• Center on Health Equity & Access
  • Clinical
  • Health Care Cost
  • Health Care Delivery
  • Insurance
  • Policy
  • Technology
  • Value-Based Care

Factors Affecting Medication Adherence Trajectories for Patients With Heart Failure

Publication
Article
The American Journal of Managed CareMarch 2015
Volume 21
Issue 3

Nonwhite race, smoking, and increasing body mass index were associated with the lowest adherence trajectories for patients with heart failure, with adherence dropping off within the first year.

Objectives: To examine the relationship between patient characteristics and medication adherence trajectories for patients with congestive heart failure (CHF).

Study Design: Historical prospective study.

Methods: We conducted a secondary analysis of data assembled for the Practice Variation and Care Outcomes (PRAVCO) study, which examined patterns of cardiovascular care. We used groupbased trajectory modeling to define medication adherence trajectories, and then modeled factors associated with belonging to a trajectory group during the 6year period from 2005 to 2010 (n = 10,986). We focused on the use of angiotensinconverting enzyme (ACE) inhibitors or angiotensin II receptor blockers (ARBs) for secondary prevention of CHF.

Results: Four trajectory groups were optimal in characterizing adherence level patterns: 1) low adherence group, with an initial average adherence rate of 62% that dropped to between 40% and 50%; 2) increasing adherence group, with an initial average adherence rate of 55% that increased to 90%; 3) decreasing adherence group, with an initial average adherence rate above 90% that decreased to 60%; 4) high adherence group, with an average adherence rate consistently above 90%. Age, region, education, smoking, and race were all significantly associated with the likelihood of belonging to a particular trajectory. Nonwhites were less likely to be in the high adherence group, and smoking was more common in the low adherence group (22%) than in the high group (10%); increasing body mass index and Charlson Comorbidity Index (CCI) scores were also associated with being in the low adherence group.

Conclusions: Population characteristics associated with sustained low adherence might be used to target interventions and improve vulnerable patients’ prospects of heart health.

Am J Manag Care. 2015;21(3):e197-e205

  • Four distinct longitudinal patterns characterize medication adherence for patients with heart failure: 1) low; 2) increasing; 3); decreasing; and 4) high.
  • Nonwhites and patients who smoke or with increasing body mass index are less likely to be in the high adherence group and may need targeted interventions.
  • The factors that explain why nonwhite patients are more likely to follow a clinically inferior adherence trajectory should be examined in future research.

Often caused by coronary artery disease, congestive heart failure (CHF) is a complex clinical syndrome of symptoms and signs that occur when the heart is not pumping efficiently. Currently, more than 5 million people in the United States suffer from CHF, and the prevalence is expected to rise over time as a result of an aging population, improved survival of people with ischemic heart disease, and more effective treatments for CHF.1,2 Clinical guidelines suggest offering angiotensinconverting enzyme (ACE) inhibitors or angiotensin receptor blockers (ARBs) along with betablockers as firstline treatment for patients with CHF.3

Despite widespread knowledge among providers of the efDespite widespread knowledge among providers of the effectiveness of ACE inhibitors/ARBs in patients with CHF, a study by McGlynn and colleagues found that only 64% of patients with CHF were receiving recommended care.4 Other studies have found that even among people who see their providers regularly, 50% are nonadherent to prescribed medications after 6 months.5 This nonadherence results in thousands of premature deaths, preventable hospitalizations, and increased healthcare costs.6,7 Of all medicationrelated hospital admissions, 33% to 69% are due to poor medication adherence, with an approximate cost of $100 billion annually.811 Greater adherence to medications for chronic conditions including hypertension, diabetes, CHF, and hypercholesterolemia has been associated with higher medication costs but net overall reduction in healthcare costs.12,13

We examined data developed from the electronic medical records of 2 large integrated healthcare systems to identify longitudinal medication adherence trajectories that captured the patterns of change in adherence observed in this population over 6 years. Trajectories are simply patterns of change in medication adherence: high adherence sustained over time, for example, or rapid decline following high initial adherence. Two recently published studies have found advantages of groupbased trajectory models (GBTMs) over traditional adherence measures in analyzing patterns of medication adherence. Franklin et al found that GBTMs summarized adherence patterns better than traditional approaches, including average medication possession ratios (MPRs),14 while a study by Li et al demonstrated that GBTMs captured dynamic adherence behavior over time better than percent of covered days, a more commonly used measure.15

Our goal was to determine whether any demographic or behavioral factors were associated with the probability of belonging to a trajectory group that poses a risk to cardiovascular health. After identifying the optimal subset of adherence trajectory groups, we determined the population characteristics that were associated with clinically suboptimal adherence trajectory groups. Such factors could be used to focus adherence promotion interventions and thereby prevent cardiovascular morbidity and mortality.

METHODS

This historical prospective study involved secondary analysis of data assembled for the Practice Variation and Care Outcomes (PRAVCO) study.16 The PRAVCO study used electronic medical and administrative records over an 11year period in 2 large Kaiser healthcare systems— Hawaii and the Pacific Northwest—covering approximately 750,000 persons in geographically and ethnically diverse populations. PRAVCO examined variation in the secondary prevention and management of cardiovascular disease to determine its relationship to outcomes of care for physicians. The goal of PRAVCO was to develop electronic medical record—based qualityofcare indices for 11 cardiovascular primary care services. It related physicians’ prior index scores to subsequent disease incidence and to care utilization in their patients.

Our study focused on factors affecting adherence to ACE inhibitors or ARBs for secondary prevention of CHF during the 6year period of 2005 to 2010 (n = 10,986). To be included in the study, an enrollee needed to have been hospitalized for a cardiovascular event, have a prior diagnosis of CHF, be at least 21 years old, and have been prescribed an ACE inhibitor or ARB. We excluded data prior to 2005 due to inconsistencies in the reporting of race. In order to have enough data to create medication adherence trajectories, only patients with 3 or more years of relevant adherence data were included. These 3 years were not required to be consecutive.

Factors Affecting Adherence

For the PRAVCO study, race information was abstracted from the electronic medical records. We grouped race into 5 categories: 1) white (n = 7562); 2) black (n = 235); 3) Asian (n = 1599); 4) Pacific Islander (n = 208); and 5) other (n = 1382). We used white as the reference group. We separated Asians from Pacific Islanders because there is increasing evidence that highlights differences in quality of care and health outcomes between the 2 groups.17-19

Our models also included age, gender, smoking status, education level, and region (Pacific Northwest or Hawaii). Enrollment, region, age, and gender information were from health plan administrative data. Education level was obtained through geocoding based on year 2000 census data. Body mass index (BMI) and smoking status were obtained at inperson clinical encounters and retrieved from electronic medical record data sources. BMI was calculated every year using height and weight; if more than 1 measure existed in a year, the latter one was used. BMI was measured as a continuous variable; and any change was calculated as BMI (time period 2) — BMI (time period 1). The Charlson Comorbidity Index (CCI) score was constructed from inpatient International Classification of Diseases, Ninth Revision, Clinical Modification diagnoses that were modified by removal of the cardiovascular component because all patients had cardiovas-cular disease.20

Medication Adherence

Pharmacy dispensing data from participating healthcare organizations were used to calculate an annual MPR21,22 that represented the days in a calendar year that patients possessed ACE inhibitors and ARBs and were not hospitalized, and that were subsequent to a primary hospital discharge diagnosis of CHF. The MPR was computed using the remainder of the nonhospitalized days in the year as the denominator and the number of days covered by the dispensing in the numerator. Days covered by dispensing in a prior calendar year were carried over and contributed to the numerator in the next year. Days within the calendar year were excluded from both the numerator and denominator if they were prior to the first ever known inpatient diagnosis of CHF or if they were spent in the hospital.

Statistical Analysis

We summarized demographic characteristics at baseline and calculated mean medication adherence scores for each racial group over time. For the main analyses, we used a GBTM approach. GBTM23 is an application of finite mixture modeling24,25 that assumes a population is composed of a mixture of distinct groups defined by their longitudinal trajectories. The first stage of model development defines groups, with a group being a set of individuals whose outcomes (medication adherence, in this case) follow approximately the same longitudinal course.

The collection of GBTM trajectory groups approximates a continuous population distribution of trajectories that is of unknown shape. Each group summarizes the trajectories in a distinctive region of this continuous distribution. To determine the best model, we used the Bayesian information criteria26,27 and Akaike information criterion,28 as well as consideration of the strengths and weakness of alternative specifications to select the most parsimonious model by balancing model complexity (number of parameters) versus goodness of fit to the sample data.

Figure 1

The model that best fit our data (depicted in ) assumes that the population we observed is composed of 4 distinct medication adherence trajectories. In a second stage, we examined how the probability of trajectory group membership depends on individuals’ preexisting sociodemographic characteristics, such as race, or on characteristics that change during the period of time covered by the trajectory such as their health (ie, comorbid illness) and health risk characteristics (ie, BMI).

Model parameters representing these characteristics quantify the risk and protective factors associated with membership in a trajectory group. We estimated trajectory groups in a model assuming a normal distribution in the first step.

Then we examined the effect of participant characteristics on trajectories in an extension of the basic model in the second step.

The interpretation of the model parameter estimates for individual characteristics depends on whether they are fixed or timevarying. Estimates for fixed characteristics (eg, race) represent the probability of trajectory group membership given a characteristic relative to a reference group. The probability estimates of trajectory group membership vary depending on that fixed characteristic. For timevarying characteristics, they estimate the change in adherence associated with a unit of increase (eg, the change in adherence among individuals within a trajectory group when their CCI score increases from 0 to 1). Hence, our approach yields 3 kinds of results: the shapes of the trajectories, the proportion of the population that is best characterized by each adherence trajectory group, and the probability of belonging to each trajectory group given personlevel attributes.

Groupbased trajectory modeling was carried out using the PROC TRAJ procedure in SAS (Cary, North Carolina).29,30

RESULTS

Medication Adherence Trajectories

The model with 4 medication adherence trajectory groups had the best fit to the data based on the methodology described above. We have labeled these 4 adherence groups as: group 1, low adherence (5.2%, n = 572); group 2, increasing adherence (15.7%, n = 1728); group 3, decreasing adherence (14.4%, n = 1257); and group 4, high adherence (67.6%, n = 7429).

As seen in Figure 1, the low adherence group had an average adherence rate of approximately 62% in the first year, dropping to between 40% and 50% in subsequent years. The high adherence group—the largest group— had mean adherence rates above 90% throughout the 6 years. The decreasing adherence group had a rate that started above 90% but dropped consistently over time, ending at approximately 60% in year 6. The increasing adherence group had a mean adherence rate of approximately 55% in year 1, which increased gradually, reaching almost 90% in year 6.

Patient Characteristics

Table 1 displays patient characteristics by trajectory group. All patient characteristics, except gender and morbidity level, differed significantly by trajectory group. The racial and ethnic composition of each group differed considerably. For instance, whites accounted for 74% of patients who sustained high adherence compared with 35% of those who sustained low adherence. In contrast, Asians represented only 12% of patients who sustained high adherence, but 33% of those who sustained low adherence.

Mean age differed considerably by group, with those having sustained low adherence being approximately 10 years younger than those with sustained high adherence. Substantial differences were also seen by region, with 75% of patients in the high adherence group being from the Pacific Northwest and 25% being from Hawaii, while only 43% of patients in the low adherence group were from the Pacific Northwest. Percent of patients with at least a high school education also varied significantly by group, ranging from 61% for those with low adherence to 71% for patients with high adherence. Smoking status also differed significantly among groups, with 22% of those in the low adherence group categorized as smokers compared with 10% of patients in the high adherence group.

Mean Medication Adherence Related to Race

We also summarized mean adherence rates over time by race (Figure 2). Mean adherence declined over time for all groups. For each year, whites maintained a higher mean adherence rate compared with other racial groups (blacks, Asians, Pacific Islanders), whose mean adherence tended to overlap. In general, Pacific Islanders had the lowest adherence.

Impact of TimeDependent Factors on Medication Adherence Trajectories

Table 2

Changes in BMI and CCI were significantly associated with likelihood of being in particular medication adherence trajectory groups (). Patients whose BMI increased were significantly more likely to be in the low or declining adherence groups and less likely to be in the increasing adherence group, than those without such an increase. A patient with a 1point increase in BMI was 5 times more likely to be in the low adherence or decreasing adherence group, and 4 times less likely to be in the increasing adherence group, than patients without an increase in BMI. Patients with increasing CCI scores were more likely to be in the low adherence group and less likely to be in the high or declining adherence groups.

Race Related to Medication Adherence Trajectories

Table 3

displays the relationship between patient race and likelihood of being in groups other than the low adherence group. Blacks, Asians, Pacific Islanders, and patients of other race were all significantly less likely than whites to be in the high adherence group. Compared with whites, patients of other race were less likely to be in the declining adherence group.

DISCUSSION

The 2002 Institute of Medicine report Unequal Treatment brought attention to the issue of racial and ethnic disparities in healthcare in the United States.31 Prior studies have suggested that improved medication adherence might help to diminish these disparities.32-34 Our study of medication adherence trajectories for almost 11,000 patients with CHF enrolled in an integrated healthcare system over a 6 year period sought to identify patterns of medication nonadherence that might help guide interventions to reduce health disparities.

To our knowledge, our study is the first to explore differences in medication adherence trajectories in CHF patients related to race and ethnicity as well as to changing BMI and CCI scores. The value of our contribution is having identified: 1) typical trajectory groups that characterize this high risk population, and 2) attributes associated with the subset of the population at greatest risk of not benefiting from a very effective treatment.

There are several advantages to using GBTMs over more common approaches such as average MPRs. First, trajectory modeling allows us to identify distinct patterns of medication adherence that actually exist for the population. A typical analysis using MPRs would examine average adherence for the whole population. If we had merely examined average adherence, we would not have known that there are, in fact, 4 common patterns: high, low, increasing, and decreasing. The distinct trajectory of interest (sustained low adherence) would most likely not have been an outcome in other approaches; deviations from the overall population average would have been. Overall population average trajectories are often a poor fit for a large portion of the actual trajectories used to calculate them, and they hide disparities. Letting the data define the distinct trajectory groups, initially, leads to a much better fit. Consequently, the whole procedure of drawing inferences is tied much more meaningfully to real observed phenomena.

We discovered 4 distinct medication trajectories: low, increasing, decreasing, and high, and we found significant racial differences in likely trajectories. Compared with blacks, Asians, Pacific Islanders, and whites were more likely to have sustained low adherence, beginning in the first year post discharge. This suggests that targeted interventions may need to be initiated during the initial year of treatment to ensure that low adherence patients understand the importance of taking their medications and have adequate access to ambulatory care—a finding consistent with prior studies. Nonwhite populations may not be prescribed necessary medications or may be less likely to take medications as prescribed. Numerous studies have documented that African Americans and Hispanics35-42 have worse medication adherence than whites, as do some Asian and Pacific Islander subgroups, including Filipinos, Native Hawaiians, and Koreans.43,44 In these studies, medication adherence was commonly measured using either patient selfreport or administrative claims data, and the studies usually assessed factors related to medication adherence in a given year.

Our analyses also reveal that patients with increasing BMI or CCI scores are also more likely to be in the low adherence group regardless of race and ethnicity. It may be that several negative behavioral changes are occurring at once for these patients. For instance, over time, they may have stopped taking their medication on a daily basis and stopped eating a healthy diet. As these patients are already at higher risk of adverse events, attention needs to be given to improving their medication adherence.

A recent Cochrane Review found that effective interventions to improve medication adherence tend to be comprehensive, including combinations of “more convenient care, information, reminders, selfmonitoring, reinforcement, counseling, family therapy, psychological therapy, crisis intervention, manual telephone follow up, and supportive care.”45 As the Cochrane Review found that even effective interventions did not tend to have a large impact on adherence and clinical outcomes, we aimed to better understand how to effectively improve adherence, particularly among patients with other risk factors, as seen in our study, including increasing BMI and smoking. Further research is needed to investigate the possibility of developing an algorithm, including race, BMI, smoking, and comorbidities, to identify patients at high risk of a poor adherence trajectory who may benefit from targeted interventions.

Limitations

There are several limitations to this study. First, data are from 2 integrated healthcare systems: one in Hawaii, and the other in the Pacific Northwest. Although diverse, these patients who were treated under an integrated care model may have adherence trajectories different from those of the general population. Hence, our results may not generalize to other populations or geographic areas. We may find even greater disparities in studies in which not everyone has comprehensive insurance. Second, although there was enough racial diversity to conduct analyses of 5 distinct groups, we did not have enough observations to break out specific ethnic subgroups. Third, as this was a secondary analysis of existing data, we did not have the opportunity to collect other information that might be relevant, such as type of medication and access to care. Another limitation of this study is that we did not have individual data on socioeconomic factors that have been shown to affect medication adherence, including income level.46 Moreover, the information on education that was available for the study was census-block—level rather than self reported individual level information. While this is a limitation, prior studies have shown that census level and individual level socioeconomic measures are similarly associated with health outcomes.47,48

CONCLUSIONS

Only 20% of the population showed clinically suboptimal adherence patterns, including low adherence (5.2%) and decreasing adherence (14.4%). Our findings suggest the possible advantages of focusing adherence promotion on the subset of patients with attributes that indicate risk of sustained poor medication adherence. This would in-clude patients who are nonwhite, are smokers, have an increasingBMI, and have a significant comorbid condition.

The authors would like to acknowledge the work of the investigators of the PRAVCO study for their efforts in compiling the data set, and the statistical support of Sixiang Nie, whose efforts helped to make this manuscript possible. William Vollmer provided critically important expertise in defining the medication possession ratio algorithm. Adrianne Feldstein, Richard Meenan, and Mary Ann McBurnie provided important scientific input and operational oversight of the PRAVCO project. Mark Schmidt, Aleli Vinoya, and Weiming Hu provided outstanding data warehouse and programming expertise. Aileen Uchida and Sharyn Nuha were essential in expediting the completion of the study.

Author Affiliations: University of Hawaii at Hilo, Daniel K. Inouye College of Pharmacy (DTJ); Kaiser Permanente Center for Health Research (AEW, YGD, SKT, CMT, TMV), Honolulu, HI; Kaiser Permanente Center for Health Research (CC), Portland, OR.

Source of Funding: Agency for Healthcare Research and Quality Grant Award No. [5 R18 HS 017016] and Agency for Healthcare Research and Quality [K01 HS018072].

Author Disclosures: The authors report no relationship or financial interest with any entity that would pose a conflict of interest with the subject matter of this article.

Authorship Information: Concept and design (YGD, DTJ, AEW, CMT, TMV); acquisition of data (CC, AEW, TMV); analysis and interpretation of data (CC, YGD, DTJ, AEW, CMT, TMV); drafting of the manuscript (DTJ); critical revision of the manuscript for important intellectual content (CC, YGD, DTJ, AEW, CMT, TMV); statistical analysis (CC, AEW); provision of patients or study materials (SKT); obtaining funding (TMV); administrative, technical, or logistic support (YGD, SKT); and supervision (AEW).

Address correspondence to: Deborah Taira Juarez, ScD, University of Hawaii at Hilo, Daniel K. Inouye College of Pharmacy, 677 Ala Moana Blvd, Ste 1025, Honolulu, HI 96813. Email: dtjuarez@hawaii.edu.

1. Go AS, Mozaffarian D, Roger VL, et al; American Heart Association Statistics Committee and Stroke Statistics Subcommittee. Heart disease and stroke statistics—2013 update: a report from the American Heart Association. Circulation. 2013;127(1):e6-e245.

2. Owan TE, Hodge DO, Herges RM, Jacobsen SJ, Roger VL, Redfield MM. Trends in prevalence and outcome of heart failure with preserved ejection fraction. N Engl J Med. 2006;355(3):251-259.

3. Yancy CW, Jessup M, Bozkurt B, et al; American College of Cardiol-ogy Foundation; American Heart Association Task Force on Practice Guidelines. 2013 ACCF/AHA guideline for the management of heart failure: a report of the American College of Cardiology Founda-tion/American Heart Association Task Force on Practice Guidelines. J

Am Coll Cardiol. 2013;62(16):e147-e239.

4. McGlynn EA, Asch SM, Adams J, et al. The quality of health care delivered to adults in the United States. N Engl J Med. 2003;348(26): 2635-2645.

5. Adherence to long-term therapies: evidence for action. WHO/ MNC/3.01. WHO website. http://www.who.int/chp/knowledge/publica-tions/adherence_report/en/. Published January 2003. Accessed May 20, 2012.

6. Col N, Fanale JE, Kronholm P. The role of medication noncompliance and adverse drug reactions in hospitalizations in the elderly. Arch Intern Med. 1990;150(4):841-845.

7. Schiff GD, Fung S, Speroff T, McNutt RA. Decompensated heart fail-ure: symptoms, patterns of onset, and contributing factors. Am J Med. 2003;114(8):625-630.

8. McDonnell PJ, Jacobs MR. Hospital admissions resulting from preventable adverse drug reactions. Ann Pharmacother. 2002;36(9): 1331-1336.

9. Senst BL, Achusim LE, Genest RP, et al. Practical approach to deter-mining costs and frequency of adverse drug events in a health care network. Am J Health Syst Pharm. 2001;58(12):1126-1132.

10. Berg JS, Dischler J, Wagner DJ, Raia JJ, Palmer-Shevlin N. Medica-tion compliance: a healthcare problem. Ann Pharmacother.1993;27 (9 suppl):S1-S24.

11. Osterberg L, Blaschke T. Adherence to medication. N Engl J Med. 2005;353(5):487-497.

12. Sokol MC, McGuigan KA, Verbrugge RR, Epstein RS. Impact of medication adherence on hospitalization risk and healthcare cost. Med Care. 2005;43(6):521-530.

13. Roebuck MC, Liberman JN, Gemmill-Toyama M, Brennan TA. Medication adherence leads to lower health care use and cost despite increased drug spending. Health Aff (Millwood). 2011;30(1):91-99.

14. Franklin JM, Shrank WH, Pakes J, et al. Group-based trajectory models: a new approach to classifying and predicting long-term medi-cation adherence. Med Care. 2013;51(9):789-796.

15. Li Y, Zhou H, Cai B, et al. Group-based trajectory modeling to as-sess adherence to biologics among patients with psoriasis. Clinico-econ Outcomes Res. 2014;6:197-208.

16. Williams AE, Vogt TM. Using Information Technology to Improve the Quality of Cardiovascular Disease Prevention and Management. Grant final report, Grant ID: 5R18HS017016. Rockville, MD: Agency for Healthcare Research and Quality; June 2010.

17. Sarnquist CC, Grieb EM, Maldonado YA. How racial and ethnic groupings may mask disparities: the importance of separating Pacific Islanders from Asians in prenatal care data. Matern Child Health J. 2010;14(4):635-641.

18. Srinivasan S, Guillermo T. Toward improved health: disaggregating Asian American and Native Hawaiian/Pacific Islander data. Am J Public Health. 2000;90(11):1731-1734.

19. Shabbir S, Kwan D, Wang MC, Shih M, Simon PA. Asians and Pacific Islanders and the growing childhood obesity epidemic. Ethn Dis. 2010;20(2):129-135.

20. Charlson ME, Pompei P, Ales KL, MacKenzie CR. A new method of classifying prognostic comorbidity in longitudinal studies: development and validation. J Chron Dis. 1987;40(5):373-383.

21. Andrade SE, Kahler KH, Frech F, Chan KA. Methods for evaluation of medication adherence and persistence using automated databases. Pharmacoepidemiol Drug Saf. 2006;15(8):565-574; discussion 575-577.

22. Steiner JF, Koepsell TD, Fihn SD, Inui TS. A general method of compliance assessment using centralized pharmacy records. description and validation. Med Care. 1988;26(8):814-823.

23. Nagin D. Group Based Modeling of Development. Cambridge, MA: Harvard University Press; 2005.

24. McLachlan G, Peel D. Finite Mixture Models. New York, NY: Wiley; 2004.

25. Titterington DM, Smith AFM, Makov UE. Statistical Analysis of Finite Mixture Models. New York, NY: Wiley; 1985.

26. Raftery AE. 1995. Bayesian model selection in social research. In: Marsden PV, ed. Sociological Methodology. Cambridge, MA: Blackwell; 1995:111-163.

27. Schwartz G. Estimating the dimension of a model. Ann Stat. 1978; 6(2):461-464.

28. Akaike H. A new look at the statistical model identification. IEEE Trans Automatic Control. 1974;19(6):716-723.

29. Jones BL, Nagin DS, Roeder K. A SAS Procedure Based on Mixture Models for Estimating Developmental Trajectories. Sociological Meth-ods & Research. 2001;29(3):374-393.

30. Nagin DS, Odgers CL. Group-based trajectory modeling in clinical research. Annu Rev Clin Psychol. 2010;6:109-138.

31. King RK, Green AR, Tan-McGrory A, Donahue EJ, Kimbrough-Sugick J, Betancourt JR. A plan for action: key perspectives from the racial/ ethnic disparities strategy forum. Milbank Q. 2008;86(2):241-272.

32. Gifford AL, Bormann JE, Shively MJ, Wright BC, Richman DD, Bozzette SA. Predictors of self-reported adherence and plasma HIV concentrations in patients on multidrug antiretroviral regimens. J Acquir Immune Defic Syndr. 2000;23(5):386-95.

33. Muma RD, Ross MW, Parcel GS, Pollard RB. Zidovudine adherence among individuals with HIV infection. AIDS Care. 1995;7(4):439-447.

34. Laine C, Newschaffer CJ, Zhang D, Cosler L, Hauck WW, Turner BJ. Adherence to antiretroviral therapy by pregnant women infected with human immunodeficiency virus: a pharmacy claims-based analysis. Obstet Gynecol. 2000;95(2):167-173.

35. Adams AS, Trinacty CM, Zhang F, et al. Medication adherence and racial differences in A1C control. Diabetes Care. 2008;31(5):916-921.

36. Adeyemi AO, Rascati KL, Lawson KA, Strassels SA. Adherence to oral antidiabetic medications in the pediatric population with type 2 diabetes: a retrospective database analysis. Clin Ther. 2012;34(3): 712-719.

37. Trinacty CM, Adams AS, Soumerai SB et al. Racial differences in long-term adherence to oral antidiabetic drug therapy: a longitudinal cohort study. BMC Health Serv Res. 2009;9:24.

38. Blackwell SA, Baugh DK, Montgomery MA, Ciborowski GM, Waldron CJ, Riley GF. Noncompliance in the use of cardiovascular medications in the Medicare Part D population. Medicare Medicaid Res Rev. 2011;1(4):E1-E27.

39. Charles H, Good CB, Hanusa BH, Chang CC, Whittle J. Racial differ-ences in adherence to cardiac medications. J Natl Med Assoc. 2003; 95(1):17-27.

40. Dickson M, Plauschinat CA. Racial differences in medication com-pliance and healthcare utilization among hypertensive Medicaid recipi-ents: fixed-dose vs free-combination treatment. Ethn Dis. 2008;18(2): 204-209.

41. Gebregziabher M, Lynch CP, Mueller M, et al. Using quantile regres-sion to investigate racial disparities in medication non-adherence. BMC Med Res Method. 2011;11:88.

42. Gellad WF, Haas JS, Safran DG. Race/ethnicity and nonadherence to prescription medications among seniors: results of a national study. J Gen Intern Med. 2007;22(11):1572-1578.

43. Taira DA, Gelber RP, Davis J, Gronley K, Chung RS, Seto TB. Antihy-pertensive adherence and drug class among Asian Pacific Americans. Ethn Health. 2007;12(3):265-281.

44. Lee R, Taira DA. Adherence to oral hypoglycemic agents in Hawaii. Prev Chronic Dis. 2005;2(2):A09.

45. Haynes RB, Ackloo E, Sahota N, McDonald HP, Yao X. Interventions for enhancing medication adherence. Cochrane Database Syst Rev. 2008; 2:CD000011.

46. Krieger N. Overcoming the absence of socioeconomic data in medical records: validation and application of a census-based method-ology. Am J Public Health. 1992;82(5):703-710.

47. Domínguez-Berjón F, Borrell C, Rodríguez-Sanz M, Pastor V. The usefulness of area-based socioeconomic measures to monitor social inequalities in health in Southern Europe. Eur J Public Health. 2006; 16(1):54-61.

48. Rehkopf DH, Haughton LT, Chen JT, Waterman PD, Subramanian SV, Krieger N. Monitoring socioeconomic disparities in death: compar-ing individual-level education and area-based socioeconomic mea-sures. Am J Public Health. 2006;96(12):2135-2138.

Related Videos
Related Content
© 2024 MJH Life Sciences
AJMC®
All rights reserved.