Evaluating the Validity and Reliability of the Beliefs About Medicines Questionnaire in Low-Income, Spanish-Speaking Patients with Diabetes in the United States

Corresponding Author: John Billimek, PhD, University of California, Irvine, Health Policy Research Institute, 100 Theory Suite 110, Irvine, CA 92697 (ude.icu@emillibj). Phone: +1 (949) 824-3065. Fax: +1 (949) 824-3388

Abstract

Purpose

The purpose of this study was to examine the reliability and validity of a Spanish version of the Beliefs About Medicines Questionnaire (BMQ) as a measure to evaluate beliefs about medications and to differentiate adherent from nonadherent patients among low-income Latino patients with diabetes in the United States.

Methods

Seventy-three patients were administered the BMQ and surveyed for evidence of medication nonadherence. Internal consistency of the BMQ was assessed by Cronbach’s alpha along with performing a confirmatory factor analysis. Criterion validity was assessed by comparing mean scores on three subscales of the BMQ (General Overuse, General Harm, and Specific Necessity-Concerns difference score) between adherent patients and patients reporting nonadherence for three different reasons (unintentional nonadherence, cost-related nonadherence, and nonadherence due to reasons other than cost) using independent samples t-tests.

Results

The BMQ is a reliable instrument to examine beliefs about medications in this Spanish-speaking population. Construct validity testing shows nearly identical factor loading as the original construct map. General Overuse scores were significantly more negative for patients reporting each reason for nonadherence compared to their adherent counterparts. Necessity-concerns difference scores were significantly more negative for patients reporting nonadherence for reasons other than cost compared to those who did not report this reason for nonadherence.

Conclusions

The Spanish version of the BMQ is appropriate to assess beliefs about medications in Latino patients with type 2 diabetes in the United States, and may help identify patients who become nonadherent to medications for reasons other than out of pocket costs.

Introduction

Adherence to a medication regimen is central to the proper management of chronic health conditions in general, and is critical to improving glycemic control and preventing complications for individuals with diabetes, specifically. 1–4 However, adherence to medications is not a simple task considering its multifactorial nature. 5 Studies have reported a variety of factors that may lead a patient to be nonadherent, including increasing costs of medications 6 , shared or personal beliefs about the potential harms of medications 7 , and lack of social support. 7–10 Such barriers to adherence are especially taxing on patients from low socioeconomic backgrounds 7,11,12 , most likely due to a lack of resources to support disease management. Latinos, particularly immigrants of Mexican descent, have lower levels of socioeconomic status and decreased levels of access to health insurance in comparison to their non-Hispanic White counterparts 13 . Understanding barriers to medication adherence may improve health outcomes for ethnically diverse patients with chronic illness.

Central to improving medication adherence is the understanding of a patient’s reasons for nonadherence. 14 Existing measures of nonadherence can identify patients who exhibit unintentional nonadherence (due to forgetfulness or difficulty getting to the pharmacy 15 ), intentional nonadherence due to cost 16 , or intentional nonadherence due reasons other than cost (concerns about side effects or a lack of benefit from the medication 16,17 ). Unintentional nonadherence could possibly be remedied through automated reminders, social support or mail-order refills 18–20 . Likewise, intentional nonadherence due to cost can be reduced by lowering the out of pocket costs of medications 21,22 . Intentional nonadherence due to reasons other than cost, however, can be difficult to address as many patients do not communicate these negative beliefs to their health providers 23–25 . For this reason, identifying patients with negative beliefs about medications may help providers address these concerns and prevent future nonadherence, especially for the many patients who never initiate a newly prescribed medication 26 or discontinue the medications soon after starting 27 .

The Beliefs about Medicines Questionnaire (BMQ) has been used to identify patients with negative beliefs about medications that may lead to nonadherence 28,29 . The BMQ has been administered in various patient populations with different diagnoses, including chronic conditions such as asthma, hypertension, and HIV 30–32 and across diverse cultural groups 33 . Although the BMQ has been translated into Spanish and validated in previous studies conducted in Spain 34–36 , it has not yet been validated for use with Spanish-speaking Latino patients with diabetes living in medically underserved communities within the U.S. Because negative beliefs about medications are an important contributor to nonadherence in this population 7 , this study aims to assess the reliability and validity of a Spanish translation of the BMQ in order to (1) evaluate patient beliefs about medications, and (2) discriminate adherent from nonadherent patients among low-income, Latino adults with diabetes living in the U.S.

Methods

Design

The present study is a cross-sectional observational study of a subset of patients participating in the EMPATHy Toolkit randomized clinical trial 37 .

Setting and Sample

Parent study

The sample for the current study consisted of a subset of participants in a larger parent study, described in detail elsewhere 37 . In short, participants for the parent study were recruited from a university-affiliated Federally Qualified Health Center (FQHC) in Santa Ana, California that predominantly serves low-income, ethnic minority patients, mostly of Hispanic ethnicity. Patients who met the following criteria were recruited: (1) age 18 and older; (2) poorly controlled type 2 diabetes (as indicated by A1C >7.5% (53 mmol/mol), LDL cholesterol >100 mg/dl (2.59 mmol/l) or systolic blood pressure >140 mmHg), (3) Hispanic ethnicity, and (4) English or Spanish speaking. Two hundred eighty-five eligible patients were approached before their regularly scheduled diabetes appointments. Of these, 210 (73.7%) consented to participate in the parent study.

BMQ validation subset

A subset of participants (the “BMQ validation subset”), recruited consecutively at the end of the parent study (N=73), was asked to complete an additional questionnaire about their medication-related beliefs (BMQ) and adherence behaviors. Participants in this BMQ validation subset comprise the analytic sample for this study.

Measures

Participants in the BMQ validation subset were asked to complete a Spanish language version of the BMQ, adapted from a version developed and validated for use in Spain in a previous study 35 , during the baseline visit, in addition to obtaining measures collected as part of the parent study protocol.

The Beliefs About Medicines Questionnaire (BMQ)

The BMQ is an 18-item questionnaire consisting of two sections—the BMQ-General, which assesses the patient’s beliefs about medications in general, and the BMQ-Specific, which assesses beliefs about the patient’s specific medications 38 . The questionnaire has been utilized and validated in numerous clinical studies since then, including those investigating nonadherence in patients with diabetes 38–41 . The BMQ-General examines a person’s ideas about medicines in general, and consists of two sub-scales: General-Overuse and General-Harm. The BMQ-Specific, on the other hand, focuses on a person’s ideas about specific medicines that are prescribed for personal use, and also consists of two sub-scales: Specific-Necessity and Specific-Concern.

In this study, responds to the BMQ items were collected using a Likert scale (1 through 5), allowing the study participants to choose where they lie in agreement with each of a series of statements (“1” meant “Strongly Agree,” whereas “5” meant “Strongly Disagree”). The responses from each subscale were reverse coded as needed and averaged so that higher scores correspond with a higher endorsement of beliefs described by the scale. Furthermore, following the Necessity-Concerns Framework posited by Horne and colleagues 33 , the difference between Specific-Necessity and Specific-Concern subscale scores was calculated for each patient in order to evaluate the patient-perceived benefit-cost analysis of medications.

Prior to data collection, the Spanish language items from a version of the BMQ that was previously validated in Spain 35 were reviewed for reading level and idiomatic compatibility with Latin American Spanish by a research staff member who is a native speaker of Latin American (Mexican) Spanish and is experienced working with Spanish-speaking patients in medically underserved communities in the U.S. Items for which the translation was not deemed appropriate for the target population were re-translated from English into Latin American Spanish by the staff member, and back-translated for verification by a second native Latin American (Mexican) Spanish speaker on the study team. The adapted questionnaire is compared to the original English version in Table 1 .

Table 1

English to Spanish translation of the BMQ a .

a The Beliefs about Medicines Questionnaire (BMQ) is an 18-item questionnaire, with each item eliciting ratings of agreement with a statement on a Likert scale (1 through 5, where “1” meant “Strongly Agree (Muy de Acuerdo), ”2” “Agree”(De Acuerdo), “3” “Neutral” (Neutral), “4” “Disagree” (No de Acuerdo) whereas “5” meant “Strongly Disagree” (Muy en Desacuerdo)” 33,36 .

Characterization of Nonadherence Behaviors

A brief baseline survey was administered to all participants prior to the study intervention. It included a 15-item questionnaire that asked about medication nonadherence behaviors within the past three months prior to the study visit. The questionnaire was adapted from a measure used in a national survey of Medicare beneficiaries 16 that has been used in previous research on contributors to suboptimal diabetes outcomes in low-income Latino patients. 7,42 Such responses were then used to characterize the nonadherence behaviors into three main types: (1) cost-related nonadherence, (2) nonadherence due to reasons other than cost, such as doubts about the medications’ effectiveness or concerns about side effects, and (3) unintentional nonadherence, such as forgetting to take their medication.

Out of pocket costs

The patient’s total monthly out of pocket costs for medications were assessed using a single question, “On average, how much do you spend on your medications per month?”

Statistical analysis

All statistical analyses were performed using SPSS version 23.0 (IBM Corp, Armonk, NY). Participant characteristics were compared between adherent and nonadherent groups using univariate statistics. Internal consistency was assessed via the Cronbach’s alpha for each BMQ domain. The construct validity of the BMQ was then evaluated using a confirmatory factor analysis with a two-factor structure for each of the BMQ sections, which replicated the original BMQ validation study 38 .

Finally, the convergent and discriminate criterion validity was tested by comparing BMQ mean scores, and out of pocket medication costs, for patients that did and did not report each type of nonadherence behavior. Specifically, it was posited that the convergent and discriminant criterion validity of the measures would be supported by the following pattern of associations (summarized in Table 2 ).

Table 2

Hypothesized associations between BMQ subscales, out of pocket costs and types of nonadherence outcomes a

a Cells marked with a ‘+’ correspond to pairs of measures and nonadherence outcomes for which a statistically significant association would support the convergent criterion validity of the measure. Blank cells correspond to pairs of measures and nonadherence outcomes for which the absence of a statistically significant association would support the discriminant criterion validity of the measure.

Patients exhibiting any type of nonadherence (unintentional, intentional related to cost and intentional for reasons other than cost) would report more negative beliefs on the BMQ-General Overuse and BMQ-General Harms subscales than adherent patients,

Patients reporting intentional nonadherence for reasons other than cost, but not patients reporting unintentional nonadherence or intentional nonadherence related to costs would report significantly more negative beliefs on the BMQ Specific Necessity-Concerns measure than their adherent counterparts on those measures, and

Patients reporting intentional nonadherence related to cost, but not patients reporting unintentional nonadherence or intentional nonadherence for reasons other than cost would report significantly higher out of pocket costs than their adherent counterparts on those measures.

Results

Characteristics of the BMQ Validation subset sample (N = 73) are described in Table 3 . Of these patients, 61% reported unintentional nonadherence, 75% reported intentional nonadherence for reasons other than cost, and 26% reported cost-related nonadherence.

Table 3

Patient demographic information.

* Based on self-reported behaviors of any of the three types of nonadherence. Patients who did not report nonadherence were placed in the adherent group.

None of the results above were statistically significant (P>0.05).

Internal Consistency

The internal consistency of this Spanish version of the BMQ was examined in order to assess whether items under a particular construct produce similar scores to one another. The Cronbach’s alpha was found to be .678 for the BMQ-Specific and .774 for the BMQ-General scales. The Cronbach’s alpha for the subdomains of the BMQ-General were .786 and .694 for General-Overuse and General-Harm, respectively.

Construct Validity

The construct validity of the BMQ in a Latino, Spanish-speaking population in Southern California was assessed through confirmatory factor analysis. The original measure development study 38 used exploratory factor analysis of 34 novel items to identify the factor structure that informed the selection of the final 18 items that comprise the BMQ, and guided the grouping of those items into the four BMQ subscales. Then, in the present study, confirmatory factor analysis of the Spanish translation of this measure was performed to examine whether this original factor structure of the BMQ is retained in the responses of U.S.-dwelling, Spanish-speaking Latino patients with diabetes.

Table 4 contains the two-factor structure of the 10-item BMQ-Specific. Each item loaded onto a single factor and did not display split loading. All five of the Specific-Necessity items loaded onto Factor 1. Similarly, all five of the Specific-Concern items loaded onto Factor 2. This 2-factor structure accounted for 48.6% of the total variance explained for the components of the BMQ.

Table 4

Confirmatory Factor Analysis of BMQ-Specific a

Similarly, Table 5 shows the two-factor structure factor analysis of the 8-item BMQ-General. For the General-Overuse subdomain, three items loaded onto Factor 1. The four General-Harm items loaded onto Factor 2. Thus, most items of the BMQ-General loaded appropriately. Only item Q4, ‘Natural remedies are safer than medications,’ demonstrated split loading between Factor 1 and 2. This two-factor structure accounted for 59% of the total variance explained for the components of the BMQ.

Table 5

Confirmatory Factor Analysis of BMQ-General a

Criterion Validity

Criterion validity was examined by comparing the mean scores of the BMQ subscales between adherent versus nonadherent respondents for each adherence outcome. Figure 1 shows the results of the independent samples t-tests comparing the General-Overuse mean score ( Figure 1a ), General-Harm mean score ( Figure 1b ), Specific-Necessity-Concern difference score (denoted as Sn-Sc, see Figure 1c ) and mean monthly out of pocket cost of medications ( Figure 1d ) for adherent versus nonadherent respondents for each type of nonadherence. General-Overuse mean scores are higher in the groups that demonstrated unintentional nonadherence (P=0.003), nonadherence not due to cost (P=0.001), and cost-related nonadherence (P=0.001) compared to their counterparts that reported adherence to their medications. General-Harms scores did not differ between adherent vs nonadherent patients for any adherence outcome. There is a significant difference in Sn-Sc score in patients who reported nonadherence not due to cost versus those that reported adherence (mean ± SD=0.77 ± 0.90 vs. 1.30 ± 1.03; P=0.04). Additionally, out of pocket monthly cost of medications differed between those who reported cost-related nonadherence compared to those who did not (mean ± SD=$78.58 ± $80.17 vs. $29.39 ± $61.56; P<0.001).

Comparing BMQ subscale scores and out of pocket medication costs between adherent vs. nonadherent patients on three measures of nonadherence

Error bars represent 95% confidence intervals. P-values computed from independent samples t-tests. (a.) BMQ General-Overuse scores were significantly higher (more negative) in patients reporting unintentional nonadherence, nonadherence related to reasons other than cost, and nonadherence related to costs compared to their adherent counterparts; (b.) None of the three nonadherence behaviors were associated with BMQ General-Harm scores; (c.) BMQ Specific necessity-concerns difference scores were significantly lower (more negative) in patients who reported nonadherence related to reasons other than costs compared to those who did not; (d.) Monthly out-of-pocket medication costs were significantly higher in patients who reported nonadherence related to cost compared to those who did not.

Discussion

This study assessed the validity and reliability of the Beliefs about Medicines Questionnaire in a patient population of Spanish-speaking Latinos with uncontrolled diabetes. The results of this study showed that the Spanish-language 18-item BMQ is reliable in this study population, as evidenced by acceptable Cronbach’s alpha scores. The factor structure of the translated items is comparable to the original concept map proposed by Horne, et al 38 , and supports the construct validity of the measure in this population. Furthermore, the results of the criterion validity testing demonstrate that the domains of the BMQ are able to predict certain types of nonadherence outcomes and therefore may be useful in targeting patients that may be at risk of medication nonadherence.

Internal consistency testing demonstrated fair to moderate Cronbach’s alpha values (.678 to .786). These values are consistent with other studies that have tested the BMQ in other patient populations 28,43,44 . Thus it can be concluded that the BMQ is a useful and reliable measure of patient beliefs about medications, including for a Spanish-speaking Latino population.

Confirmatory factor analysis of the BMQ demonstrated adequate factor loading for both domains. For the BMQ-Specific in particular, the items that loaded onto Factors 1 and 2 are well represented by the established constructs of Specific-Necessity and Specific-Concern, respectively. The items of the BMQ-General, conversely, did not load as strongly. The items that loaded onto Factors 1 and 2 most likely represent General-Overuse and General-Harm, respectively; however, the factor analysis demonstrated split loading of the item “Natural remedies are safer than medicines.”

This deviation from the original concept map may be explained in part by cultural variations in usage of natural therapies. According to the 2002 National Health Interview Survey, Hispanics were the second largest ethnic group to report the utilization of complementary and alternative medicine (CAM). Hispanic older adults in particular had greater odds of utilizing CAM compared to their White counterparts 45 . Increased likelihood of utilization of CAM may reflect greater trust in natural remedies, which can in turn translate into mixed beliefs regarding the usage of allopathic medications, thus skewing the construct validity of this particular item. Overall, however, the factor analysis of the Spanish BMQ demonstrated valid measurement of the constructs, signifying that the BMQ scores can provide accurate information regarding the nature of the beliefs that affect medication adherence.

The intention of the criterion validity testing was to evaluate the accuracy of the BMQ scores to reflect an established measure of medication nonadherence 16 . According to the hypotheses of this study, if the Spanish BMQ were valid, stronger negative beliefs towards medications (defined by low Specific-Necessity mean scores or high Specific-Concern, General-Overuse, and General-Harm mean scores) would correlate highly with the presence of nonadherence behaviors. Particularly, the results of the criterion validity testing demonstrate that the General-Overuse subdomain was significantly associated with all three of the nonadherence behaviors. In other words, patients that reported nonadherence behaviors had greater mean General-Overuse scores compared to patients that reported adherence to their medications. Contrary to the study hypotheses, the General-Harm measure did not discriminate adherent from nonadherent patients on any of the nonadherence measures.

Additionally, the Specific-Necessity-Concern difference score was significantly associated with intentional nonadherence due to reasons other than cost. Patients that reported nonadherence for other reasons had a narrower difference between the Specific-Necessity and Specific-Concern mean scores. This implies that patients who believed that the risks of taking their medications (i.e. concerns) outweighed the benefit (i.e. necessity) were more likely to report intentional nonadherence due to reasons other than cost. This is consistent with the current literature, particularly in reference to the Necessity-Concerns Framework, which contextualizes the effect of this risk-benefit analysis on medication adherence. According to a meta-analysis inclusive of 94 studies that utilized the BMQ, higher adherence was associated with stronger perceptions of necessity of treatment and fewer concerns about medications 33 .

Furthermore, as a reference for the criterion validity testing, out of pocket monthly cost of medications was compared across adherence categories for all three forms of nonadherence. As expected, the results demonstrated that cost was significantly associated with cost-related nonadherence. Patients that reported cost-related nonadherence paid on average $50 more per month for their medications than their adherent counterparts. A systematic review by Briesacher, et al. reported an established link between medication adherence due to cost, particularly in the presence of underlying financial burden 46 . This review also revealed that heavy disease burden is also a contributing factor toward nonadherence 46 . Therefore, a possible next step for this measurement study would be to evaluate the effect of disease severity and patient-perceived burden on nonadherence behaviors, cost, and BMQ scores.

There are several limitations to this study. First, data were collected using a single cross-sectional survey. Therefore, a causal relationship between long-term beliefs and perceptions of medications cannot necessarily be inferred. For example, it is unclear if beliefs about medications lead to nonadherence or nonadherent behaviors persuade patients to alter their beliefs. Second, the focus of the study is to assess beliefs about medication adherence and not other adherence behaviors that are also critical for diabetes management, such as diet and exercise modifications. Third, the results are based on self-reported nonadherence, which is subjective by nature. For future studies, more objective measures of medication adherence, such as medication electronic monitoring system (MEMS) bottle caps that record each time a medication bottle is opened 5 , can be used. Fifth, because the study population consisted of Latino, Spanish-speaking patients with diabetes living in the United States, the results may not generalize to other populations.

Finally, although the content validity (i.e. the extent to which the measure represents all relevant facets of a construct) of the original BMQ was established through the item-generation process described in the original measure development study 38 , the content validity was not formally re-evaluated in the present study. Instead, consistent with other studies demonstrating the reliability and validity of the BMQ in novel populations and languages 34–36 , the study provides evidence for the internal consistency, construct validity and criterion validity to guide other investigators’ evaluation of the BMQ for use in U.S.-dwelling Latinos with diabetes. It is possible, however, that additional tailoring of the item content may further improve the performance of the measure in this population.

Implications for practice

The BMQ has been used extensively in clinical practice to predict adherence 30–33 . It is a practical and useful instrument that can help guide the patient-clinician discussion regarding disease management and can facilitate the communication of concerns that the patient may have regarding taking medications 33 . In this sense, the BMQ is effective in identifying patients at risk of medication nonadherence in order to further provide support and intervention. The assessments of validity and reliability that were employed in this study suggest that the measure is appropriate for use in Latino patients with diabetes living medically underserved communities in the U.S.

Studies on medication adherence and barriers to treatment are particularly lacking for minority and underrepresented patient populations, especially those that speak a language other than English 35 . Health beliefs—whether they are shared among community members or only held by the individual—have critical implications on an individual’s disease management and perceptions of medication use. The findings from this study suggest that the beliefs that medications are overused in general may contribute to multiple types of nonadherence behavior, and supports the well-documented finding that higher out of pocket costs contribute to cost related nonadherence. 22 An additional, important implication of the findings is that negative beliefs about the tradeoff of benefits to harms about a specific medication may contribute to nonadherence for reasons that cannot be addressed simply by reducing out of pocket costs. These negative beliefs can be measured reliably in a high risk population of Latino patients with diabetes living in medically underserved communities the U.S., and may represent an important target for intervention to reduce disparities in diabetes outcomes.

Just as providers can anticipate—and attempt to mitigate—cost-related nonadherence when prescribing expensive medications or unintentional nonadherence for patients with complex regimens or limited support, measures like the BMQ may be useful to help providers anticipate intentional nonadherence for reasons other than cost, and take steps to engage patients in a discussion about those beliefs. The findings support the need for future research to better understand the effects of Latino culture on beliefs about medications, and to develop interventions aimed to promote more positive beliefs about medications in underserved Latino patient populations with great potential to benefit from optimal therapy.

Acknowledgments

The project described was supported by the National Center for Research Resources and the National Center for Advancing Translational Sciences, National Institutes of Health, through Grant UL1 TR001414. The content is solely the responsibility of the authors and does not necessarily represent the official views of the NIH. All authors have no conflicts of interest to report. Dr. Billimek had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

References

1. American Diabetes Association. Summary of Revisions to the 2014 Clinical Practice Recommendations. Diabetes Care. 2014; 37 :S4–S4. [PubMed] [Google Scholar]

2. Ho PM, et al. Effect of medication nonadherence on hospitalization and mortality among patients with diabetes mellitus. Arch Intern Med. 2006; 166 :1836–1841. [PubMed] [Google Scholar]

3. Egede LE, et al. Differential impact of longitudinal medication non-adherence on mortality by race/ethnicity among veterans with diabetes. J Gen Intern Med. 2013; 28 :208–215. [PMC free article] [PubMed] [Google Scholar]

4. Egede LE, et al. Regional, Geographic, and Ethnic Differences in Medication Adherence Among Adults with Type 2 Diabetes. Ann Pharmacother. 2011; 45 :169–178. [PubMed] [Google Scholar]

5. Osterberg L, Blaschke T. Adherence to Medication. N Engl J Med. 2005; 353 :487–497. [PubMed] [Google Scholar]

6. Ngo-Metzger Q, Sorkin DH, Billimek J, Greenfield S, Kaplan SH. The effects of financial pressures on adherence and glucose control among racial/ethnically diverse patients with diabetes. J Gen Intern Med. 2012; 27 :432–437. [PMC free article] [PubMed] [Google Scholar]

7. Billimek J, August KJ. Costs and beliefs: understanding individual- and neighborhood-level correlates of medication nonadherence among Mexican Americans with type 2 diabetes. Health Psychology. 2014; 33 :1602–1605. [PMC free article] [PubMed] [Google Scholar]

8. Mishra SI, Gioia D, Childress S, Barnet B, Webster RL. Adherence to medication regimens among low-income patients with multiple comorbid chronic conditions. Health Soc Work. 2011; 36 :249–258. [PMC free article] [PubMed] [Google Scholar]

10. Kim C-J, et al. The role of social support on the relationship of depressive symptoms to medication adherence and self-care activities in adults with type 2 diabetes. J Adv Nurs. 2015; 71 :2164–2175. [PubMed] [Google Scholar]

11. Silverman J, et al. The Relationship Between Food Insecurity and Depression, Diabetes Distress and Medication Adherence Among Low-Income Patients with Poorly-Controlled Diabetes. J Gen Intern Med. 2015; 30 :1476–1480. [PMC free article] [PubMed] [Google Scholar]

12. Ursem CJ, et al. Adherence to adjuvant endocrine therapy for breast cancer: importance in women with low income. J Womens Health. 2015; 24 :403–408. [PubMed] [Google Scholar]

13. Williams DR, Mohammed SA, Leavell J, Collins C. Race, socioeconomic status, and health: complexities, ongoing challenges, and research opportunities. Ann N Y Acad Sci. 2010; 1186 :69–101. [PMC free article] [PubMed] [Google Scholar]

14. Voils CI, et al. Initial validation of a self-report measure of the extent of and reasons for medication nonadherence. Med Care. 2012; 50 :1013–1019. [PMC free article] [PubMed] [Google Scholar]

15. Unni E, Farris KB. Determinants of different types of medication non-adherence in cholesterol lowering and asthma maintenance medications: a theoretical approach. Patient Educ Couns. 2011; 83 :382–390. [PubMed] [Google Scholar]

16. Safran DG, et al. Prescription drug coverage and seniors: findings from a 2003 national survey. Health Aff Proj Hope. 2005:W5–152–166. Suppl Web Exclusives. [PubMed] [Google Scholar]

17. Lowry KP, Dudley TK, Oddone EZ, Bosworth HB. Intentional and unintentional nonadherence to antihypertensive medication. Ann Pharmacother. 2005; 39 :1198–1203. [PubMed] [Google Scholar]

18. Park LG, Howie-Esquivel J, Chung ML, Dracup K. A text messaging intervention to promote medication adherence for patients with coronary heart disease: A randomized controlled trial. Patient Educ Couns. 2014; 94 :261–268. [PubMed] [Google Scholar]

19. Huang H-L, et al. Effects of and satisfaction with short message service reminders for patient medication adherence: a randomized controlled study. BMC Med Inform Decis Mak. 2013; 13 :127. [PMC free article] [PubMed] [Google Scholar]

20. Ovbiagele B, et al. Mobile health medication adherence and blood pressure control in recent stroke patients. J Neurol Sci. 2015; 358 :535–537. [PubMed] [Google Scholar]

21. Kirkman MS, et al. Determinants of adherence to diabetes medications: findings from a large pharmacy claims database. Diabetes Care. 2015; 38 :604–609. [PMC free article] [PubMed] [Google Scholar]

22. Choudhry NK, et al. Eliminating medication copayments reduces disparities in cardiovascular care. Health Aff Proj Hope. 2014; 33 :863–870. [PubMed] [Google Scholar]

23. Ha JF, Longnecker N. Doctor-Patient Communication: A Review. Ochsner J. 2010; 10 :38–43. [PMC free article] [PubMed] [Google Scholar]

24. Weiner SJ, et al. Patient-Centered Decision Making and Health Care Outcomes: An Observational Study. Ann Intern Med. 2013; 158 :573–579. [PubMed] [Google Scholar]

25. Weiner SJ, et al. Content coding for contextualization of care: evaluating physician performance at patient-centered decision making. Med Decis Mak Int J Soc Med Decis Mak. 2014; 34 :97–106. [PubMed] [Google Scholar]

26. Fischer MA, et al. Primary Medication Non-Adherence: Analysis of 195,930 Electronic Prescriptions. J Gen Intern Med. 2010; 25 :284–290. [PMC free article] [PubMed] [Google Scholar]

27. Steiner JF, et al. Socio-Demographic and Clinical Characteristics are Not Clinically Useful Predictors of Refill Adherence in Patients with Hypertension. Circ Cardiovasc Qual Outcomes. 2009; 2 :451–457. [PMC free article] [PubMed] [Google Scholar]

28. Andersson Sundell K, Jönsson AK. Beliefs about medicines are strongly associated with medicine-use patterns among the general population. Int J Clin Pract. 2016; 70 :277–285. [PubMed] [Google Scholar]

29. Driesenaar JA, De Smet PAGM, van Hulten R, Noordman J, van Dulmen S. Cue-responding behaviors during pharmacy counseling sessions with patients with asthma about inhaled corticosteroids: Potential relations with medication beliefs and self-reported adherence. Health Commun. 2016:1–10. doi: 10.1080/10410236.2015.1062974. [PubMed] [CrossRef] [Google Scholar]

30. Schoenthaler A, et al. A practice-based randomized controlled trial to improve medication adherence among Latinos with hypertension: study protocol for a randomized controlled trial. Trials. 2015; 16 :290. [PMC free article] [PubMed] [Google Scholar]

31. Sofianou A, et al. Asthma beliefs are associated with medication adherence in older asthmatics. J Gen Intern Med. 2013; 28 :67–73. [PMC free article] [PubMed] [Google Scholar]

32. Lopez M. Increasing medication adherence among HIV+ Hispanics: A culturally tailored application of the imb model. Poster presented at: 141st APHA Annual Meeting; 2013 Nov 2–6; Boston, MA. [Google Scholar]

33. Horne R, et al. Understanding patients’ adherence-related beliefs about medicines prescribed for long-term conditions: a meta-analytic review of the Necessity-Concerns Framework. PloS One. 2013; 8 :e80633. [PMC free article] [PubMed] [Google Scholar]

34. De las Cuevas C, et al. Adaptation and validation study of the Beliefs about Medicines Questionnaire in psychiatric outpatients in a community mental health setting. Hum Psychopharmacol. 2011; 26 :140–146. [PubMed] [Google Scholar]

35. Beléndez-Vázquez M, Hernández-Mijares A, Horne R, Weinman J. Assessment beliefs about treatment: Validity and reliability of the Spanish version of the beliefs about medicines questionnaire [Evaluación de las creencias sobre el tratamiento: Validez y fiabilidad de la versión española del Beliefs about Medicines Questionnaire] Int J Clin Health Psychol. 2007; 7 :767–779. [Google Scholar]

36. Perpiñá Tordera M, Moragón EM, Fuster AB, Bayo AL, Císcar CP. Spanish Asthma Patients’ Beliefs About Health and Medicines: Validation of 2 Questionnaires. Arch Bronconeumol Engl Ed. 2009; 45 :218–223. [PubMed] [Google Scholar]

37. Billimek J, Guzman H, Angulo MA. Effectiveness and feasibility of a software tool to help patients communicate with doctors about problems they face with their medication regimen (EMPATHy): study protocol for a randomized controlled trial. Trials. :16. 20015. [PMC free article] [PubMed] [Google Scholar]

38. Horne R, Weinman J, Hankins M. The beliefs about medicines questionnaire: The development and evaluation of a new method for assessing the cognitive representation of medication. Psychol Health. 1999; 14 :1–24. [Google Scholar]

39. Sweileh WM, et al. Influence of patients’ disease knowledge and beliefs about medicines on medication adherence: findings from a cross-sectional survey among patients with type 2 diabetes mellitus in Palestine. BMC Public Health. 2014; 14 :94. [PMC free article] [PubMed] [Google Scholar]

40. Kreyenbuhl J, et al. A comparison of adherence to hypoglycemic medications between Type 2 diabetes patients with and without serious mental illness. Psychiatry Res. 2011; 188 :109–114. [PMC free article] [PubMed] [Google Scholar]

41. Aflakseir A. Role of illness and medication perceptions on adherence to medication in a group of Iranian patients with type 2 diabetes. J Diabetes. 2012; 4 :243–247. [PubMed] [Google Scholar]

42. Kaplan SH, Billimek J, Sorkin DH, Ngo-Metzger Q, Greenfield S. Reducing Racial/Ethnic Disparities in Diabetes: The Coached Care (R2D2C2) Project. J Gen Intern Med. 2013; 28 :1340–9. [PMC free article] [PubMed] [Google Scholar]

43. Mårdby A-C, Akerlind I, Jörgensen T. Beliefs about medicines and self-reported adherence among pharmacy clients. Patient Educ Couns. 2007; 69 :158–164. [PubMed] [Google Scholar]

44. Mårdby A-C, Akerlind I, Hedenrud T. Different development of general beliefs about medicines during undergraduate studies in medicine, nursing and pharmacy. Patient Educ Couns. 2009; 75 :283–289. [PubMed] [Google Scholar]

45. Arcury TA, et al. Complementary and alternative medicine use among older adults: ethnic variation. Ethn Dis. 2006; 16 :723–731. [PubMed] [Google Scholar]

46. Briesacher BA, Gurwitz JH, Soumerai SB. Patients at-risk for cost-related medication nonadherence: a review of the literature. J Gen Intern Med. 2007; 22 :864–871. [PMC free article] [PubMed] [Google Scholar]