B. (2005). Promoting influenza vaccination in older people: Rationale and reality. British Journal of Community Nursing, 10 (9), 427–430.

Wiggers, J. H., & Sanson-Fisher, R. (1997). Practitioner provision of preventive care in general practice consultations: Association with patient educational and occupational status. Social Science & Medicine, 44(2), 137–146.

Wiggers, J. H., Sanson-Fisher, R. W., & Halpin, S. J. (1995). Prevalence and frequency of health service use: Associations with occupational prestige and educational attainment. Australian Journal of Public Health, 19(5), 512–519.

Wilkinson, R., & Marmot, M. (2003). Social determinants of health: The solid facts. Copenhagen, Denmark: WHO.

Wilkinson, R. G. (2001). Social status, inequality and health. In T. Heller, R. Muston, M. Sidell & C. Lloyd (Eds.), Working for health. London: SAGE Publications.

Williams, D. R., & Collins, C. (1995). US socioecnomic and racial differences in health: Patterns and explanations. Annual Review of Sociology, 21, 349–386.

Wilson, A. J., Oldenburg, B. F., & Lopez, A. D. (2003). Targeted approaches for reducing inequities in chronic disease. Medical Journal of Australia, 179, 231–232.

Windsor, R., Baranowski, T., Clark, N., & Cutter, G. (1994). Evaluation of health promotion, health education, and disease prevention programs. Mountain View: Mayfield.

Winett, R. A., King, A. C., & Altman, D. G. (1989). Health psychology and public health : An integrative approach. New York: Pergamon Press.

Winkleby, M. A., Kraemer, H. C., Ahn, D. K., & Varady, A. N. (1998). Ethnic and socioeconomic differences in cardiovascular disease risk factors: Findings for women from the Third National Health and Nutrition Examination Survey, 1988–1994. JAMA, 280(4), 356–362.

Wooden, M., & Watson, N. (2000). The household, income and labour dynamics in Australia (HILDA) survey: An introduction to the proposed survey design and plan. Melbourne: The Melbourne Institute of Applied Economic and Social Research, University of Melbourne.

World Health Organization. (1986). Ottawa Charter for health promotion. First International Conference on Health Promotion. Retrieved 16 August, 2004, from http://www.who.int/hpr/NPH/docs/ottawa_charter_hp.pdf.

World Health Organization. (2002). The World Health Report 2002: Reducing risks, promoting healthy life. Geneva: WHO.

286

World Health Organization. (2003a). International conference on primary health care, Alma-Ata: Twenty-fifth anniversary. Geneva: Author.

World Health Organization. (2003b). Primary health care: A framework for future strategic directions. Geneva: Author.

Young, A. F., Dobson, A. J., & Byles, J. E. (2000). Access and equity in the provision of general practitioner services for women in Australia. Australian and New Zealand Journal of Public Health, 24(5), 474–480.

Young, A. F., Dobson, A. J., & Byles, J. E. (2001). Determinants of general practioner use among women in Australia. Social Science & Medicine, 53, 1641–1651.

Yusuf, S., & Cutler, J. (1987). Single factor trials: Drug studies. In AG Olsson et al. (Eds), Atherosclerosis (1st. ed., pp. 393–398). New York: Churchill Livingstone.

Zorbas, H. M. (2003). Breast cancer screening. Medical Journal of Australia, 178(12), 651–652.

socioeconomic position and utilisation of preventive health … · 2010-06-09 · preventive health...

Documents