The Center for Medicare and Medicaid Services (CMS) uses its Medicare STAR program to monitor and reward the quality of care in health plans with Medicare Advantage enrollees, and provides substantial incentives to health plans that perform well on its Medicare STAR metrics.1 In 2012, CMS introduced 3 new Medicare STAR metrics to measure medication adherence for cardiovascular disease (CVD) risk factor medications: angiotensin converting enzyme inhibitors and angiotensin receptor blockers (ACEI/ARBs) to control hypertension; statins to control LDL-cholesterol; and oral antihyperglycemics to control glycosylated hemoglobin (HbA1c). The implementation of these new measures emphasizes the responsibility of health plans to monitor and improve medication adherence in their Medicare populations.2,3
Overall performance on these Medicare STAR measures is highly dependent on the adherence of patients with diabetes, who account for almost all oral antihyperglycemic use and much antihypertensive and statin use.4–6 Most research on medication adherence to these CVD risk factor medications has focused on younger populations,7–11 and there is limited information on the correlates of adherence in diabetes patients aged 65 and older.10 Medication adherence research traditionally focuses on patient-level characteristics such as age and sex,7–12 despite the fact that patient characteristics have limited ability to predict adherence.13 In contrast, interventions focused on health care system-level characteristics may hold the most hope for improving appropriate CVD risk factor medication use at the population level,2,14–16 as these characteristics are modifiable through health care benefit and policy changes. Despite the potential for system-level interventions to improve adherence in high-risk patients, the system-level correlates of adherence to CVD risk factor medications in Medicare-aged diabetes patients, and their combined effect on adherence levels, are largely unknown.
The purpose of this study is to examine the relationship between Medicare STAR medication adherence metrics and modifiable health system-level characteristics in a cohort of Medicare-aged diabetes patients.
Study Setting and Population
The population and data for this study were derived from the Surveillance, Prevention, and Management of Diabetes Mellitus (SUPREME-DM) DataLink, a multisite data resource for research that accesses electronic health record, clinical, and administrative data from 2005 to 2011.17–20 The current study drew data from 3 SUPREME-DM integrated, group-model health care delivery systems sites that collectively serve 4.1 million members: Kaiser Permanente Northern California (KPNC), Kaiser Permanente Colorado (KPCO), and Kaiser Permanente Northwest (KPNW; Northwest Oregon and Southwest Washington). Patients were eligible for the current study if they had diabetes in 2010, and were 65 years of age and above as of January 1, 2010. Patients were defined as having diabetes if they had ≥2 outpatient diabetes ICD-9 diagnosis codes (250.xx) within a 2-year window since the start of 2000.21–23
Medicare STAR Medication Adherence
We calculated the Medicare STAR adherence metrics following CMS specifications to obtain the Proportion of Days Covered (PDC) measure in 2010 for all patients for each of the 3 therapeutic groupings covered by the measures: ACEI/ARBs, statins, and oral diabetes medications.3,24 PDC is defined as the percent of days in the measurement period “covered” by prescription fills for the same medication or medications in the same therapeutic category.25 Per CMS specifications, the measurement period began with patients’ first fill in 2010 through December 31, 2010, and applies only to patients with ≥2 fills in the therapeutic grouping within that period; also per CMS specifications, the PDC calculation for oral diabetes medications excludes patients taking insulin.3 The PDC ranges from 0 to 1; the Medicare STAR adherence measure considers patients to be “adherent” if their PDC is ≥0.8.
Health System-level Independent Variables
This study focused on 4 health system-level predictors of Medicare STAR medication adherence: mean days’ supply of medications, annual individual out-of-pocket maximum costs, generic drug copayment, and percent of medications delivered through mail order pharmacy. These variables are considered “system level” because they are directly influenced by the benefit and pharmacy delivery structures of a patient’s health plan that are potentially modifiable. Days’ supply was derived from the pharmacy electronic medication dispensing record. Annual individual out-of-pocket cost maximum and generic drug copayments per 30 days of medication supply were obtained from membership files. The rate of mail order pharmacy use to fill a specific medication in 2010 was calculated as the percent of time a patient used the mail order service (vs. a KP “brick and mortar” pharmacy) to fill a medication in the 12-month period; details on the KP mail order pharmacy system have been published elsewhere.14,15
To examine the relationship between adherence and health system-level characteristics and patient characteristics, we obtained relative risks using 3 Poisson regression models, 1 for each of the 3 therapeutic groupings, using PDC≥0.8 (vs. <0.8) as the dichotomous-dependent variable; this method is appropriate for determining risk in cross-sectional data where the outcome is relatively common.26 Each model was adjusted for the 4 health system-level independent variables defined above, and also adjusted for patient age, sex, race/ethnicity, census block-level median household income and education, delivery system region, and total number of medications the patient was concurrently taking in 2010. To test the combined effect of all 4 health system-level variables on adherence, we obtained predicted percentages27 of patients achieving “good” (ie, PDC≥0.8) Medicare STAR adherence in each of the therapeutic groupings when all 4 system-level variables were at the specified values “optimized” at the cutpoints used in the regression models (days’ supply >90; 30 day copayment ≤$10; out-of-pocket maximum ≤$2000; >50% of fills delivered through mail order pharmacy), versus when all 4 system-level variables were at the least optimized specified values (days’ supply <31; copay >$10; out-of-pocket maximum >$2000; no fills through mail order pharmacy) after adjusting for patient demographic, clinical, and socioeconomic characteristics. To further our understanding of the role of the health system-level variables that may affect potential disparities in adherence, we tested whether any of our health system-level variables had a statistically significant (P<0.01) interaction with race/ethnicity; significant interaction terms were included in the final models. All analyses were performed using Stata v10.1.
This study was approved by the KPNC Institutional Review Board; KPCO and KPNW ceded IRB oversight to KPNC.
Table 1 presents the demographic and clinical characteristics of the patients in the sample by each CVD risk factor medication therapeutic grouping. On the basis of the Medicare STAR metric definition, 81%, 82%, and 79% of patients were considered adherent to medications in the therapeutic groupings for ACEI/ARBs, oral diabetes medications, and statins, respectively.
Table 2 shows predictors of whether a patient achieved the Medicare STAR adherence target of PDC≥0.8 in each therapeutic grouping. After controlling for individual-level predictors of adherence, the strongest predictor of achieving STAR-defined medication adherence was a median prescribed days’ supply of >90 days of medication in each therapeutic group (RR=1.61 for each; all P<0.001). Using mail order pharmacy to fill medications >50% of the time was independently associated with better adherence with these therapeutic groupings (RR=1.07, 1.06, 1.07 for antihypertensives, oral diabetes medications, and statins, respectively; P<0.001). Medication copays ≤$10 (RR=1.02, 1.02, 1.02; P<0.001) and annual out-of-pocket maximums ≤$2000 (RR=1.02, 1.01, 1.02; P<0.001) were also significantly associated with higher adherence for all 3 therapeutic groups. Hispanic and black patients were significantly less adherent across each therapeutic grouping. In addition, there were also a significant positive interactions between mail order pharmacy use and black race/ethnicity (RR=1.05, P<0.05; 1.05, P<0.01 for oral diabetes medications and statins, respectively), and a significant positive interaction between mail order pharmacy use and Hispanic race/ethnicity for ACEI/ARBS and statins (RR=1.03, 1.04 respectively, both P<0.001).
Figure 1 shows that the predicted percentage of patients who achieved medication adherence with optimized values for each of these system-level predictors was 91%, 90%, and 90% for antihypertensives, oral diabetes medications, and statins, respectively, compared with 51%, 51%, and 50%, respectively, in patients with the least optimized values for these factors.
We found that the strongest predictor of greater Medicare STAR adherence in 3 large Medicare Advantage providers was a greater days’ supply of medications. System-level interventions that focus on increasing the standard days’ supply of medications, especially low-cost generic medications, may be an important lever for increasing access to medications for patients at risk for low adherence.2,28–31 Having lower copays and lower annual individual out-of-pocket maximums were also associated with higher levels of medication adherence, consistent with prior work showing that diabetes patients may be vulnerable to cost-related nonadherence.32 This suggests that “value-based” insurance designs that are sensitive to the out-of-pocket costs of diabetes care may improve medication adherence for patients,33 and subsequently improve Medicare STAR adherence scores for the health plans that offer such benefit designs.
We found that medication adherence was higher among individuals with greater use of mail order pharmacy to deliver medications. Although this relationship has been demonstrated in previous research in younger patients,14–16,34 our study is the first to indicate that mail order pharmacy use is related to better adherence in Medicare-aged diabetes patients as well.
To the best of our knowledge, this study is the first to examine the cumulative effect of these 4 system-level factors on medication adherence. The predicted percentage of patients who achieved Medicare STAR adherence on these medications was approximately 90% in patients with all 4 system-level factors optimized: a level of adherence at the population level that exceeds what is required to achieve a 5-star Medicare STAR rating on these measures,24 and is almost twice the level of patients predicted to have good adherence when these health system-level factors were at the least optimized level. This suggests that multifactorial efforts to lower system-level barriers to obtaining medications can profoundly impact health plan Medicare STAR scores for CVD risk factor medication adherence. Within our study cohort, <15% of patients (data not shown) had the defined maximized values for all 4 system-level variables, suggesting that close to 90% of patients with diabetes could potentially benefit if health system-level barriers to optimizing their use of these chronic illness medications were addressed.
In this study, black and Hispanic patients derived an even greater adherence benefit from mail order pharmacy use than white patients, as demonstrated by the positive interaction between mail order pharmacy use and race/ethnicity. Mail order pharmacy use may mitigate disparities in adherence16: as mail order use eliminates the need for travel to a local pharmacy, it may have a stronger benefit for patients with transportation issues, time constraints, or other barriers to access that may particularly affect minority patients.35 However, despite this potential for increased benefit of mail order use by black and Hispanic patients compared with non-Hispanic whites, previous research suggests that nonwhite patients with diabetes are the least likely to use the mail order pharmacy.14 Understanding the reasons for these disparities in the use of mail order pharmacy, and developing effective interventions to encourage the use of mail order pharmacy among Medicare-aged patients, particularly among nonwhite patients, are important areas of future research.
Our study included only diabetes patients; Medicare STAR metrics are also applied to hypertension or hyperlipidemia patients without diabetes. Adherence based on Medicare STAR measures in this study setting was generally high, which is consistent with this system’s high scores on other STAR metrics3; the relationship between adherence and system-level factors may be different in other settings. It is possible that the Medicare-aged diabetes patients in our study may have characteristics that differ from those in other settings, which could impact the generalizability of our findings. These findings are based on cross-sectional data: while the relationship between system-level factors and adherence suggest approaches for a trial intervention, we are unable to establish whether change in these same factors would in fact improve STAR ratings.
We found that system-level factors had a consistent relationship with medication adherence in Medicare-aged patients, and that optimizing these factors almost doubled the level of “good” adherence in this population. Interventions that focus on improving adherence and reducing medication adherence disparities by providing behavioral “nudges” to large populations36 should focus modifying these types of system-level barriers to obtaining evidence-based medications.
The authors would like to gratefully acknowledge the contributions of Dr Michael Chase, Associate Medical Director for Quality at Kaiser Permanente Colorado; Dr Alan Whippy, Medical Director of Quality and Safety, The Permanente Medical Group, Oakland, CA; and Dr Elizabeth A. McGlynn, Director of the Kaiser Permanente Center for Effectiveness and Safety Research.
2. Steiner JF. Rethinking adherence
. Ann Intern Med. 2012;157:580–585.
3. Schmittdiel J, Raebel M, Dyer W, et al.. The Medicare STAR adherence
measure excludes diabetes patients with poor CVD risk factor control. Am J Manag Care. 2015;20:e581–e589.
4. Wienbergen H, Senges J, Gitt AK. Should we prescribe statin and aspirin for every diabetic patient? Is it time for a polypill? Diabetes Care. 2008;31suppl 2S222–S225.
5. Johnson ML, Singh H. Patterns of antihypertensive therapy among patients with diabetes. J Gen Intern Med. 2005;20:842–846.
6. Konzem SL, Devore VS, Bauer DW. Controlling hypertension in patients with diabetes. Am Fam Physician. 2002;66:1209–1214.
7. Schmittdiel JA, Uratsu CS, Karter AJ, et al.. Why don’t diabetes patients achieve recommended risk factor targets? Poor adherence
versus lack of treatment intensification. J Gen Intern Med. 2008;23:588–594.
8. Schmittdiel JA, Uratsu CS, Fireman BH, et al.. The effectiveness of diabetes care management in managed care. Am J Manag Care. 2009;15:295–301.
9. Ho PM, Rumsfeld JS, Masoudi FA, et al.. Effect of medication nonadherence on hospitalization and mortality among patients with diabetes mellitus. Arch Intern Med. 2006;166:1836–1841.
10. Wiegand P, McCombs JS, Wang JJ. Factors of hyperlipidemia medication adherence
in a nationwide health plan. Am J Manag Care. 2012;18:193–199.
11. Walker EA, Molitch M, Kramer MK, et al.. Adherence
to preventive medications: predictors and outcomes in the Diabetes Prevention Program. Diabetes Care. 2006;29:1997–2002.
12. Yang Y, Thumula V, Pace PF, et al.. Predictors of medication nonadherence among patients with diabetes in Medicare Part D programs: a retrospective cohort study. Clin Ther. 2009;31:2178–2188discussion 2150-2171.
13. Raebel MA, Ellis JL, Carroll NM, et al.. Characteristics of patients with primary non-adherence
to medications for hypertension, diabetes, and lipid disorders. J Gen Intern Med. 2012;27:57–64.
14. Duru OK, Schmittdiel JA, Dyer WT, et al.. Mail-order pharmacy use and adherence
to diabetes-related medications. Am J Manag Care. 2010;16:33–40.
15. Schmittdiel JA, Karter AJ, Dyer W, et al.. The comparative effectiveness of mail order pharmacy use vs. local pharmacy use on LDL-C control in new statin users. J Gen Intern Med. 2011;26:1396–1402.
16. Adams AS, Uratsu C, Dyer W, et al.. Health system factors and antihypertensive adherence
in a racially and ethnically diverse cohort of new users. JAMA Intern Med. 2013;173:54–61.
17. Nichols GA, Desai J, Elston Lafata J, et al.. Construction of a multisite DataLink using electronic health records for the identification, surveillance, prevention, and management of diabetes mellitus: the SUPREME-DM project. Prev Chronic Dis. 2012;9:E110.
18. Raebel MA, Xu S, Goodrich GK, et al.. Initial antihyperglycemic drug therapy among 241 327 adults with newly identified diabetes from 2005 through 2010: a surveillance, prevention, and management of diabetes mellitus (SUPREME-DM) study. Ann Pharmacother. 2013;47:1280–1291.
19. Schmittdiel J, Raebel M, Dyer W, et al.. Prescription medication burden in patients with newly-diagnosed diabetes: the SUrveillance, PREvention, and ManagEment of Diabetes Mellitus (SUPREME-DM) cohort. J Am Pharm Assoc. 2014;54:374–382.
20. Raebel MA, Ellis JL, Schroeder EB, et al.. Intensification of antihyperglycemic therapy among patients with incident diabetes: a Surveillance Prevention and Management of Diabetes Mellitus (SUPREME-DM) study. Pharmacoepidemiol Drug Saf. 2014;23:609–618.
21. Miller DR, Safford MM, Pogach LM. Who has diabetes? Best estimates of diabetes prevalence in the Department of Veterans Affairs based on computerized patient data. Diabetes Care. 2004;27suppl 2B10–B21.
22. O’Connor PJ, Rush WA, Pronk NP, et al.. Identifying diabetes mellitus or heart disease among health maintenance organization members: sensitivity, specificity, predictive value, and cost of survey and database methods. Am J Manag Care. 1998;4:335–342.
23. Zgibor JC, Orchard TJ, Saul M, et al.. Developing and validating a diabetes database in a large health system. Diabetes Res Clin Pract. 2007;75:313–319.
26. Zou G. A modified poisson regression approach to prospective studies with binary data. Am J Epidemiol. 2004;159:702–706.
27. Cameron AC, Trivedi PK. Microeconometrics Using Stata, Revised Edition. 2010.College Station, TX: STATA Press.
28. Shrank WH, Stedman M, Ettner SL, et al.. Patient, physician, pharmacy, and pharmacy benefit design factors related to generic medication use. J Gen Intern Med. 2007;22:1298–1304.
29. Steiner JF, Robbins LJ, Roth SC, et al.. The effect of prescription size on acquisition of maintenance medications. J Gen Intern Med. 1993;8:306–310.
30. Batal HA, Krantz MJ, Dale RA, et al.. Impact of prescription size on statin adherence
and cholesterol levels. BMC Health Serv Res. 2007;7:175.
31. Foster DG, Hulett D, Bradsberry M, et al.. Number of oral contraceptive pill packages dispensed and subsequent unintended pregnancies. Obstet Gynecol. 2011;117:566–572.
32. Schmittdiel JA, Ettner SL, Fung V, et al.. Medicare Part D coverage gap and diabetes beneficiaries. Am J Manag Care. 2009;15:189–193.
33. Choudhry NK, Avorn J, Glynn RJ, et al.. Full coverage for preventive medications after myocardial infarction. N Engl J Med. 2011;365:2088–2097.
34. Choudhry NK, Fischer MA, Smith BF, et al.. Five features of value-based insurance design plans were associated with higher rates of medication adherence
. Health Aff (Millwood). 2014;33:493–501.
35. Duru OK, Gerzoff RB, Selby JV, et al.. Identifying risk factors for racial disparities in diabetes outcomes: the translating research into action for diabetes study. Med Care. 2009;47:700–706.
36. Thaler RH, Sunstein CR. Nudge: Improving Decisions About Health, Wealth, and Happiness. 2008.New Haven, CT: Yale University Press.