Persistence With Biologic Therapies in the Medicare Coverage Gap

, , , , , ,
The American Journal of Managed Care, November 2011, Volume 17, Issue 11

For patients who reached the Medicare Part D coverage gap, discontinuation was more likely for patients taking osteoporosis medication.


To describe persistence with teriparatide and other biologic therapies in Medicare Part D plans with and without a coverage gap.

Study Design:

Retrospective (2006) cohort study of Medicare Part D prescription drug plan beneficiaries from a large benefits company. Two plans with a coverage gap (defined as “basic”) were combined and compared with a single plan with coverage for generic and branded medications (defined as “complete”).


Patients taking alendronate (nonbiologic comparator), teriparatide, etanercept, adalimumab, interferon β-1a, or glatiramer acetate were selected for the study. For patients with complete coverage, equivalent financial thresholds were used to define the “gap.” The definition of discontinuation was failure to fill the index prescription after reaching the gap.


For alendronate, 27% of 133,260 patients had enrolled in the complete plan. Patients taking biologic therapies had more commonly enrolled in complete plans: teriparatide (66% of 6221), etanercept (58% of 1469), adalimumab (52% of 824), interferon β-1a (60% of 438), and glatiramer acetate (53% of 393). For patients taking either alendronate or teriparatide, discontinuation rates were higher in the basic, versus complete, plan (adjusted odds ratios, 2.02 and 3.56, respectively). Discontinuation did not significantly vary by plan type for etanercept, adalimumab, interferon β-1a, or glatiramer acetate.


For patients who reached the coverage gap, discontinuation was more likely for patients taking osteoporosis (OP) medication. Not having a coverage gap was associated with improved persistence with OP treatment.

(Am J Manag Care. 2011;17(11):753-759)

For patients who reached the coverage gap, discontinuation was more likely for patients taking osteoporosis medication:

  • Not having a coverage gap was associated with improved persistence with osteoporosis treatment.

  • Patients being treated for the symptomatic diseases of rheumatoid arthritis and multiple sclerosis had no significant increase in discontinuation of therapy in the coverage gap.

  • Health policy makers should be aware that a sudden increase in patient cost-sharing was associated with medication discontinuation for teriparatide.

Medicare Part D began in January of 2006 with implementation of the Medicare Prescription Drug Improvement and Modernization Act.1 The standard benefit has an initial deductible followed by coinsurance up to a threshold. Those exceeding this threshold enter the “donut hole,” or coverage gap, where they are responsible for all medication costs until the catastrophic threshold is reached, beyond which the patient pays a reduced level of coinsurance. In 2006, 4% of plans provided coverage for branded drugs without a gap,2 but such plans ceased to exist after 2006.2,3

The cost-sharing features of Part D have been associated with reduced adherence.4-8 Characteristics such as costly medications, a greater number of prescriptions, or higher total monthly drug expense have been associated with cost-coping behaviors.5,7 Patients with serious chronic illness such as osteoporosis (OP), rheumatoid arthritis (RA), and multiple sclerosis (MS) who are treated with biologic therapies, which can be expensive relative to other medications, might be at particularly high risk for discontinuation in the gap. Our primary objective was, for a given medication, to compare persistence in Part D plans. At Humana, a large benefits company, plans were offered in 2006 either with or without a coverage gap, which allowed us the unique opportunity to compare persistence between these 2 plan designs.


Study Design and Patient Cohorts

This was a retrospective cohort study of beneficiaries enrolled in a Medicare Part D Prescription Drug Plan (PDP) through Humana. The institutional review board at the University of Miami (Florida) approved the study protocol. Two databases were linked to conduct the analysis. The first file contained enrollment and demographic information: age, gender, type of insurance, and geographic region for each beneficiary. The second file contained pharmaceutical claims data (frequency and cost) for each beneficiary.

Study cohorts were defined by medication usage, characteristics of which are defined in Table 1. Patients were required to have filled at least 2 prescriptions of 1 of the study medications between January 1, 2006, and December 31, 2006. The medications were chosen to be representative of treatment for OP (alendronate, teriparatide), RA (etanercept, adalimumab), or MS (interferon β-1a, glatiramer acetate). Eli Lilly, the manufacturer of teriparatide, was interested in assessing adherence to teriparatide during the Medicare coverage gap in a real-world setting. Alendronate was selected as a nonbiologic comparator to teriparatide to evaluate adherence during the coverage gap for an OP medication with oral administration and potentially lower out-of-pocket (OOP) expenses. The other medications are injectable biologics used for other disease states, but like teriparatide, could be associated with high OOP expenses during the coverage gap, and therefore were also included. The within-medication, between-plan comparisons of persistence were of primary interest. Comparisons of persistence between different medications were only descriptive and should be considered exploratory.

Definition of the Medicare Plans

In 2006, Humana offered 3 stand-alone Medicare PDPs. The 2 plans with a coverage gap differed somewhat in their plan designs (eg, one had higher premiums and a lower patient cost per filled prescription) but were comparable on the characteristic of primary interest (having a gap). They were pooled in the analysis and are referred to as the “basic” plan. The plan without a gap in coverage for both generic and branded medications is described as the “complete” plan. Therefore, for each medication, 1 cohort with and 1 cohort without a gap in coverage were available for analysis.

In 2006, the mandated Part D plan1 included cost-sharing or actuarially equivalent benefits according to the following basic format for individuals not receiving a low-income subsidy. Patients eligible for Medicaid or a lowincome subsidy were excluded because they did not experience a gap in coverage regardless of plan selection.

Deductible: Patient paid 100% up to $250

Coinsurance: Patient paid 25% until an initial coverage limit of $750 in total patient OOP spending ($2250 total drug costs)

Gap: No coverage between patient OOP costs of $750 ($2250 total drug costs) and $3600 ($5100 total drug costs)

Catastrophic: Patient paid 5% of drug cost after OOP spending exceeded $3600 ($5100 total drug costs)

Definition of the Outcome Variable (Persistence)

The date of a member’s first recorded claim for the index medication was considered the index date. The number of days spent in each phase of Part D was tallied, beginning with the index date. The financial thresholds from the basic plan were assigned to the complete plan to create a “gap equivalent” for comparison purposes only. For example, $2250 was defined as the beginning of the “gap” phase for both the basic and complete patients even though the complete patients did not experience an actual gap in coverage. Patients were excluded if they initiated the index medication after reaching the threshold due to the cost of medications other than the index medication.

For each prescription, we used the fill date and the number of days supplied to calculate persistence. Discontinuation was defined as the absence of any index medication fills after entering the gap. No maximum refill time window was required. To be considered eligible for discontinuation, patients had to reach the coverage gap or equivalent threshold and have filled at least 1 prescription for the index medication prior to doing so.

Evaluation of Possible Confounders

Patient demographics included age, gender, chronic disease score (CDS), and pill burden. The CDS is a validated aggregate comorbidity measure in which disease severity and complexity are determined based on use of specific drug classes.9,10 Pill burden was quantified as the number of unique medications (identified by 10-digit Generic Product Identifiers) filled during the investigation period. Possible unmeasured confounders are discussed as limitations.

Statistical Analysis

The baseline characteristics of basic and complete members were summarized by mean and standard deviation (SD) for age, percentages for gender, and median with interquartile ranges (IQR) for the CDS and pill burden. The number of patients beginning each phase, OOP cost during each phase, total days spent in each phase, and persistence were summarized by plan type for each drug. Multivariable logistic regression models included lack of persistence (discontinuation after reaching the gap) as the outcome variable. Coverage during the gap (complete vs basic plan) was the independent variable of interest. The odds ratios were adjusted for the following covariates: age, gender, CDS, pill burden, and calendar quarter of enrollment. All analyses were performed using SAS 8.0 (SAS Institute Inc, Cary, North Carolina). Significance tests were 2-sided at an alpha level of .05 unless otherwise noted. Two-sided P values based on 2 independent sample t tests were used for the comparison of the mean age between plans. Whenever the assumption of equality of variances was violated, the Satterthwaite method was used. Fisher exact test was used to compare gender and discontinuation between plans. Wilcoxon-Mann-Whitney test was used to compare the median of CDS and unique medications between plans. A 2-sided P value of .017 (Bonferroni-adjusted for multiple comparisons) was used to compare the median of number of days and OOP cost in each phase between plans (Wilcoxon-Mann-Whitney test).


Baseline Characteristics

Demographic characteristics of the 142,605 patients included in the study are shown in Table 2. The mean ages of the OP, RA, and MS cohorts were approximately 56, 67, and 77 years, respectively. The vast majority of patients in every cohort was female. The RA cohort had the highest median CDS scores. Complete plan beneficiaries had higher CDS scores and more concomitant medication usage than basic plan beneficiaries. In the alendronate cohort, beneficiaries enrolled in the basic plan outnumbered those in the complete plan. In contrast, the reverse was true for the 5 biological therapies.

Costs and Time in the Medicare Plan

The number of patients filling medication before, during, and after the gap is shown in Table 3 along with the median number of days spent in each phase and corresponding OOP cost. With the complete plan, the financially equivalent thresholds are used to define the “gap” for comparison purposes. The median number of days before the coverage gap is significantly lower for individuals enrolled in complete plans for each medication cohort. Complete, versus basic, plan beneficiaries taking either of the 2 OP medications had more days in the coverage gap. Alendronate-treated patients spent less time in the catastrophic phase if they were enrolled in complete plans. Complete beneficiaries taking either of the OP medications had significantly higher OOP costs before reaching the gap. For each medication cohort, patients in the complete plan had significantly lower OOP costs in the gap.

Effect of the Coverage Gap on Discontinuation

Unadjusted rates of discontinuation while in the gap are shown in Table 4. Among patients taking RA or MS medications,less than 10% discontinued their index medication after reaching the gap, regardless of plan type. Among patients taking OP medications, discontinuation rates ranged from 8% to 35% depending on plan type and medication cohort. For patients in basic plans, discontinuation was substantially higher than in the complete plan for patients taking OP medications. After adjusting for covariates, patients taking alendronate or teriparatide in the basic plan were twice as likely to discontinue as the controls in the complete plan (Figure).


The Medicare Part D coverage gap was associated with discontinuation among patients being treated with alendronate and especially teriparatide, a biologic OP therapy. For patients taking biologic therapy for RA or MS, no significant discontinuation was observed, but the smaller samples limited our ability to draw definitive conclusions for these cohorts. Overall, our finding that financial issues are associated with discontinuation but that nonfinancial factors also likely play an important role is consistent with previous reports.11,12

The majority of patients receiving biologic therapy had preferentially enrolled in the complete plan, which had a higher premium to help finance its enhanced benefits. The enrollment of these patients taking high-cost biologics suggests these patients were knowledgeable about their plan selection. However, any difference between the complete and basic plan cohorts beyond pharmacy utilization and demographics was unmeasured and is a study limitation. We included 6 different treatments across 3 diseases to minimize the likelihood of a single spurious result. The overwhelming majority (97%) of Medicare patients who enrolled in a stand-alone PDP in 2006 elected a plan that did not offer coverage of branded drugs in the gap,13 but for patients like those in this study (patients taking biologics or with an anticipated higher pill burden), the choice of a complete plan appears logical.

A detailed 2006 analysis of Part D plans reported that 94% of PDPs and71% of Medicare Advantage with Prescription Drug (MA-PD) plans lacked coverage in the gap.13 The remaining plans offered coverage for generic (3% of PDPs and 22% of MA-PDs) and/or branded (3% of PDPs and 6% of MAPDs) medications while in the gap. Of those patients enrolled in a complete plan, 52% were enrolled in Humana. Branded and generic coverage in the gap applied to just 3% of PDP enrollees overall but accounted for 12% of Humana’s PDP enrollees.2 By 2008, no PDPs and only about 1% of MA-PDs offered full brand and generic coverage in the gap.13 In our analysis, most patients taking biologic therapy reached the coverage gap, and many, particularly those with RA and MS, exceeded the catastrophic threshold. With monthly costs that exceed $1300 for a single medication, patients taking RA or MS therapy reach catastrophic spending levels quite quickly. In comparison, only 18% of the alendronate basic cohort and 25% of patients in a generalized Medicare population without gap coverage reached the donut hole.2 Patients treated with expensive biologic therapies clearly have special monetary challenges.

For patients taking the OP drugs, those in the basic, versus complete, plan were much more likely to discontinue. However, even though their OOP cost would have exceeded $600 per refill, 75% of teriparatide patients in the basic plan persisted with treatment while in the gap. The unadjusted discontinuation rates seen after reaching the gap for the OP basic cohorts (25% for teriparatide and 35% for alendronate) were somewhat higher than the 18% rate previously reportedfor OP medications.13 In prior work, we found that discontinuation was more common among teriparatide users than among beneficiaries with other chronic conditions with claims for hypertension and cholesterol medications.14 Another study across multiple disease states found that Medicare patients who reached the gap reduced their drug use by 17%.6

Reaching the gap without coverage was not associated with significantly increased discontinuation in patients with RA or MS. It should be noted that there were substantially fewer events (discontinuations) for the RA (<34) and MS (<16) cohorts, thereby limiting power to detect a difference between the complete and basic cohorts. Also, these patients were in the coverage gap an average of only 2 months, so the OOP costs would have been concentrated into this relatively short time frame. They would have spent more than $1300 out of pocket per prescription for the RA or MS medication in addition to the costs of any other concomitant medications until they reached catastrophic coverage. Despite higher OOP costs for the patients taking RA or MS medications, their rates of discontinuation were generally lower than those of the teriparatide- or alendronate-treated patients.

The differences between patients with RA or MS and patients with OP might be explained by the latter patients experiencing fewer physical symptoms than those with RA or MS. However, Briesacher et al showed that low adherence was observed for both OP (with alendronate) and the physically symptomatic disease gout (with allopurinol).15 It is worth noting that allopurinol is not associated with immediate relief of gout symptoms. Interestingly, other studies have found that copaymentincreases in medications that treat symptoms, versus the underlying disease, were associated with larger reductions in medication utilization.16 In a study before the implementation of Part D, specialty medications for RA and MS were found to be largely insensitive to cost sharing.17 Another possible explanation is that while MS and RA are managed almost exclusively by neurologists and rheumatologists, respectively, OP is managed by a more heterogeneous group of generalists and specialists whose overall approach to OP might be less aggressive.18 Also, patients with OP were older than their counterparts in the RA and MS cohorts, and factors such as lower income or a bias toward less aggressive care in older adults might partially explain our findings.19 In 2007, the median yearly income of individuals 75 years and older was two-thirds that of individuals 65 to 74 years of age.20 In a 2006 survey, roughly 5% of seniors and disabled individuals reported going without basic necessities in order to pay for medication while in the gap.5 Clearly, difficult choices are being faced by patients who reach the gap and incur large OOP expenses. Some appear to be making substantial sacrifices in order to continue treatment.

An important limitation is that the RA and MS cohorts were relatively small, which reduced our power, as noted above. Also, our study was based only on pharmacy claims data, with no associated medical records that would have allowed us to compare healthcare provider visits and hospitalizations. The true indication for which a medication was prescribed was not available to us, nor was information regarding the clinical outcomes of patients who discontinued after reaching the gap. Furthermore, we did not know whether a decision to discontinue was made by the clinician or patient, and whether the decision was based on financial concerns or other factors not associated with the coverage gap. For example, side effects might contribute to discontinuation independently of financial issues. Nonadherence is common with chronic medications, and background discontinuation was probably present. Furthermore, patients could have switched therapies. Our primary interest was discontinuation of the index medications, so we considered switching a relevant outcome, but our methodology simplified the actual situation. The complete plan served as an important control for the background discontinuation and switching limitations. An issue specific to teriparatide is that it is recommended for only 2 years of use per patient.21 Some patients could have completed 2 years of teriparatide while in the gap and discontinuation was a medical, rather than financial, decision. However, we expect that patients in the complete, versus basic, plan would have been at least as likely to reach the 2-year maximum in 2006. We hypothesize that in late 2005 or early 2006, teriparatide users would have been more likely to choose the complete plan to gain coverage within the gap.

Any patient in the study sample who discontinued filling prescriptions after reaching the gap might have continued therapy by using clinician-distributed samples or prescription assistance programs (PAPs), which could lead us to overestimate the actual discontinuation rates.22 While we cannot be sure if use of samples or participation in PAPs would differ by plan type, given the greater OOP burden in the gap for patients in the basic plan, we hypothesize that participation would be greater among basic plan enrollees. One possible refill pattern that we were unable to observe is restarting medications in 2007 after discontinuing in 2006, but we have found this to be uncommon.23 Our study was based exclusively on 2006 claims because the complete plan did not provide brand coverage during the gap in subsequent years. In the first year, 2006, that Medicare Part D was offered, patients could enroll without penalty through May.1 While a patient enrolling later in the year would have less time available to reach the gap, we controlled for a quarter of enrollment in the regression models and the patients typically reached the gap relatively quickly after initiating therapy. Even though a randomized study is probably not feasible, our observational study design limits our ability to conclude that the gap caused the discontinuation. While the recently enacted Patient Protection and Affordable Care Act24 includes provisions for discounts and then gradually eliminates the gap over the next decade, cost sharing might still be substantial. Even the standard 25% coinsurance rate can lead to high patient expenses with costly medications. Health policy makers need to understand the medication refill patterns during the original design and what the implications of more complete coverage are on utilization, patient care, and budgetary impact.


Appreciation is expressed to Tamara Ball, MD, for editorial assistance. Dr Ball is a scientific writer employed full-time by i3 Statprobe, a division of Ingenix, which is a subsidiary of UnitedHealth Group.

Author Affiliations: From Miller School of Medicine, University of Miami, Miami, FL (LT); Competitive Health Analytics (CLU), Humana Health Services Research Center (JL, JWH), Humana Inc, Miramar, FL; Global Health Outcomes (DEB, ESM), Lilly USA (KK), Lilly Corporate Center, Indianapolis, IN.

Funding Source: This study was funded by Eli Lilly and Company.

Author Disclosures: Drs Meadows, Ball, and Krohn report employment and stock ownership with Eli Lilly, manufacturer of teriparatide, one of the prescription medications included in this study. Dr Uribe reports employment with Competitive Health Analytics, Humana Inc, which was contracted by Eli Lilly for this study. The other authors (LT, JL, JWH) 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 (ESM, LT, CLU, JL, JWH, DEB, KK); acquisition of data (LT, JL, JWH, KK); analysis and interpretation of data (ESM, LT, CLU, JL, JWH, DEB, KK); drafting of the manuscript (ESM, JL, CLU, JWH, KK); critical revision of the manuscript for important intellectual content (ESM, LT, CLU, JWH, DEB, KK); statistical analysis (LT, JL); provision of study materials or patients (ESM); obtaining funding (LT, JWH); administrative, technical, or logistic support (JL, CLU); and supervision (ESM, CLU).

Address correspondence to: Eric S. Meadows, PhD, MedMining, a Geisinger Health System Business, 285 Mill St, Danville, PA 17822-4012. E-mail:

1. US Congress. The Medicare Prescription Drug Improvement and Modernization Act of 2003. Public Law 108-173. Published 2003.

2. Cubanski J, Neuman P. Status report on Medicare Part D enrollment in 2006: analysis of plan-specific market share and coverage. Health Aff (Millwood). 2007;26(1):w1-w12.

3. Hoadley J, Thompson J, Hargrave E, Merrell K, Cubanski J, Neuman T. The Henry J. Kaiser Family Foundation. Medicare Part D 2008 Data Spotlight: Premiums. Published November 2007. Accessed March 9, 2010.

4. Reed M, Brand R, Newhouse JP, Selby JV, Hsu J. Coping with prescription drug cost sharing: knowledge, adherence, and financial burden. Health Serv Res. 2008;43(2):785-797.

5. 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.

6. 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.

7. Tseng CW, Brook RH, Keeler E, Steers WN, Mangione CM. Cost-lowering strategies used by Medicare beneficiaries who exceed drug benefitcaps and have a gap in drug coverage. JAMA. 2004;292(8):952-960.

8. Neuman P, Strollo MK, Guterman S, et al. Medicare prescription drug benefit progress report: findings from a 2006 national survey of seniors. Health Aff (Millwood). 2007;26(5):w630-w643.

9. Von Korff M, Wagner EH, Saunders K. A chronic disease score from automated pharmacy data. J Clin Epidemiol. 1992;45(2):197-203.

10. Clark DO, Von Korff M, Saunders K, Baluch WM, Simon GE. A chronic disease score with empirically derived weights. Med Care. 1995;33(8):783-795.

11. Piette JD, Heisler M, Wagner TH. Medication characteristics beyond cost alone influence decisions to underuse pharmacotherapy in response to financial pressures. J Clin Epidemiol. 2006;59(7):739-746.

12. Briesacher BA, Gurwitz JH, Soumerai SB. Patients at-risk for costrelated medication nonadherence: a review of the literature. J Gen Intern Med. 2007;22(6):864-871.

13. Hoadley J, Thompson J, Hargrave E, Merrell K, Cubanski J, Neuman T. The Henry J. Kaiser Family Foundation. Medicare Part D 2008 Data Spotlight: The Coverage Gap. pdf. Published November 2007. Accessed September 30, 2009.

14. Conwell LJ, Esposito D, Garavaglia S, et al. Out-of-pocket drug costs and drug utilization patterns of postmenopausal Medicare beneficiaries with osteoporosis [published online ahead of print May 26, 2011]. Am J Geriatr Pharmacother. In press.

15. Briesacher BA, Andrade SE, Fouayzi H, Chan KA. Comparison of drug adherence rates among patients with seven different medical conditions. Pharmacotherapy. 2008:28(4):437-443.

16. Goldman DP, Joyce GF, Escarce JJ, et al. Pharmacy benefits and the use of drugs by the chronically ill. JAMA. 2004;291(19):2344-2350.

17. Goldman DP, Joyce GF, Lawless G, Crown WH, Willey V. Benefit design and specialty drug use. Health Aff (Millwood). 2006;25(5): 1319-1331.

18. Morris CA, Cheng H, Cagral D, Solomon DH. Predictors of screening and treatment of osteoporosis: a structured review of the literature. Endocrinologist. 2004;14(2):70-75.

19. Robb C, Hongbin C, Haley WE. Ageism in mental health and health care: a critical review. J Clin Geropsychol. 2002;8(1):1-12.

20. US Census Bureau. Current Population Survey, 2007 Annual Social and Economic (ASEC) Supplement. Washington, DC: US Census Bureau; March 2007. pdf. Accessed March 9, 2010.

21. Forteo [package insert]. Indianapolis, IN: Eli Lilly and Company; 2009. Accessed March 9, 2010.

22. Tjia J, Briesacher BA, Soumerai SB, et al. Medicare beneficiaries and free prescription drug samples: a national survey. J Gen Intern Med. 2008;23(6):709-714.

23. Colby M, Esposito D, Goldfarb S, et al. Medication discontinuation and reinitiation among Part D beneficiaries taking costly medications. Am J Pharm Benefits. In press.

24. The Patient Protection and Affordable Care Act. http://www.gpo. gov/fdsys/pkg/PLAW-111publ148/pdf/PLAW-111publ148.pdf. Accessed July 26, 2011.