The coverage gap in the Medicare Part D program has a significant negative impact on adherence to diabetes medications.
To evaluate the impact of Medicare Part D coverage gap (donut hole) on adherence to diabetes medications.
Retrospective cohort analysis based on pharmacy claims data.
The sample included 12,881 Medicare Part D beneficiaries with diabetes who entered the coverage gap in 2008. Sample patients had 3 different levels of coverage in the donut hole: no coverage, generic drug coverage only, and both generic and brand-name drug coverage. Adherence was measured by the proportion of days covered. We used a difference-in-difference model to evaluate the effect of coverage gap on adherence.
In the donut hole, the average copayment for diabetes medications increased substantially for beneficiaries with no coverage and beneficiaries with generic drug coverage only, whereas the average copayment for beneficiaries with both generic and brand-name medication coverage declined slightly. Compared with beneficiaries with full coverage of both generic and brand-name drugs, beneficiaries with no coverage (odds ratio[OR] = 0.617, P <.0001, 95% confidence interval [CI] = 0.523, 0.728) and beneficiaries with generic drug coverage only (OR = 0.702, P <.0001, 95% CI = 0.604, 0.816) were significantly less likely to be adherent after entering the donut hole. The difference between having generic coverage and no coverage was not significant (P = .1586).
The coverage gap in the Medicare Part D program has a significant negative impact on medication adherence among beneficiaries with diabetes. Availability of brand-name drug coverage in the donut hole is critical to adherence to diabetes medications.
(Am J Manag Care. 2010;16(12):911-918)
The Medicare Part D coverage gap has a significant negative impact on adherence to diabetes medications.
The Medicare Part D program, introduced on January 1, 2006, provides prescription drug coverage for Medicare beneficiaries. One unique feature of the Part D benefit design is the coverage gap (or donut hole). The defined standard benefit in 2008 started with a $275 deductible and a 25% copayment for drug spending between $275 and $2510. After the initial coverage period, beneficiaries entered a coverage gap, in which they paid 100% of the drug cost, until their true out-of-pocket drug spending reached the catastrophic limit of $4050 (or total drug spending of $5726.25). Under the catastrophic coverage, beneficiaries pay the greater of a 5% or a $2.25/$5.60 (generic/brand-name) copayment.
The donut hole is a controversial component of the Part D design because a lack of benefit may negatively impact patients’ drug utilization and subsequent clinical outcomes. Studies have shown that beneficiaries of private drug benefit plans with annual benefit limits are more likely to discontinue their medications and have a higher rate of hospitalization compared with beneficiaries who have no benefit limits.1,2 Sun et al reported that the donut hole was associated with reductions in medications used for potentially disabling and life-threatening conditions.3 Another study by Zhang et al indicated that beneficiaries with no coverage in the donut hole decreased their use of monthly prescriptions by about 14%.4
In this study, we examine the effects of the Part D coverage gap on adherence to diabetes medications. We focused on diabetes because diabetes is a serious chronic medical condition with considerable morbidity and mortality. The prevalence rate of diabetes is high among senior citizens (23.1%), compared with 7.8% in the US population.5 In 2007, over half of the $174 billion economic cost of diabetes was attributed to people 65 years and older, and a large portion of these costs were borne by Medicare.6
Previous studies on the Part D coverage gap generally used the number of prescriptions filled as the outcome variable.3,4 It is difficult to link the number of prescriptions with important outcomes such as hospitalizations and glycosylated hemoglobin (A1C) levels to help decision making in Part D benefit design. Our research focused on adherence to diabetes medications, a critical component in the management of diabetes. Aggressive glycemic management, as measured by A1C, can reduce long-term complications7 and medical costs associ-ated with diabetes treatment.8,9 Poor adherence to diabetes medications is a major contributor to unsatisfactory glycemic control10-13 and increases the risk of adverse health events associated with diabetes.14-16 Inadequate adherence to diabetes medications in the donut hole could have serious long-term consequences for both Medicare and Medicare beneficiaries.
Data Source and Sample Selection
This study is a retrospective cohort analysis based on prescription drug claims data provided by MedImpact Healthcare Systems, Inc, a national pharmacy benefit management company. Compared with the national data, Part D beneficiaries in MedImpact had the same sex distribution but were less likely to be under age 65 years ().
The sample in this research included patients over age 65 years who were continuously enrolled in a Medicare Part D plan from January 1, 2007, to December 31, 2008, and had at least 2 diabetes medication claims in each year. This study required patients to use diabetes medications in 2007 to ensure that all patients had used diabetes medications prior to the study period, January 1, 2008, to December 31, 2008. Utilization data in 2007 were used to acquire comorbidity information. Patients who received low-income subsidies for their premium and copayment were excluded from the sample because of their unique benefit structure.
A total of 35,426 patients were identified using these criteria. Among these patients, 37.10% reached the donut hole in 2008 but did not reach the catastrophic limit, 13.38% reached both the donut hole and the catastrophic limit, 49.44% never reached the donut hole, and the remaining 0.09% reached the catastrophic limit directly without entering the donut hole. We selected only patients who had reached the donut hole but not the catastrophic limit for this research.3,4 The final sample contained 12,881 patients from more than 30 different health plans.
Coverage in Donut Hole
Part D patients can have different types of Part D plans, including the defined standard benefit (DSB) plan, the actuarially equivalent (AE) plan, the basic alternative (BA) plan, and the enhanced alternative (EA) plan. The AE and BA plans are equivalent to DSB plans in value and do not provide coverage in the donut hole. The EA plans charge a supplemental premium and provide supplemental benefits. Many EA plans cover either generic drugs only or both generic and brand-name drugs as supplemental benefits.
In our analysis, we designated 3 groups of Part D beneficiaries based on their coverage in the donut hole. The first group consisted of beneficiaries with no donut hole coverage (“no coverage” group). This group included beneficiaries with DSB, AE, or BA plans and beneficiaries with EA plans that did not offer coverage in the donut hole. The second group comprised beneficiaries with only generic drug coverage in the donut hole (“generic coverage only” group) from EA plans. The third group included beneficiaries with both generic and brand-name drug coverage in the donut hole (“full coverage” group) from EA plans. All Part D plans identified in this research had $2510 and $5726.25 for the start and end of the donut hole.
The outcome measure of interest was adherence to diabetes medications. In this analysis, adherence to diabetes medications was measured by the proportion of days covered (PDC), calculated as the number of days covered by at least 1 diabetes medication in a certain period of time, divided by the number of days in that period. To calculate a PDC, each day in that period was evaluated as covered or as not covered by a medication based on fill date and days of supply. If a new claim for the same diabetes medication occurred before the end date of the last claim, the starting date of the new claim was pushed forward to the end date of the last claim. We calculated 2 PDCs for each patient: one for the period before reaching the coverage gap (pregap PDC) and the other for the period after reaching the coverage gap (postgap PDC). Patients with a PDC of 80% or more were classified as adherent.10,14-18 In addition to a dichotomous adherence variable, we used continuous PDC as the outcome variable to test the robustness of the results.
Control variables include age, sex, whether patient used insulin in the prior 12 months, geographic region (Northeast, South, Midwest, and West), and comorbidities in the prior 12 months. Comorbidity indicators were constructed based on the RxRisk model of comorbid conditions.19 A series of dummy variables for each major comorbid condition were
created for the analysis.
We used a difference-in-difference regression analysis to measure the effect of coverage gap on adherence to diabetes medications. The difference-in-difference method used patients with full coverage in the donut hole as the control group because these patients experienced little copayment change in the donut hole. The experimental group consisted of patients in the other 2 groups (no coverage and generic coverage only), who experienced substantial copayment increases in the donut hole.
Two challenges in measuring the effect of the donut hole are that (1) patients may self-select different levels of coverage and (2) they may have different propensities for being adherent to diabetes medications. The difference-in-difference method addresses this issue by having repeated observations for the same patient in both the pregap period and postgap periods. A group fixed effect was used to control for the permanent differences across different groups, and a time fixed effect was used to control for the secular trend within the study period. The true donut hole effect is the difference in adherence to diabetes medications between the control and the experimental groups in the donut hole, adjusting for the difference between the control and the experimental groups before entering the donut hole.
presents the patients’ characteristics. The average age for the entire sample was 74.97 years, and 48.06% of patients in the sample were male. A typical patient spent around 104 days in the donut hole. The vast majority of patients in the sample lived in the Midwest (82.85%). The most significant comorbidities were hypertension (87.41%), hyperlipidemia (81.54%), and heart disease/hypertension (69.47%).
About a third (33.47%) of patients used insulin in the prior 12 months. Among patients who used insulin in 2007, the percentages using short-acting, intermediate-acting, long-acting, and mixed insulin were 32.27%, 25.68%, 55.56%, and 27.67%, respectively. These categories are not mutually exclusive because a patient could have used more than 1 type of insulin in 2007.
In the sample, 19.63% of the 12,881 beneficiaries had no coverage in the donut hole, 20.08% of beneficiaries were covered for generic drugs only, and 54.28% of beneficiaries had full coverage. Patients with no coverage in the donut hole were slightly older (75.40 years) than patients with generic coverage only (74.85 years) and patients with full coverage (74.87 years). Patients with generic coverage only had the longest stay in the donut hole (107.96 days), and patients with no coverage in the donut hole spent the shortest time in the donut hole (99.91 days); patients with full coverage fell in between (103.54 days).
shows the adherence rates and average copayments for patients in the sample. The adherence rate before reaching the donut hole for the no-coverage group was 0.82, which was close to the adherence rates of 0.81 for the group with generic coverage only and 0.85 for the group with full coverage. After reaching the donut hole, the adherence rate of patients with no coverage dropped 8 percentage points to 0.74. Similarly, the adherence rate of patients with generic coverage dropped 6 percentage points to 0.75. In contrast, patients with full coverage maintained their adherence rate at 0.85 in the donut hole.
The coverage gap had a substantial impact on copayment. The average copayment (per 30 days of supply, the same thereafter) of all diabetes medications before reaching the donut hole was $18.63 for patients with no coverage, $18.63 for patients with generic coverage only, and $16.48 for patients with full coverage. After reaching the donut hole, the average copayment increased to $69.20 (a 271% increase) for the group with no coverage and to $67.18 (a 261% increase) for the group with generic coverage only. Partially because of the increased use of 90 days of supply, the average copayment for patients with full drug coverage actually decreased slightly to $14.33 after reaching the donut hole.
Table 3 breaks down the change in average copayment by generic and brand-name diabetes medications. For patients with no coverage, the average copayment for generic drugs almost doubled after reaching the donut hole ($12.13 vs $6.42). In contrast, the average copayment for generic drugs did not change for patients with generic coverage only ($6.64
vs $6.52) and even decreased from $6.37 to $4.63 for patients with full coverage. The average copayment for brand-name drugs quadrupled after reaching the donut hole for patients with no coverage ($126.05 vs $31.79) and patients with generic coverage only ($133.34 vs $32.47).
presents the changes in the market share by diabetes medication classes before and in the donut hole. Market share was defined as the sum of days of supply in each class of diabetes medications divided by the sum of days of supply for all diabetes medications. Overall, the market shares of different diabetes medications changed little before and in the donut hole. There was no evidence that patients with no brand-name drug coverage tried to increase the use of genericdrugs in response to increases in copayment for brand-name drugs in the donut hole. In fact, the market share of insulin and dipeptidyl peptidase-4 inhibitors, 2 classes of medications dominated by brand-name medications, increased in the donut hole for patients in all 3 groups.
Difference-in-Difference Regression Analysis
We conducted a multivariable logistic regression to evaluate the probability of adherence to diabetes medications, defined by a dummy PDC greater than or equal to 0.8. indicates that age, sex, and use of insulin in the prior 12 months all contributed to adherence to diabetes medications. Men were more likely to adhere than women (odds ratio [OR] = 1.122, P = .001). Compared with patients aged 65 to 74 years, those aged 75 to 84 years and aged ≥85 years were less likely to be adherent (OR = 0.838, P <.0001 for age 75-84; OR = 0.747, P <.0001 for age ≥85). Patients using insulin in 2007 were less likely to be adherent (OR = 0.698, P <.0001). Of the 4 regions, patients in the South were less likely to be adherent and patients in the West were more likely to be adherent compared with patients in the Midwest. The effects of most comorbidities were negative, indicating that patients’ overall sickness and the number of comorbidities might negatively influence adherence to diabetes medications.
The regression model indicates that patients with no coverage in the donut hole and patients with only generic coverage in the donut hole had a 38% reduction (OR = 0.617, P <.0001, 95% confidence interval [CI] = 0.523, 0.728) and 30% reduction (OR = 0.702, P <.0001, 95% CI = 0.604, 0.816) in the odds of being adherent after reaching the donut hole, respectively, compared with patients with full coverage (Table 5). The difference between the effect of no coverage and the effect of having generic drug coverage was not statistically significant (P = .1586).
We performed a robustness analysis by using continuous PDC as the outcome variable. When the outcome variable was continuous PDC, patients with no coverage had a 9.8 percentage point drop in PDC (P <.0001, 95% CI = -11.5, -8.1), whereas patients with generic coverage had only an 8.6 percentage point drop in PDC (P <.0001, 95% CI = -10.2, -7.0) compared with patients with full coverage. The difference between no coverage and generic coverage only also was statistically insignificant (P = .2449).
This research shows that the donut hole had a large impact on patients’ copayments. After reaching the donut hole, the average copayment for diabetes medications increased substantially for patients with no coverage and patients with generic drug coverage only. Corresponding to the copayment increase, adherence to diabetes medications in the donut hole dropped substantially for patients in both groups.
The decrease in adherence to diabetes medications should be of concern for policy makers and healthcare providers. Diabetes is a highly prevalent disease among senior citizens. In 2005, it was estimated that 3.40 million people aged 65 to 74 years and 2.88 million people 75 years and older in the United States had diabetes.20 By 2050 in the United States, the number of patients with diabetes aged 65 to 74 years is projected to increase by 220% to 10.88 million, and the number of patients with diabetes 75 years and older is projected to increase by 449% to 15.81 million.20 A decrease in adherence to diabetes medications could have serious long-term consequences for morbidity, mortality, and healthcare costs.5,6,12-16
This research also showed that offering only generic coverage in the donut hole may have a limited impact on improving patients’ adherence to diabetes medications. Compared with patients with no coverage, patients with generic drug coverage had higher adherence levels, but the difference was not statistically significant. The availability of brand-name drug coverage in the donut hole is important to maintain patients’ adherence to their diabetes medications. The reason may be that brand-name medications play a critical role in the treatment of diabetes. Currently, thiazolidinediones (TZDs), glucagon-like peptide-1 (GLP-1) agonists, dipeptidyl peptidase-4 inhibitors, and insulin are predominately brand-name medications, whereas generic substitutions are readily available for metformin and sulfonylureas. The current type 2 diabetes treatment algorithm from the American Diabetes Association and the European Association for the Study of Diabetes recommends metformin for first-line pharmacotherapy. If patients cannot lower their A1C to 7% or below within the first 2-3 months, then they need to upgrade to combination medications such as metformin + basal insulin, metformin + sulfonylurea, metformin + TZD, or metformin + GLP-1 agonist. If patients still cannot reach their A1C goal, metformin + intensive insulin should be considered.21 Because many patients with diabetes in Part D are senior citizens, most of them may have progressed beyond the first-line pharmacotherapy treatment and need brand-name medications including TZD, GLP-1 agonists, and insulin to maintain A1C control. Consequently, the lack of coverage for brand-name medications has a large impact on the treatment of diabetes.
This research has several limitations. First, this research is based on data from 1 pharmacy benefit management company and the findings may not be necessarily generalized to other populations. Second, this analysis relies exclusively on pharmacy claim data. Although pharmacy claim data are useful to examine medication adherence, we cannot examine the potential impact of coverage gap on other important clinical outcomes such as A1C level and diabetes-related hospitalizations. In addition, we cannot differentiate type 1 and type 2 diabetes because of the lack of medical claim data. Data on duration of diabetes also are not available. Third, adherence to diabetes medication is only part of diabetes management. The impact of the donut hole on other aspects of diabetes management such as self-management is unclear. Further analysis is needed to understand the effect of coverage gap on hospitalization risks and A1C control for Part D beneficiaries with diabetes.
This research found that Part D patients without coverage of brand-name medications experienced substantial increase in copayments for diabetes medications and substantial decrease in adherence to diabetes medication in the coverage gap. The coverage of brand-name medications was critical for patients to maintain adherence in the donut hole. The findings in this research supported the recently passed healthcare reform on Part D coverage gap.22 Starting from 2011, Part D will provide a 50% discount on brand-name medication costs in the coverage gap. New legislations will provide additional discounts in the coverage gap each year through 2020, when the donut hole will be closed completely. In 2014, copayment for brand-name medications in the coverage gap will be further reduced until it reaches 25% in 2020. Copayment for generic medications in the coverage gap will also begin to gradually decrease by 7% in 2011 until it reaches 25% in 2020. These reform measures should help improve adherence for Part D patients with diabetes.
Author Affiliations: From School of Policy, Planning, and Development (QG), University of Southern California, Los Angeles, CA; MedImpact Healthcare Systems, Inc (FZ, BVP, LCT), San Diego, CA.
Funding Source: The authors report no external funding for this research.
Author Disclosures: The authors (QG, FZ, BVP, LCT) 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 (QG, FZ, BVP); acquisition of data (FZ, BVP); analysis and interpretation of data (QG, FZ); drafting of the manuscript (QG, FZ); critical revision of the manuscript for important intellectual content (QG, FZ, BVP, LCT); statistical analysis (QG, FZ); provision of study materials or patients (FZ); administrative, technical, or logistic support (LCT); and supervision (FZ, BVP, LCT).
Address correspondence to: Feng Zeng, PhD, MedImpact Healthcare Systems Inc, 10680 Treena St, San Diego, CA 92131. E-mail: firstname.lastname@example.org.
1. Joyce GF, Goldman DP, Karaca-Mandic P, Zheng Y. Pharmacy benefit caps and the chronically ill. Health Aff (Millwood). 2007;26(5):1333-1344.
2. Hsu J, Price M, Huang J, et al. Unintended consequences of caps on Medicare drug benefits. N Engl J Med. 2006;354(22):2349-2359.
3. Sun SX, Lee KY, Aruru M. Examining Part D beneficiaries' medication use in the doughnut hole. Am J Pharm Benefits. 2009;1(1):19-28.
4. Zhang Y, Donohue JM, Newhouse JP, Lave JR. The effects of the coverage gap on drug spending: a closer look at Medicare Part D. Health Aff (Millwood). 2009;28(2):w317-w325.
5. Centers for Disease Control and Prevention. National Diabetes Fact Sheet: General Information and National Estimates on Diabetes in the United States, 2007. http://www.cdc.gov/diabetes/pubs/pdf/ndfs_2007. pdf. Accessed October 7, 2009.
6. American Diabetes Association. Economic costs of diabetes in the U.S. in 2007 [published correction appears in Diabetes Care. 2008; 31(6):1271]. Diabetes Care. 2008;31(3):596-615.
7. UK Prospective Diabetes Study (UKPDS) Group. Intensive blood-glucose control with sulphonylureas or insulin compared with conventional treatment and risk of complications in patients with type 2 diabetes (UKPDS 33) [published correction appears in Lancet. 1999;354(9178):602]. Lancet. 1998;352(9131):837-853.
8. Stephens JM, Botteman MF, Hay JW. Economic impact of antidiabetic medications and glycemic control on managed care organizations: a review of literature. J Manag Care Pharm. 2006;12(2):130-142.
9. Shetty S, Secnik K, Oglesby AK. Relationship of glycemic control to total diabetes-related costs for managed care health plan members with type 2 diabetes. J Manag Care Pharm. 2005;11(7):559-564.
10. Pladevall M, Williams LK, Potts LA, Divine G, Xi H, Lafata JE. Clinical outcomes and adherence to medications measured by claims data in patients with diabetes. Diabetes Care. 2004;27(12):2800-2805.
11. Krapek K, King K, Warren SS, et al. Medication adherence and associated hemoglobin A1c in type 2 diabetes. Ann Pharmacother. 2004; 38(9):1357-1362.
12. Schectman JM, Nadkarni MM, Voss JD. The association between diabetes metabolic control and drug adherence in an indigent population. Diabetes Care. 2002;25(6):1015-1021.
13. Dalewitz J, Khan N, Hershey CO. Barriers to control blood glucose in diabetes mellitus. Am J Med Qual. 2000;15(1):16-25.
14. Hepke KL, Martus MT, Share DA. Costs and utilization associated with pharmaceutical adherence in a diabetic population. Am J Manag Care. 2004;10(2 pt 2):144-151.
15. 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(17):1836-1841.
16. Lau DT, Nau DP. Oral antihyperglycemic medication nonadherence and subsequent hospitalization among individuals with type 2 diabetes. Diabetes Care. 2004;27(9):2149-2153.
17. Cramer JA. A systematic review of adherence with medications for diabetes. Diabetes Care. 2004;27(5):1218-1224.
18. Odegard PS, Capoccia K. Medication taking and diabetes: a systematic review of the literature. Diabetes Educ. 2007;33(6):1014-1029.
19. Fishman PA, Goodman MJ, Hornbrook MC, Meenan RT, Bachman DJ, O'Keeffe Rosetti MC. Risk adjustment using automated ambulatory pharmacy data: the RxRisk model. Med Care. 2003;41(1):84-99.
20. Narayan KM, Boyle JP, Geiss LS, Saaddine JB, Thompson TJ. Impact of recent increase in incidence on future diabetes burden: U.S., 2005- 2050. Diabetes Care. 2006;29(9):2114-2116.
21. DM, Holman RR, Buse JB, et al; American Diabetes Association; European Association for Study of Diabetes. Medical management of hyperglycemia in type 2 diabetes: a consensus algorithm for the initiation and adjustment of therapy: a consensus statement of the American Diabetes Association and the European Association for the Study of Diabetes. Diabetes Care. 2009;32(1):193-203.
22. Patient Protection and Affordable Care Act. HR 3590. Pub L No. 111-148. 124 Stat 119.