Impact of medicare part D on access to and cost sharing for specialty biologic medications for beneficiaries with rheumatoid arthritis

Abstract

Objective

Many worry that the use of specialty tiering for biologic disease-modifying antirheumatic drugs (DMARDs) by Medicare Part D plans imposes a heavy financial burden on beneficiaries with rheumatoid arthritis (RA). To date, no one has examined the cost-sharing structures for biologic DMARDs in Part D plans or the resulting cost burden for patients.

Methods

We followed 14,929 vulnerable, low-income patients with RA who were enrolled in the Medicare Replacement Drug Demonstration (MRDD) in 2005. As the MRDD population transitioned into Part D in 2006, we examined correlates of Part D enrollment and compared the cost-sharing provisions for biologic DMARDs in the Medicare Advantage and stand-alone plans. We simulated the out-of-pocket costs of beneficiaries under 3 cost-sharing scenarios.

Results

Eighty-one percent of MRDD beneficiaries with RA enrolled in Part D. Enrollment predictors were female sex (odds ratio [OR] 1.48, 95% confidence interval [95% CI] 1.32–1.67), prior MRDD benefit use (OR 2.29, 95% CI 2.04–2.58), other self-reported drug coverage (OR 1.53, 95% CI 1.36–1.71), and receiving an MRDD subsidy (OR 2.00, 95% CI 1.74–2.30). Compared with stand-alone plans, Medicare Advantage plans had lower deductibles, lower premiums, and fewer prior authorization, step therapy, and quantity limit restrictions. However, ∼75% of all plans used coinsurance as the preferred form of cost sharing. Out-of-pocket costs exceeded $4,000 annually in all cost-sharing scenarios.

Conclusion

Most MRDD beneficiaries with RA enrolled in Part D. Although plans assume some costs for biologic DMARDs, the majority of costs are shifted to beneficiaries and to Medicare. Such cost shifting may place these medications out of the beneficiary's financial reach and expose Medicare to high financial liability.

INTRODUCTION

Adalimumab, etanercept, infliximab, and other biologic disease-modifying antirheumatic drugs (DMARDs) are effective in reducing symptoms of and slowing disease progression in patients with rheumatoid arthritis (RA) (1–3). Compared with nonbiologic DMARDs, the biologic DMARDs have a quicker onset of action, require less regular laboratory monitoring, and are better tolerated (3). Biologic DMARDs typically cost 10–100 times more than the alternate nonbiologic DMARDs, placing them out of reach for many patients (4). Insured patients experience reduced financial burden, but plans differ greatly in their coverage of and cost sharing for biologic DMARDs, at times shifting a large portion of the cost to patients (5, 6).

Recognizing both the importance of biologic agents in treating RA and other diseases and patients' financial burdens, Congress created the Medicare Replacement Drug Demonstration (MRDD), a program that would provide temporary drug insurance until the start of Medicare Part D, the new voluntary federal drug insurance program. The MRDD began in September 2004 and ended December 31, 2005. The MRDD provided coverage for several medications and biologics that could replace those covered under Medicare Part B. For beneficiaries with RA, the demonstration covered 3 self-injected medications: adalimumab, etanercept, and anakinra. For the purposes of the MRDD, these medications would replace infliximab, which was already covered under Medicare Part B. Low-income vulnerable Medicare beneficiaries with select conditions, including RA, without comprehensive drug insurance coverage were targeted for MRDD enrollment. The MRDD was structured to have similar patient cost-sharing arrangements as the anticipated standard Medicare Part D benefit. Evaluation results showed that the MRDD reduced financial barriers, improved health outcomes, and reduced the need to make outpatient visits to receive drug infusions (7).

Medicare Part D began on January 1, 2006, and numerous insurance plan providers administer Part D. Stand-alone providers offer prescription coverage only. Medicare Advantage–based plans expanded their Medicare Part A and Medicare Part B managed-care plans to offer Part D coverage. Medicare Advantage–based plans may have different incentives than stand-alone plans in terms of coverage and benefit design, and may attract enrollees with different characteristics.

Although plans may offer differing benefit structures, each plan must be actuarially equivalent to Congress' standard design. Figure 1 depicts the 2006 standard benefit design and details cost-sharing responsibilities for plans, Medicare, and patients as patients move through 4 distinct coverage periods (8). Low-income beneficiaries were eligible for subsidies to reduce costs. Unlike in the MRDD, however, in Part D, a beneficiary's assets were included as income in eligibility testing for low-income subsidies. In 2006, each Part D plan was required to provide coverage for at least 2 medications in each approved drug category and class; to be competitive, most offered many more (9, 10). Plans could and often did vary in cost sharing, restrictions, and coverage for each medication in a particular class, often assigning different medications in the same class to different tiers of cost-sharing responsibility. Unlike in the MRDD, Part D plans could place high-cost items (>$500) such as biologic DMARDs in a specialty tier, where they are subject to higher patient cost sharing (9, 11).

Figure 1. 2006 Medicare Part D standard benefit design.

In this study, we highlight the experience of vulnerable, low-income patients with RA as they transitioned from the MRDD to Part D. Part D has been criticized as being difficult to understand, and we were concerned that the MRDD population might not enroll. We thought that fewer beneficiaries might receive low-income subsidies under Part D. To address these concerns, we examine the enrollment of MRDD beneficiaries in Part D plans by demographic and health utilization characteristics and assess these characteristics as predictors of Part D enrollment. We contrast the generosity of the benefit designs of the stand-alone versus Medicare Advantage Part D plans, hypothesizing that beneficiaries in Medicare Advantage Part D plans would experience fewer cost-sharing obligations but more treatment restrictions (e.g., increased prior authorization requirements) for biologic DMARDs than beneficiaries enrolled in stand-alone plans. We explore coverage of and restrictions for adalimumab and etanercept across the wide array of plans. Finally, we estimate beneficiaries' cost-sharing burden under 3 coverage scenarios to test the hypothesis that the addition of a specialty tier by Part D plans would shift substantial costs to patients and determine whether beneficiaries enrolled in a plan that covered the biologic DMARD they used during the MRDD.

In selecting a Part D plan, Medicare beneficiaries with RA face a dizzying number of choices: plans offer different benefit designs, different biologic DMARDs, and different cost-sharing mechanisms. This study describes the characteristics of beneficiaries with RA that are most likely to enroll in Part D, the cost-sharing obligations they face, and the extent of coverage for and restricted access to biologic DMARDs across different plan types. Our findings will assist both patients with RA and rheumatologists prescribing biologic DMARDs as they evaluate and then select a Part D plan that best meets the patient's clinical and financial needs.

PATIENTS AND METHODS

Study sample.

A report describing the MRDD source population has been previously published (12). Briefly, patients were eligible for the MRDD if Medicare was their primary source of insurance, if they lived in one of the 50 US states or the District of Columbia and were not enrolled in a comprehensive drug coverage plan, and if upon enrollment they were presented a signed physician's prescription for one of the medications to treat an MRDD-allowed condition. For RA, the MRDD covered adalimumab, etanercept, and anakinra; we excluded the 34 patients who reported use of anakinra because it was rarely used for RA at that time. We included only those patients whose physician signed an affidavit declaring a diagnosis of RA on the MRDD enrollment form, who had at least 30 days of MRDD participation, and who were alive as of January 1, 2006.

Patient characteristics.

Patients' demographic characteristics were drawn from the June 2006 Medicare eligibility file. We measured disease burden using Medicare's proprietary Hierarchical Condition Category score, removing age and sex from the score in order to separate out the effects of these predictors in our regression models (13, 14). Using beneficiaries' zip codes, we determined county of residence and median household income as recorded in the 2000 Census (15). The Area Resource File from the Health Resources and Services Administration was used to assign rural/urban codes to each zip code (16). Using MRDD drug claims data, patients were categorized as to whether or not they filled at least one prescription during the MRDD benefit period.

Part D enrollment and other creditable coverage.

The MRDD ended on December 31, 2005. To obtain drug insurance from January 1, 2006 and later, patients needed to either enroll in a Part D plan or obtain other creditable coverage. To determine a patient's drug insurance status, we obtained proprietary Medicare files of Part D plan enrollment and creditable coverage for the period January 1, 2006 through July 1, 2006. Available creditable coverage data included drug insurance benefits from a current/former employer, Veteran's Administration, TriCare, a retiree drug subsidy, or federal employee health benefits. We grouped patients into 1 of 3 drug coverage options: 1) enrollment in a Part D plan further stratified by a Medicare Advantage or stand-alone plan, 2) continuation or acquisition of other creditable coverage, or 3) unknown coverage. If a patient enrolled in a Part D plan also had other creditable coverage, the Part D plan was considered the primary source of drug coverage. Patients who did not enroll in Part D or who did not appear in our creditable coverage data were classified as having unknown coverage. These patients either had no drug insurance or had drug insurance through a provider whose data were not available to us (e.g., an employer-sponsored plan that did not apply for the retiree drug subsidy). We used patients' dates of Part D plan enrollment to calculate whether there was a gap, and the length of the gap between the end of the MRDD and Part D enrollment.

Part D plan characteristics and comparisons.

We next examined the benefit design of each plan, in which MRDD beneficiaries with RA enrolled using the same publicly available plan files. Variables included each plan's deductible amount, the monthly premium payment amount, and whether coverage was continuous or included a coverage gap (17). The coverage gap is the period of time during which a beneficiary enrolled in a Part D plan is responsible for his or her own medication costs. In the 2006 standard benefit legislated by Congress, the coverage gap began when patient medication costs reached $750 during a calendar year period and ended when patient medication costs reached $3,600, at which time prescription drug coverage resumed. Both stand-alone and Medicare Advantage Part D plans had the option of covering a beneficiary during the coverage gap period through alternative benefit structuring, as long as a plan was actuarially equivalent to the standard benefit. Plans for which data were missing (129 Medicare Advantage and 108 stand-alone) were noted and not included in calculations of frequency measures. We determined each plan's formulary coverage of and cost-sharing mechanisms for adalimumab and etanercept. The plans' rules for prior authorization, step therapy (a process in which a patient must try and fail treatment with another medication before being eligible for reimbursement of the requested drug), and quantity limits were assessed.

Cost-sharing scenarios.

To examine potential differences in beneficiaries' annual out-of-pocket costs for biologic DMARDs under different Part D plans, we created 3 coverage scenarios for etanercept: 1) worst copayment, 2) modal coinsurance, and 3) worst coinsurance. We explored each scenario separately among Medicare Advantage plans and stand-alone plans, for a total of 6 models. The first scenario describes the worst copayment. From those plans with the highest copayment for etanercept, we selected the plan with the highest deductible and lowest initial coverage amount to model costs. The initial coverage limit is the total drug spending threshold at which patients move from the initial coverage period to the coverage gap period. During the initial coverage period, patients pay only a portion of medication costs, whereas in the coverage gap period, patients pay 100% of medication costs. In the 2006 standard benefit legislated by Congress, the initial coverage limit was $2,250. Some plans chose to raise the initial coverage limit, which prolongs a patient's time in the initial coverage period. The modal coinsurance scenario uses the modal values for the plans' deductible amount, initial coverage amount, and coinsurance percentage. Finally, for the worst coinsurance scenario, we selected the plans with the highest coinsurance percentage for etanercept and then modeled costs using the plan with the highest deductible and the lowest initial coverage amount. We used the 2005 MRDD average cost for 1 dispensing of etanercept (defined as 4 doses of the medication, 1 to be administered each week) and assumed that the beneficiary purchased the medication monthly over 1 year. We based our calculations on plan cost-sharing data, which is available in Medicare public use files (18, 19).

Changes in out-of-pocket spending.

Changes in out-of-pocket costs (MRDD costs − Part D costs) were calculated for each beneficiary who enrolled in Part D. Out-of-pocket costs included any deductible, copayments, and/or coinsurance that the beneficiary would spend to obtain a 1-year's supply of his or her biologic DMARD. All calculations used 2005 MRDD-based medication prices.

Statistical analysis.

Descriptive frequencies were calculated for patient characteristics as well as for plan benefit design attributes. An analysis of variance and chi-square tests were used to identify differences among patients with different prescription drug coverage types following Part D implementation. We examined beneficiary characteristics associated with Part D enrollment using multivariate logistic regressions. Predictors in the model included female sex, age, race, region of the US, urban residence, disease burden score, use of the MRDD benefit, MRDD subsidy status, and self-report of other medication coverage during the MRDD. We compared plan attributes and coverage for each of the biologic DMARDs between stand-alone and Medicare Advantage plans using Wilcoxon's rank sum tests (for non-normally distributed data), chi-square tests and Fisher's exact tests as appropriate for categorical data, and t-tests for continuous data. Analyses were conducted using SAS, version 9.1 (SAS Institute, Cary, NC).

RESULTS

The majority of MRDD beneficiaries with RA were white women (Table 1). Among MRDD beneficiaries with RA, 12,150 (81%) enrolled in Part D plans during the first half of 2006, which is ∼56% higher than the rate of enrollment in the general Medicare population (20). Part D enrollees with RA were significantly more likely to have used the MRDD benefit (55%) than beneficiaries who had other creditable coverage (37%) or for whom coverage was unknown (29%). Ninety-two percent of Part D enrollees had no gap in coverage between the MRDD and Part D.

Table 1. Demographic and health care characteristics of 14,929 MRDD participants with rheumatoid arthritis and their prescription drug coverage following Part D implementation*

	Enrolled in a Part D plan			Total enrolled in a Part D plan	Other creditable coverage	Did not enroll or coverage unknown	P
	Stand-alone	Medicare Advantage	P
* Values are the number (percentage) unless otherwise indicated. MRDD = Medicare Replacement Drug Demonstration. † Percentages may not add up to 100 due to rounding. ‡ The MRDD took place between September 2004 and December 2005.
Total	9,865 (81)	2,285 (19)		12,150 (81)	958 (6)	1,821 (12)
Demographics
Age at Part D implementation,   mean ± SD years	65 ± 11	66 ± 11	0.1244	66 ± 11	69 ± 10	67 ± 12	< 0.0001
Women	7,724 (78)	1,826 (80)	0.0898	9,550 (79)	625 (65)	1,353 (74)	< 0.0001
Race
White	8,518 (86)	1,891 (83)	< 0.0001	10,409 (86)	846 (88)	1,614 (89)	0.0004
Other/unknown	1,347 (14)	394 (17)		1,741 (14)	112 (12)	207 (11)
Urban residence	6,806 (69)	2,047 (90)	< 0.0001	8,853 (73)	720 (75)	1,348 (74)	0.0207
Region†
Northeast	892 (9)	342 (15)	< 0.0001	1,234 (10)	110 (11)	279 (15)	< 0.0001
North central	2,096 (21)	337 (15)		2,433 (20)	211 (22)	423 (23)
South	5,368 (54)	867 (38)		6,235 (51)	492 (51)	828 (46)
West and outlying areas	1,497 (15)	735 (32)		2,232 (18)	144 (15)	287 (16)
Household income,   median ± SD dollars	44,680 ± 21,056	45,868 ± 18,985	0.0201	44,915 ± 20,667	46,745 ± 19,411	45,798 ± 22,000	0.0332
Deceased as of July 1, 2006	36 (0.4)	4 (0.2)	0.3706	40 (0.3)	95 (10)	100 (5)	< 0.0001
Health care characteristics and   utilization
Disease burden in the   previous year, mean ± SD	1.30 ± 0.99	1.25 ± 0.88	0.0333	1.29 ± 0.97	1.42 ± 1.02	1.20 ± 0.91	< 0.0001
Disability as original reason for   Medicare	5,538 (56)	1,305 (57)	0.5594	6,843 (56)	387 (40)	924 (51)	< 0.0001
At least 1 prescription drug   claim for an MRDD-covered   drug during the   demonstration‡	5,514 (56)	1,130 (49)	< 0.0001	6,644 (55)	352 (37)	520 (29)	< 0.0001
Subsidy status
Received MRDD subsidy	4,304 (44)	795 (35)	< 0.0001	5,099 (42)	196 (20)	390 (21)	< 0.0001
Received Part D subsidy	4,139 (42)	651 (28)	< 0.0001	4,790 (39)	28 (3)	33 (2)	< 0.0001
Self-reported prescription drug  coverage in addition to MRDD  benefit in 2005	1,813 (18)	985 (43)	< 0.0001	2,798 (23)	463 (49)	473 (26)	< 0.0001

The majority (81%) of Part D enrollees selected stand-alone plans over Medicare Advantage plans. Stand-alone enrollees were less likely to live in urban areas (69% versus 90%) and have higher disease burden, and more likely to have used the MRDD benefit (56% versus 49%), have received an MRDD subsidy (44% versus 35%), and/or have received a Part D subsidy (42% versus 28%).

Female sex (odds ratio [OR] 1.48, 95% confidence interval [95% CI] 1.32–1.67), MRDD benefit use (OR 2.29, 95% CI 2.04–2.58), other self-reported drug coverage during the MRDD (OR 1.53, 95% CI 1.36–1.71), and receiving an MRDD subsidy (OR 2.00, 95% CI 1.74–2.30) were associated with Part D enrollment (Table 2). Region of the US was an important predictor as well. Compared with living in the South, beneficiaries who lived in the Northeast were less likely to enroll in Part D (OR 0.77, 95% CI 0.66–0.90); beneficiaries who lived in the West and outlying areas were more likely to enroll in Part D (OR 1.41, 95% CI 1.21–1.65).

Table 2. Multivariate predictors of Part D enrollment among 14,929 MRDD beneficiaries with rheumatoid arthritis*

	OR (95% CI)
* MRDD = Medicare Replacement Drug Demonstration; OR = odds ratio; 95% CI = 95% confidence interval. † The MRDD took place between September 2004 and December 2005.
Female sex	1.48 (1.32–1.67)
Age	1.00 (0.99–1.01)
Race
White	Reference
Other/unknown	0.98 (0.83–1.15)
Household income, median  dollars
<31,310	0.98 (0.84–1.14)
31,310–40,845	0.95 (0.82–1.10)
40,846–53,845	1.03 (0.89–1.19)
≥53,846	Reference
Urban residence	1.00 (0.88–1.13)
US region
South	Reference
Northeast	0.77 (0.66–0.90)
Midwest	0.94 (0.83–1.08)
West and outlying areas	1.41 (1.21–1.65)
Health care characteristics and  utilization
Disability as original reason  for Medicare	1.05 (0.92–1.21)
Used MRDD-covered drug  ≥1 time during  demonstration†	2.29 (2.04–2.58)
Self-reported other drug  coverage during the MRDD	1.53 (1.36–1.71)
Disease burden in the  previous year	1.05 (0.99–1.12)
MRDD subsidy status
No subsidy	Reference
Received subsidy	2.00 (1.74–2.30)

Table 3 shows beneficiaries' enrollment experiences in Part D plans. The majority enrolled in plans with no deductible (65% among beneficiaries in stand-alone plans, 85% among beneficiaries in Medicare Advantage plans) and a low or no monthly premium (71% among beneficiaries in stand-alone plans, 84% among beneficiaries in Medicare Advantage plans). Although we saw wide variation in plan offerings, beneficiaries tended to enroll in plans with lower premiums. Approximately 20% of Part D beneficiaries enrolled in Medicare Advantage plans versus stand-alone plans. In the general population, ∼28% of Part D enrollees were enrolled in Medicare Advantage plans (20).

Table 3. Medicare Replacement Drug Demonstration beneficiaries' (n = 12,150) enrollment experiences in Part D plans*

	Stand-alone (n = 9,865)	Medicare Advantage (n = 2,285)	P
* Values are the number (percentage) unless otherwise indicated. See Materials and Methods for definition of initial coverage limit and coverage gap. IQR = interquartile range. † A total of 201 beneficiaries enrolled in Medicare Advantage plans and 81 beneficiaries enrolled in stand-alone plans were missing information on drug premium, drug deductible, and initial coverage limit.
Plan structure
Deductible, median (IQR) dollars†	0 (0–250)	0 (0–0)	< 0.0001
$0	6,398 (65)	1,935 (85)	< 0.0001
$1–100	187 (2)	35 (2)
$101–250	3,199 (32)	195 (9)
Monthly premium, median (IQR) dollars†	28.27 (23.66–35.29)	0 (0–23.00)	< 0.0001
$0	0 (0)	1,269 (56)	< 0.0001
$1–34	7,045 (71)	643 (28)
≥$34	2,739 (28)	253 (11)
Initial coverage limit, median (IQR) dollars†	2,250 (2,250–2,250)	2,250 (2,250–2,250)	< 0.0001
$2,250	9,779 (99)	2,007 (88)	< 0.0001
>$2,250	5 (0.05)	158 (7)
Drug coverage during the coverage gap
Generic drugs only	76 (0.77)	70 (3)	< 0.0001
Generic and trade name drugs	16 (0.16)	19 (0.83)

Table 4 shows beneficiaries' plan enrollments and resulting access to and cost sharing for the biologic DMARDs available during the MRDD. Most beneficiaries enrolled in plans with coinsurance as the preferred form of cost sharing, with average coinsurance ranging from 26–28% of the medication price. The maximum coinsurance across plans was 75% for etanercept. For both medications, at least half of the beneficiaries faced prior authorization requirements; step therapy and quantity limit requirements were less popular.

Table 4. Beneficiaries' plan enrollments and resulting access to and cost sharing for 2 rheumatoid arthritis biologic medications*

Enrolled in a plan with the following characteristics	Adalimumab (n = 7,271)			Etanercept (n = 4,879)
	Stand-alone plans	Medicare Advantage plans	P	Stand-alone plans	Medicare Advantage plans	P
* Values are the number (percentage) unless otherwise indicated.
Total	5,921 (81)	1,350 (19)		3,944 (81)	935 (19)
Prior authorization	5,240 (88)	1,149 (85)	0.5273	3,404 (86)	743 (79)	< 0.0001
Step therapy	190 (3)	3 (0.2)	< 0.0001	189 (5)	4 (0.4)	< 0.0001
Quantity limits	1,929 (33)	288 (21)	< 0.0001	1,282 (33)	180 (19)	< 0.0001
Copayment required	1,225 (21)	157 (12)	< 0.0001	797 (20)	254 (27)	< 0.0001
Copayment, mean ± SD   (range) dollars	36 ± 17 (17–125)	36 ± 26 (3–125)	0.7684	36 ± 13 (18–125)	33 ± 21 (1–125)	0.0024
<$20	107 (11)	17 (12)	0.6767	29 (5)	19 (8)	0.0353
$20–50	689 (73)	109 (76)		418 (78)	204 (83)
>$50	144 (15)	18 (13)		86 (16)	24 (10)
Coinsurance required	4,510 (76)	1,091 (81)	< 0.0001	3,111 (79)	656 (70)	< 0.0001
Coinsurance, mean ± SD   (range) percentage	26 ± 3 (25–50)	27 ± 4 (20–50)	< 0.0001	28 ± 8 (25–75)	27 ± 4 (10–50)	< 0.0001
<10%	0 (0)	0 (0)	< 0.0001	0 (0)	1 (0.2)	< 0.0001
10–20%	0 (0)	46 (4)		0 (0)	25 (4)
21–30%	4,223 (94)	827 (76)		2,647 (86)	489 (75)
>30%	266 (6)	217 (20)		444 (14)	139 (21)

Figure 2A depicts beneficiaries' cost-sharing experiences in either a stand-alone or Medicare Advantage plan, where the highest copayment was $125 for etanercept. In the initial coverage period (dispensing 1), a beneficiary pays the copayment, with the remaining cost burden shifted to the Part D plan. Due to the high cost of the medication, the beneficiary enters the coverage gap period during dispensing 2 and remains there until his or her out-of-pocket costs total $3,600. In dispensing 5, the beneficiary transitions into the catastrophic coverage period and costs decrease to 5% of the price of etanercept, whereas Medicare pays for 80% and the Part D plan pays for 15%. In this scenario, a patient's annual cost burden is $4,068 (27% of the total cost).

Figure 2. Cost-sharing scenarios for biologic disease-modifying antirheumatic drugs. A, worst copayment in a stand-alone or Medicare Advantage Part D plan, B, modal coinsurance in a stand-alone or Medicare Advantage Part D plan, C, worst coinsurance in a Medicare Advantage Part D plan, D, worst coinsurance in a stand-alone Part D plan.

In Figure 2B, the beneficiary pays the modal coinsurance percentage (25% [$311.50]) and experiences high costs with the first dispensing of etanercept. By dispensing 2, the beneficiary has again entered the coverage gap and is facing even greater costs because he or she pays 100% of the cost of the drug. By the time the beneficiary fills 12 dispensings, he has paid $19 more than the beneficiary in Figure 2A who has a copayment.

Figures 2C and 2D show the worst coinsurance scenarios. Beneficiaries in Medicare Advantage plans pay a coinsurance of up to 50% (Figure 2C). After dispensing 2, the beneficiary has already paid $1,488 for etanercept. Over the course of 12 dispensings, the beneficiary will pay $4,117, which is $49 more annually than the beneficiary paying a high copayment. Stand-alone plans have even higher coinsurance percentages, the highest being 75% (Figure 2D). Even during the initial coverage period (dispensing 1), the beneficiary is paying $934.50 for a single dispensing. With 12 dispensings, the beneficiary spends $4,149. As plans increasingly shift costs to the beneficiary, Medicare's costs to provide coverage in the catastrophic period rise. Under the worst copayment scenario, Medicare pays $7,482 (50%). Under the worst stand-alone coinsurance scenario, Medicare pays $8,777 (58%).

Out-of-pocket costs for a 1-year supply of the beneficiary's biologic DMARD at the point of sale decreased slightly with the transition to Part D. Under the MRDD, the annual out-of-pocket cost for every beneficiary without an MRDD subsidy was $4,213 for adalimumab and $4,093 for etanercept. In stand-alone Part D plans, beneficiaries without a Part D subsidy paid a median of $4,126 (interquartile range [IQR] $4,126–4,136) for etanercept and a median of $4,251 (IQR $4,251–4,260) for adalimumab. Estimated costs in Medicare Advantage plans were nearly identical. Slightly more beneficiaries who enrolled in Medicare Advantage Part D plans selected a plan that covered the biologic DMARD he or she received during the MRDD, which is 2,139 (94%) as compared with 9,053 beneficiaries who enrolled in stand-alone plans (92%; P = 0.0032) (data not shown).

DISCUSSION

In this study, we followed poor and disabled Medicare beneficiaries with RA who participated in an MRDD as they transitioned into Medicare Part D. Among this group, 81% enrolled in Part D plans by July 2006. Beneficiaries who had used their MRDD benefit, received an MRDD low-income subsidy, and had worse disease burden, i.e., those who appeared to need prescription drug coverage the most, were more likely to enroll.

Compared with stand-alone Part D plans, Medicare Advantage plans offered lower deductibles, lower premiums, and were more likely to require copayments rather than coinsurance for biologic DMARDs, which confirms our hypothesis. Contrary to our hypothesis, Medicare Advantage plans placed significantly fewer restrictions on the biologic DMARD reimbursement, using fewer prior authorizations, step therapy, and quantity limit restrictions. In spite of the greater generosity and lesser restrictions of Medicare Advantage plans, the most sick and most financially needy patients enrolled in these plans less often than they did in stand-alone plans. Patients enrolled in Medicare Advantage plans had lower disease burden, were less likely to use their MRDD benefit, and had fewer subsidies than patients enrolled in stand-alone plans. It is unclear to what extent the differences between enrollees in Medicare Advantage plans and stand-alone plans are the result of the enrollment efforts of the plans versus patient self-selection.

Although Part D enrollees selected plans with a wide range of benefit designs, most enrolled in plans that placed biologic DMARDs on high-cost specialty tiers and used high coinsurance proportions, the highest reaching 75% (the statutory maximum) (9). The specialty tier was created to ensure that beneficiaries receiving high-cost biologic agents were not discriminated against but rather faced the same cost sharing as all of the other beneficiaries. Despite this fact, many observers have been concerned about the monetary impact of specialty tiering, especially the widespread use of high coinsurance proportions (21–23). The scenarios depicted in Figures 2A, 2B, 2C, and 2D show that observers' concerns are justified. Part D plans that require coinsurance instead of copayments for biologic DMARDs shift the financial burden of these high-cost medications from the plan to the patient and to Medicare. The larger the coinsurance proportion, the worse patients and Medicare fare: patients in Part D plans with 75% coinsurance for etanercept pay an estimated $81 more annually than those with a $125 copayment for the medication, whereas Medicare pays an estimated $1,295 more annually.

The preferences for coinsurance of the plans affect not only patients' total out-of-pocket spending but also when that spending takes place. One question is whether differences in spending trajectories across dispensings impact whether patients fill their prescriptions at all. Previous studies have shown that higher out-of-pocket costs result in poorer adherence to medications (24), and poor adherence can impact health outcomes. In plans where cost sharing is particularly high, e.g., plans with coinsurance proportions of 75%, patients may delay or not even begin therapy with prescribed medications because of the cost. In plans where cost sharing is steep but manageable, e.g., plans with high copayments, patients may begin therapy but then discontinue when they reach the coverage gap period and are confronted with paying the full cost of the medication out of pocket. Neither scenario is optimal for patients who may benefit from biologic DMARDs.

Rheumatologists must be mindful of these financial obstacles to biologic DMARD treatment initiation and adherence in their patients with Part D medication insurance. The patient and the rheumatologist must talk frankly about both the clinical benefits and risks of these medications, as well as their financial impact. Rheumatologists can educate and encourage patients receiving biologic DMARDs to enroll in Part D plans that charge copayments. When considering adding a biologic DMARD to the treatment plan of a patient enrolled in Part D, the rheumatologist can question the patient about the cost-sharing provisions of his or her Part D plan and discuss what implications cost will have on use and adherence. If adherence is unlikely due to prohibitive costs, the rheumatologist may decide that a prescription for infliximab, an infused biologic DMARD covered more generously under Medicare Part B, is a more affordable and prudent option.

Our study has several limitations. Because the MRDD population was already identified it was easier to enroll these beneficiaries in Part D, and so the high proportion of enrollment among the MRDD population cannot be extrapolated to the general Medicare population. However, the characteristics of those who did enroll, i.e., those who had worse disease burden, were women, and used their MRDD benefit, suggest that patients in the larger Medicare population can be better targeted for Part D enrollment. Whereas our study assessed the most vulnerable Medicare beneficiaries whose incomes and/or need for drug insurance qualified them for the MRDD, the high costs associated with the biologic DMARDs and the commonplace specialty tiering by Part D plans will impact any elderly or disabled person who fills a prescription for one of them. Our study focused on the 2 self-injectable biologic DMARDs that were covered by the MRDD. Other self-injectable biologic DMARDs are similarly expensive, and our results regarding out-of-pocket costs and cost sharing for these medications can be extrapolated to other biologic DMARDs. Beneficiaries are likely to make their decisions about plan enrollment on the basis of the full array of medications they are taking. However, biologic DMARDs may account for a large proportion of drug spending, thus leading beneficiaries to base their plan enrollment decisions on coverage for these drugs. In order to characterize enrollees' cost-sharing experiences under Part D, we assumed that enrollees filled one prescription per month and did not switch to other medications or discontinue use. These assumptions are likely to approximate actual use but cannot be verified because Part D drug claims data were not available at the time that we conducted this research.

In this study of vulnerable Medicare beneficiaries with RA who received biologic DMARDs, most (81%) enrolled in Medicare Part D during the first 6 months of 2006. The demographic and health care characteristics that predicted their enrollment suggest that these beneficiaries most needed a drug insurance plan and offer evidence that such beneficiaries can be targeted by programs like the MRDD for enrollment in Part D. Enrollees chose a wide range of plans and benefit designs, and overall, the Medicare Advantage plans were more generous and less restrictive than their stand-alone counterparts.

However, the majority of both types of plans placed biologic DMARDs on the specialty tier and used coinsurance as the preferred means of cost sharing, resulting in estimated annual expenditures that exceeded $4,000 despite drug insurance coverage. Alarmingly, more Part D plans have adopted specialty tiering over time. In 2006, 60% of the national stand-alone plans used specialty tiering. By 2008, 87% used specialty tiering (23). Coinsurance proportions have also grown. Between 2006 and 2008, the number of plans charging 33% coinsurance increased more than 5-fold (23). Lee and Emanuel describe the use of specialty tiering and coinsurance for biologic medications as the “fraying of the social compact,” arguing that by shifting large expenses to a small number of patients, we endanger not only those patients, but the very insurance system that cares for the sick (23). Patients assume up to 28% and Medicare assumes more than 58% of the costs of biologic DMARDs in our scenarios, yet neither is in a position to sustain such financial burden. As more biologic DMARDs are approved and used for RA and more plans use the specialty tier system, both beneficiaries and Medicare face costs they may be increasingly unable to afford.

AUTHOR CONTRIBUTIONS

All authors were involved in drafting the article or revising it critically for important intellectual content, and all authors approved the final version to be published. Ms Polinski had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Study conception and design. Polinski, Mohr, Johnson.

Acquisition of data. Polinski, Mohr, Johnson.

Analysis and interpretation of data. Polinski, Mohr, Johnson.