1 b.a.j. (336200).doc · web viewscholars as sasser, olsen and wyckoff (1978), gronroos (1982)...

Treatment compliance of chronicdiseased patients

Perception or cost and benefit deliberation?

Bart A.J. ZoetStudentnumber: 336200

Rotterdam, The NetherlandsSeptember 9, 2010

MASTER’S THESIS IN MARKETINGERASMUS SCHOOL OF ECONOMICSERASMUS UNIVERSITY ROTTERDAM

SUPERVISOR:N.M. Almeida Camacho MSc

Treatment compliance of chronic diseased patients

Perceptions or cost and benefit deliberation?

Table of content

Erasmus University RotterdamSchool of Economics

6

Abstract

This research examines factors that affect “patient compliance”. A model of “patient compliance” is developed as a function of the constructs (1) “expected treatment quality”, (2) “perceived treatment quality” and (3) “confirmed/ disconfirmed treatment quality” to examine if this independent variables affect the dependent variable “patient compliance. Tested is also if the positive effect of “confirmed treatment quality” is larger than the negative effect of “disconfirmed treatment quality”. The model is test using several multiple regression models, whereas asthma patients participate by an online survey. Data aggregation due to a small sample size results in separation of “expected treatment quality” into two underlying constructs: “expected treatment benefits” and “expected treatment costs”. The result show that “expected treatment benefits” drives “patient compliance” significantly and that “perceived treatment quality” drives “confirmed/ disconfirmed treatment quality” significantly. All other results were statistically not significant. This research provide valuable insights for physicians and pharmaceutical companies to increase “patient compliance”. Shared decision making and clarifying the patients role before starting a treatment is important to let patients shape realistic expectations. Salience and mindfulness is important to know what the perceptions of patients are. The expectations and perceptions that patients have are important to optimize future treatments. Finally limitations and future research directions are provided.

Keywords: “patient compliance” “expected treatment quality”; “perceived treatment quality”, “confirmed/ disconfirmed treatment quality”

Table of content

1 INTRODUCTION........................................................................................................................... 61.1 RESEARCH IMPORTANCE..............................................................................................................61.2 RESEARCH OBJECTIVE AND DEFINITION.........................................................................................61.3 RESEARCH QUESTIONS................................................................................................................61.4 METHODOLOGY........................................................................................................................... 6

2 THEORY AND CONCEPTUAL MODEL.......................................................................................62.1 CONSTRUCTS.............................................................................................................................. 6

2.2.1 Patient compliance............................................................................................................62.2.2 Expected treatment quality and perceived treatment quality.............................................62.2.3 Confirmed/ disconfirmed treatment quality........................................................................6

2.3 HYPOTHESIS................................................................................................................................ 62.4 CONCEPTUAL FRAMEWORK...........................................................................................................62.5 MODEL........................................................................................................................................ 6

2.5.1 Hypothesis 1, 2 and 3.......................................................................................................62.5.2 Hypothesis 4..................................................................................................................... 62.5.3 Hypothesis 5 and 6...........................................................................................................6

3 DATA AND MEASUREMENT.......................................................................................................63.2 Sample size........................................................................................................................... 63.3 Method.................................................................................................................................. 63.4 Questionnaire........................................................................................................................ 6

4 RESULTS...................................................................................................................................... 64.1 Descriptive statistics..............................................................................................................64.2 Hypothesis testing................................................................................................................. 64.2.1 Testing hypothesis 1, 2 and 3...........................................................................................64.2.3 Testing hypothesis 4.........................................................................................................64.2.5 Testing hypothesis 5 and 6...............................................................................................6

5 CONCLUSION, IMPLICATIONS, LIMITATIONS AND FUTURE RESEARCH DIRECTIONS......65.1 Conclusion and implications..................................................................................................65.1.1 Conclusions...................................................................................................................... 65.1.2 Implications....................................................................................................................... 65.2 Limitations and future research directions.............................................................................6

REFERENCES....................................................................................................................................... 6APPENDIX A OVERVIEW QUALITY HEALTHCARE CRITERIA......................................................6APPENDIX B QUESTIONNAIRE........................................................................................................6Appendix C Supportive output..........................................................................................................6

Acknowledgement Erasmus University Rotterdam

School of Economics7

This research presents the results of the effects of expectations and perceptions that chronic diseased patients have about the quality of a treatment in relation to treatment compliance. This research showed me insight of one of the major problems in healthcare and increased my knowledge. It also showed me what problems chronic diseased patient face when they follow a treatment as provided by their medical expert.

This topic could not be explored without help of many people. I would like therefore to thank my supervisor Nuno Camacho for his critical notes during the meetings and value adding advises. I would like to thank the Foundation “Nederland-Davos”, “Astma & COPD Nieuws” and “Zorgportaal” for their help in publishing the survey on their website. Finally, I would like to thank family and friends for their support and interest in this topic.

Bart A.J. Zoet

Leiden, September 2010


8

1 IntroductionPeriodically, no matter the discipline, new fields of research emerge. Marketing is no different. The number of papers on health and marketing published in marketing journals has been increasing rapidly over the last five years (Stremersch, 2008, p.1). Although marketing scholars often seek to contribute new knowledge that is applicable across industries (Stewart, 2002), most industries have unique characteristics that require industry-specific knowledge development (Eliashberg, Elberse and Leenders, 2006). The life science industry spans companies in pharmaceuticals, biotechnology and therapeutic medical devices and it forms the innovative producer side of the health care industry. Two fundamental dimensions underlie the life sciences industry: science-based knowledge and quality of life (Stremersch and Van Dyck, 2009, p.4). A life science therapy is launched into regular health care after scientific research of its impact on patients quality of life based on safety, cost effectiveness and efficacy. This process can take up to 15 years. To make their therapy profitable, firms promote their therapies and marketers face unique challenges in launching and promotion decisions. Not only the fields of marketing and life science span their disciplines, the worlds of physicians and patients are nearing each other too. There is a trend towards more participatory decision-making in which doctors and patients together bear responsibility for medical decisions (Camacho, Landsman and Stremersch, 2008). There is a movement from the old “white coat model”, in which a physician uses his or her knowledge to prescribe treatments (Morgan, 2003), to a “dialogue model” where scientific evidence is connected to patient needs and preferences (Charles, Gafni and Wehlan, 1999). The change from a “white coat model” to a “dialogue model” implicates that patient empowerment increases. Research found that patients who are more involved in the process or have more empowerment generate better therapy results and are more compliant.

1.1 Research importanceThis research is based on earlier work of Stremersch and Van Dyck. Figure 1.1, developed by Stremersch and Van Dyck (2009), shows future research possibilities.

Figure 1.1: Future research possibilities in life science (Stremersch and Van Dyck, 2009, p.10)

The authors state that research directions in the top right are important contributions to academic knowledge and of high, immediate and practical relevance to business performance and/or patient welfare. This research focus on stimulating patient compliance. Earlier research focus mostly on measurement, extent and determinants of “patient compliance”. Morris (1992) argues that very little consistent information is available and that the patient’s perspective needs to be examined. “Patient compliance” is an important issue in pharmaceutical business and scientific literature. Global pharmaceutical industry lose $ 30 billion each year simply by patients who do not take their medicines as advised (Bates, 2008). Seventy percent of patients who begin with a medical treatment discontinue with one year (Bates, 2008, p.28). DiMatteo (2004) argues that this percentage is about 25%. The


5

greatest drop in patient compliance occurs after the first six months of treatment. This results in significant loss for pharmaceutical companies who spent millions to get those initial prescriptions (Bates, 2008, p.28) Non-compliant patients are very expensive for the society as well. Non-compliant patients influence the profit of health insurers, waiting lists, are responsible for $ 290 billion a year of increased medical costs (Ziegler, 2009) and are also expensive for their employers because they create less economic value. Beside this, poor compliance is associated with a lower quality of life, poor clinical outcomes, increased hospitalizations and higher overall healthcare costs (Hosken, 2006, p.6). Beside quantifying patient compliance in money and percentages, patient compliance is also quantified in a specific death rate. In 1984, The National Pharmaceutical Council, an association of pharmaceutical companies, estimated that misuse of drug prescription cause 125,000 deaths a year from hearth disease alone (Zuger, 1998).

1.2 Research objective and definitionThe objective of this research is to work further on previous work of Morris (1992) and Stremersch and Van Dyck (2009). Cochrane, Horne and Chanez (1999) also provide a future research possibility. The authors state that it is unclear what the influence is of patients treatment quality expectations and patients treatment quality perceptions on “patient compliance”. Asch et al (2006) argues that the quality of most treatments fall short of expectations. Oliver (1977; 1980; 1993) argues that expectations and perceptions lead to disconfirmation. In the satisfaction literature research, disconfirmation occupies a central position as a crucial intervening variable (Churchill and Surprenant, 1982, p.492). Referring to Oliver (1977), Churchill and Surprenant (1982, p.493) state that measuring disconfirmation is important because the construct has an independent, additive effect on satisfaction. This research is based on four constructs: 1. “patient compliance”, 2. “expected treatment quality” 3. “perceived treatment quality” and 4. “confirmed/ disconfirmed treatment quality”. Construct means; the main factors that are investigated.

The World Health Organization define the first construct, “patient compliance”, as “the extent to which a person’s behavior (taking medication, following a diet, and/or executing lifestyle changes) corresponds with agreed recommendations from a health care provider” (WHO, 2003, p.136).

The literature does not provide one single definition or description about the second construct, “expected quality”. The literature only pay attention to “quality” and “expectations separately where 1. Parasuraman, Zeithaml and Berry (1985, p.42) use the definition of Crosby (1979) to define “quality” as “conformance to requirements” and 2. where “expectations” is “pretrial beliefs about a product Olson and Dover (1979, p.181) or service that serve as standard or reference point against which a product or service performance is judged” (Zeithaml, Berry and Parasuraman, 1993, p.1). Combining this two definitions, “expected treatment quality” is during this research defined as “pretrial belief that the treatment will be conform the requirements which serves as standard and reference point”.

“Perceived quality”, the third construct, is defined as “the consumer’s judgment about a product’s overall excellence or superiority” (Zeithaml, 1988, p.3). Zeithaml (1988) state that “perceived quality” is different than “objective quality”. “Objective quality” is the actual technical superiority or excellence of the product. Zeithaml (1988, p.5) also argues that “objective quality” may not exist because all quality is perceived by someone, be it consumer or managers or researchers.

“Expected quality” and “perceived quality” lead to the fourth construct, “confirmed/ disconfirmed treatment quality”. “Confirmed/ disconfirmed treatment quality” is the difference between perceptions and expectations (Oliver, 1993). Oliver (1993) defines “disconfirmation” as “combining perceptions and expectations to form satisfaction judgments”. Oliver (1980) distinguish negative disconfirmation (performance poorer than expected) and positive disconfirmation (performance better than expected). This research labeled “confirmed/ disconfirmed treatment quality” as one single construct, where “confirmed treatment quality


6

refers to quality performance better than expected and disconfirmed treatment quality refers to quality performance poorer than expected.

1.3 Research questionsThe previous discussion lead to the following research questions: 1. Do “expected treatment quality”, “perceived treatment quality” and “confirmed/ disconfirmed treatment quality” have an effect on “patient compliance”?

Following that if there is an effect: 2. What is the effect of “expected treatment quality”, “perceived treatment quality” and “confirmed/ disconfirmed treatment quality” on “patient compliance”?3.Will the negative effect of “disconfirmed treatment quality” be larger than the positive effect of “confirmed treatment quality”?4. Does “confirmed/ disconfirmed treatment quality” mediate the effect of “expected treatment quality” and “perceived treatment quality” on “patient compliance”? 5. Which dimensions of a treatment’s quality have a stronger influence on “patient compliance”?

To answer the research questions, the following operational questions need to be answered during this research process;1. What are the definitions of the different constructs?2. What are the determinants of the different constructs?3. How can we model the different constructs?4. Which target group is used to test the model?5. What are the conclusions, implementations, limitations and future research possibilities?

1.4 MethodologyAfter this chapter, the research follow with the theoretical framework and conceptual framework. The theoretical framework give a literature overview based on the four constructs “expected treatment quality”, “perceived treatment quality”, “confirmed/ disconfirmed treatment quality” and “compliant behavior”. After the theoretical framework hypothesis are stated and a conceptual model is developed. This conceptual model is developed to link the role of the different construct with each other. Also will be explained how the different hypothesis are going to be measured and tested.

The third chapter discuss data and measurement. Explained is what the target group is for this research. Also the method, questionnaire and how to avoid research bias is discussed.

The fourth chapter discuss each hypothesis. Results shows which hypothesis are not rejected and which hypothesis are rejected in order to answer the stated research questions.

Finally the fifth chapter gives a general conclusion by answering the research questions. The answered research question lead to implications. Beside this, chapter five also discuss limitations and provides future research directions.


7

2 Theory and conceptual modelThis chapter provides a literature overview of the different constructs. After this literature review, hypothesis are stated and the conceptual model is discussed. Finally, the different models to test the hypothesis are discussed.

2.1 Constructs

2.2.1 Patient compliance“Patient compliance” receive special attention of the WHO. The WHO states that across diseases, compliance is the single most important modifiable factor that compromises treatment outcome (WHO, 2003, p. 135). “Patient compliance” is a complex behavioral process determined by several determinants as attributes of the patient, the patients environment and characteristics of the disease and its treatment (WHO, 2003). Morris (1992) states that a lot of research is done to the determinants, measurement and the extent of non-compliance. Morris (1992) argues also that very little consistent information is available and that the patient’s perspective needs to be examined. In making patients compliant to their prescribed treatment, there is also a shift to shared decision making between physician and patient. This “patient empowerment” as stated by Christensen and Hwang (2009) will increase patient satisfaction (Camacho, Landsman and Stremersch, 2008). Some scientists find a connection between patient empowerment and an increasing patient compliance. For example, research of hypertension suggests that a better understanding of patients’ attitudes (part of shared decision model) could improve programs designed to increase their compliance to treatment regimes (Hopfield, Linden and Tevelow, 2006). Another example are diabetics who take their health in own hand and find that cost of care decreases dramatically, while the quality increases. It is more effective than relying on an expert whom they may see only every few months. Other research find that there are benefits in patient participation in decision making. One study of 15 overweight children found that 7 children who thought they had chosen their own treatment program lost more weight than the group who thought they had no choice (Mendonca and Brehm, 1983). Ashcroft, Leinster and Slade (1986) surveyed 31 patient with early-stage breast cancer. They find that patients who choose their own surgical therapy did not experience the high levels op depression and loss of self-esteem that have been recorded for breast-cancer patients in other studies. Dellande, Gilly and Graham (2004, p.79) state that expert providers can take a more active role in structuring the information environment and in clarifying the patient’s role. With the requisite role clarity, ability, and motivation, compliance is more likely (Dellande, Gilly and Graham, 2004, p.79).

There are several studies that determine factors which affect “patient compliance”. These factors range from individual characteristics, social circumstances and health knowledge (Sherbourne, Hays, Ordway, DiMatteo and Kravitz, 1992) to satisfaction with care, age, gender, work disruption and plenty of other factors (Fincham, 2007). As mentioned earlier, the World Health Organization pays special attention to “patient compliance”. The organization describes several important variables that are behavioral in nature and are also dynamic, and therefore amenable to intervention (WHO, 2003, p.137). The variables are separate in three key determinants: (1.) provider behaviors: variables relate to how health care providers communicate and interact with their patients are important determinants of compliance and health care success. Important factors in this are providing information, “positive talk” and asking specific questions to the patient. Patients who consider themselves as partners in the health care process and who are actively engaged are more compliant. In sum it is important there is warmth and empathy from the providers side towards the patient (WHO, 2003). (2.) health system factors: the health care delivery system influences the compliance of patients. The policies and procedures of the health system itself controls access to, and quality of care (WHO, 2003, p.138). The system variables include the


8

accessibility and availability of services, support for education of patients, information management and data collection, providing feedback to health care providers and patients, supporting the availability of communities and the training provided to health care providers (WHO, 2003, p. 138). System direct providers schedule, dictate appointment lengths, allocate resources, set fee structures and established organizational priorities (WHO, 2003, p.138). Finally, (3.) patient attributes: the characteristics of patients are determinants which are the basis of numerous researches of “patient compliance”. The WHO report state that demographic, economic and social factors have not definitely been associated with compliance. It is about illness relevant cognitions, perceptions and expectations of disease factors that have stronger relations with patient compliance (WHO, 2003). McHorney (2009) developed a three item compliance estimator. The author state that the estimator is readily scored and easily interpretable. The estimator focus on three pillars; 1. medication, 2. non-fulfillment and 3. non-persistence (McHorney, 2009, p.215) which are causes that influence and relate compliant behavior of patients. But the best way to measure “patient compliance” is to ask patients if they are completely compliant in a specific period (Kaptein, 2006). This is also one of the recommendations by Grundmeijer, Reenders and Rutten (2004) to measure “patient compliance”. Single questions are sensitive for social desirability bias. This research deal with this which is discussed later this research.

2.2.2 Expected treatment quality and perceived treatment qualityExpected treatment qualityConceptualization of patient expectations of care has two aspects; what patients expect as a result of their own or others’ experiences (normative/ comparative expectations) and the care they would like and/ or hope for (idealized expectations) (McKinley, Stevenson, Adams and Manku-Scott, 2002, p.334). Fulfillment of patient expectations may affect visit satisfaction, may influence health care utilization and costs and can be used as a indicator of the quality of care (Peck et al., 2001, p.101). Patients develop expectations by “learning” from different type of sources with varying validity (Jayasankar, 2009). Past experiences are also a contributor to future expectations. Important is the way the patient understands the information and translate this information in actionable beliefs. Expert providers must create role clarity about the treatment that is advised and the skills to perform the treatment (Dellande, Gilly and Graham, 2004). Expectations are pretrial beliefs (Olson and Dover, 1979) and are important in being satisfied or dissatisfied (Parasuraman, Zeithaml and Berry, 1985). Oliver (1981, p.33) state “that it is generally agreed that expectations are consumer-defined probabilities of the occurrence of positive and negative events if the consumer engages in some behavior”. According to this, positive events lead to satisfied behavior and negative events lead to dissatisfied behavior. Referring to Oliver (1981), patients are assumed to be non-compliant if the quality of the treatment is more poorly than the quality expected, and patients are compliant if the quality of the treatment is better than the quality expected.

Perceived treatment qualityVerhage (2004) state that there are three types of perceptions: subjective, cognitive and selective. Monroe and Krishnan (1985) argue that perceived quality is the perceived ability of a product or service to provide satisfaction “relative” to alternatives. The definition provided by Zeithaml (1988) is a user based approach (Sirohi, McLaughlin and Wittink, 1998). To form perceptions, people use intrinsic and extrinsic cues (Zeithaml, 1988; Sirohi, McLaughlin and Wittink, 1998). Intrinsic cues are derived directly from the product or service and extrinsic cue are not directly related to the product or service (Rao and Monroe, 1989). Scholars as Sasser, Olsen and Wyckoff (1978), Gronroos (1982) and Lethinen and Lethinen (1982) unambiguously support the notion that product or service quality, as perceived by consumers, stems from a comparison of what they feel a product or service should offer (Parasuraman, Zeithaml and Berry, 1988, p. 16). This comparison leads to judgments about the product’s or service overall performance excellence or superiority (Zeithaml, 1988).


9

Key determinants of expected and perceived treatment qualityA lot of research is done to investigate the key quality determinants of a treatment. Campbell, Roland and Buetow (2000, p.1615) provide an overview of the most important criteria concerning treatment quality, appendix A. The overview shows that there is a major overlap in the criteria. Within this overview, the seven key determinants by Donabedian (1990) are most influential and find most citations in other literature. The overview shows that the concept “legitimacy” and “optimality” provided by Donabedian (1990) find no support by other authors. This gives that the criteria “legitimacy” and “optimality” are not used as quality determinants to test “expected treatment quality” and “perceived treatment quality”. Beside the mentioned criteria, there is strong support that side effects (Vermeire et al, 2001; Guyatt et al, 1993; Bowman, Heilman and Seetharaman, 2004) and convenience (Heather, Amit and Haynes, 2002; Bowman, Heilman and Seetharaman, 2004) are important quality aspects in health care. These two aspects complete the criteria provided by Donabedian (1990) in testing “expected treatment quality” and “perceived treatment quality”. Figure 2.1 provides an overview of the selected criteria and their explanations.

Determinant DescriptionEfficiency The treatment obtain the greatest health improvements at the lowest costAcceptability The treatment will be conform to patient preferencesEfficacy The treatment is able to improve healthEffectiveness The degree to which the treatment is attainable to realize health improvementsEquity There are no differences in treatment compared to othersConvenience The treatment is easy to follow and easy to useSide effects The treatment create a peripheral or secondary effect

Figure 2.1: key determinants in testing “expected treatment quality” and “perceived treatment quality”

2.2.3 Confirmed/ disconfirmed treatment qualityThe basic premise of the marketing concept is that marketers should strive to create customer satisfaction and should remedy dissatisfaction (Blodgett, Granbois and Walters, 1993). In line with “expected treatment quality” and “perceived treatment quality” it is possible to measure “confirmed/ disconfirmed treatment quality” where “confirmed treatment quality” leads to satisfaction which lead to increasing “patient compliance’ and “disconfirmed treatment quality” leads to dissatisfaction which leads to increasing “patient non-compliance”. Woodruff, Cadotte and Jenkins (1983) state that if perceived performance match expected performance it results in “conformation”. A mismatch leads to “disconfirmation”. Oliver (1980, 1993) provides a framework to measure “confirmed/ disconfirmed treatment quality”. “Confirmed/ disconfirmed treatment quality” can be measured by one single control question (Niedrich, Kiryanova and Black, 2005; Oliver 1980) with a “better than expected or worse than expected question, which find a lot of support by other authors. Baumeister, Bratlavsky, Finkenauer and Vohs (2001) argue that negative events in daily life have a stronger effect than positive events. Bad emotions, bad parents and bad feedback have more impact than good ones and bad information is more thoroughly than good information. Bad impressions and bad stereotypes are easier to form and more resistant to “disconfirmation” than good ones to “confirmation” (Baumeister, Bratlavsky, Finkenauer and Vohs, 2001, p.323). The authors argue that negative events have a larger impact than positive events. The assumption is that a mismatch between “perceived treatment quality” and “expected treatment quality” leads to a larger negative effect of “disconfirmed treatment quality” than the positive effect of “confirmed treatment quality”.

In sum this research describes “patient compliance” as “the extent to which a person’s behavior (taking medication, following a diet, and/or executing lifestyle changes) corresponds with agreed recommendations from a health care provider” (WHO, 2003, p.136). It describe “expected treatment quality” as “ a pretrial belief that the treatment will be


10

conform the requirements”. This research describe “perceived treatment quality” as “the consumer’s judgment about the overall excellence or superiority of the treatment” (Zeithaml, 1988). Finally this research describe “confirmed/ disconfirmed treatment quality” as “combining perceptions and expectations to form satisfaction judgments” (Oliver, 1993).

2.3 HypothesisBased on the literature research the following hypothesis are constructed to answer the research questions:H1: There is a positive effect of “expected treatment quality” on “patient compliance”.

H2: There is a positive effect of “perceived treatment quality” on “patient compliance”.

H3: There is a positive effect of “confirmed/ disconfirmed treatment quality” on “patient compliance”.

H4: The negative effect of “disconfirmed treatment quality” is larger than the positive effect of “confirmed treatment quality”.

H5: The positive effect between “expected treatment quality” and “patient compliance” will be fully mediated by “confirmed/ disconfirmed treatment quality”.

H6: The positive direct effect between “perceived treatment quality” and “patient compliance” will be fully mediated by “confirmed/ disconfirmed treatment quality”.

2.4 Conceptual frameworkFigure 2.3 shows the conceptual framework. This model is a graphic overview of the stated hypothesis and present their underlying connections. The conceptual framework is based on earlier work of Oliver (1977; 1980; 1993). Where Olivier (1977; 1980; 1993) study the effects of expectations, perceptions, disconfirmation on satisfaction, this research study the effect of “expected treatment quality”, “perceived treatment quality” and “confirmed/ disconfirmed treatment quality” is on “patient compliance”. Oliver (1977; 1980; 1993) study also if the effect of expectations and perceptions on satisfaction is mediated by disconfirmation. Referring to Oliver’s (1977; 1980; 1993) mediation analysis, this study test if “confirmed/ disconfirmed treatment quality” mediates the possible effect of “expected treatment quality” and “perceived treatment quality” on “patient compliance”. This to explain how external physical events take on internal psychological significance (Baron and Kenny, 1986, p.1176).


11

Expected treatment Perceived treatmentquality quality

Confirmed/ disconfirmed treatment quality

Patient compliance

Figure 2.2: Conceptual model


Acceptability

Convenience

Efficiency

Effectiveness

Side effects

Efficacy

Efficiency

Acceptability

Convenience

Side effects

Effectiveness

Efficacy

One single control question

EquityEquity

One single control question

12

Independent variable X

Mediator variable M

Dependent variable Y

2.5 Model

2.5.1 Hypothesis 1, 2 and 3Multiple regression analysis measure the effect of the constructs “expected treatment quality”, “perceived treatment quality” and “confirmed/ disconfirmed treatment quality” on “patient compliance”:

Patient compliancei = γ0+ γ1Confirm/DisconfirmTreatmQuali + γ2PercTreatmQuali + γ3ExpTreatmQuali + e1i (1)

Where “patient compliance” is the dependent variable of the multiple regression analysis for individual i, γn is the regression coefficient of the independent variables “expected treatment quality”, “perceived treatment quality” and “confirmed/ disconfirmed treatment quality” and e1 is the error term to capture all other unobserved factors that influence the outcome. With this multiple regression analysis hypothesis 1, 2 and 3 are tested.

2.5.2 Hypothesis 4Hypothesized has been that the negative effect of “disconfirmed treatment quality” is larger than the positive effect of “confirmed treatment quality”. This because negative events have a stronger impact in daily life than positive events (Baumeister, Bratlavsky, Finkenauer and Vohs 2001). The bipolar scale is changed into a unipolar scale. The categories ”totally disagree”, “disagree” and “disagree a bit” are recoded as 1, 2 and 3 which refers to “disconfirmed treatment quality” and “agree a bit”, “agree” and “totally agree” are recoded as 5, 6 and 7 which refers to “confirmed treatment quality”. The following multiple regression analysis is developed to test hypothesis four:

Patient compliancei = β0+β1ConfirmTreatmQuali + β2DisconfirmTreamQuali + β3PercTreatmQuali + β4ExpTreatmQuali + e2i (2)

Where “patient compliance” is the dependent of the multiple regression analysis for individual i, βn is the regression coefficient of the independent variables “expected treatment quality”, “perceived treatment quality”, “confirmed treatment quality” and “disconfirmed treatment quality” and e2 is the error term to capture all other unobserved factors that depend the outcome.

2.5.3 Hypothesis 5 and 6A mediation analysis tests if “confirmed/ disconfirmed treatment quality” mediate the direct effect of “expected treatment quality” and “perceived treatment quality” on “patient compliance”. Figure 2.3 provide an overview of the mediation analysis where (1) the independent variable X is “expected treatment quality” and “perceived treatment quality”, (2) the mediator variable M is “confirmed/ disconfirmed treatment quality” and (3) the dependent variable Y is “patient compliance”.

A


B

C’

13

Figure 2.3: Graphic overview mediation analysis (Zhao, Lynch jr. and Chen, 2010)

The mediation effect is measured with the steps provided by Baron and Kenny (1986). The authors state that there are multiple methods to extent an mediation effect (multiple regression, logistic regression or structural equal modeling) but the steps necessary for testing mediation are the same. Baron and Kenny (1986) discuss three steps based on multiple regression. First, one should regress the mediator on the independent variable. This is done with the following multiple regression analysis:

Confirm/DisconfirmTreatmQuali = η0 + η1PercTreatmQuali + η2ExpTreatmQuali + e3i (3)

Where “confirmed/ disconfirmed treatment quality” is the dependent variable for the multiple regression analysis for individual i, ηn is the regression coefficient of the independent variables “expected treatment quality” and “perceived treatment quality”, and e3 is the error term to capture all other unobserved factors that influence the outcome.

The second step is regressing the dependent variables with the indepent variables “expected treatment quality” and “perceived treatment quality” (Baron and Kenny, 1986):

Patient compliancei= δ0+ δ1PercTreatmQuali + δ2ExpTreatmQuali + e4i (4) Where “patient compliance” is the dependent variable of the multiple regression analysis for individual i, δn is the regression coefficient of the independent variables “expected treatment quality” and “perceived treatment quality”, and e4 is the error term to capture all other unobserved factors that influence the outcome.

The third step is regressing the dependent variable on both the independent variables controlling for the mediator variable (Baron and Kenny, 1986, p. 1177):

Patient compliancei = θ0+ θ1ExpTreatmQuali + θ2PercTreatmQuali + θ3Confirm/DisconfirmTreatmQuali + e5i (5)

Where patient compliance is the dependent variable of the multiple regression analysis for individual i, γn is the regression coefficient of the independent variables “expected treatment quality”, “perceived treatment quality” and the controlled mediator variable “confirmed/ disconfirmed treatment quality” and e5 is the error term to capture all other unobserved factors that influence the outcome.

These three regression enquations provide the tests of the linkages of the mediational model (Baron and Kenny, 1986, p. 1177). To test mediation, the following conditions must hold: (1.) the independent variables must affect the mediator in the third equation, (2.) the independent variables must be shown to affect the dependent variable in the fourth equation and (3.) The mediator must affect the dependent variable in the fifth equation (Baron and Kenny, 1986). If all these conditions hold, than the effect of the independent variable on the dependent variable must be less in the fifth equation then in the third equation. Full mediation holds if the independent variables have no effect when the mediator is controlled for. Partial mediation holds if the independent variable is reduced but still different from zero when the mediator is controlled (Baron and Kenny, 1986).


14

3 Data and measurementThis chapter discusses the data collection. First, the target group is discussed followed by discussing the sample size, the method and the questionnaire.

3.1 Target groupAsthma patients participate in this research. The WHO defines asthma as “a chronic disease characterized by recurrent attacks of breathlessness and wheezing, which vary in severity and frequency from person to person (WHO, 2010). Asthma is a frequent problem and occurs in every age category. About 1.5 - 2% of the world population has asthma (Joos and Bel, 1999). Asthma is a hereditary disease, but there are also risk factors that lead to the development of asthma. These factors and their role are a bit unclear (Astmafonds, 2007). In 2007, the Dutch Asthma Fund calculate that in the Netherlands there are 519.800 people with asthma. Of this population, 45,5% is male and 54.5% is female. The real amount might be higher, because not everybody with symptoms goes to their general practitioner and not every general practitioner recognizes (direct) all cases of asthma. Compared to the growth of the Dutch population, the Dutch Asthma Foundation expects that the total amount of Asthma patients will grow to 2% of the total population in 2025 (Astmafonds, 2007).

Asthma has an influence on the quality of life of patients. Complaints can occur in a minute. The disease has an influence on the daily life of patients but will also influence their emotional and social behavior (Astmafonds, 2007). Al these negative effects may lead low “patient compliance”. Non compliance with asthma treatment is about 10% - 46% (Spector, 2000).

3.2 Sample sizeLarger sample sizes increase power and decrease estimation error. However, the practical realities of conducting research such as time, access to samples and financial costs restrict the sample size. The balance is generating a sample large enough to provide sufficient power while allowing for the ability to actually garner the sample (VanVoorhis and Morgan, 2007, p.46). The sample size should lead to power. A way to do so is by rules of thumb for numbers of respondents needed for common statistical procedures (VanVoorhis and Morgan, 2007). Green (1991) provide two rules of thumb. The author state that if a model is tested overall, a minimum sample size of 50+8k is recommended, where k is the number of predictor variables. If individual predictors are tested than a minimum sample size of 105+k is recommended.

3.3 MethodThe asthma patients participate by an online survey, to reduce social desirability bias (Nederhof 1985). The survey is published among other things on the website of “Verening Nederland-Davos”. This is an important patient association that focus on the rights and importance of patients with asthma. The foundation collaborates close with the Asthma Centre in Davos in facilitating asthma treatment. A lot of asthma patients visit the website of the foundation in search for the latest news and information relating to asthma. The second website that publish the survey is “Astma & COPD Nieuws”. This is a website where asthma patients can find the latest news about their disease. The third website is “Zorgportaal”. This is also a website where asthma patients can find the latest news and information about their disease.

3.4 Questionnaire Appendix B shows the questionnaire. To avoid recall bias, first a recall question is asked. After this, year of birth, education level and gender are asked. This questions result in basic


15

sample characteristics. After this, the different constructs are tested. The constructs are separated in concepts which are discussed during the literature review. Discussed earlier, the constructs “expected treatment quality” and “perceived treatment quality” exist out of five criteria provided by Donabedian (1990) and other scholars. Beside this five criteria, side effects (Vermeire et al, 2001; Guyatt et al, 1993; Bowman, Heilman and Seetharaman, 2004) and convenience (Heather, Amit and Haynes, 2002; Bowman, Heilman and Seetharaman, 2004) will complete the criteria. The literature supports the fact that these factors are important quality aspects. The constructs “confirmed/ disconfirmed treatment quality” and “patient compliance” are measured by one single control question. This single question measurement find also support in literature.

In measurement, the characteristics of validity and reliability are important factors. The reliability is discussed in the results section of this research. There are several arguments that are important in choosing the right rating scale (Cooper and Schindler, 2006). In developing the measurement scales for the constructs “expected treatment quality” and “perceived treatment quality”, earlier contributions of Oliver (1977) and Parasuraman, Zeithaml and Berry (1988) are important. In line with these authors, the constructs “expected treatment quality” and “perceived treatment quality” are measured on a 7-point scale. Oliver (1977) and Parasuraman, Zeithaml and Berry (1988) measure expectations and perceptions on a 7-point bipolar Likert scale ranging from 3 (totally agree) to -3 (totally disagree) where 0 is neutral (neither agree, nor disagree). This result in high validity and reliability (Bearden and Netemeyer, 1999). Diefenbach, Weinstein O’Reilly (1993) also argue that no other scale performed significant better in their research than the 7-point Likert scale in measuring perceptions and expectations. Except the item “side effects” all other statements are positively worded. This means that higher scores reflect positive attitudes toward the statement. To interpret all results consistently, the scale of “side effects” is reversed. The category “totally disagree” reflect a positive attitude toward the statement and the category “totally agree” reflect a negative attitude toward the statement (Polit and Beck, 2004). After measuring “expected treatment quality” and “perceived treatment quality”, “confirmed/ disconfirmed treatment quality” is measured. This construct is measured on a 7-point bipolar scale ranging from 3 (totally agree) to -3 (totally disagree) where 0 is the neutral midpoint (neither agree, nor disagree) which is in line with prior work of Oliver (1977; 1980). Finally “patient compliance” is measured. One question is asked about “patient compliance” of the prescribed therapy of last month, which is measured on a 7 point bipolar scale ranging from 3 (completely) to -3 (not). Measuring “patient compliance” by one single question can lead to recall bias and social desirability bias (Kaptein, 2006; Grundmeijer, Reenders and Rutten, 2004). Measurement bias can lead to the reporting of spurious or misleading research results (Fisher, 1993). To cope with recall bias, patients are asked to recall their last month treatment. Furthermore, the question is asked in a polite and a no threatening way suggested by Grundmeijer, Reenders and Rutten (2004) and Sudman and Brandburn (1974) to avoid social desirability bias. Because the questions are asked in a online setting, patients can answer the question anonymous. This is also a manner to cope with social desirability bias (Nederhof, 1985). These techniques to avoid social desirability bias are completed with two other techniques: (1) a general instruction about key facts of “patient compliance” and the need of this research to make patients feel less guilty if they are non compliant is implemented and (2) a short introduction is implemented to motivate patients to fill in the question honestly. Finally the cause of “patient compliance” is measured. This measurement is highly based on the original paper by McHorney (2009) completed with some extra questions based on earlier work of Vermeire, Heartshaw, van Royen and Denekens (2001), Becke and Maiman (1980) and Zuger (1998). The measurement scale here is a 6-point bipolar rating scale ranging from 3 (totally agree) to -3 (totally disagree) without a zero point. This scale find support by McHorney (2009).


16

4 ResultsThis chapter discuss the results of this research. It starts with a brief discussion of the descriptive statistics. After this, the results per hypothesis are discussed.

4.1 Descriptive statistics Testing the model is based on a sample size of 106 respondents. Figure 4.1, 4.2 and 4.3 show the sample characteristics. The minimum age of the sample group is 17 years and maximum age of 76 years old. Most of the patients are HBO graduated (32%) followed by MBO graduation (27%). The smallest group is lower educated (2%) or not educated (1%).Of the sample group, 34% is male and 66% .

Minimum age 17 yearsMaximum age 76 yearsMean age 50,5 yearsMode 50 years

Figure 4.1: Sample characteristics age

Level of graduation Percentage WO 9%HBO 32%VWO 5%HAVO 12%MBO 28%MAVO 11%Lower education 2%No education 1%

Figure 4.2: Sample characteristics level of graduation

Gender Percentage Male 34%Female 66%

Figure 4.3: Sample characteristics gender

To determine if the constructs can be interpreted consistently across different situations (Field, 2009) the reliability is measured for the constructs “expected treatment quality” and “perceived treatment quality”. “Expected treatment quality” is reliable at .715 and “perceived treatment quality” is reliable at .868. Both constructs can not be improved if individual variables are excluded out of the model.

Cronbach's Alpha Cronbach's Alpha Based on Standardized Items N of Items.715 .725 7

Figure 4.4: Reliability construct “expected treatment quality”

Cronbach's Alpha Cronbach's Alpha Based on Standardized Items N of Items.868 .877 7

Figure 4.5: Reliability construct “perceived treatment quality”


17

Based on the rules of thumb provided by Green (1991), this sample size (n=106) is not large enough to test the effects of all quality components on “patient compliance”. A necessary step to deal with this is to aggregate data by factor analysis. A factor analysis is conducted for the constructs “expected treatment quality” and “perceived treatment quality” separately. Due to the small sample size, a significance level of 10% is used in this research where 5% is usual in research. Figure 4.6 and 4.7 show the KMO and Bartletts Test of Sphericity. The KMO represents the ration of the squared correlation between variables to the squared partial correlation between variables (Green, 1991, p.647). The KMO varies between 0 and 1 (Green, 1991). A value of 0 indicates that the sum of partial correlations is large relative to the sum of correlations, factor analysis is likely to be inappropriate (Green, 1991, p.647). A value close to 1 indicates that patterns of correlations are relatively compact and so factor analysis should yield distinct an reliable factors (Green, 1991, p.647). KMO for “expected treatment quality is .717, which is good, and for “perceived treatment quality it is .866, which is great. Bartlett’s Test of Sphericity tests that the original correlation matrix is an identity matrix. If the test is significant, factor analysis is appropriate (Green, 1991, p.660). Bartlett’s test of sphericity X² is respectively (21)=149,332 and (21)=333,379, which is significant for both constructs (p < .1). The results indicate that correlations between items were sufficiently, the correlation matrix is not an identity matrix and therefore factor analysis is appropriate (Green, 1991).

Kaiser-Meyer-Olkin Measure of Sampling Adequacy. .717Bartlett's Test of Sphericity Approx. Chi-Square 149.332

df 21.Sig. 0.000

Figure 4.6: KMO and Bartlett’s test “expected treatment quality”

Kaiser-Meyer-Olkin Measure of Sampling Adequacy. .866Bartlett's Test of Sphericity Approx. Chi-Square 333.379

df 21.000Sig. 0.000

Figure 4.7: KMO and Bartlett’s test “perceived treatment quality”

Figure 4.8 and 4.9 show the extraction methods for the constructs “expected treatment quality” and “perceived treatment quality”. The construct “expected treatment quality” has two components with eigenvalues over Kaiser’s criterion of 1. These two components are labeled as “expected treatment costs” and “expected treatment benefits” and are discussed later this research. Both components explain 55% of variance in “patient compliance. The construct “perceived treatment quality” has one component with eigenvalue over Kaiser’s criterion of 1, and therefore there is no need to re-lable this construct. This component explains 57% of variance in “patient compliance”.

Component Initial EigenvaluesTotal % of Variance Cumulative %

1 2.660 37.995 37.9952 1.302 18.606 56.6013 .863 12.328 68.9294 .686 9.797 78.7265 .636 9.079 87.8056 .498 7.114 94.9197 .356 5.081 100

Figure 4.8: Extraction method “expected treatment quality”


18

ComponentPercTreatmQual

Efficacy .818Effectiveness .814Convenience .784Acceptability .758Side effects .718Efficiency .715Equity .696

Component Initial EigenvaluesTotal % of Variance Cumulative %

1 4.003 57.617 57.6172 .828 11.822 69.4383 .590 8.434 77.8734 .544 7.771 85.6445 .422 6.028 91.6726 .327 4.669 96.3417 .256 3.659 100

Figure 4.9: Extraction method “perceived treatment quality”

4.2 Hypothesis testing

4.2.1 Testing hypothesis 1, 2 and 3Figure 4.10 show the rotated component matrices. The construct “expected treatment quality” is based on two underlying components. Bowman, Heilman and Seetharaman (2004, p.326) argue that normative decision theory suggests that any decision strategy has benefits and costs associated with its use. The authors argue that the benefits from compliance follow directly from higher performance evaluations for and satisfaction with the product’s efficacy and effectivenss (Bowman, Heilman and Seetharaman, 2004, p.326). The costs of compliance include monetary, time and usage costs (Bowman, Heilman and Seetharaman, 2004, p.327). Based on Bowman, Heilman and Seetharaman (2004), efficacy and effectiveness are labeled as “expected treatment benefits” and convenience, equity, efficiency and side effects are labeled as “expected treatment costs”. The construct “perceived treatment quality” is based on one underlying component. Steenkamp (1990) argues that perceived quality is overall unidimensional evaluative judgment. It is a global summary construct (Steemkamp, 1997). Therefore, there is no need to re-lable the construct “perceived treatment quality”.

ComponentExpTreatmCosts ExpTreatmBenefits

Convenience .788Equity .731Side effects .694Efficiency .629Acceptability .513Effectiveness .901Efficacy .848

Figure 4.10: Rotated component matrices “expected treatment quality and “perceived treatment quality”

Data aggregation result in a new multiple regression model where the underlyingcomponents of “expected treatment costs” and “expected treatment benefits” are implemented:

Patient compliance i = γ0+ γ1Confirm/DisconfirmTreamQuali + γ2PercTreatmQuali + γ3ExpTreatmCostsi +γ4ExpTreatmBenefitsi +e1i

To draw conclusions about a population based on regression analysis done on a sample, several assumptions must be true (Field, 2009, p.220). The literature is not consistent about the amount of assumptions that must be met. Field (2009) provide nine assumptions that


19

must be met and Allison (1999) argues that five assumptions have to be and several other authors argue that there are only four assumptions that have to be met. A short literature review result in five assumptions which are quit consistent in literature. The assumptions are: (1) multicollinearity, (2) homoscedasticity, (3) independent errors, (4) Normally distributed errors and (5) linearity. The first assumption, multicollinearity, is met. The correlation matrix does not show correlation higher than 0.80 or 0.90, based on this rule there is no concern for multicollinearity (Field, 2009). Another multicollinearity diagnose is the variance inflation factor (VIF). Although there are no hard and fast rules about what value of the VIF should cause concern, Myers (1990) suggests that a value of 10 is a good value at which to worry Field, 2009, p.224). This is not the case for this research, all values are below 10. Related to the VIF is the tolerance statistics, which is its reciprocal (1/VIF). As such, values below 0.2 or 0.1 indicate serious problems (Field, 2009, p.224) which is not the case for this research. The second assumption, homoscedasticity, is met. The predictor variables “confirmed/ disconfirmed treatment quality” (F (28, 22.347) = 2.55, ns), “perceived treatment quality (F (28, 23) = .170, ns), “expected treatment costs” (F (28, 16) = 5.53, ns) and “expected treatment benefits” (F (28, 12) = .172, ns) all show results that are statistically not significant. This indicates that residuals at each level of the predictor(s) have the same variance (Field, 2009, p.220). The third assumption, independent errors, is met. The residual terms are uncorrelated. Durbin-Watson test ( Durbin-Watson = 1.301) lays between the rule of thumb of one and three. Finally the results show that the fourth and fifth assumptions are met. The residuals in de model are normally distributed and the predictors lay along a straight line.

Figure 4.11 shows the model summary. The first model consists out of the dependent variable “patient compliance” and the independent variable “confirmed/ disconfirmed treatment quality”. The second model consist out of the dependent variable “patient compliance” and the constructs “confirmed/ disconfirmed treatment quality” and “perceived treatment quality”. Finally the third model is the same as the second model completed with the constructs “expected treatment costs” and “expected treatment benefits” as independent variable. The results for the third model are significant (p < .1). The third model shows the largest correlation (R = .291) and take account for 8.5% of the variation in “patient compliance”. This indicates that 91.5% of the variance in “patient compliance” is unexplained which indicates that there are a large number of factors affect “patient compliance” The adjusted R square, which shows how well the model generalizes and give a more realistic indication of the variance explained (Voeten and van den Bercken, 2004), generalizes better in the third model (adjusted R² = .085) than in the first and second model.

Model R R Square Adjusted Std. Error R Square of the Estimate

1 .038 .001 -.008 1.5592 .117 .014 -.005 1.5563 .291 .085 .048 1.514

Change Statistics Durbin-Watson

R Square Change F Change df1 df2 Sig. F Change.001 .154 1 104 .696.012 1.285 1 103 .260.071 .071 2 101 .023 1.301

Figure 4.11: Model summary


20

Figure 4.12 shows that ANOVA is significant for the third model (p < .1). The third model has the highest F-ratio (2,335) and significantly improve the ability to predict the outcome variable. Based on model summary and ANOVA, testing hypothesis one, two and three is based on the third model.

Model Sum of Squares df Mean Square F Sig.1 Regression .374 1 .374 .154 .696

Residual 252.617 104 2.429Total 252,991 105

2 Regression 3.487 2 1.743 .720 .489Residual 249.504 103 2.422Total 252.911 105

3 Regression 21.411 4 5.353 2.335 ,061Residual 231.580 101 2.293Total 252.911 105

Figure 4.12: ANOVA

Figure 4.13 shows the coefficients of the different constructs. The construct “confirmed/ disconfirmed treatment quality” show a negative direct effect (γ1 = -.079) which is not statistically significant. “Perceived treatment quality” show a negative direct effect which is not statistically significant (γ2= -.068). “Expected treatment costs” show a negative direct effect (γ3= .209) and is not statistically significant. Finally, the results show a positive direct effect of “expected treatment benefits” which is statistically significant (γ4= .438, p < .1). Because of not significant results, there is no support for hypothesis 1, 2 and 3. The results show that “confirmed/ disconfirmed treatment quality”, “perceived treatment quality” and “expected treatment quality dot not affect “patient compliance” significantly. Despite the not significant results, the negative effect of “confirmed/ disconfirmed treatment quality” indicates that there is a mismatch between “expected treatment quality” and “perceived treatment quality” (Oliver, 1980; 1993). Patients seem “disconfirmed” which lead to decreasing “patient compliance”. The results show that “expected treatment costs” has a positive effect where a negative effect is normally expected. This indicates that people want to make costs in exchange of a treatment of good quality. The marketing literature argue that price can be seen as one of the indicators of quality and that price is correlated with quality (Dodds, Monroe and Grewal, 1991; Zeithaml, 1988; Gerstner, 1985).

Model UC* SC* t Sig.γ Std. Error Beta

3 (Constant) 2.034 .157 12.972 ,000Confirm/ DisconfirmTreatmQual -.079 .099 -.085 -.793 .430PercTreatmQual -.068 .217 -.044 -.315 .754ExpTreatmCosts .209 .195 .135 1.072 .286ExpTreatmBenefits .438 .158 .282 2.762 .007

95% Confidence Interval for B Correlations Collinearity StatisticsLower Bound Upper Bound Zero-order Partial Part Tolerance VIF1.723 2.345-.275 .118 -.038 -.079 -.075 .797 1.255-.498 .362 .082 -.031 -.030 .264 2.154-.178 .597 .082 .106 .102 .572 1.750.123 .752 .262 .265 .263 .689 1.150

* UC = Unstandardized Coefficients, SC = Standardized Coefficients Figure 4.13: Coefficients different constructs


21

4.2.3 Testing hypothesis 4The bipolar rating scale to measure “confirmed/ disconfirmed treatment quality” is changed in a unipolar scale. Ratings ranging from 1 till 3 are labeled as “disconfirmation” and ratings ranging from 5 till 7 are labeled as “confirmation”. Data aggregation results in a new multiple regression model:

Patient compliance i = β0 + β1ConfirmTreatmQuali + β2DisconfirmTreatmQuali + β3PercTreatmQuali + β4ExpTreatmCostsi +β5ExpTreatmBenefitsi

To draw conclusions about a population based on a regression analysis, the assumptions are test (Field, 2009). The first assumption, multicollinearity, is met. The correlation matrix does not show correlation higher than 0.80 or 0.90, based on this rule there is no concern for multicollinearity (Field, 2009). All VIF values are below 10 and the tolerance statistics are not below 0.2 or 0.1. The second assumption, homoscedasticity, is met. The predictor variables “confirmed treatment quality” (F (28, 19) = 2.377, ns), “disconfirmed treatment quality” (F (28, 21) = 1.698, ns) “perceived treatment quality (F (28, 23) = .170, ns), “expected treatment costs” (F (28, 16) = .172, ns) and “expected treatment benefits” (F (28, 12) = .5.534, ns) all show results that are statistically not significant. This indicates that residuals at each level of the predictor(s) have the same variance (Field, 2009, p.220). The third assumption, independent errors, is met. The residual terms are uncorrelated. Durbin-Watson test (Durbin-Watson = 1.342) lays between the rule of thumb of one and three. Finally the results show that the fourth and fifth assumptions are met. The residuals in de model are normally distributed and the predictors lay along a straight line.

Figure 4.14 shows the model summary. The results for the third model are significant (p < .01). The third model show the highest correlation (R = .326) and explains 10.6% of the variance in “patient compliance”. This indicates that 89.4% of the variance in “patient compliance” is unexplained, which indicates that there are a lot of factors affect “patient compliance”. The explanation in variance of the complete model for hypothesis 4 increase with 24.7% ((10.6% - 8.5%) / 8.5% = 24.7%) compared to the complete model of hypothesis 1, 2 and 3. The results indicate that the separation of “confirmation/disconfirmation” is mainly responsible for this increase. The first model of hypothesis four 4, where “confirmed/ disconfirmed treatment quality” is treated separately, explains 10% more of the variance in “patient compliance” compared to the first model of hypothesis 1, 2 and 3 where “confirmed/ disconfirmed treatment quality” is not treated separately. The absolute explanation power of implementing “expected treatment costs” and “expected treatment benefits” to complete the third model of hypothesis 4 (10.6% - 2.5% = 8.1%) is also larger than the absolute explanation power of implementing ‘”expected treatment costs” and “expected treatment benefits” to complete the third model of hypothesis 1, 2 and 3 (8.5% - 1.4% = 7.1%).

Model R R Square Adjusted Std. Error R Square of the Estimate

1 .103 .011 -.009 1.5592 .157 .025 -.004 1.5553 ,326 .106 ,061 1.504

Change Statistics Durbin-Watson

R Square Change F Change df1 df2 Sig. F Change.011 .547 2 103 .580.014 1.464 1 102 .229.082 4.561 2 100 .013 1.342

Figure 4.14: Model summary


22

Figure 4.15 shows that ANOVA is significant for the third model (p < .1). The third model has the highest F-ratio (2,335). The third model significantly improves the ability to predict the outcome variable. Based on the model summary and ANOVA, testing hypothesis four is based on the third model.

Model Sum of Squares df Mean Square F Sig.1 Regression 2.660 2 1.330 .547 .580

Residual 250.331 103 2.430Total 252.991 105

2 Regression 6.201 3 2.067 .854 .467Residual 246.789 102 2.420Total 252.991 105

3 Regression 26.830 5 5.366 2.373 .044Residual 226.160 100 2.262Total 252.991 105

Figure 4.15: ANOVA model

Figure 4.16 shows the coefficients for the third model. “Confirmation” show a positive statistically significant direct effect (β1 = .111, p <.1). “Disconfirmation” shows a negative direct effect (β2 = -.141) which is not statistically significant. Because of the not significant result of “disconfirmed treatment quality”, there is no support for hypothesis 4. The results indicate that the negative effect of “disconfirmed treatment quality” is larger than the positive effect of “confirmed treatment quality”. This is in line with Baumeister, Bratlavsky, Finkenauer and Vohs (2001) who argue that negative events in daily life have a larger impact than positive events. Negative events related to the quality of the treatment have a larger impact than positive events related to the quality of the treatment. It is not completely sure due to the fact that “disconfirmed treatment quality” is not statistically significant and due to the fact that “disconfirmed treatment quality” shows a larger standard error than “confirmed treatment quality” (.165 > .064). This larger variance indicates that “disconfirmed treatment quality” is more variable and less representative to the population than “confirmed treatment quality”.

Model UC* SC* t Sig.β Std. Error Beta

3 (Constant) 2.419 .306 7.898 ,000ConfirmTreatmQual .111 .064 -.218 -1.738 .085DisconfirmTreatmQual -.141 .165 -.103 -.857 .394PercTreatmQual -.029 .213 -.019 -.136 ,892ExpTreatmCosts .179 .196 .116 .914 .363ExpTreatmBenefits .478 .159 .308 3.008 .003

95% Confidence Interval for B Correlations Collinearity StatisticsLower Bound Upper Bound Zero-order Partial Part Tolerance VIF1.811 3.026-.237 .016 -.074 -.171 -.164 .570 1.626-.469 .186 -.013 -.085 -.081 .615 1.755-.452 .394 .082 -.014 -.013 .474 2.110-.210 .568 .082 .091 .086 .560 1.785.163 .793 .262 .288 .284 .854 1.170

* UC = Unstandardized Coefficients, SC = Standardized CoefficientsFigure 4.16: Coefficients


23

4.2.5 Testing hypothesis 5 and 6The first step in testing hypothesis five and six is to regress the mediator variable with the independent variables (Baron and Kenny, 1986). Because changes are made due to data aggregation the new multiple regression model is:

Confirm/DisconfirmTreatmQuali = η0 + η1PercTreatmQuali + η2ExpTreatmCostsi + η3ExpTreatmBenefits + e3i

Figure 4.17 shows the coefficients. “Perceived treatment quality” shows a positive effect (η1 =.753, p < .1) which is statistically significant. ”Expected treatment costs” shows a positive direct effect (η2= .020) which is not statistically significant. Finally, “expected treatment benefits” shows negative direct effect which is not statistically significant (η3= -.057).

The second step regresses the dependent variable “patient compliance” with the independent variables “expected treatment quality” and “perceived treatment quality” (Baron and Kenny, 1986). Because changes are made due to data aggregation, the new multiple regression model is:

Patient compliancei= δ0 + δ1PercTreatmQuali + δ2ExpTreatmCostsi + δ3iExpTreatmBenefits + e4i

The coefficients shows a negative direct effect of “perceived treatment quality” on “patient compliance (δ1 = -.127) which is not statistically significant. “Expected treatment costs” shows a positive direct effect which is not statistically significant (δ2 = .208) and “expected treatment benefits” shows a positive direct effect (δ3 = .442, p < .1) which is statistically significant.

The third step suggest by Baron and Kenny (1986) regresses the dependent variable on both the independent variables and the mediator variable (Baron and Kenny, 1986). “Patient compliance” is treated as outcome variable. “Expected treatment costs”, “expected treatment benefits”, “perceived treatment quality” and “confirmed/ disconfirmed treatment quality” are treated as predictor variables. Because changes are made a new model is developed:

Patient compliancei = θ0+ θ1PercTreatmQuali + θ2ExpTreatmCosts i + θ3iExpTreatmBenefits + θ4Confirm/DisconfirmTreatmQuali +e5i

Figure 4.17 shows that “perceived treatment quality” has a negative effect controlling for mediation (θ1 = -.068) which is not statistically significant. “Expected treatment costs” shows a positive effect controlling for mediation which is not statistically significant (θ2 = .268). “Expected treatment benefits” shows a positive direct effect which is statistically significant (θ3 = .438, p < .1) controlling for mediation. The results show that the effect of “perceived treatment benefits” on “patient compliance” is neither fully mediated neither partial mediated by “confirmed/ disconfirmed treatment quality” due to not significant results. The results show that the effect of “expected treatment costs” on “patient compliance” is neither fully neither partially mediated by “confirmed/ disconfirmed treatment quality”. The results do not show significant results to the conditions that must hold. Finally, the results show that the effect of “expected treatment costs” on “patient compliance” is not fully mediated by “confirmed/ disconfirmed treatment quality”. The results do not show significant results to the conditions that must hold. The results show no support for hypothesis 5 and 6. Despite not significant results the direct effects of “perceived treatment quality”, “expected treatment costs” and “expected treatment benefits” are not fully mediated by “confirmed/ disconfirmed treatment quality”. The results show that “confirmed/ disconfirmed treatment quality” is highly driven by “perceived treatment quality” (η1 = .753, p < .1) and that “patient compliance” is highly driven by “expected treatment benefits (δ3 = .442, p < .1) which is statistically significant. Despite not


24

significant results, “expected treatment costs” and “expected treatment benefits” seem to affect “confirmed/ disconfirmed treatment quality” respectively slightly positive (η2 = .020) and slightly negative (η3 = -.057). The results show that “patient compliance” is highly driven by “expected treatment benefits” (δ3 = .442, p < .1). Despite not significant results, “expected treatment costs” (δ2 = .208) and “perceived treatment quality (δ1 = -.127) seem to affect “patient compliance” respectively positive and negative.

The results show that “expected treatment benefits”, and despite non significance the results indicate also “expected treatment costs”, drive “patient compliance” find support by Wosinka (2005) and Bowman, Heilman and Seetharaman (2004). These authors argue that the cost and benefits paradigm is associated with “patient compliance”. Patients compare the costs and benefits of a prescribed treatment in as one of the factors to decide to be compliant or to be non-compliant. The results show also the “perceived treatment quality” drives “confirmed/ disconfirmed treatment quality” statistically significant. This result find support by de Ruyter, Bloemer and Peeters (1997). The authors argue that perceptions of quality drive disconfirmation more than expectations. This is also the results of this research. “Perceived treatment quality” drive “comfirmation/ disconfirmation” more than “expected treatment costs” and “expected treatment benefits”. This indicates that patients base their “confirmed/ disconfirmed treatment quality” mainly on comparing quality performance judgments between two or more treatments instead of letting their expectations affect “confirmed/ disconfirmed treatment quality”.

Steps in testing mediation UC* SC* t Sig.η Std. Error Beta

Testing step 1 (path a) Outcome: Confirm/ DisconfirmTreatmQual(Constant) .547 .147 3.728 .000PercTreatmQual .753 .203 .451 3.705 .000ExpTreatmCosts .020 .195 .012 .100 .920ExpTreatmBenefits -.057 .158 -.034 -.361 .719

δTesting step 2 (path c) Outcome: Patient compliance

(Constant) 1.991 .147 13.559 .000PercTreatmQual -.127 .203 -.082 -.627 .532ExpTreatmCosts .208 .195 .134 1.066 .289ExpTreatmBenefits .442 .158 .285 2.797 .006

θTesting step 3 (paths b and c') Outcome: Patient compliance

(Constant) 2.034 .157 12.972 .000PercTreatmQual -.068 .217 -.044 -.315 .754ExpTreatmCosts .209 .195 .135 1.072 .286ExpTreatmBenefits .438 .158 .282 2.762 .007Mediator: Confirm/ DisconfirmTreatmQual (path b) -.079 .099 -.085 -.793 .430

* UC = Unstandardized Coefficients, SC = Standardized Coefficients Figure 4.17: Coefficients mediation effect


25

5 Conclusion, implications, limitations and future research directions

This chapter discuss the different research questions stated earlier. After answering the different research questions, implications and future research directions are provided.

5.1 Conclusion and implications

5.1.1 ConclusionsThe research extend earlier research of Morris (1992), Cochrane, Horne and Chanez (1999) and Stremersch and van Dyck (2009) by investigating “patient compliance”. Furthermore this research deals with future research directions of Cochrane, Horne and Chanez (1999) by investigating the role of “expected treatment quality” and “perceived treatment quality” and “confirmed/ disconfirmed treatment quality” on “patient compliance”.

This research faces a small sample size. A necessary step to avoid non-significant results is to aggregate the data and develop new measurement models. The results show that after data aggregation, only “expected treatment benefits” show a positive direct effect on “patient compliance” which is statistically significant. The results provide not enough significant evidence to determine if the negative effect of “disconfirmed treatment quality” is larger than the positive effect of “confirmed treatment quality”. Furthermore, this research shows that “confirmed/ disconfirmed treatment quality” does not mediate the effects of “perceived treatment quality”, “expected treatment costs” and “expected treatment benefits” on “patient compliance”. Finally, the rotated component matrices show that the different underlying quality concept have different influence on “patient compliance”. For example, the quality concepts “effectiveness” and “equity” show influences larger than ,8. This means that these quality concepts are assessed as very important.

5.1.2 ImplicationsThe pharmaceutical business face major problems with non-compliance patients. Beside the billions of costs caused by non compliant patients, in the US 125,000 patient with heart disease alone died because of not being compliant towards their prescribed treatment. In the Netherlands, each year hundreds of patient die because of non-compliance (Pfizer, 2008). Appendix C shows the compliance rates and the causes of non-compliance based on this research. Of the patients that participate this research, 60% is fully compliant to their prescribed treatment. 40% is not (fully) compliant to their prescribed treatment ranging from “very frequently” to “very rarely”. This results are in line with earlier research of Spector (2000). The results show that most patients (34%) agree in some manner that out of pocket costs caused non-compliant behavior. Other causes do not show major differences in agreement compared to each other. Most of the patients disagree with the statements. This indicates that there are a lot more causes that affect “patient compliance”.

The main question is; how can pharmaceuticals and physicians increase patient compliance? This research provide two areas that can provide solutions for this problem. (1) Patients form expectations about the costs and benefits of the quality of a treatment, whereas the benefits significantly drives “patient compliance”. Dellande, Gilly and Graham (2004) suggest that role clarity can improve “patient compliance”. Pharmaceuticals and physicians could dialog with patients to shape their expectations from a cost and benefit point of view. This role clarity can also help to motivate patients to follow a treatment as is prescribed. A second method to shape expectations that lead to increasing “patient compliance” is the movement from the traditional white coat model to a shared decision making model (Camacho, Landsman and Stermersch, 2008). Physicians can discuss during a clinical intervention what patients expect of the new prescribed treatment and increase the patients


26

knowledge about the need of the treatment (Hopfield, Linden and Tevelow, 2006). The physician can also observe if patients form expectations that are not realistic. If this is the case, physicians can provide more realistic expectations. Bridging this gaps at the begin of the treatment will have a better effect on “patient compliance” rather than the gaps must be bridged during the treatment. (2) The perceptions that patients form about the quality of a treatment affect “confirmed/ disconfirmed treatment quality” significantly. Salience and mindfulness can have a positive effect on the patients perceptions, which positively affect the satisfaction judgments and “patient compliance”. Bowman, Heilman and Seetharaman (2004) state that salience/ mindfulness peaks at the time of a service provider intervention (Bowman, Heilman and Seetharaman, 2004, p. 326). Literature support the fact that between clinical intervention “patient compliance” decrease and that “patient compliance” increase at the time of clinical intervention (Cramer, Scheyer and Mattson, 1990). In line with this, salience increase “patient compliance” (Ailawadi and Scott, 1998) because patients want to answer yes when asked if they followed their treatment as prescribed, and more frequent clinical interventions increase the patients mindfulness of being non-compliant (Bowman, Heilmann and Seetharaman, 2004). Especially clinical interventions are very useful to increase “patient compliance”. During clinical interventions, pharmaceuticals and physicians get the opportunity to receive feedback from the patients side which can be used to improve treatment aspects which are judged as negative from patients side. Implementing this feedback can positively affect patients perceptions in the future which lead to satisfaction and increasing “patient compliance”. One company that is dealing with this is Eliza Corporation. Lucas Merrow, CEO of Eliza Corporation state that patients must be contacted when they start their treatment and stay in contact during their treatment. Patients need to know what they can perceive. The aim of this, is to build trust and make them feel confident. With this, it is possible to learn from patients and what patients motivate (Tolve, 2010). This can improve the perceptions that patients have of a treatment.

The provided solutions to increase “patient compliance” have one overall similarity; physicians must account for patients opinions about a treatment and their knowledge about a treatment There must be an open dialogue between the physicians and patients to positively affect “patient compliance”.

5.2 Limitations and future research directionsThere are several limitations about this research. First, most results are non significant due to a small sample size. To get a sample size that create significant results, beside the three earlier mentioned websites, the “Dutch Asthma Foundation” has been asked to publish the survey but the foundation rejected this. From their experience, asthma patients and other chronic diseased patients are asked for too many investigations, which results in a lack of motivation from patient side to collaborate in researches. Future research should provide a sample size that creates significant results. A way to do so is to use panel data or collaborate closely with physicians to let patients participate. Another option to create a sufficient large sample size is to have a medic ethical declaration. For a lot of physicians and patient organizations, a medical ethical declaration is a first priority in the decision to collaborate in a research.

Second, this research measures “patient compliance” of the last month. This has been done to avoid recall bias, but it also has drawbacks. For instance, asthma patients might be more compliant during summer than during winter and there are probably more different factors that influence “patient compliance” during a period of time. Future research should deal with this to pinpoint the degree of “patient compliance” in a period of time.

Third, this research focus only on the effects of expectations and perceptions about quality aspects on “patient compliance”. Expectations, perceptions and quality come forth out of other future research directions. Future research should consider determinants and aspects from a larger pool.

Fourth, data aggregation result in two underlying components for the construct “expected treatment quality” and one underlying component for the construct “perceived


27

treatment quality”. This research indicate that “expected treatment costs” and “expected treatment benefits” are the two underlying components that form “expected treatment quality” (Bowman, Heilman and Seetharamen, 2004) and the “perceived treatment quality” is a unidimensional evaluative judgment (Steemkamp, 1990). Probably there are other causes that results in this difference in components for the two constructs. Future research could pay attention to this and provide a answer for this difference.

Fifth, this research is highly based on prior work of Baron and Kenny (1986). Zhao, Lynch jr. and Chen (2010) argue that Baron and Kenny’s (1986) framework has several limitations. The authors argue that (1) Baron and Kenny (1986) hold full mediation as a golden standard where much research report partial mediation, (2) that the only requirement for mediation is that the indirect effect must be significant. Baron and Kenny (1986) argue that the direct and indirect effect must be significant and (3) to test the significance of an indirect effect, the more rigorous and powerful bootstrap test must be used and not Sobel’s test. Future research should consider the paper of Zhao, Lynch jr. and Chen (2010, p.197) which provide a decision tree and a step-by-step procedure for testing mediation, classifying its type and interpreting the implications of findings for theory building and future research.

Finally, this research measured the different constructs in general for a asthma treatment. There are different types of medications to treat asthma like for instance “inhaled steroids”, “leukotriene modifiers” and “immunomodulators”. Future research could focus on one of the medications and test if there are differences in “patient compliance” between medications. One problem that possibly arise then, is to generate a sample size that is large enough to collect significant results.


28

ReferencesAilawadi, K. and Scott, Neslin A. (1998), “The effect of promotion on consumption: buying more and consuming it faster”, Journal of Marketing Research, 35 (August), 390-398.

Allison, Paul D. (1999), Multiple regression: a primer, California: Pine Forge Press.

Asch, Steven M., Kerr, Eve A., Keesey, J., Adams, John L., Setodji, Claude M., Malik, S. and McGlynn, Elizabeth A. (2006), “Who is at greatest risk for receiving poor-quality health care?, The New England Journal of Medicine, vol. 354, 1146-1156.

Ashcroft, J.J., Leinster, S.J. and Slade, P.D. (1986), “Mastectomy vs breast conservation: Psychological effects op patient choice of treatment”, in Psychological Issues in Malignant Disease, Pergamon Press, Oxford.

Astmafonds (2007), “Astma”, (accessed April 7, 2010), [available at http://www.astmafonds.nl].

Bates, Andree K. (2008), “Patient compliance programs: how to ensure they are not doomed to fail”, DTC Perspectives, 38-40.

Baumeister, Roy F., Bratlavsky E., Finkenauer, C. and Vohs, Kathleen D.(2001), “Bad is stronger than good”, Review of General Psychology, vol. 5 (4), 323-370.

Baron, R.M. and Kenny, D.A. (1986), “The moderator-mediator variable distinction in social psychological research: conceptual, strategic and statistical considerations, Journal of Personality and Social Psychology, vol. 51, 1173-1182.

Becker, Marshall H. and Maiman, Lois A. (1980), “Strategies for enhancing patient compliance”, Journal of Community Health, vol. 6 (2), 113-135.

Bearden, William O. and Netemeyer, Richard G. (1996), Handbook of marketing scales: multiple item measures for marketing and consumer behavior research, California: Sage publication Inc., second edition.

Blodgett, Jeffrey G., Granbois, Donald H. and Walters, Rockney G. (1993), “The effects of perceived justice on complainants’ negative word-of-mouth behavior and repatronage intentions”, Journal of Retailing, vol. 69 (4), 399-428.

Bloom, Bernard S. (2001), “Daily regimen and compliance with treatmen”, British Medical Journal,vol. 323, 647.

Brink, van den A., and Dulmen, van A.M. (2004), “Factors related pharmacy therapy compliance of chronicle diseases (translated)”, Nivel, 1-37.

Bowman, D., Heilman, Carrie M. and Seetharaman, P.B. (2004), “Determinants of product-use compliance behavior”, Journal of Marketing Research, vol. 41 (3), 324-338.

Camacho, N., Landsman, V. and Stremersch, S. (2008), “The connected patient”, in The Connected Customer: The chancing nature of consumer and business markets, Stefan H.K. Wuyts, ed, New-York: Routledge Academic.

Campbell, S.M., Roland, M.O. and Buetow, S.A. (2000), “Defining quality of care”, Social Science & Medicine, vol. 51, 1611-1625.


29

Charles, C., Gafni, A. and Whelan, T. (1999), “Decision-making in the physician-patient encounter: revisiting the shared treatment decision-making model”, Social Science & Medicine, vol. 49 (5), 651-661.

Christensen, Clayton M. and Hwang, J. (2009), “Power to the patients”, (accessed February 17, 2010), [available at http://www.theatlantic.com].

Churchill, Gilber A. and Surprenant, C. (1982), “An investigation into the determinants of customer satisfaction”, Journal of Marketing Research, vol. 19 (4), 491-504.

Cochrane G.M., Horne, R. and Chanez, P. (1999), “Compliance in Asthma”, Respiratory Medicine, vol. 93, 763-769.

Cooper, Donald R. and Schindler, Pamela S. (2006), Business Research Methods, New York: McGraw-Hill, ninth edition.

Cramer, Joyce A., Richard, Scheyer D. and Richard, Mattson H. (1990), “Compliance declines between clinical visits”, Archives of Internal Medicine, 150 (7), 1377-1378.

Crosby, Philip B. (1979), Quality is free: the art of making quality certain, New York: New American Library.

Dellande, S., Gilly, Mary C. and Graham, John L. (2004), “Gaining compliance and losing weight: the role of the service provider in health care services”, Journal of Marketing, vol. 68 (7), 78-91.

Diefenbach, Michael A., Weistein, Neil D. and O’Reilly, J. (1993), “Scales for assessing perceptions of health hazard susceptibility”, Health Education Research, vol. 8 (2), 181-192.

DiMatteo, M.R. (2004), “Variations in patients’ adherence to medical recommendations: A quantitative review of 50 years of research”, Medical Care, vol. 42 (3), 200-209.

Dodds, William B., Monroe, Kent B. and Grewal, D. (1991), “Effects of price, brand and store information on buyers’ product evaluations”, Journal of Marketing Research, vol. 28 (3), 307-319.

Donabedian, A. (1990), “The seven pillars of quality”, Archives of Pathology & Laboratory Medicine, vol. 114 (11), 1115-1118.

Eliashberg, J., Elberse, A., Leenders, M. (2006), “The motion picture industry: Critical issues in practice, current research and new research directions”, Marketing Science, vol. 25 (6), 638-661.

Field, A. (2009), Discovering statistics using SPSS, Sage, London, third edition.

Fincham, Jack E. (2007), Compliance with medications: issues and opportunities, Pharmaceutical Products Press, Binghamton.

Fisher, Robert J. (1993), “Social desirability bias and the validity of indirect questioning”, Journal of Consumer Research, vol. 20 (2), 303-315.

Gerstner, E. (1985), “Do higher prices signal higher quality?”, Journal of Consumer Research, vol.22 (2), 209-215.


30

Green, S.B. (1991), “How many subjects does it take to do a regression analyses?”, Multivariate Behavioral Research, vol. 26, 499-510.

Gronroos, C. (1982), Strategic management and marketing in the service sector, Helsingfors, Swedish School of Economics and Business Administration.

Grundmeijer, H.G.L.M. , Reenders, K. and Rutten, G.E.H.M. (2004). The medical process: clinical reasoning from complaint to thereapy (translated), Maarssen: Elsevier Healthcare, fourth edition, 181.

Guyatt, Gordon H., Feeny, David H. and Patrick, Donald L., (1993), “Measuring health-related quality of life”, Annals of Internal Medicine, vol. 118, 622-629.

Heather, McDonald P., Amit, Garg X. and Haynes, Brian R. (2002), “Interventions to enhance patient adherence to medication prescriptions”, Journal of the American Medical Association, vol 288 (22), 2868-2879.

Higginbottom, Jayne M. and Hurst, K. (2001), “Quality assuring a therapy service”, International Journal of Health Care Quality Assurance, vol. 14 (4/5), 149-156.

Hopfield, J., Linden, R. and Tevelow, B. (2006), “Getting patients to take their medicine”, The McKinsey Quarterly, nr. 4, 12-15.

Hosken, J. (2006), “Patient compliance Europe: in-depth report from an eyeforpharma conference”, (accessed July 9, 2010), [available at http://www.keywordpharma.com].

Hubbert, Amy R., Sehorn, Annetta G. and Brown, Stephan W. (1995), “Service expectations: the consumer versus the provider”, International Journal of Service Industry Management, vol. 6 (1), 6-21.

Jayasankar, Jay S. (2009), “Patient expectations: How do they matter?”, (accessed February 27, 2010), [available at http://www.aaos.org].

Joos, G.F. and Bel, E.H.D. (1999), “Asthma at adults”, in Learning book lung sicknesses, H.C. Hoogsteden, e.d. Maarssen: Elsevier, 65-81.

Kaptein, A.A. (2006), Medical Psychology (translated), Bohn Stafleu van Loghum, first edition, Houten.

Lehtinen, U. and Lehtinen, Jarma R. (1982), “Service quality: a study of quality dimensions”, unpublished working paper, Helsinki: Service management institute, Finland OY.

Luce, M.F. and Khan, Barbara E. (1999), “Avoidance or vigilance: the psychology of false-positive test results”, Journal of Consumer Research, vol. 26 (3), 242-259.

Maxwell, Robert J. (1992), “Dimensions of quality revisited: from thought to action”, Quality in Health Care, vol. 1, 171-177.

McCluskey, Jill J. and Swinnen, Jonathan F.M. (2004), “Political economy of the media and consumer perceptions of biotechnology”, American Journal of Agricultural Economics, vol. 86 (5), 1230-1237.


31

McHorney, A. Colleen (2009), “The adherence estimator: a brief, proximal screener for patient propensity to adhere to prescription medications for chronic disease”, Current Medical Research and Opinion, vol. 25 (1), 215-238.

McKinley, R.K., Stevenson, K., Adams, S. and Manku-Scott, T.K. (2002), “Meeting patient expectations of care: the major determinant of satisfaction with out-of-hours primary medical care?”, Family Practice, vol. 19 (4), 333-338.

Mendonca, P.J. and Brehm, S.S. (1983), “Effects of choice on behavioral treatment of overweight children”, Journal of Social Clinical Psychology, vol. 1, 343-358.

Morris, L.S. (1992), “Patient compliance: an overview”, Journal of clinical pharmacy and therapeutics, vol. 17 (5), 283-295.

Monroe, Kent B. and Krishnan, R. (1985), “The effect of price on subjective product evaluations”, in Perceived Quality, Jacob Jacob and Jerry C. Olson, eds. Lexington, MA: Lexington Books, 209-232.

Morgan, M. (2003), The doctor-patient relationship, in Sociology as applied to medicine, Graham Scambler, Philadelphia: Saunders Ltd., fifth edition, 49-66.

Myers, R. (1990), Classical and modern regression with application, Boston MA: Duxbury, second edition.

Nederhof, A.J. (1985), “Methods of coping with social desirability bias: a review”, European Journal of Psychology, vol. 15 (3), 263-280.

Niedrich, Ronald W., Kiryanova, E. and Black, William C. (2005), “The dimensional stability of standards used in the disconfirmation paradigm”, Journal of Retailing, vol. 81 (1), 49-57.

Oliver, Richard L. (1977), “Effect of expectation and disconfirmation of postexposure product evaluations: an alternative interpretation”, Journal of Applied Psychology, vol. 62 (4), 480-486.

Oliver, Richard L. (1980), “A cognitive model of antecedents and consequence of satisfaction decisions”, Journal of Marketing Research, vol. 17, 460-469.

Oliver, Richard L. (1981), “Measurement and evaluation of satisfaction process in retail settings”, Journal of Retailing, vol. 57 (Fall), 25-28.

Oliver, Richard L. (1993), “Cognitive, affective and attribute bases of the satisfaction response”, The Journal of Consumer Research, vol. 20 (3), 418-430.

Olson, J.C. and Dover, P. (1979), “Disconfirmation of consumer expectations through product trial”, Journal of Applied Psychology, vol. 64 (April), 179-189.

Parasuraman, A., Zeithaml, Valerie A. and Berry, Leonard L. (1985), “A conceptual model of service quality and its implications for future research”, Journal of Marketing, vol. 49 (fall 1985), 41-50.

Parasuraman, A., Zeithaml, Valerie A. and Berry, Leonard L. (1988), “SERVQUAL: a multiple-item scale for measuring consumer perceptions of service quality”, Journal of Retailing, vol. 64 (1), 12-37.


32

Peck, Mitchell B. et al. (2001), “Measuring patient expectations: Does the instrument affect satisfaction or expectations?”, Medical care, vol 39 (1), 100-108.

Pfizer (2008), “The people do not swallow it any more (translated), (accessed 9 July, 2010), [available at http://www.pfizer.nl].

Polit, Denise F. and Beck, Cheryl T. (2004), Nursing research: principles and methods, Philadelphia: Lippincott Williams & Wilkins, seventh edition, 356.

Rao, Akshay R. and Kent, Monroe B. (1989), “The effect of price, brand name and store name on buyers’ perceptions of product quality: an integrative review”, Journal of Marketing Research, vol. 26 (August), 351-357.

Ruyter, de K., Bloemer, J. and Peeters, P. (1997), “Merging service quality and service satisfaction. An empirical test of an integrative model”, Journal of Economic Psychology, vol.18 (4), 387-406.

Sasser, Earl W., Olsen, Paul R. and Wyckoff, Daryl D. (1978), Management of service operations: text and cases, Boston: Allyn & Bacon.

Sirohi, N., McLaughlin, Edward W. and Wittink, Dick R. (1998), “A model of consumer perceptions and store loyalty intentions for a supermarket retailer”, Journal of Retailing, vol. 74 (2), 223-245.

Sherbourne, Donald C., Hays, Ron D., Ordway, L., DiMatteo, Robin M. and Kravitz, Richard L. (1992), “Antecedents of Adherence to Medical Recommendations: Results from the medical outcomes study”, Journal of Behavioral Medicine, vol. 15 (5), 447-468.

Spector, S. (2000), “Noncompliance with Asthma Therapy- Are there any solution?”, Journal of Asthma, vol. 37 (5), 381-388.

Steenkamp, J.B.E.M. (1990), “Conceptual model of the quality perception process”, Journal of Business Research, vol. 21 (4), 309-333.

Steenkamp, J.B.E.M.(1997), “Dynamics in consumer behavior with respect to agricultural and food product”, in Agricultural marketing and consumer behavior in a changing world, Massachusetts: Kluwer Academic Publishers.

Stewart, David W. (2002), “Getting published: Reflections of an old editor”, Journal of Marketing, vol. 66 (10), 1-6.

Stremersch, S. (2008), “Health and Marketing: The emergence of a new field of research”, International Journal of Research in Marketing, vol. 25 (4), 1-19.

Stremersch, S. and Van Dyck, W (2009), “Marketing of the life sciences: a new framework and research agenda for a nascent field”, Journal of Marketing, vol. 73 (4), 4-30.

Sudman, S. and Bradburn, N.M. (1979), Response effects in surveys, Aldine, Chicago, Illinois.

Tolve, A. (2010), “Improving adherence by listening and learning”, (accessed 9 July, 2010), [available at http://www.eyeforpharma.com].


33

VanVoorhis, Wilson C.R. and Morgan, Betsy L. (2007), “Understanding power and rules of thumb for determining sample sizes”, Tutorials in Quantitative Methods for Psychology, vol. 3 (2), 43-50.

Verhage, B. (2004), Principles of marketing (translated), Groningen: Stenfert Kroese, fourth edition.

Vermeire, E., Hearnshaw, H., Royen, van P. and Denekens, J. (2001), “Patient adherence to treatment: three decades of research. A comprehensive review”, Journal of Clinical Pharmacy and Therapeutics, vol. 26, 331-342.

Voeten, M.J.M. and Bercken, van den J.H.L. (2004), “Regression analysis with SPSS: a guide for linear regression analysis with SPSS”, (accessed Augus 27, 2010), [available at http://www.socsci.ru.nl].

Woodruff, Robert B., Ernest, Cadotte R. and Jenkins, Roger L. (1983) “Modeling consumer satisfaction processes using experience-based norms, Journal of Marketing Research, vol. 20 (3), 296-304.

World Health Organization (2003), Adherence to Long-Term Therapies: Evidence for action, Geneva, Switzerland: World Health Organization.

World Health Organization (2010), “Definition of Asthma”, (accessed April 7, 2010), [available at http://www.who.int].

Wosinka, M. (2005), “Direct-to-consumer advertising and drug therapy compliance”, Journal of Marketing Research, vol. 42 (3), 323-332.

Zhao, H., Lynch jr., John G. and Chen, Q. (2010), “Reconsidering Baron and Kenny: myths and truths about mediation analysis”, Journal of Consumer Research, vol. 37 (august 2010), 197-206.

Ziegler, A. (2009), “Study: Patient non-compliance with drugs costs $290B a year”, (accessed February 19, 2010), [available at http://www.fiercehealthcare.com].

Zeithaml, Valerie A. (1989), “Defining and relating price, perceived quality and perceived value”, Journal of Marketing, vol. 52 (7), 2-22.

Zeithaml, Valerie A., Berry, Leonard L. and Parasuraman, A. (1993), “The nature and determinants of customer expectations of service”, Journal of the Academy of Marketing Science, vol. 21 (1), 1-12.

Zuger, A. (1998), “The “other” drug problem: forgetting to take them”, In: The New York , Times, quire Science, June 2 1998.


34

Appendix A Overview quality healthcare criteriaMaxwell HSRG Donabedian O'Leary & O'Leary1992 1992 1990 1992

Accessibility Accessibility AccesbilityPatient-centredness Patient-centredness

Effectiveness Effectiveness Effectiveness EffectivenessEfficiency Efficiency Efficiency Efficiency

Continuity ContinuityEfficacy Efficacy

Acceptability AcceptabilityEquity Equity

LegitimacyComprehensiveness

RelevanceOptimality

Figure A.1: Overview of criteria regarding quality of healthcare (Campbell et al, 2000, p.1615)


35

Appendix B Questionnaire

General instructionThe Erasmus University Rotterdam is investigating compliant behavior of patients. Patient compliance is an essential element within the medical process. Earlier investigation show that 70% of patients with a chronical disease is not compliant to their prescribed therapy.

The goal of this research is to better understand the antecedents of patient compliance. Your sincere answers are crucial to allow us to have a better understanding of this important issue, so we thank you in advance for your collaboration.

Next the survey will start. The total survey will take about 5 minutes. The results will only be used for social aims. Commercial aims are excluded. There are no right or wrongs answers. The results will be processed confidential and anonymous.

1 Have you been prescribed an asthma treatment within the last four months?0 Yes0 No**we thank you very much for your collaboration but this survey does not apply to you.

2 Year of birth:_________

3 Level op graduation:0 WO0 HBO0 VWO0 HAVO0 MBO0 MAVO0 Lower graduation0 No education

4 Gender:0 Male0 Female


36

The next questions are about your expectations prior to your treatment. Under treatment can be understood; all the activities advised by your medical specialist to improve your health.5. My expectations concerning the treatment were…

Acceptability...that it will be conform my preferences for an asthma treatment0 Totally disagree 0 Disagree 0 Disagree a bit 0 Neutral 0 Agree a bit 0 Agree 0 Totally Agree

Efficacy...that it will be directly operative to improve my health0 Totally disagree 0 Disagree 0 Disagree a bit 0 Neutral 0 Agree a bit 0 Agree 0 Totally Agree

Effectiveness...that is will be effective in realizing health improvements0 Totally disagree 0 Disagree 0 Disagree a bit 0 Neutral 0 Agree a bit 0 Agree 0 Totally Agree

Equity...that there are differences in treatment compared to other patients0 Totally disagree 0 Disagree 0 Disagree a bit 0 Neutral 0 Agree a bit 0 Agree 0 Totally Agree

Side effects (scale reversion)...that the treatment will cause side effects0 Totally disagree 0 Disagree 0 Disagree a bit 0 Neutral 0 Agree a bit 0 Agree 0 Totally Agree

Convenience...that it will be easy to follow0 Totally disagree 0 Disagree 0 Disagree a bit 0 Neutral 0 Agree a bit 0 Agree 0 Totally Agree

Efficiency...that it will improve my health maximum against the lowest costs0 Totally disagree 0 Disagree 0 Disagree a bit 0 Neutral 0 Agree a bit 0 Agree 0 Totally Agree


37

The next questions are about the judgments you have toward your asthma treatment. Under treatment can be understood; all the activities advised by your medical specialist to improve your health.6. Please rate the overall performance of the treatment quality of the treatment you follow compared to other treatment’s you followed.

Acceptability...it is conform my preferences for an asthma treatment0 Totally disagree 0 Disagree 0 Disagree a bit 0 Neutral 0 Agree a bit 0 Agree 0 Totally Agree

Efficacy...it is directly operative to improve my health0 Totally disagree 0 Disagree 0 Disagree a bit 0 Neutral 0 Agree a bit 0 Agree 0 Totally Agree

Effectiveness...it is effective in realizing health improvements0 Totally disagree 0 Disagree 0 Disagree a bit 0 Neutral 0 Agree a bit 0 Agree 0 Totally Agree

Equity...there are differences in treatment compared to other patients0 Totally disagree 0 Disagree 0 Disagree a bit 0 Neutral 0 Agree a bit 0 Agree 0 Totally Agree

Side effects (scale reversion)...the treatment cause side effects0 Totally disagree 0 Disagree 0 Disagree a bit 0 Neutral 0 Agree a bit 0 Agree 0 Totally Agree

Convenience...the treatment is easy to follow0 Totally disagree 0 Disagree 0 Disagree a bit 0 Neutral 0 Agree a bit 0 Agree 0 Totally Agree

Efficiency...the treatment improve my health maximum against the lowest costs0 Totally disagree 0 Disagree 0 Disagree a bit 0 Neutral 0 Agree a bit 0 Agree 0 Totally Agree


38

The next question will be about your confirmation towards the performance of the treatment.

7. Considering my expectations for an asthma treatment, the performance of this treatment was better than I expected0 Totally disagree 0 Disagree 0 Disagree a bit 0 Neutral 0 Agree a bit 0 Agree 0 Totally Agree

The next question will be about your compliant behavior. In the interest of the reliability of this research you are requested to fill in the question honestly.

8. Regarding your therapy of the last month, how compliant have you been?0 Not 0 Very Rarely 0 Rarely 0 Occasionally 0 Frequently 0 Very frequently 0 Completely

The last question will be about the causes of compliant behavior. In the interest of the reliability of this research you are requested to fill in the questions honestly.

9. Please indicate to what extent the causes below influence your degree of being compliant to your prescribed treatment

I forget to follow my treatment prescribed by my medical expert.0 Totally disagree 0 Disagree 0 Disagree a bit 0 Agree a bit 0 Agree 0 Totally Agree

It is hard to follow my prescribed treatment complete and correct.0 Totally disagree 0 Disagree 0 Disagree a bit 0 Agree a bit 0 Agree 0 Totally Agree

I forget to follow my treatment because of holidays or not being at home.0 Totally disagree 0 Disagree 0 Disagree a bit 0 Agree a bit 0 Agree 0 Totally Agree

.I am not convinced of the importance of my prescribed treatment.0 Totally disagree 0 Disagree 0 Disagree a bit 0 Agree a bit 0 Agree 0 Totally Agree

I worry that my prescription medication will do more harm than good to me.0 Totally disagree 0 Disagree 0 Disagree a bit 0 Agree a bit 0 Agree 0 Totally Agree

I feel financially burdened by my out-of-pocket expenses for my prescribed treatment.0 Totally disagree 0 Disagree 0 Disagree a bit 0 Agree a bit 0 Agree 0 Totally Agree


39

Category PercentageVery rarely 2%Rarely 3%Occasionally 5%Frequently 10%Very frequently 20%Completely 60%

Appendix C Supportive output

Figure C.1: Percentage of patients that comply to treatment

Totally disagree Disagree Disagree a bit Agree a bit Agree Totally AgreeForget 4% 56% 19% 11% 8% 2%Hard to follow 2% 47% 30% 10% 10% 1%Holidays, not at home 3% 49% 23% 18% 7% 0%Not convinced 5% 53% 21% 13% 6% 2%More harm than good 3% 51% 29% 12% 4% 1%Financialy burdened 2% 47% 17% 18% 11% 5%

Figure C.2: Causes of patient compliance


40

1 b.a.j. (336200).doc · web viewscholars as sasser, olsen and wyckoff (1978), gronroos (1982)...

Documents