<<

J Behav Med (2009) 32:278–284 DOI 10.1007/s10865-009-9202-y

Predictors of adherence to : the role of disease and beliefs

Devin M. Mann Æ Diego Ponieman Æ Howard Leventhal Æ Ethan A. Halm

Received: November 5, 2008 / Accepted: January 9, 2009 / Published online: January 30, 2009 Ó Springer Science+Business Media, LLC 2009

Abstract Despite the effectiveness of drug therapy in suboptimal beliefs are potentially modifiable and are log- diabetes management high rates of poor adherence persist. ical targets for educational interventions to improve dia- The purpose of this study was to identify potentially betes self-management. modifiable patient disease and medication beliefs associ- ated with poor medication adherence among people with Keywords Diabetes Á Medication adherence Á Health diabetes. A cohort of patients with diabetes was recruited beliefs Á Self-regulation model from an urban primary-care clinic in New York City. Patients were interviewed in English or Spanish about: disease beliefs, medication beliefs, regimen complexity, Introduction diabetes knowledge, depression, self-efficacy, and medi- cation adherence (Morisky scale). Logistic regression was Type 2 diabetes is an epidemic affecting approximately one used to identify multivariate predictors of poor medication in ten US adults at an estimated cost in 2007 of 174 billion adherence (Morisky [ 1). Patients (n = 151) had diabetes dollars (American Diabetes Association 2008a, b; National for an average of 13 years with a mean HgA1C of 7.6 (SD Diabetes Information Clearinghouse. National Diabetes 1.7). One-in-four (28%) were poor adherers to their dia- Statistics: NIDDK 2007). The rising epidemic of diabetes betes medicines. In multivariate analyses, predictors of threatens to increase the prevalence and severity of car- poor medication adherence were: believing you have dia- diovascular disease particularly among disproportionately betes only when your sugar is high (OR = 7.4;2–27.2), afflicted urban minority communities (National Diabetes saying there was no need to take medicine when the glu- Information Clearinghouse. National Diabetes Statistics: cose was normal (OR = 3.5;0.9–13.7), worrying about NIDDK 2007). Diabetes accelerates the natural course of side-effects of diabetes medicines (OR = 3.3;1.3–8.7), atherosclerosis and requires treatment of , lack of self-confidence in controlling diabetes (OR = hyperlipidemia and hyperglycemia to reduce the risk of 2.8;1.1–7.1), and feeling medicines are hard to take cardiovascular disease (American Diabetes Association (OR = 14.0;4.4–44.6). Disease and medication beliefs 2008a, b). Effective medical therapy in conjunction with inconsistent with a chronic disease model of diabetes were lifestyle changes in diet and physical activity are the cor- significant predictors of poor medication adherence. These nerstones of diabetes therapy (American Diabetes Associ- ation 2008a, b). The past decade has seen the development of many & D. M. Mann ( ) Á D. Ponieman Á E. A. Halm simple and effective drug therapies for diabetes (American Division of General Internal Medicine, Mount Sinai School of Medicine, 1 Gustave Levy Place, Box 1087, Diabetes Association 2008a, b). However, their clinical New York, NY 10029, USA impact has been limited by poor rates of adherence e-mail: [email protected] (Osterberg and Blaschke 2005). Rates of adequate/good adherence to diabetes medicines vary widely with esti- H. Leventhal Department of Psychology, Rutgers University, mates from 36 to 93% including studies that assessed New Brunswick, NJ, USA medication use using dispensing databases or more 123 J Behav Med (2009) 32:278–284 279 rigorous electronic monitoring (Cramer 2004; DiMatteo 2008). Moreover, in a study of over 800 patients using 2004; Walker et al. 2006). Suboptimal medication adher- diabetes medications, concerns about the medicines as- ence has been implicated as a major factor in poor glyce- sessed using the Beliefs about Medicines Questionnaire mic control (Guillausseau 2003). Socio-demographic and (Horne et al. 1999) (derived from self-regulation theory) medical factors such as age, race, education, and disease were associated with higher rates of cost-related and cost- severity represent largely unmodifiable and often incon- unrelated medication underuse measured by single item sistent predictors of poor adherence to drug therapy (Os- self-report measures (Aikens and Piette 2009). terberg and Blaschke 2005). At present, there is limited evidence identifying disease Understanding how patients’ beliefs about their disease and medication beliefs associated with diabetes medication and its treatment affect health behaviors such as medication adherence among minority populations. The purpose of this adherence represent important opportunities for improving study was to use self-regulation theory to identify poten- diabetes medication adherence (Cerkoney and Hart 1980). tially modifiable disease and medication beliefs associated In a meta-analysis of 26 studies, psychological factors such with diabetes medication adherence among minority pa- as emotional stability, internal and external motivations, tients with diabetes. The primary hypothesis was that dis- perceived benefit, and supportive structure were associated ease and medication beliefs discordant with the chronic with better adherence to diabetes medicines while per- disease nature of diabetes would be associated with worse ceived barriers and negative social environment were cor- medication adherence. related with poor adherence (Nagasawa et al. 1990). Using the Health Belief Model as a framework, a survey of 445 predominately white patients with diabetes and depression Methods noted that the relationship between increasing depression severity and worsening diabetes medication adherence was Study population in part mediated through higher perceived barriers and lower self-efficacy (Chao et al. 2005). After Institutional Review Board approval, study partici- Drawing on this foundation, Leventhal’s self-regulation pants were recruited from an outpatient general internal theory has been increasingly used to identify how several medicine clinic in New York City between January and July domains of health beliefs are associated with medication 2007. Trained bilingual staff identified patients using a adherence (Brewer et al. 2002; Horne and Weinman 1999). computer generated list of adults with diabetes coming in for According to this theory, patient beliefs about their disease visits each day and approached these potential participants in (chronicity, cause, consequences, controllability, among the waiting room. All English or Spanish speaking patients others), and their medicines (necessity, concerns) are reporting a history of Type 2 diabetes for at least 6 months important drivers of decisions about whether or not to take who were prescribed diabetes medication were eligible. medicine and under what circumstances (Leventhal et al. Exclusion criteria included a new diagnosis of diabetes and 2003). For example, beliefs that are inconsistent with the terminal illness with life expectancy of\1 year. chronic model of disease such as believing that the con- dition is only present when people feel bad (are symp- Data collection tomatic) have been shown to predict poor adherence to medications in , hypercholesterolemia and coronary Each consented participant was interviewed in a private disease (Brewer et al. 2002; Halm et al. 2006; Horne and room by a bilingual trained study member using a ques- Weinman 1999). Self-regulation theory has also been tionnaire in English or Spanish. The interview took applied to diabetes care. In a study of diabetes beliefs *45 min and participants were given $20 to cover their among Tongan compared to Europeans with diabetes in time and travel expenses. New Zealand, Tongans perceived their disease to be acute and cyclical in nature, uncontrollable with less perceived Measures need for medications all of which were associated with lower adherence to diet and medication taking (Barnes Socio-demographic factors, diabetes history, and comor- et al. 2004). In a case-control study, diabetes related foot bidities (medical and psychiatric) were self-reported. The ulcers and retinopathy were more common in patients who most recent hemoglobin A1C was ascertained using elec- perceived lower diabetes treatment control, had worse ill- tronic medical record review of the prior 6 years. The pri- ness coherence (e.g., understanding of their disease) and mary outcome measure of adherence to diabetes medicines viewed diabetes as a cyclical rather than chronic progres- was determined using a modified version of the four sive disease (as measured by the Revised Illness Perception items, self-reported Morisky medication adherence scale Questionnaire) (Moss-Morris et al. 2002; Searle et al. (Morisky et al. 1986). Each item is in a yes/no format with a 123 280 J Behav Med (2009) 32:278–284 maximum possible score of four equating very poor were used to identify disease and medication beliefs that adherence and 0 or 1 typically considered as good adher- were univariate predictors of poor adherence. The final ence. The Morisky scale has been used across many chronic multivariable logistic regression model was developed to diseases, including diabetes, as a self-reported measure of identify independent predictors of poor adherence based on adherence to medications and has demonstrated good reli- the variables associated with adherence in the univariate ability and predictive validity (Krapek et al. 2004; Krousel- analysis using a stepwise elimination method. Wood et al. 2004). We also asked patients if they used their Associations between adherence rates and the four belief medications when their glucose was low, normal, and high. groups (i.e., skeptical vs. accepting) were examined using chi square tests. All statistical analyses were performed using STATA 9.0 statistical software. Disease and medication beliefs

Patient’s disease beliefs were measured with 19 items Results assessing beliefs about the chronicity, cause, consequences and controllability of their diabetes using the Brief-Illness The 151 study subjects were all clinic-attending patients Perception Questionnaire (test–retest reliability across with type 2 diabetes who were predominantly Latino and domains 0.42 –0.72) as a framework (Broadbent et al. African–American and low in self reported socioeconomic 2006). Medication beliefs were assessed using the five status with 64% born in the United States, 31% born in items of greatest relevance to diabetes medication adapted Puerto Rico and 80% receiving Medicaid. Participants had from the Beliefs about Medicines Questionnaire (Cronbach longstanding diabetes (average of 13 years) and half (55%) alpha:necessity = 0.74, concerns = 0.80) (Horne and were using insulin. Overall, glycemic control was moder- Weinman 1999). Two of the questions were from the ately good (mean A1C 7.6), though 25% had an A1C [ 8.5 ‘necessity’ of using diabetes medications domain (patients (Table 1). Respondents reported high levels of co-morbid belief about the importance of using a medicine) and three conditions commonly associated with diabetes, e.g., 80% were from the ‘concerns’ about medications domain reported hypertension and 61% hyperlipidemia. Self-re- (worries about side-effects, addiction, etc.). The relation- ported depression rates were high (43%) and were con- ship of medication beliefs to adherence was examined in sistent with those identified by the PHQ-9. Self-reported two ways; using necessity and concern items as predictors, rates of anxiety were also notable (23%). and by generating a variable in which the scores from the most robust necessity and concerns items (‘‘importance of Univariate predictors of poor medication adherence taking medicine when the glucose is normal’’ and ‘‘worries about side-effects’’) were split at the median to form four Approximately one-quarter (28%) of the patients reported groups: patients who were skeptical (low necessity and poor adherence with their diabetes medication (Mori- high concerns), ambivalent (high necessity and high con- sky [ 1). Five beliefs about diabetes were endorsed by cerns), indifferent (low necessity and low concerns), and participants who reported poor medication adherence accepting of medication (high necessity and low concerns); (Table 2): the belief that you only have diabetes when your the approach is based upon the empirical and theoretical blood sugar is high, the consequences of diabetes are framework of (Aikens et al. 2005). minimal, diabetes has few symptoms, and perceiving Additional questions assessed confidence in controlling themselves as having little control over diabetes, and diabetes (disease-specific self-efficacy) and self-reported reporting that diabetes interferes with their social lives. difficulty in taking diabetes medications as prescribed Several medication beliefs were also correlated with (regimen complexity). Because depression can influence poor adherence including: necessity (no need to take adherence, we also assessed depressive symptoms using diabetes medicines when sugar is normal), concerns (side- the Patient Health Questionnaire (PHQ-9) which uses a cut- effects and addiction) and regimen complexity (medica- off of C10 for depression (Kroenke et al. 2001). tions are hard to take). In addition, low confidence in controlling their diabetes and depressive symptoms were Analysis both associated with higher rates of poor adherence.

Categorical variables are reported as percentages and con- Multivariable predictors of poor medication adherence tinuous variables as means. Socio-demographics, medical history, beliefs and knowledge rates were calculated using Five variables predicted poor adherence in multivariable descriptive statistics. Poor adherence was defined as a analysis (Table 3): have diabetes only when the glucose is Morisky score of[1 (Morisky et al. 1986). Chi-square tests high (disease belief), not taking meds when sugar normal

123 J Behav Med (2009) 32:278–284 281

Table 1 Socio-demographic and clinical characteristics of study (necessity related medication belief), worrying about side- patients (n = 151) effects (concerns related medication belief), reporting the % medicines were hard to take (regimen complexity), and lack of self-confidence (disease specific self-efficacy), Socio-demographics Mean age (SD) 57 (11) Female 68 Associations between necessity/concern belief subtypes Married 15 and adherence Employed 11 \High school education 51 The distribution of patients into the four medication belief Income \ $30,000 89 subtypes were as follows: 6% were labeled as ‘‘skeptical’’, English native language 65 34% ‘‘ambivalent’’, 5% ‘‘indifferent’’, and 55% ‘‘accept- Latino 58 ing’’. These distinctions appeared to be important because Black 34 they were related in a linear fashion with the rates of poor medication adherence (Fig. 1). Patients holding skeptical Insurance beliefs were significantly more likely to be poorly adherent Medicaid alone 53 than those holding ambivalent (p = .02), indifferent Medicaid + medicare 27 (p = .03) or accepting beliefs (p \ .001). Medicare alone 11 Commercial insurance 9 Diabetes history Mean diabetes duration, years (SD) 13 (11) Discussion Mean hemoglobin A1C (SD) 7.6 (1.7) Family history of diabetes 82 Our findings demonstrate that inner-city patients with Using insulin 55 diabetes, despite having longstanding disease and regular Co-morbidities outpatient diabetes care, frequently hold disease and Hypertension 80 medication beliefs that are inconsistent with a chronic disease model of diabetes. These misconceptions are High cholesterol 61 important for two main reasons. First, from a face validity Heart attack 17 standpoint, they are likely to be major barriers to having CHF 7 patients engage in guideline-recommended self-manage- History of depression 43 ment behaviors. Second, our data show that even in a Depressive symptoms (PHQ-9 C 10) 30 modest size sample, several of these suboptimal beliefs History of anxiety 23 were robust predictors of poor medication adherence.

Table 2 Proportion of patients % who are poorly % who are poorly p value who are poorly adherent adherent if agree adherent if disagree according to disease and with belief with belief medication beliefs Disease beliefs Have diabetes only when sugar is high 56 24 .006 Consequences of diabetes are low 36 19 .03 Symptoms of diabetes are minimal 39 16 .002 Have low control over diabetes 40 17 .003 Medication beliefs Don’t need diabetes medicines when sugar is normal 53 25 .02 Worried about side-effects of medicines 42 18 .001 Worried about addiction to medicines 46 25 .04 Medicines are hard to take 74 18 .001 Other Little confidence in ability to control diabetes 48 18 .001 Significant depressive symptoms 40 23 .03 Diabetes significantly interferes with social life 43 22 .01

123 282 J Behav Med (2009) 32:278–284

Table 3 Multivariate predictors of poor medication adherence berg and Blaschke 2005; Walker et al. 2006). As a result, Belief OR SE C.I. studies have begun to explore more modifiable predictors of adherence such as depression, provider–patient com- Have diabetes only when sugar high 7.5 5.0 2.0–27.2 munication, regimen complexity, cost, , and Not need to take medications when sugar 3.6 2.5 0.9–13.7 health beliefs (Gazmararian et al. 2006; Mann et al. 2007; is normal Rieckmann et al. 2006). Worried about side-effects 3.4 1.7 1.3–8.7 Low confidence in controlling diabetes 2.7 1.3 1.1–7.1 Medicines are hard to take 14.3 8.4 4.4–44.6 Association between health beliefs and adherence

Patients’ disease and medication beliefs have been corre-

90 78 lated with medication use in hyperlipidemia (Brewer et al. 80 70 2002; Horne and Weinman 1999), hypertension (Ross et al. 60 50 2004), asthma (Halm et al. 2006), heart disease (Horne and 40 36 30 25 Weinman 1999; Sud et al. 2005), depression (Aikens et al. 20 17 10 2005; Chao et al. 2005) and chronic disease medications in % Poorly Adherent 0 Skeptical Ambivalent Indifferent Accepting general (Phatak et al. 2006), but have received little attention in diabetes (Barnes et al. 2004). In a cross-sec- Fig. 1 Rates of poor adherence according to four belief subtypes. tional study of 324 patients with several different chronic Definitions: Skeptical = low necessity + high concerns; ambiva- diseases (including some with diabetes), individuals with lent = high necessity + high concerns; Indifferent = low neces- sity + low concerns; accepting = high necessity + low concerns. beliefs about the necessity of therapy that outweighed p values: Skeptical compared to: ambivalent (.02), indifferent (.03), concerns about them had higher rates of adherence (Horne accepting (\.001) and Weinman 1999). The combination of these positive and negative medication beliefs were much more powerful As predicted by the Self-regulation theory upon which predictors of behavior than socio-demographic or clinical the study was based, several different domains of beliefs factors. In another study focusing on disease beliefs, pa- were independent predictors of medication taking behavior. tients who viewed the consequences of hypercholesterol- These included beliefs about the chronicity and omni- emia as less severe reported lower rates of adherence presence of disease (believing you only had diabetes when (Brewer et al. 2002). In a study of 81 patients examining the glucose was high), as well as medication beliefs about the relationship between medication beliefs and depression the necessity of taking medication when their glucose was medication adherence higher levels of concern beliefs and normal, and concerns about side-effects. Patients’ percep- lower levels of necessity beliefs were associated with lower tion about the complexity of their regimen and their self- adherence (Aikens et al. 2005). Similar to our data, the confidence in controlling diabetes were also important investigators noted the same pattern of declining adherence correlates of adherence. Interestingly, the disease-specific in the four belief domains (skeptical, ambivalent, indif- self-efficacy item (confidence in controlling their diabetes) ferent and accepting) with patients who had little belief in was much more important than other more traditional the need for drug therapy and high concerns about side- generic self-efficacy questions that we asked (confidence in effects (skeptical) being far more likely to be poorly controlling their future health). Together these factors tend adherent (Aikens et al. 2005). These studies and other to undermine a chronic disease model of diabetes and the health belief literature supports similar conclusions to our need for constant treatment. study in that disease and medication beliefs that are at odds with a chronic disease model of therapy such as indefinite Predictors of medication adherence treatment with medication are associated with poor adherence. Furthermore, our data extend to diabetes the Numerous studies have explored potential predictors of concept that patients estimates of the need for and concerns adherence to medicines across a variety of conditions. about treatment predict adherence. However, the majority of studies have explored largely unmodifiable variables due to the retrospective databases Clinical implications that are often used to measure adherence. Frequently cited predictors include age, sex, ethnicity, income, education, One of the most common challenges physicians face with a and comorbidity though their relationship to adherence has patient with poorly controlled diabetes is to try to and been inconsistent due to variations in study designs and figure out if the patient’s hyperglycemia is due to non- sample populations (Cramer 2004; DiMatteo 2004; Oster- adherence or is occurring despite proper medication use

123 J Behav Med (2009) 32:278–284 283

(i.e., therapy needs to be intensified). Since patients may be nificant predictors of poor medication adherence. However, more willing to report suboptimal beliefs about medication, these suboptimal beliefs are potentially modifiable and so than admit to poor adherence itself, probing the handful of would be logical targets for tailoring educational messages strongly predictive factors we have identified should be both in real world clinical practice, as well as rigorously useful for two reasons. First, it can help identify those evaluated in future interventions to improve diabetes self- highly likely to be poor adherers. Second, it can direct the management. physician on which aspects of diabetes and its management they should focus their patient education efforts. For Acknowledgments The authors thank Jessica Lorenzo, MPH, Ju- example, a patient with signs of poor adherence who notes lian Baez, John Marcel and Manuel Vilchez for their work throughout this project. Data from this study were presented at the bi-annual high concerns and high necessity (ambivalence) would be International Congress of Behavioral Medicine (August 25, 2008; given a message tailored to reducing their concerns rather Tokyo, Japan). This study was funded by the National Institute on than trying to further enhance already substantial necessity Aging (R24 AG023958) and the Center for the Study of Health Be- beliefs. Clinicians may want to be most alert for the liefs and Behaviors. Disclosure The authors have no relevant conflict of interest to disclose. ‘skeptical’ subgroup of patients with diabetes (who did not feel that medications were important and worried a lot about their side-effects) who were nearly all non-adherent References in our study. Similarly, three-quarters of those who said the medications were hard to take were poor adherers. Aikens, J. E., Nease, D. E., Jr, Nau, D. P., Klinkman, M. S., & Schwenk, T. L. (2005). Adherence to maintenance-phase Limitations medication as a function of patient beliefs about medication. Annals of Family Medicine, 3, 23. doi:10.1370/afm.238. Aikens, J. E., & Piette, J. D. (2009). Diabetic patients’ medication Our results should be viewed with consideration of several underuse, illness outcomes, and beliefs about antihyperglycemic limitations. While our inner city patient population is of and antihypertensive treatments. Diabetes Care, 32(1), 19–24. particular interest due to its high burden of diabetes mor- American Diabetes Association. (2008a). Standards of medical care in diabetes–2008. Diabetes Care, 31, S12–S54. doi:10.2337/dc08- bidity, the generalizability of our observations to other S012. settings is unknown, and should be explored in future work. American Diabetes Association. (2008b). The economic costs of The use of a self-reported medication adherence scale diabetes in the US in 2007. Diabetes Care, 31, 1–20. doi: represents a potential limitation of most studies in this 10.2337/dc08-S001. Barnes, L., Moss-Morris, R., & Kaufusi, M. (2004). Illness beliefs and field. However, the Morisky medication adherence scale is adherence in diabetes mellitus: A comparison between Tongan well validated and one of the most widely used self-re- and European patients. The New Zealand Medical Journal, 117, ported measures of adherence. The modest sample size U743. limits our ability to detect weaker associations. Finally, our Brewer, N., Chapman, G., Brownlee, S., & Leventhal, E. (2002). Cholesterol control, medication adherence and illness cognition. data do not explain ‘why’ so many patients had these British Journal of Health Psychology, 7, 433–447. doi: suboptimal health beliefs. Since most patients had long 10.1348/135910702320645408. standing diabetes, were in the regular care of a physician, Broadbent, E., Petrie, K. J., Main, J., & Weinman, J. (2006). The brief Journal of Psychosomatic and were largely insured with drug coverage, these dis- illness perception questionnaire. Research, 60, 631–637. doi:10.1016/j.jpsychores.2005.10.020. connects are not attributable to simple insurance or access Cerkoney, K. A., & Hart, L. K. (1980). The relationship between the problems. Whether patients were given recommended health belief model and compliance of persons with diabetes diabetes education (by physicians or allied health profes- mellitus. Diabetes Care, 3, 594–598. doi:10.2337/diacare.3.5.594. sionals) but still remained skeptical, or they never clearly Chao, J., Nau, D. P., Aikens, J. E., & Taylor, S. D. (2005). The mediating role of health beliefs in the relationship between received these health messages in a manner they could depressive symptoms and medication adherence in persons with understand (due to inadequate health literacy or ineffective diabetes. Research in Social & Administrative , 1, explanation by providers), is unknown. Cultural differences 508–525. doi:10.1016/j.sapharm.2005.09.002. which were not measured may also play a significant role Cramer, J. A. (2004). A systematic review of adherence with medications for diabetes. Diabetes Care, 27, 1218–1224. doi: in the relationship between beliefs and adherence. 10.2337/diacare.27.5.1218. DiMatteo, M. R. (2004). Variations in patients’ adherence to medical recommendations: A quantitative review of 50 years of research. Conclusion Medical Care, 42, 200–209. doi:10.1097/01.mlr.0000114908. 90348.f9. In summary, we found that disease and medication beliefs Gazmararian, J. A., Kripalani, S., Miller, M. J., Echt, K. V., Ren, J., & Rask, K. (2006). Factors associated with medication refill inconsistent with a chronic disease model of diabetes were adherence in cardiovascular-related diseases: A focus on health common among a population of people living in the inner literacy. Journal of General Internal Medicine, 21, 1215–1221. city who have diabetes, and that these beliefs were sig- doi:10.1111/j.1525-1497.2006.00591.x.

123 284 J Behav Med (2009) 32:278–284

Guillausseau, P. J. (2003). Influence of oral antidiabetic drugs (IPQ-R). Psychology & Health, 17, 1–16. doi:10.1080/0887 compliance on metabolic control in type 2 diabetes. A survey in 0440290001494. general practice. Diabetes & Metabolism, 29, 79–81. doi: Nagasawa, M., Smith, M. C., Barnes, J. H., Jr, & Fincham, J. E. 10.1016/S1262-3636(07)70011-3. (1990). Meta-analysis of correlates of diabetes patients’ com- Halm, E. A., Mora, P., & Leventhal, H. (2006). No symptoms, no pliance with prescribed medications. The Diabetes Educator, 16, asthma: the acute episodic disease belief is associated with poor 192–200. doi:10.1177/014572179001600309. self-management among inner-city adults with persistent asthma. National Diabetes Information Clearinghouse. National Diabetes Chest, 129, 573–580. doi:10.1378/chest.129.3.573. Statistics: NIDDK. (2007). National diabetes statistics fact sheet: Horne, R., & Weinman, J. (1999). Patients’ beliefs about prescribed General information and national estimates on diabetes in the medicines and their role in adherence to treatment in chronic United States. US: National Institute of Health. http://diabetes. physical illness. Journal of Psychosomatic Research, 47, 555– niddk.nih.gov/dm/pubs/statistics/index.htm. 567. doi:10.1016/S0022-3999(99)00057-4. Osterberg, L., & Blaschke, T. (2005). Adherence to medication. The Horne, R., Weinman, J., & Hankins, M. (1999). The beliefs about New England Journal of Medicine, 353, 487–497. doi:10.1056/ medicines questionnaire: The development and evaluation of a NEJMra050100. new method for assessing the cognitive representation of medi- Phatak, H. M., & Thomas, J., I. I. I. (2006). Relationships between cation. Psychology & Health, 14, 1–24. doi:10.1080/0887044 beliefs about medications and nonadherence to prescribed 9908407311. chronic medications. The Annals of Pharmacotherapy, 40, Krapek, K., King, K., Warren, S. S., George, K. G., Caputo, D. A., 1737–1742. doi:10.1345/aph.1H153. Mihelich, K., et al. (2004). Medication adherence and associated Rieckmann, N., Kronish, I. M., Haas, D., Gerin, W., Chaplin, W. F., hemoglobin A1c in Type 2 diabetes. The Annals of Pharmaco- Burg, M. M., et al. (2006). Persistent depressive symptoms lower therapy, 38, 1357–1362. doi:10.1345/aph.1D612. adherence after acute coronary syndromes. American Kroenke, K., Spitzer, R. L., & Williams, J. B. W. (2001). The PHQ-9. Heart Journal, 152, 922–927. doi:10.1016/j.ahj.2006.05.014. Validity of a brief depression severity measure. Journal of Ross, S., Walker, A., & MacLeod, M. (2004). Patient compliance in General Internal Medicine, 16, 606–613. doi:10.1046/j.1525- hypertension: Role of illness perceptions and treatment beliefs. 1497.2001.016009606.x. Journal of Human Hypertension, 18, 607–613. doi:10.1038/ Krousel-Wood, M., Thomas, S., Muntner, P., & Morisky, D. (2004). sj.jhh.1001721. Medication adherence: A key factor in achieving blood pressure Searle, A., Wetherell, M. A., Campbell, R., Dayan, C., Weinman, J., control and good clinical outcomes in hypertensive patients. & Vedhara, K. (2008). Do patients’ beliefs about type 2 diabetes Current Opinion in Cardiology, 19, 357–362. doi:10.1097/01.hco. differ in accordance with complications: An investigation into 0000126978.03828.9e. diabetic foot ulceration and retinopathy. International Journal of Leventhal, H., Brissette, I., & Leventhal, E. (2003). The common Behavioral Medicine, 15, 173–179. doi:10.1080/107055008 sense models of self-regulation of health and illness. London: 02212940. Taylor & Francis Books, Ltd. Sud, A., Kline-Rogers, E. M., Eagle, K. A., Fang, J., Armstrong, D. F., Mann, D. M., Allegrante, J. P., Natarajan, S., Halm, E. A., & Rangarajan, K., et al. (2005). Adherence to medications by Charlson, M. (2007). Predictors of adherence to for patients after acute coronary syndromes. The Annals of Pharma- primary prevention. Cardiovascular Drugs and Therapy, 21, cotherapy, 39, 1792–1797. doi:10.1345/aph.1G249. 311–316. doi:10.1007/s10557-007-6040-4. Walker, E. A., Molitch, M., Kramer, M. K., Kahn, S., Ma, Y., Morisky, D., Green, L., & Levine, D. (1986). Concurrent and Edelstein, S., et al. (2006). Adherence to preventive medications: predictive validity of a self-reported measure of medication Predictors and outcomes in the diabetes prevention program. adherence. Medical Care, 24, 67–74. doi:10.1097/00005650- Diabetes Care, 29, 1997–2002. doi:10.2337/dc06-0454. 198601000-00007. Moss-Morris, R., Weinman, J., Petrie, K., Horne, R., Cameron, L., & Buick, D. (2002). The revised illness perception questionnaire

123