ARTICLE IN PRESS

0277-9536/$ - se

doi:10.1016/j.so

�Correspondfax: +441223 7

E-mail addr

Social Science & Medicine 61 (2005) 2460–2465

www.elsevier.com/locate/socscimed

Sociodemographic and attitudinal correlates of cervicalscreening uptake in a national sample of women in Britain

Stephen Sutton�, Celia Rutherford

Institute of Public Health, University of Cambridge, Forvie Site, Robinson Way, Cambridge CB2 2SR, UK

Available online 18 August 2005

Abstract

Sociodemographic and attitudinal correlates of self-reported cervical screening uptake were investigated among 1307

women in the target age group who participated in two national surveys conducted in Britain in 1999. Evidence for

inequalities in screening uptake was mixed. Of the socioeconomic indicators, only age of completed full-time education

showed a significant effect in the multivariate analysis. The strong effects of car ownership and housing tenure in the

univariate analyses were eliminated by controlling for marital status, which showed a robust association with uptake.

Uptake was highest among married and separated women and lowest among single and widowed women. There was no

evidence that the effects of marital status and education were mediated by the attitudinal variables. Anticipated

embarrassment and attitudes to screening (e.g., ‘‘There’s no point going for screening if you don’t have any symptoms’’)

were significant independent predictors of uptake. These findings suggest that information campaigns need to address

feelings of embarrassment and lack of understanding of the rationale for screening and that efforts should be made to

encourage single and widowed women to attend.

r 2005 Elsevier Ltd. All rights reserved.

Keywords: Cervical screening; Socioeconomic status; Marital status; UK

Introduction

Mortality figures for the UK show a sharp gradient

towards higher mortality from cervical cancer in women

of lower socioeconomic status (SES) (Quinn, Babb,

Brock, Kirby, & Jones, 2000). Socioeconomic differ-

ences in cervical screening uptake may be one factor

contributing to this gradient (Wardle et al., 1999), and it

is therefore important to investigate the extent of, and

reasons for, such differences.

In the UK, the relationship between SES and cervical

screening coverage or uptake has been examined at three

e front matter r 2005 Elsevier Ltd. All rights reserve

cscimed.2005.07.017

ing author. Tel.: +441223 330594;

62515.

ess: srs34@medschl.cam.ac.uk (S. Sutton).

different levels: health district (e.g., Baker & Middleton,

2003), general practice (e.g., Ibbotson, Wyke, McEwen,

Macintyre, & Kelly, 1996) and individual (e.g., Coulter

& Baldwin, 1987). This paper reports an individual level

study that differed from previous studies in examining a

range of indicators of SES, as well as other socio-

demographic variables. In addition, we included mea-

sures of women’s beliefs about cervical cancer and

cervical screening and attitudes towards screening and

health. The selection of variables was guided by the

Health Belief Model (Janz & Becker, 1984) and by

previous studies of cervical screening uptake (e.g., Bish,

Sutton, & Golombok, 2000; Orbell, Crombie, &

Johnston, 1996).

We hypothesised that any associations between socio-

demographic variables and uptake would be explained

d.

ARTICLE IN PRESSS. Sutton, C. Rutherford / Social Science & Medicine 61 (2005) 2460–2465 2461

at least in part by attitudes and beliefs. To our

knowledge, the present study is the first to estimate the

independent contributions of these different classes of

variables to cervical screening uptake in a national

sample of women in Britain.

Methods

Participants, design and procedure

Two similar surveys of adults aged 16 and over living

in private households in Britain were conducted in

March and May 1999 through the Office for National

Statistics. The response rate to both surveys was 70%.

There were 1469 women in the target age group for

cervical screening (20–64 in England and Wales; 20–60

in Scotland). Analysis was restricted to the 1307 women

who stated that they had not had a hysterectomy and

who said whether or not they had had a smear test in the

last 5 years.

Measures

(1)

Sociodemographic factors: age; marital status; re-

gion; income; age completed full-time education;

social class by occupation; employment status;

number of cars; housing tenure.

(2)

Attitudes and beliefs about cervical screening:

perceived effectiveness of cervical screening; per-

ceived risk of cervical cancer; worry about cervical

cancer; anticipated embarrassment and pain.

(3)

Attitudes to screening and health. Principal compo-

nents analysis of responses to nine statements

yielded three components. Component I reflected

negative attitudes to screening (e.g., ‘‘There’s no

point going for screening if you don’t have

symptoms’’); component II, attitudes to health

(‘‘It’s better to live from day to day than to worry

about what might happen in 10 years’ time’’); and

component III, optimistic outlook (‘‘Where my

health’s concerned, I’m always optimistic about my

future’’).

(4)

Whether or not the respondent had had a smear test

in the last 5 years (1 ¼ yes, 0 ¼ no). This was derived

from response to the question ‘‘When did you last

have a smear test. Was ity Less than 3 years ago,

3–5 years ago, 6–10 years ago, or have you never had

one?’’ Reported uptake was high (1200/1307; 92%).

Results

Five sociodemographic variables showed a significant

association with uptake in univariate logistic regression

analyses: age, marital status, age completed full-time

education, number of cars and housing tenure (Table 1).

Age showed a curvilinear effect. Of the attitude and

belief variables, perceived effectiveness, anticipated pain

and embarrassment, and the three attitudinal compo-

nents were significantly associated with uptake.

In a multiple logistic regression analysis of the

sociodemographic variables, only age, marital status

and education were significant predictors. The attenua-

tion of the effects of car access and housing tenure was

due largely to controlling for marital status. The three

significant demographic variables from this analysis

together with the attitude and belief variables that were

significant in the univariate analyses were entered in a

multiple logistic regression analysis (Table 1). The

significant predictors in this model were marital status,

age completed full-time education, perceived effective-

ness, anticipated embarrassment and the first attitudinal

component. Apart from the attenuation of the age

effect, there was little evidence that the effects of the

sociodemographic variables were mediated by the atti-

tudinal variables. It was possible to simplify the model

without significant loss of fit by dropping the third

attitudinal component and anticipated pain and repre-

senting anticipated embarrassment as a single degree of

freedom linear component (model w2 ¼ 104:83, df ¼ 18,

po0:0005).

Discussion

This study had a number of limitations. First, it used a

cross-sectional design. Second, although the response

rate was quite high (70%), we had no information about

non-responders. We therefore cannot rule out the

possibility of selection bias. Non-responders may be of

lower SES on average and may be less likely to attend for

screening. Third, the measure of behaviour was based on

self-report. It was not possible to verify self-reports

against medical records. Strengths of the study are that it

was based on a relatively large national sample and

included a range of socioeconomic indicators.

Evidence for inequalities in screening uptake was

mixed. Of the SES indicators, only education showed a

significant effect in the multivariate analyses. The strong

effects of car ownership and housing tenure in the

univariate analyses were eliminated by controlling for

marital status. By contrast, marital status itself showed a

robust association with screening uptake (cf. Orbell et

al., 1996). Marital status is related to sexual behaviours

that are well-established risk factors for cervical cancer,

in particular number of partners and age at first

intercourse (Brinton, 1992). Thus, the women who were

at lowest risk for cervical cancer (i.e., married women)

had the highest uptake of cervical screening.

ARTIC

LEIN

PRES

S

Table 1

Sociodemographic and attitudinal correlates of cervical screening uptake: summary of univariate and multivariate logistic regression analyses

Variable Categories Frequencies % Smear test Univariate analyses (N ¼ 1266� 1307) Multivariate analysis (N ¼ 1259)

p Odds ratio (95% CI) p Odds ratio (95% CI)

Sociodemographic variables

Age o0.0005 0.124

55–64 210 88.6 0.60 (0.34,1.06) 0.62 (0.28,1.34)

45–54 255 94.9 1.44 (0.73,2.83) 1.25 (0.53,2.94)

35–44 345 95.1 1.49 (0.80,2.77) 1.23 (0.61,2.49)

25–34 391 92.8 1.00 1.00

20–24 106 76.4 0.25*** (0.14,0.45) 0.50 (0.23,1.07)

Marital status o0.0005 0.007

Married 687 95.8 1.00 1.00

Cohabiting 125 90.4 0.42* (0.21,0.84) 0.48 (0.21,1.09)

Single 255 82.4 0.21*** (0.13,0.34) 0.40** (0.21,0.78)

Divorced 144 93.1 0.59 (0.28,1.24) 0.91 (0.37,2.25)

Separated 58 96.6 1.23 (0.29,5.31) 1.74 (0.29, 10.52)

Widowed 38 76.3 0.14*** (0.06,0.33) 0.20** (0.07,0.59)

Age completed full-time education 0.004 0.004

Over 25 65 95.4 1.57 (0.48,5.12) 0.58 (0.16,2.09)

19–25 267 88.4 0.58* (0.37,0.91) 0.40** (0.22,0.70)

15–18 949 92.9 1.00 1.00

Up to 14 26 76.9 0.25** (0.10,0.65) 0.27* (0.08,0.92)

Number of cars o0.0005Two or more 458 95.0 1.69* (1.01,2.83)

One 584 91.8 1.00

None 265 86.4 0.57* (0.36,0.90)

Housing tenure 0.001

Own outright 222 89.6 0.51* (0.30,0.88)

Owns mortgage 714 94.4 1.00

Rent (non-private) 245 90.2 0.55* (0.32,0.93)

Rent (private) 125 84.0 0.31*** (0.18,0.55)

Income 0.210

1st quintile 268 89.6 0.52 (0.26,1.03)

2nd quintile 233 92.3 0.73 (0.35,1.52)

3rd quintile 227 94.3 1.00

4th quintile 256 90.6 0.59 (0.29,1.18)

5th quintile 238 94.1 0.97 (0.45,2.12)

Other 85 88.2 0.46 (0.19,1.08)

S.

Su

tton

,C

.R

uth

erford

/S

ocia

lS

cience

&M

edicin

e6

1(

20

05

)2

46

0–

24

65

2462

ARTIC

LEIN

PRES

SSocial class 0.204

I 35 85.7 0.46 (0.17,1.25)

II 383 92.7 0.96 (0.57,1.63)

IIIN 454 93.0 1.00

IIIM 100 91.0 0.77 (0.35,1.66)

IV 237 91.1 0.78 (0.44,1.39)

V 54 92.6 0.95 (0.32,2.79)

Other 44 81.8 0.34* (0.15,0.80)

Employment status 0.421

In employment 855 92.5 1.00

Unemployed 46 89.1 0.66 (0.25,1.74)

Economically inactive 406 90.6 0.78 (0.52,1.19)

Region 0.742

The North 333 90.4 0.77 (0.45,1.32)

Midlands+East Anglia 345 92.5 1.00

London 157 91.1 0.83 (0.42,1.64)

South East 188 94.7 1.45 (0.68,3.07)

South West 118 90.7 0.79 (0.38,1.66)

Wales 54 92.6 1.02 (0.34,3.04)

Scotland 112 91.1 0.83 (0.39,1.78)

Attitudinal variables

Perceived effectiveness 0.005 0.048

Very effective 508 94.7 1.41 (0.86,2.29) 1.31 (0.76,2.26)

Quite effective 643 92.7 1.00 1.00

Not very effective 96 89.6 0.68 (0.33,1.39) 1.64 (0.67,3.99)

Not at all effective 19 73.7 0.22** (0.08,0.64) 0.24* (0.07,0.85)

Anticipated pain o0.0005 0.254

Not at all painful 624 95.0 1.00 1.00

A bit painful 514 91.4 0.56* (0.35,0.90) 0.59 (0.34,1.02)

Quite painful 111 87.4 0.36** (0.19,0.71) 0.91 (0.38,2.23)

Very painful 43 79.1 0.20*** (0.09,0.45) 0.59 (0.19,1.87)

Anticipated embarrassment o0.0005 0.005

Not at all embarrassing 592 95.4 1.00 1.00

A bit embarrassing 422 92.4 0.58* (0.34,0.99) 0.60 (0.33,1.09)

Quite embarrassing 145 89.0 0.39** (0.20,0.74) 0.63 (0.28,1.40)

Very embarrassing 143 79.0 0.18*** (0.10,0.31) 0.26*** (0.13,0.54)

Worry 0.252

Very worried 105 93.3 1.00 (0.43,2.31)

Quite worried 158 91.1 0.73 (0.38,1.40)

A bit worried 527 93.4 1.00

Not at all worried 514 90.1 0.65 (0.41,1.01)

S.

Su

tton

,C

.R

uth

erford

/S

ocia

lS

cience

&M

edicin

e6

1(

20

05

)2

46

0–

24

65

2463

ARTIC

LEIN

PRES

STable 1 (continued )

Variable Categories Frequencies % Smear test Univariate analyses (N ¼ 1266� 1307) Multivariate analysis (N ¼ 1259)

p Odds ratio (95% CI) p Odds ratio (95% CI)

Perceived risk 0.299

Much higher 21 90.5 0.78 (0.18,3.43)

Higher 49 93.9 1.26 (0.38,4.17)

A bit higher 101 93.1 1.10 (0.49,2.49)

About the same 790 92.4 1.00

A bit lower 157 93.6 1.21 (0.60,2.41)

Lower 114 86.8 0.54* (0.30,0.99)

Much lower 51 86.3 0.52 (0.22,1.20)

Attitude component I (five-point scale) o 0.0005 0.33*** (0.25,0.43) o 0.0005 0.44*** (0.30,0.63)

Attitude component II (five-point scale) o 0.0005 0.65*** (0.53,0.80) 0.052 0.77 (0.59,1.00)

Attitude component III (five-point scale) 0.023 1.29* (1.04,1.62) 0.375 0.87 (0.65,1.18)

�po0:05; ��po0:01; ���po0:001.Multivariate model: w2 ¼ 111.43, df ¼ 24, po0:0005. Pseudo-R2 (proportional reduction in w2 compared to constant-only model) ¼ 0.18.

The reference category for each categorical variable is shown in bold.

S.

Su

tton

,C

.R

uth

erford

/S

ocia

lS

cience

&M

edicin

e6

1(

20

05

)2

46

0–

24

65

2464

ARTICLE IN PRESSS. Sutton, C. Rutherford / Social Science & Medicine 61 (2005) 2460–2465 2465

The finding that anticipated embarrassment was a

significant independent predictor is broadly consistent with

previous studies that have shown the importance of

perceived barriers to screening uptake (Murray & McMil-

lan, 1993; Orbell et al., 1996). This problem needs to be

addressed in information campaigns and other aspects of

the screening service, especially as a substantial minority of

women (22%) in the present study thought that going for

smear test would be quite or very embarrassing.

Attitudes to screening (component I) were also

strongly associated with uptake. Agreement with state-

ments such as ‘‘There’s no point going for screening if

you don’t have any symptoms’’ may reflect a failure to

understand the rationale for screening. Previous studies

have identified a similar lack of understanding (e.g.,

Nicoll, Narayan, & Paterson, 1991). Clearly, this needs

to be addressed in information campaigns and materials.

There was no evidence that attitudinal variables

mediated the effects of marital status and education on

uptake. Future studies should investigate other attitu-

dinal variables (such as perceived social pressure or

approval from the woman’s partner) that may mediate

the effect of marital status.

The relatively weak and inconsistent effects of the SES

indicators in this study may be partly explained by the

fact that a national call–recall programme has been in

operation in the UK since 1988. It is possible that a

system based on sending women regular postal invita-

tions has had the effect of reducing socioeconomic

differentials in cervical screening uptake by comparison

with other health services where participation requires an

unprompted response. Although we cannot formally

generalise our findings beyond Britain, we suggest that

similar findings may be expected in countries where

organised screening programmes based on postal invita-

tions have been in place for some years, e.g., Finland,

Iceland and The Netherlands. However, we are not aware

of any studies that have investigated sociodemographic

and attitudinal correlates of uptake in these countries.

Acknowledgements

This study was funded by the National Breast and

Cervical Screening Programmes. We thank Mrs. Julietta

Patnick, the National Screening Coordinator, for her

interest and support.

References

Baker, D., & Middleton, E. (2003). Cervical screening and

health inequality in England in the 1990s. Journal of

Epidemiology and Community Health, 57, 417–423.

Bish, A., Sutton, S., & Golombok, S. (2000). Predicting uptake

of a routine cervical smear test: A comparison of the health

belief model and the theory of planned behaviour.

Psychology and Health, 15, 35–50.

Brinton, L. A. (1992). Epidemiology of cervical cancer—

overview. In N. Munoz, F. X. Bosch, K. V. Shah, & A.

Meheus (Eds.), The epidemiology of human papillomavirus

and cervical cancer (pp. 3–23). Lyon: International Agency

for Research on Cancer.

Coulter, A., & Baldwin, A. (1987). Survey of population

coverage in cervical cancer screening in the Oxford region.

Journal of the Royal College of General Practitioners, 37,

441–443.

Ibbotson, T., Wyke, S., McEwen, J., Macintyre, S., & Kelly, M.

(1996). Uptake of cervical screening in general practice:

Effect of practice organisation, structure, and deprivation.

Journal of Medical Screening, 3, 35–39.

Janz, N. K., & Becker, M. H. (1984). The health belief model: A

decade later. Health Education Quarterly, 11, 1–47.

Murray, M., & McMillan, C. (1993). Health beliefs, locus of

control, emotional control and women’s cancer screening

behaviour. British Journal of Clinical Psychology, 32,

87–100.

Nicoll, P. M., Narayan, K. V., & Paterson, J. G. (1991).

Cervical cancer screening: Women’s knowledge, attitudes

and preferences. Health Bulletin, 49, 184–190.

Orbell, S., Crombie, I., & Johnston, G. (1996). Social cogni-

tion and social structure in the prediction of cervical

screening uptake. British Journal of Health Psychology, 1,

35–50.

Quinn, M., Babb, P., Brock, A., Kirby, L., & Jones, J. (2000).

Cancer trends in England and Wales 1950–1999. London:

The Stationery Office.

Wardle, J., Farrell, M., Hillsdon, M., Jarvis, M., Sutton, S., &

Thorogood, M. (1999). Smoking, drinking, physical activity

and screening uptake and health inequalities. In D. Gordon,

M. Shaw, D. Dorling, & G. Davey-Smith (Eds.), Inequalities

in health: The evidence presented to the independent inquiry

into inequalities in health, chaired by Sir Donald Acheson

(pp. 213–239). Bristol: The Policy Press.