During the Medicare open enrollment period, which runs from October 15 to December 7 each year, beneficiaries have the opportunity to enroll in a plan that provides Part D prescription drug coverage, either a stand-alone prescription drug plan (PDP) as a supplement to traditional Medicare, or a Medicare Advantage drug plan (MA-PD), which provides all Medicare-covered benefits including prescription drugs. Of the nearly 40 million beneficiaries enrolled in Part D plans, about 6 in 10 are in PDPs and the rest in MA-PDs.¹ 

This issue brief provides an overview of the 2016 PDP marketplace, focusing on key changes from 2015, based on our analysis of data from the Centers for Medicare & Medicaid Services (CMS).^2 ,3  In 2016, the typical Medicare beneficiary will have a choice of more than two dozen PDPs. Many will see higher premiums and deductibles if they remain in their current plan. As in the past, Part D enrollees need to consider that their plan may require coinsurance, rather than flat copayments, for brand-name and specialty drugs, and differential cost sharing for prescriptions depending on their choice of pharmacy. These factors, as well as which drugs are included on a plan’s formulary, may affect out-of-pocket costs as much as premiums. Other highlights include:

• In 2016, beneficiaries in each region will have a choice of 26 PDPs, on average, down by 4 from 2015.
• The average PDP premium (weighted by 2015 plan enrollment) is projected to increase by 13 percent from 2015 to 2016, from $36.68 to $41.46 per month. Even if a number of beneficiaries switch or are reassigned to lower-premium plans, the average premium increase for 2016 is likely to be the largest since 2009.
• More than one-third of the 11.2 million PDP enrollees who do not receive Low-Income Subsidies (LIS) would pay premiums of $60 or more per month in 2016 if they stay in the same plan.
• Nearly 4.4 million of these enrollees not receiving the LIS face a premium increase of at least $10 per month in 2016 if they stay in the same plan.
• Two-thirds of all PDPs will have deductibles in 2016, a higher share than in previous years. A growing share of PDPs will impose the maximum deductible allowed by law, which increased from $320 in 2015 to $360 in 2016, the largest increase in the deductible since the start of the program.
• Most PDPs charge coinsurance, rather than flat copayments, for non-preferred brand-name and specialty drugs, which could lead to higher out-of-pocket costs for those who use high-cost drugs.
• Nearly all PDPs (84 percent in 2016) use tiered pharmacy networks, with lower cost sharing in selected network pharmacies and higher cost sharing in other network pharmacies, a significant increase from just a few years ago when only a small share of plans used this type of preferred pricing (7 percent in 2011).
• Beneficiaries receiving the LIS will have access to 7 plans for no monthly premium in 2016, on average, fewer than in any past year. About 1.9 million LIS enrollees who are slated to be in PDPs where they will pay premiums must switch plans or be reassigned by CMS in order to have premium-free coverage in 2016.

Part D Plan Availability

In 2016, beneficiaries across the country continue to have a substantial number of Part D plan choices.

• The average beneficiary will have a choice of 26 PDPs in 2016, down from 30 in 2015 and 55 at the peak in 2007 (Figure 1).⁴  Beneficiaries will also have access to 16 Medicare Advantage prescription drug plans, on average.⁵ 

• The number of PDPs per region in 2016 will range from a low of 19 PDPs in the Alaska region to a high of 29 PDPs in the Pennsylvania/West Virginia region (Figure 2; Appendix 1, Table A1). The number of plans is lower in all but one region (Pennsylvania/West Virginia) compared to 2015; there are five fewer PDPs in Alaska, Florida, and Illinois in 2016.

• A total of 886 PDPs will be offered nationwide in 2016, down by 11 percent from the 1,001 PDPs offered in 2015 and the lowest number of PDPs in the program’s history (Figure 3). The total number of PDPs in 2016 represents less than half the number offered at the peak level in 2007 of 1,875 plans. The lower number of plans reflects both the cumulative effect of mergers among plan sponsors and the response to CMS policies that encourage plan sponsors to eliminate low-enrollment plans and to drop multiple PDPs that are not meaningfully different from each other.

• In 2016, there are 11 plan sponsors offering 23 PDPs on a national or near-national basis. The number of national or near-national plan sponsors is down from 18 in 2009, as a result of corporate acquisitions and some firms exiting the market after not achieving a substantial market share. In addition, more sponsors are offering two PDPs, whereas it was more common to offer three in the program’s early years.
• In July 2015, the PDP sponsor Torchmark Corporation was placed under a sanction that banned marketing activities and new enrollments. Its PDPs, marketed under the United American name, have 208,000 enrollees in 91 PDPs across all regions in September 2015.

For 2016, 31 new PDPs are entering the market, and 146 PDPs are exiting the program.⁶  (See Appendix 2 for more detail on market dynamics.)

• Two new sponsors are entering the PDP market in 2016: Magellan Rx America, offering PDPs in 20 regions, and QualChoice Advantage, offering a PDP only in Arkansas.
• Among the 146 exiting plans, 143 are offered by sponsors that offer other PDPs, and enrollees in most of them (137 of 143) will be automatically assigned to another PDP offered by that sponsor. This process should ensure a seamless transition and minimize disruption that would otherwise result from plan terminations. The other 3 exiting PDPs are offered by sponsors (each operating in a single region) leaving the market entirely

Premiums

National Premium Trends

The national average monthly PDP premium is expected to increase for 2016, marking a significant departure from recent years when premiums were essentially flat. Even if a number of beneficiaries switch or are reassigned to lower-premium plans, effective in January 2016, the average premium increase for 2016 is likely to be the largest amount in any year since 2009.

• The projected average monthly PDP premium for 2016 will be $41.46 (weighted by September 2015 enrollment, assuming beneficiaries remain in their current plan) (Figure 4).⁷  This is a 13 percent projected increase ($4.66) from the weighted average monthly premium of $36.68 in 2015, and a 60 percent increase from $25.93 in 2006, the first year of the Medicare Part D drug benefit. As a point of comparison, since 2006, the medical care consumer price index (CPI) has increased 34 percent and the CPI for all items has increased 18 percent.

• In prior years, the average premium after the end of the open enrollment period has been 4 percent to 5 percent lower than the projection made during the open enrollment period, due to existing enrollees switching to a lower premium plan, new enrollees choosing low-premium plans, and reassignments of some LIS beneficiaries to cheaper plans. For example, the average premium in February 2015 was $37.20,which is about 4 percent ($1.63) below the projected premium of $38.83 calculated prior to enrollment changes.⁸  Applying a similar adjustment to our projected premium for 2016 suggests that the premium increase, after open enrollment, is likely to be lower, but still larger than in any year since 2009.
• For PDPs offering only the basic benefit, the 2016 weighted average premium is projected to be 8 percent higher in 2016 than in 2015 (based on current enrollment patterns), whereas premiums for enhanced PDPs are projected to rise by about 17 percent.⁹ 

A growing share of PDP enrollees are paying premiums that are well above the national average.

• In 2016, over one-third of PDP enrollees not receiving the LIS (36 percent) are projected to have premiums of at least $60 per month if they stay in their plans—a substantial increase from the 10 percent of enrollees paying premiums at this level in 2015 (Figure 5). Among them, nearly 400,000 are projected to pay premiums of at least $100 per month. By contrast, 14 percent of these PDP enrollees are scheduled to pay premiums of less than $20 per month.

CMS reported that the average premium for standard Part D coverage offered by PDPs and Medicare Advantage drug plans will “remain stable” at an estimated $32.50 in 2016.¹⁰  The higher premium reported here is based on PDPs only, including PDPs offering both basic and enhanced coverage. Enhanced plans typically have higher premiums than basic PDPs. Our estimate does not assume any voluntary switches by beneficiaries and reassignments by CMS of Low-Income Subsidy enrollees to different plans, nor do we make any assumptions about plan choices made by new enrollees in 2016.

Monthly Premium Changes for Enrollees

Underneath these national trends, premium changes for individual PDPs from 2015 to 2016 vary widely (Figure 6).

• Among the 11.2 millionPart D PDP enrollees who are responsible for paying the entire premium (which excludes LIS recipients), most (83 percent) are projected to have a higher premium in 2016 if they stay in their current plans, including 4.4 million (39 percent) who are projected to have a premium increase of at least $10 per month. Conversely, nearly half a million (448,000) are projected to experience a premium decrease of $10 or more.
• Taking into account all PDP enrollees, including the LIS enrollees whose premiums are partially paid by the government, 80 percent are in plans in which premiums are projected to rise in 2016 if they stay in their current plans; this includes 28 percent in plans that will have an increase of $10 or more in their monthly premium in 2016 assuming no enrollment changes. By contrast, 2 percent of all PDP enrollees are in plans projected to have a premium decrease of $10 or more.

Premium Changes for PDPs with the Most Enrollees

Changes to premiums from 2015 to 2016, averaged across regions and weighted by 2015 enrollment, vary widely across some of the most popular PDPs (Figure 7). These variations reflect factors such as higher premiums in older plans and substantial geographic variations.¹¹  Although variation in benefits is also a factor, premium variation persists even among PDPs that offer benefits actuarially equivalent to the basic benefit defined in law.

• Average premiums for seven of the eight largest PDPs will increase for 2016. Five of these seven PDPs will have larger average percentage increases than the national average. Average premiums for three of these PDPs are projected to increase by at least 20 percent in 2016.
• The highest increases, in percentage terms, are for Humana Enhanced PDP, with 1.1 million enrollees in 2015, and AARP MedicareRx Saver Plus, with 1.4 million enrollees. On average, premiums for both PDPs will rise 25 percent (from $52.86 to $66.25 for Humana Enhanced and from $28.13 to $35.23 for AARP MedicareRx Saver Plus). The average monthly premium for the PDP with the most enrollees in 2015, AARP MedicareRx Preferred, with 3.5 million enrollees, will increase by 21 percent between 2015 and 2016, from $50.19 to $60.79.
• By contrast, enrollees in one of the largest PDPs will experience a modest decrease in their monthly premium if they stay in the same plan. The average monthly premium for SilverScript Choice will decrease by 2 percent ($22.56 in 2016 versus $23.13 in 2015).

Enrollees in some of the most popular PDPs could see bigger or smaller premium increases than the national average, depending on the region where they live.

• UnitedHealth’s AARP MedicareRx Preferred PDP is raising premiums by as little as 9 percent (in Oregon/Washington) but as high as 37 percent (in Alaska) (Figure 8). The range is even greater for CVS Health’s SilverScript Choice PDP, which is reducing premiums by as much as 36 percent (in Arkansas) and increasing premiums by as much as 24 percent (in Florida).

Monthly premiums for the most popular PDPs available since the start of the Part D program in 2006 have changed in different ways.

• The average premium for Humana PDP Enhanced, with 1.1 million enrollees in 2015, is more than four times higher than it was in 2006, having increased from $14.73 to $66.25. Of the eight most popular PDPs, this plan is projected to have the most expensive monthly premium in 2016.
• The average premium for UnitedHealth’s AARP Preferred MedicareRx PDP, the PDP with the most enrollees in 2015 (3.5 million), has more than doubled since 2006, from $26.31 to $60.79 in 2016.
• By contrast, the monthly premium for SilverScript Choice, operated by CVS Health, with 2.5 million enrollees in 2015, is 20 percent lower in 2016 ($22.56) than it was in 2006 ($28.32).

Premium Variation and Changes by Region

Average PDP monthly premiums, weighted by 2015 enrollment, will vary widely in 2016 across regions, along with the change in premiums from 2015 to 2016.

• Average PDP premiums will range from $29.70 per month in the New Mexico region (one of only three regions with an average under $35) to $49.61 per month in New Jersey and $45.31 in Florida (Appendix 1, Table A2).
• Premium changes from 2015 to 2016 vary considerably by region. For example, average premiums in Arkansas are projected to fall by about 1 percent, whereas average premiums will be at least 20 percent higher in Arizona and Florida (Figure 9).

Premiums for Higher-Income Enrollees

Average and plan-level premium amounts do not take into account the income-related Part D premium that took effect in 2011 for Part D enrollees with higher annual incomes ($85,000/individual and $170,000/couple). Established by the Affordable Care Act of 2010 (ACA), the income-related Part D premium requires higher-income enrollees to make an additional payment to the government for Part D coverage, regardless of the plan selected. In 2016, the monthly surcharge will range from $12.70 to $72.90, depending on income, in addition to the monthly premium payment for the specific Part D plan.¹²  Nearly 6 percent of Part D enrollees are expected to make these additional payments in 2016.¹³  Under current law, the income thresholds are not indexed to increase annually until 2020, meaning that by 2019, a projected 8 percent of Part D enrollees will pay income-related premiums.

Benefit Design

In 2016, for the second year in a row, all Part D PDPs will offer an alternative benefit design to the defined standard benefit, which has a $360 deductible in 2016 and 25 percent coinsurance for all covered drugs. Some plans modify or eliminate the deductible, and all PDPs use some type of varying cost-sharing tiers for covered generic and brand-name drugs.

Deductibles

The standard (maximum) Part D deductible is increasing by $40 in 2016, the largest one-year increase since the program began. It is adjusted each year by a statutory formula based on the annual percentage increase in average per capita aggregate expenditures for covered Part D drugs. Other amounts for the standard benefit design parameters are increasing as well (see Appendix 3 for the 2006-2016 standard benefit amounts).

• Two-thirds of PDPs (67 percent) will charge a deductible in 2016, up from 58 percent in 2015 (Figure 10). The share of PDPs with a deductible is the highest ever in the program. Most PDPs adding deductibles have relatively few enrollees, so the share of PDP enrollees facing a deductible is likely to change more modestly. Nearly two-thirds of all Part D enrollees are in plans that are increasing the deductible for 2016.
• Most PDPs with a deductible (53 percent) will charge the standard $360 amount, which is also the highest share in the program’s history. Another 14 percent of all PDPs have a deductible below the standard amount.

Tiered Cost Sharing

• The use of formulary tiers has been common since the program’s beginning in 2006, but for the second year in a row, all PDPs will use tiered cost sharing. Similar to 2015, most PDPs will use five tiers (two for generic drugs, two for brand-name drugs, and one for higher-cost specialty drugs).
• Over the years, there has been a trend toward the use of coinsurance in place of flat copayments for some formulary tiers. Coinsurance typically means higher out-of-pocket costs for more expensive drugs, such as those over $200 per month. Coinsurance has been used by nearly all plans for the specialty tier since 2006, and that coinsurance cannot exceed 33 percent. For all PDPs, the specialty tier coinsurance is set between 25 percent and 33 percent. In the program’s early years coinsurance was uncommon for other tiers, but in 2016, over 90 percent of PDPs are using coinsurance for their non-preferred brand tiers; about 40 percent do so for their preferred brand tiers.¹⁴ 
• For brand tiers, the typical coinsurance is 20 percent for preferred brands and 40 percent for non-preferred brands. CMS guidance allows coinsurance for these tiers to be set as high as 25 percent and 50 percent, respectively. Some PDPs have elected to set coinsurance at these maximum levels.

The Coverage Gap

The coverage gap, or “doughnut hole,” is gradually becoming a less salient feature of the Part D benefit design, as a result of changes made by the ACA. It will be fully eliminated as of 2020, when beneficiaries will only be responsible for 25 percent of their total drug costs in the gap. Since 2011, any beneficiary with drug costs high enough to reach the gap has paid less than the full cost of the price of their drugs. In 2016, manufacturer prices for brand-name drugs purchased in the gap will be discounted by 50 percent, with plans paying an additional 5 percent of the cost and enrollees paying the remaining 45 percent. Plans will pay 42 percent of the cost for generic drugs in the gap, with enrollees paying 58 percent. In 2016, the coverage gap begins after an enrollee incurs $2,960 in total drug spending and ends after an enrollee has spent a total of $4,850 out of pocket (or $7,515 in total drug costs under the standard benefit).¹⁵  At that point, catastrophic coverage begins, where enrollees generally pay only 5 percent of drug costs.

Most Part D plans will offer no gap coverage in 2016 beyond what is required by the ACA under the standard benefit. With all Part D enrollees now getting coverage for a share of their costs in the gap, the value of additional gap coverage offered by plans will become lower each year until 2020, when the gap is fully closed.

• In 2016, about 78 percent of all PDPs will offer no additional gap coverage. This is a small increase from 2015, when 74 percent of PDPs offered no additional gap coverage—meaning a slightly smaller share of plans will offer some gap coverage beyond what the ACA requires in 2016 than in 2015. One national PDP (First Health Part D Value Plus) added gap coverage, and one near-national plan (Symphonix PrimeSaver Rx) dropped gap coverage.

Tiered Pharmacy Networks

Nearly all PDPs (85 percent) use tiered pharmacy networks. In these networks, enrollees pay preferred (lower) cost sharing for their prescriptions when they use selected network pharmacies and higher cost sharing in other network pharmacies.

• The share of PDPs with tiered preferred pharmacy networks is down slightly from 87 percent in 2015, but still a significant increase from just a few years ago when only a small share of plans used this type of preferred pricing (7 percent in 2011).

The cost implications of tiered pharmacy networks for enrollees vary across PDPs.

• For example, among the national and near-national PDPs with tiered pharmacy networks, the AARP MedicareRx Saver Plus PDP charges a $20 copayment for a preferred brand drug in a pharmacy that offers preferred cost sharing and $33 in another network pharmacy that does not offer preferred cost sharing. The Humana Walmart Rx PDP charges a $1 copayment for preferred generic drugs and $4 for non-preferred generics at a pharmacy with preferred cost sharing, compared to $10 and $33, respectively, at other network pharmacies where preferred cost sharing is not offered. These differentials are somewhat smaller than in 2015.

Low-income Subsidy (Benchmark) Plans

In 2016, the total availability of premium-free (benchmark) plans—that is, PDPs available for no monthly premium to Low-Income Subsidy (LIS) enrollees—will be at its lowest level in the program’s ten years (Figure 11; Appendix 1, Table A3).

• In 2016, 226 plans will be available for enrollment of LIS recipients for zero premium. This represents a 20 percent decrease in plans for LIS recipients, or 57 fewer plans than in 2015 and 126 fewer plans than in 2014.
• On average (weighted by enrollment), LIS have 7 benchmark plans available to them for 2016, or about one-fourth the number of PDP choices available overall. All LIS enrollees retain the option to select other plans offered in their area, but if they enroll in a non-benchmark plan, they must pay a share of the monthly premium.
• Of the 226 benchmark plans in 2016, about one in four (57 plans) qualify through the “de minimis” policy—about the same number as the 54 “de minimis” plans in 2015. Policies adopted by CMS in previous years make it easier for PDPs to qualify as benchmark plans, including the “de minimis” policy that allows plans to waive a premium amount of up to $2 in order to retain their LIS enrollees.¹⁶ 
• Among the 2015 benchmark plans, 72 PDPs have lost their benchmark status due to either higher premiums in 2016 or to a lower regional benchmark in 2016 (one other exited the market completely). About 742,000 LIS beneficiaries (9 percent of LIS enrollment in PDPs in 2015) are enrolled in these 73 plans, meaning these beneficiaries may experience some disruption in their coverage for 2016. In 2016, 25 PDPs will newly qualify for benchmark status: 14 PDPs new to the program and 11 older PDPs (in nine other cases, two benchmark PDPs were consolidated by the plan sponsor).

Availability of Benchmark Plans by Region

The number of benchmark plans available in 2016 will vary by region, from 10 benchmark PDPs in the Arizona, Idaho/Utah, and Pennsylvania/West Virginia regions to just 2 benchmark PDPs in the Hawaii region (out of 21 PDPs overall) and 3 (of 22) in the Florida region (Figure 12).

• Benchmark plan availability will decline in 26 of 34 regions between 2015 and 2016, while one additional benchmark plan will be available in 3 regions (Figure 13).
• Year-to-year changes in most regions are relatively modest; the largest change is the loss of seven benchmark plans in the Hawaii region, after gaining five benchmark plans for 2015 and losing six for 2014. As a result of losing this many benchmark plans, 92 percent of Hawaii’s LIS beneficiaries are projected to pay a premium unless they are reassigned or switch plans. In five other regions, between 40 percent and 50 percent of LIS beneficiaries are in this same situation. In regions where the availability of benchmark plans is more stable, the share of affected LIS beneficiaries is much smaller.

Impact of Benchmark Plan Changes for Low-Income Subsidy Enrollees

About 1.9 million people—about one in four LIS beneficiaries (24 percent)—are enrolled in PDPs in 2015 that will not qualify as benchmark plans in 2016 (Figure 14).

• This group includes 742,000 beneficiaries who were in benchmark plans in 2015; the remaining three-fifths (60 percent) of these beneficiaries are currently enrolled in non-benchmark plans and thus paid a premium in 2015.
• About 764,000 LIS enrollees will pay premiums of at least $20 per month and about 53,000 LIS enrollees will pay premiums of at least $50 per month if they do not switch to other PDPs.
• The number of LIS beneficiaries who will potentially pay a premium in 2016 unless they enroll in (or are switched to) benchmark plans (1.9 million) is up slightly from the 1.8 million LIS beneficiaries who were in a similar situation at the time of the open enrollment period for 2015. CMS will reassign those LIS enrollees who were randomly assigned by CMS to their current plan, and several states will help reassign those enrolled in their state pharmacy assistance programs (SPAPs).¹⁷  Effective for January 2015, CMS reassigned about one-fifth of those scheduled to pay a premium. Many other LIS beneficiaries who are currently not enrolled in plans that will be premium-free in 2016 must switch plans on their own or pay a premium if they remain in their 2015 plan. Those in the latter group will not be automatically reassigned by CMS because in the past they or someone assisting them made a choice to switch plans. Most affected LIS beneficiaries will receive a letter from CMS or their SPAP either informing them of their reassignment or reminding them that they can choose a different plan and avoid paying a premium.

Benchmark Plans by Plan Sponsor

The number of premium-free plans for LIS enrollees offered by the major Part D organizations has fluctuated substantially over the years.

• For 2016, six PDP sponsors will offer benchmark plans in at least half of the 34 regions.¹⁸  No sponsor will offer benchmark plans in all 34 regions in 2016; three sponsors (Aetna, CVS Health, and Humana) offer benchmark plans in at least 33 regions.
• In 2015, about 70 percent of LIS enrollees in PDPs are in plans operated by just four plan sponsors (CVS Health, Humana, UnitedHealthcare, and Cigna) (Figure 15). The first two of these sponsors offer PDPs that qualify as benchmark plans in at least 32 of the 34 PDP regions in 2016. By contrast, UnitedHealth offers PDPs that qualify as premium-free in 15 regions in 2016, down from 33 in 2015. Cigna offers PDPs that qualify as premium-free in 12 regions in 2016, down from 23 in 2015. Over half of the LIS beneficiaries projected to pay premiums are currently enrolled in UnitedHealth or Cigna PDPs.

• The LIS enrollees in PDPs offered by these sponsors are split between premium-free benchmark plans and other plans operated by these sponsors. For example, 64 percent of UnitedHealthcare’s LIS enrollees are in non-benchmark PDPs and thus pay premiums of $21.06 on average. Many of them are in the AARP MedicareRx Preferred PDP, which qualified as a benchmark PDP in the program’s earlier years.

Discussion

This review of the Part D plan landscape for 2016 and changes over time indicates that most PDP enrollees can expect to see some changes in their prescription drug coverage and costs in 2016. For most enrollees, premiums are projected to be higher in 2016 than in 2015, and many will also see higher deductibles and more cost-sharing tiers with coinsurance. LIS enrollees will have fewer PDP options available at a zero premium, and many be reassigned to different plans or will need to switch to a different plan to continue without a premium. Beneficiaries have options during open enrollment to select plans that better meet their needs, including plans with lower premiums. Although premiums tend to be the most visible feature when comparing plans, beneficiaries need to consider other factors, especially whether their drugs are on a plan’s formulary, whether their usual pharmacy offers preferred cost sharing, and the total out-of-pocket costs (including both premiums, deductibles, and cost sharing) for coverage of their prescription drug needs.

Changes in Part D plan costs and benefit design from one year to the next have been a defining featuring of the program since its beginning. The reduced number of PDPs for 2016 could make the process of comparing and reviewing plan options a little easier for beneficiaries to undertake during the open enrollment period, but the more than two dozen PDP options in each region may still discourage many from reevaluating their choices. While the annual enrollment period is the best opportunity for people on Medicare to evaluate their plan options and make changes, our analysis of plan switching by PDP enrollees not receiving the LIS showed that about 87 percent stayed in the same plan in annual enrollment periods between 2006 and 2010.¹⁹  Even when they were projected to face large premium increases, a majority of PDP enrollees stayed in the same plan from one year to the next. Among LIS enrollees, although many are reassigned each year to new plans by CMS, few (14 percent) of those not eligible for reassignment by CMS and projected to pay a premium switched to new plans in 2010.²⁰  Finding ways to get more Part D enrollees engaged in comparing and reviewing plans and making changes that could save them money remains an ongoing challenge for CMS and policymakers.

Appendix 1: Information about PDPs by State

Appendix 2: Market Dynamics

• Among the 31 new PDPs in the market, 21 are offered by sponsors totally new to the program: Magellan Rx America (20 PDPs in 20 regions) and QualChoice Advantage (1 PDP in Arkansas). The other new PDPs are additional offerings in a few regions by three national or near-national plan sponsors (EnvisionRxPlus, Symphonix Heath, and WellCare).
• Four firms that offer PDPs on a national or near-national basis are eliminating one of the PDPs they offer in all or most regions. Aetna continued consolidating the PDPs it acquired from Coventry and now sponsors three PDPs instead of four, eliminating the Aetna Medicare Rx Premier PDP (42,000 enrollees across all regions). Two of its three offerings in each region are marketed under the First Health name. Cigna, after consolidating PDPs from its acquisition of HealthSpring, is eliminating one of its three PDPs (Cigna-HealthSpring Rx Secure-Max PDP with 22,000 enrollees). Express Scripts is eliminating the SmartD Rx Saver PDPs it acquired in 2014 (73,000 enrollees), leaving only two PDPs per region. Finally, TransAmerica eliminated its MedicareRx Choice PDPs (20,000 enrollees), leaving it with just a single PDP per region (and also dropped all PDPs in three regions, affecting another 2,000 enrollees).
• Three plan sponsors, each operating a single PDP in just one region, left the program completely: Alliance Medicare RX PDP, Colorado Access Vista Medicare PDP, Health Alliance Medicare Prescription Plan – Basic PDP. Each PDP had about 1,000 enrollees.
• Four other plan sponsors (EnvisionRxPlus, MedicareBlue Rx, SmartSaver Rx, and United American) made small adjustments to their offerings. The decision by Blue Cross Blue Shield plans in the upper Midwest region to drop their Value Plus PDP will affect about 100,000 enrollees, but the other changes have much smaller effects.

Appendix 3: Medicare Part D Standard Benefit Parameters, 2006-2016

1. These enrollment counts include nearly 7 million Part D enrollees in employer-only plans, not otherwise analyzed for this spotlight. ↩︎
2. Centers for Medicare and Medicaid Services, “Medicare Advantage Premiums Remain Stable; Enrollment at All-Time High,” September 21, 2015; 2016 PDP, MA, and SNP Landscape Source Files and related files are available at http://www.cms.hhs.gov/PrescriptionDrugCovGenIn/. ↩︎
3. For analysis of the 2015 Part D marketplace and ten-year trends, see Jack Hoadley, Juliette Cubanski, and Tricia Neuman, “Medicare Part D at Ten Years: The 2015 Marketplace and Key Trends, 2006-2015,” Kaiser Family Foundation, October 2015, available at https://modern.kff.org/medicare/report/medicare-part-d-at-ten-years-the-2015-marketplace-and-key-trends-2006-2015/; analysis of the Part D marketplace for earlier years is available at https://modern.kff.org/. ↩︎
4. The average is weighted by enrollment. ↩︎
5. Gretchen Jacobson, Anthony Damico, and Tricia Neuman, “What’s In and What’s Out? Medicare Advantage Market Entries and Exits for 2016,” Kaiser Family Foundation, October 2016. ↩︎
6. In addition, two PDPs offered by First Health (Aetna) changed contract numbers. ↩︎
7. Based on authors’ analysis using the CMS 2016 Part D Crosswalk file. ↩︎
8. During 2015, the average premium continued to drop, most likely due to plan elections by newly eligible beneficiaries, who are likely to enroll in plans with below-average premiums. The average PDP premium fell from $37.20 in February to $37.02 in April to $36.68 in September, a net drop of 1 percent in seven months. ↩︎
9. The average premium will increase from $28.18 in 2015 to $30.41 in 2016 for basic-benefit PDPs and will increase from $47.15 to $55.06 for enhanced-benefit PDPs. Enrollees in 37 PDPs will switch from enhanced to basic, but excluding them does not affect the rates of increase. ↩︎
10. CMS, “Medicare Prescription Drug Premiums Projected to Remain Stable”, available at https://www.cms.gov/Newsroom/MediaReleaseDatabase/Press-releases/2015-Press-releases-items/2015-07-29.html. ↩︎
11. Jack Hoadley, Juliette Cubanski, and Tricia Neuman, “Medicare Part D at Ten Years: The 2015 Marketplace and Key Trends, 2006-2015,” Kaiser Family Foundation, October 2015, available at https://modern.kff.org/medicare/report/medicare-part-d-at-ten-years-the-2015-marketplace-and-key-trends-2006-2015/. ↩︎
12. Centers for Medicare & Medicaid Services, “Annual Release of Part D National Average Bid Amount and Other Part C & D Bid Related Information,” July 29, 2015. ↩︎
13. Juliette Cubanski and Tricia Neuman, “Medicare’s Income-Related Premiums: A Data Note,” June 2015, available at https://modern.kff.org/medicare/issue-brief/medicares-income-related-premiums-a-data-note/. ↩︎
14. These calculations are made from the Plan Finder for one region and cover only national and near-national plans. ↩︎
15. This amount corresponds to the estimated catastrophic coverage limit for non-Low-Income Subsidy enrollees ($7,063 for LIS enrollees), which corresponds to True Out-of-Pocket (TrOOP) spending of $4,850 (the amount used to determine when an enrollee reaches the catastrophic coverage threshold). ↩︎
16. Plans qualifying through the de minimis policy are eligible for new enrollees, but will not receive auto-assigned enrollees. ↩︎
17. Estimates for the total number of beneficiaries subject to paying a premium are based on plan data from the landscape and crosswalk files, together with CMS enrollment reports. The number to be reassigned will be released by CMS in November or December 2015. ↩︎
18. These counts of benchmark plans include those designated as de minimis plans, which will not receive auto-assigned enrollees. ↩︎
19. Jack Hoadley, Elizabeth Hargrave, Laura Summer, Juliette Cubanski, and Tricia Neuman, “To Switch or Not to Switch: Are Medicare Beneficiaries Switching Drug Plans To Save Money?” Kaiser Family Foundation, October 2013, available at https://modern.kff.org/medicare/issue-brief/to-switch-or-not-to-switch-are-medicare-beneficiaries-switching-drug-plans-to-save-money/. ↩︎
20. Jack Hoadley, Laura Summer, Elizabeth Hargrave, Samuel Stromberg, Juliette Cubanski, and Tricia Neuman, “To Switch or Be Switched: Examining Changes in Drug Plan Enrollment among Medicare Part D Low-Income Subsidy Beneficiaries” Kaiser Family Foundation, July 2015, available at https://modern.kff.org/medicare/report/to-switch-or-be-switched-examining-changes-in-drug-plan-enrollment-among-medicare-part-d-low-income-subsidy-enrollees/. ↩︎

During the debate surrounding the Affordable Care Act (ACA), some questioned whether plans would exit as a result of the payment reductions to Medicare Advantage plans, leading to a drop in plan choices, and eventually, a decline in enrollment. Instead, Medicare Advantage enrollment growth has exceeded expectations, increasing from 24 percent of beneficiaries in 2010 to almost one-third (31%) of beneficiaries in 2015.¹  Between 2010 and 2015, the total number of plans declined modestly, but beneficiaries in 2015 still had the option to choose among 18 Medicare Advantage plans, on average.² 

This Data Note examines the availability of plans nationwide and by state, in 2016, and tracks changes in plan availability since 2012. The analysis uses the Centers for Medicare and Medicaid Services (CMS) 2016 Medicare Advantage Landscape Source file released September 15, 2015 and the 2016 Part C&D Plan Crosswalk file released October 1, 2015; the analysis excludes employer-based (group) Medicare Advantage plans, Special Needs Plans (SNPs) and other plans not available for general enrollment. It documents the number and share of Medicare Advantage enrollees affected by plan withdrawals each year, the characteristics of plans that will be entering or exiting the market in 2016, and the potential implications of these changes for Medicare Advantage enrollees.³ 

Our analysis finds that the Medicare Advantage market is relatively stable in terms of plan participation and beneficiary choice.

• On average, Medicare beneficiaries will be able to choose among 19 Medicare Advantage plans in 2016, a slight increase from last year (18 plans per beneficiary).
• The majority of plans available in 2015 will continue to be offered in 2016.
• A smaller number of plans will exit or enter the market than in any year since 2012.
• In most states, the total number of plans offered in 2016 will be greater than or equal to the number offered in 2015.
• Among the minority of 2015 plans exiting the market in 2016, most have relatively low enrollment, which means that relatively few beneficiaries will be affected by the market exits.
• Several firms will offer Medicare Advantage plans for the first time in 2016, including some that will be led by local, established health care systems.

Medicare Advantage Plan Availability

Nationwide

The total number of Medicare Advantage plans will be similar in 2016 to the number in 2015, increasing by 3 percent from 1,945 plans in 2015 to 2,001 plans in 2016 (Figure 1). The number of plans reflects both plan exits and entries. Between 2015 and 2016, fewer Medicare Advantage plans will exit than enter: 203 plans will exit markets across the country at the end of 2015 while 259 new plans will enter markets in 2016.   The majority of plans that will be offered in 2016 (87%) were also available in 2015.

A smaller number of Medicare Advantage plans will exit the market in 2016 than in any year since 2012 (Figure 2). Similarly, a smaller number of plans will enter the market in 2016 than in any year since 2012.

By State

In 37 states and the District of Columbia, the total number of plans offered in 2016 will be greater than or equal to the number offered in 2015. In 13 states, the number of plans offered in 2016 will equal the number offered in 2015, including five states that will have no departing plans or new plans. Ohio and Washington will experience the largest net increase in the number of plans offered between 2015 and 2016 (15 plans and 12 plans, respectively). In the remaining 13 states and Puerto Rico, there will be fewer plans offered in 2016. Beneficiaries in two states (CA and NY) will see large numbers of both new plans and departing plans (20 or more) in 2016.

Plans per Beneficiary

Medicare beneficiaries will be able to choose among 19 plans, on average, in 2016, a slight increase from the number available to beneficiaries in 2015 (Figure 3). Beneficiaries in metropolitan areas will be able to choose from 21 plans, on average, and beneficiaries in non-metropolitan areas will be able to choose from among 11 plans. Among the 19 Medicare Advantage plans available to the average beneficiary in 2016, 16 plans will offer Part D prescription drug coverage (MA-PDs).

New Firms

Several firms will offer Medicare Advantage plans for the first time in 2016. Some of these firms, such as Johns Hopkins, are led by local, established health care systems that have less experience as insurers. Other firms, such as Medical Mutual of Ohio, are large, established health insurers that are branching out to Medicare Advantage. Firms entering the Medicare Advantage market for the first time in 2016 will be offering plans in nine states (AR, FL, IA, KY, MD, MN, NC, OH, and OK). In four of these states (AR, IA, MD, and OK), the average Medicare Advantage penetration was less than 20 percent – far below the national average of 31 percent. Firms may perceive these states as ripe for Medicare Advantage enrollment growth.

Effect of Plan Exits on Enrollees

The vast majority of Medicare Advantage enrollees will not be affected by plans exiting the market in 2016. Three percent of Medicare Advantage enrollees (about 328,000 enrollees) are in plans that are exiting the market at the end of 2015 and will need to find an alternative source of Medicare coverage, either another Medicare Advantage plan or traditional Medicare (Appendix Table 1). At the end of 2014, a larger number (roughly 480,000 enrollees) and a slightly larger share (about 4%) of the Medicare Advantage population were affected by plan departures.⁴ 

The share of Medicare Advantage enrollees affected by plan departures ranges from less than 1 percent in 15 states, the District of Columbia, and Puerto Rico to more than 10 percent in 2 states (AL and WY). New York will have the largest number of enrollees in plans that will not be offered in 2016 (roughly 70,000 enrollees) – more than twice the number of enrollees in exiting plans in North Carolina (about 31,000 enrollees), the state with the second largest number of affected enrollees. However, in both of these states, the affected enrollees account for a relatively small share of the state’s total Medicare Advantage enrollment (9% and 8%, respectively).   In two states (AL and WY), a larger share (14% and 13%, respectively) but smaller number (less than 25,000 and less than 300, respectively) of enrollees are in plans that will not be offered in 2016. In 18 states, the District of Columbia, and Puerto Rico, fewer than 1,000 people are enrolled in plans in 2015 that will not be offered in 2016.

Characteristics of Medicare Advantage Market Exits and Entries

Plan Type

Among the relatively few plans exiting the Medicare Advantage market in 2016, most (61%) are HMOs and about one-quarter (26%) are local PPOs (Figure 4). These shares are similar to the overall composition of plan offerings in 2015.   PFFS plans are somewhat overrepresented among departing plans. The reduction in PFFS plans available between 2015 and 2016 continues the trend in PFFS plan exits that has been observed since the enactment of the Medicare Improvements for Patients and Providers Act (MIPPA) of 2008, which required most PFFS plans to adopt provider networks. HMOs comprise the about three-fourths (77%) of the new plans that will be offered in 2016, but a somewhat smaller share (68%) of all Medicare Advantage plans offered in 2016 (Figure 4).

Tax Status

The majority of the 203 plans departing the market at the end of 2015 are for-profit plans (77%), a share which is similar to the proportion of available plans in 2015 that were for-profit (Appendix Table 2). Similarly, most new plans are for-profit plans (71%) in 2016.

Enrollment

Among the relatively small number of plans that are departing from the market (excluding those that consolidated into new plans) before 2016, most have relatively few enrollees in 2015. Nearly three-quarters (74%) have 1,000 or fewer enrollees, and more than half (59%) have fewer than 500 enrollees in 2015 (Figure 5; Appendix Table 3). The average number of enrollees in a plan available for general enrollment was 6,164 in 2015. This analysis excludes the plans that will consolidate into new plans, because the plans that are consolidating often have different enrollment levels in the various areas in which they serve, which cannot be separately analyzed.

Quality Ratings

The majority of rated plans exiting the market at the end of 2015 have average or below average ratings (3.5 stars or fewer), or are missing star ratings (Figure 5). Plans with below average ratings (2 or 2.5 stars) represent a small share of all departing plans (13%), but among these plans with below average ratings, a disproportionate share (25%) is departing (Appendix Table 3). The departing plans that are missing star ratings never were rated because they were too new or had too few enrollees. This analysis excludes the plans that will consolidate into new plans, because the plans that are consolidating often have different star ratings in the various areas in which they serve, which cannot be separately analyzed.

Discussion

More Medicare Advantage plans will be offered in 2016 than in 2015, nationally and in many states, with the fewest market exits in any year since 2012. The vast majority of plans that were available in 2015 will continue to be available in 2016. Among the 203 plans (10% of plans) exiting the market, most had relatively low enrollment. Overall, 3 percent of all Medicare Advantage enrollees are in plans in 2015 that will not be offered in 2016. In all likelihood, beneficiaries affected by plan exits will have other Medicare Advantage options available, including new plans vying for beneficiaries, in addition to traditional Medicare.

 

1. Gretchen Jacobson, Anthony Damico, Tricia Neuman, and Marsha Gold. Medicare Advantage 2015 Spotlight: Enrollment Market Update. Menlo Park (CA): 2015. https://modern.kff.org/medicare/issue-brief/medicare-advantage-2015-spotlight-enrollment-market-update/ ↩︎
2. Gretchen Jacobson, Anthony Damico, Tricia Neuman, and Marsha Gold. Medicare Advantage 2015 Data Spotlight: Overview of Plan Changes. Menlo Park (CA): 2014. https://modern.kff.org/medicare/issue-brief/medicare-advantage-2015-data-spotlight-overview-of-plan-changes/ ↩︎
3. Plans that consolidate or withdraw from the certain counties, but remain in others, are counted among the exiting plans for areas where they will no longer offer plans in 2016. ↩︎
4. Gretchen Jacobson, Tricia Neuman, and Anthony Damico. What’s In and What’s Out? Medicare Advantage Market Entries and Exits in 2015. Menlo Park (CA): 2014. https://modern.kff.org/medicare/issue-brief/whats-in-and-whats-out-medicare-advantage-market-entries-and-exits-for-2015/ ↩︎

Geographic variation in Medicare utilization and spending has been a frequent subject of discussion and analysis among researchers and policymakers for many years. Some researchers have suggested that the differences in Medicare spending across geographic areas resulted mainly from differences in practice patterns, which could be addressed by policy interventions, such as changes in financial incentives for providers. Other researchers have emphasized differences in beneficiaries’ health and socioeconomic status as drivers of geographic variation in Medicare spending, which are less amenable to policy intervention than practice patterns.

This paper contributes to the body of research on geographic variation in Medicare spending by analyzing variation in Medicare per capita spending at the county level, using the most current data available (2013); analyzing detailed county-level data on utilization and spending for specific types of services; and examining changes over time from 2007 to 2013 in county-level Medicare per capita spending growth rates. We rank counties based on Medicare per capita spending in 2013 and spending growth rates between 2007 and 2013, and examine characteristics of counties at the top and bottom of the rankings. (A related interactive map shows Medicare per beneficiary spending, and spending growth, in counties across the U.S.) The primary data source for this analysis is the February 2015 update of the Medicare Geographic Variation Public Use File (GV PUF) from the Centers for Medicare & Medicaid Services (CMS).

Key Findings from this Analysis

Geographic Variation in 2013 Medicare Per Capita Spending

• Unadjusted Medicare per capita spending averaged $9,415 in 2013, but was nearly two times greater in the 20 counties with the highest per capita spending ($13,149) than in the 20 counties with the lowest per capita spending ($6,726).
• The 20 counties with the highest unadjusted Medicare per capita spending in 2013 were primarily in northeast, mid-Atlantic, and southern states. Compared to the 20 lowest-spending counties, and the national average, the 20 highest-spending counties have much sicker and poorer beneficiary populations, on average, and a substantially greater share of black and Hispanic beneficiaries.
• Medicare per capita spending on hospital inpatient care is more than twice as high in the highest-spending counties than in the lowest-spending counties, making it by far the most important service category in terms of explaining spending differences between the highest- and lowest-spending counties.
• When we adjust per capita spending in the 20 highest-spending counties and the 20 lowest-spending counties to account for differences in Medicare prices and beneficiaries’ health risk, we find that the gap between the averages narrows substantially—from a 96 percent difference ($13,139 versus $6,726) to a 22 percent difference ($9,344 versus $7,640)—but does not disappear.
• Ranking counties based on price- and health-adjusted spending, we find that 19 of the 20 counties with the highest adjusted per capita spending are in the south, with Texas and Louisiana together accounting for 14 of the 20 counties. The 20 highest-spending counties stand out for having significantly more post-acute care providers per capita and significantly fewer physicians than the 20 lowest-spending counties.

Geographic Variation in Medicare Per Capita Spending Growth Rates, 2007-2013

• The average annual rate of growth in unadjusted Medicare per capita spending between 2007 and 2013 was 2.2 percent nationwide, but ranged from -0.9 percent among the 20 counties with the lowest spending growth rate to 4.6 percent among the 20 counties with the highest spending growth rate.
• Fifteen of the 20 counties with the lowest spending growth rates are in southern states; the 20 counties with the highest spending growth rates are more geographically dispersed. In the counties with the lowest spending growth rates, Medicare per capita spending for hospital services, home health care, and durable medical equipment fell from 2007 to 2013.

Change in Geographic Variation, 2007-2013

• Counties with relatively high unadjusted Medicare per capita spending in 2007 tended to experience relatively low spending growth between 2007 and 2013, and vice versa. But counties at the top of the ranking of unadjusted Medicare per capita spending have tended to remain at the top over time.
• The amount of geographic variation in unadjusted Medicare spending—as measured by the coefficient of variation—began to decline after 2009, indicating a modest narrowing of geographic variation in recent years.

Our analysis shows that geographic variation in Medicare per capita spending persists, although the gap between the highest- and lowest-spending counties appears to have narrowed since 2009. Recent activities, including new efforts to change how providers deliver care and how Medicare pays for it, may be helping to curb Medicare spending in many parts of the country, including areas with some of the more notable excesses in spending. The Affordable Care Act included a number of provisions designed to encourage greater efficiency in the delivery of care for Medicare beneficiaries by modifying incentives for providers to reduce excess costs and improve quality of care. Yet even with such efforts, deep differences in per capita Medicare spending in different parts of the country remain and are likely to persist due to underlying differences in beneficiary characteristics related to poverty and poor health, along with differences in the prices that Medicare pays for services, that contribute to variations in spending.

In 2009, the physician and author Atul Gawande focused the attention of the health policy community on McAllen, Texas, linking exceptionally high Medicare spending in that city to what he characterized as a profit-oriented “culture of money” among providers.¹  Although research on geographic variation in health care utilization and spending predated that article by many years, Gawande’s article struck a nerve and attracted significant attention among policymakers and researchers.

Gawande’s work built on examinations of differences in Medicare spending per capita among hospital referral regions (HRRs)²  by researchers at Dartmouth University. Dartmouth researchers concluded that such differences could not be explained by differences in health status, and that “increased Medicare spending in high-cost regions provides no important benefits in terms of survival.”³  The Dartmouth researchers suggested that the differences in spending across HRRs resulted mainly from differences in practice patterns, which could be addressed by policy interventions, such as changes in financial incentives for providers.⁴ 

In contrast, other researchers have emphasized differences in beneficiaries’ health status as drivers of geographic variation in Medicare spending. Reschovsky et al. found that differences in disease burden were largely responsible for geographic variation in Medicare spending, based on their analysis of data from 60 communities.⁵  Similarly, Zuckerman et al. found that beneficiary demographics and health status help to explain geographic differences in Medicare spending, but also found that even after adjusting for other possibly relevant factors, such as provider supply measures, significant unexplained differences remained.⁶  Sheiner also concluded, based on a state-level analysis, that much of the geographic variation in Medicare spending can be explained by differences in health status and demographics, which are less amenable to policy intervention than practice patterns.⁷ 

In 2009, Congress directed the Institute of Medicine (IOM) to conduct a series of studies on geographic variation in Medicare spending and in the broader health care system. The IOM documented significant variation in spending even within high-spending and low-spending areas; showed that much of the geographic variation in Medicare spending is attributable to differences in post-acute care spending; and ultimately recommended against changes in payment policy designed primarily to reduce geographic variation in spending.⁸  The IOM expressed some concern that reductions in payments to providers in high-spending areas could inadvertently penalize providers practicing appropriately who happened to work in high-fraud areas.

This paper contributes to the body of research on geographic variation in Medicare spending in three ways. First, we analyze variation in Medicare per capita spending at the county level, rather than at the state or HRR level, using the most current data available (through 2013). Second, we analyze detailed data on utilization and spending for specific types of services in our comparisons of high- versus low-spending counties. Third, we examine changes over time from 2007 to 2013 in county-level Medicare per capita spending to compare counties with high versus low rates of growth, and to assess whether geographic variation in per capita spending is increasing or decreasing.

Our analysis addresses the following questions:

• How does Medicare per capita spending vary by county in 2013, and what are the characteristics of the counties with the highest and lowest Medicare per capita spending? How does the amount of variation, and the characteristics of high- versus low-spending counties, differ if rankings are based on unadjusted Medicare spending versus Medicare spending adjusted for differences in prices and beneficiary health status?
• How much variation exists across counties in the rate of growth in Medicare per capita spending between 2007 and 2013, and what are the characteristics of the counties with the highest and lowest per capita spending growth rates?
• Did the amount of geographic variation in Medicare per capita spending increase or decrease from 2007 to 2013?

The primary data source for this analysis is the February 2015 update of the Medicare Geographic Variation Public Use File (GV PUF) from the Centers for Medicare & Medicaid Services (CMS).⁹  We analyzed data at the county level (the most-granular level available), focusing on 2007 and 2013, the earliest and latest years in the 2015 GV PUF. The GV PUF reports spending and beneficiary characteristics only among Medicare beneficiaries enrolled in both Parts A and B and in traditional Medicare (i.e., excluding beneficiaries enrolled in a private Medicare Advantage plan). To avoid mistakenly identifying counties as high- or low-spending based on a few individual outliers, our analysis included only the 736 counties with an average of 10,000 or more traditional Medicare beneficiaries from 2007 through 2013. The national averages discussed in our analysis include all counties, regardless of the number of beneficiaries in any given year.

We ranked the 736 counties in our analysis using three different spending measures. The first measure is unadjusted Medicare per capita spending in 2013, which enables us to show how counties rank on their actual Medicare per capita spending levels, including the effects of differences in prices and health risk. (In this context, ‘price’ refers to Medicare payment rates for services, which are set based on formulas that take into account differences in local wages and certain provider characteristics such as add-on payments for teaching hospitals.) The second measure is price and health-risk adjusted Medicare per capita spending in 2013, which enables us to show how counties rank when differences in prices and health risk are factored out. The third measure is the average annual rate of growth from 2007 to 2013 in unadjusted Medicare per capita spending.

For each of these three spending measures, we identified and compared the counties at the top and bottom of the rankings. For ease of display, we focus on comparisons of the 20 top and bottom counties. We tested whether our results were sensitive to the selection of 20 counties in each group by replicating our analysis for groups of 50 counties; the results were not appreciably different. Our analysis compares beneficiary-weighted averages across these groups of 20 counties for Medicare beneficiary characteristics, health care provider supply measures, and service spending and use, as well as the county-level poverty rate:

• Beneficiary characteristics include percent black, percent Hispanic, percent eligible for both Medicare and Medicaid, and health risk scores (“hierarchical condition categories,” or HCCs) from the GV PUF; the share of traditional Medicare beneficiaries with five or more chronic conditions, based on our analysis of a five percent sample of Medicare claims from the 2013 CMS Chronic Conditions Data Warehouse (CCW); and the percent of county residents ages 65 and over living in poverty, based on our analysis of American Community Survey (ACS) 2008-2012 five-year pooled data from the 2013-2014 Area Health Resources File (AHRF); and the percent of beneficiaries living in metropolitan areas, based on our analysis of the AHRF.
• Measures of health care provider supply are reported per 10,000 county residents and include the number of physicians, primary care physicians as a percent of all physicians, hospital beds, skilled nursing facility beds, home health agencies, ambulatory surgical centers, and hospices. Supply measures are from our analysis of the AHRF for various years.
• Measures of spending and utilization from the GV PUF include the share of beneficiaries using specific Medicare-covered services, spending per capita and per user, and event counts (days, visits, procedures) per 1,000 beneficiaries.
• County-level poverty from the AHRF is the poverty rate among county residents of all ages.

To examine whether geographic variation in county-level Medicare per capita spending increased or decreased between 2007 and 2013, we calculated the “coefficient of variation” (COV) in county-level spending. This measure is useful for measuring trends in spending variation, while adjusting for inflation.

Our analyses use the entire population of U.S. counties and the entire population of Medicare fee-for-service beneficiaries within those counties, rather than a random sample. Although conventional tests of statistical significance are not necessary in this situation, we present results from statistical significance testing for the comparisons of county-level averages in the appendix tables. For more details on the data and methods used in this analysis, see Appendix 1: Data and Methods.

How does Medicare per capita spending vary by county in 2013, and what are the characteristics of the counties with the highest and lowest Medicare per capita spending?

Unadjusted Medicare per capita spending in 2013 averaged $9,415 nationwide, but was nearly two times greater, on average, in the 20 counties with the highest per capita spending ($13,149) than in the 20 counties with the lowest per capita spending ($6,726) (Figure 1). Unadjusted Medicare per capita spending in 2013 ranged from a low of around $6,000 in Josephine County, Oregon to a high of more than $16,000 in Miami-Dade County, Florida (Appendix 2: Table 1).

Most (13 of 20) of the counties with the highest unadjusted Medicare per capita spending in 2013 are in states in the northeastern and mid-Atlantic regions of the country (NY, MD, NJ, PA, CT, MA), five are in Southern states (FL, TX, LA), and the remaining two are in California and Michigan (Figure 2).¹⁰  In contrast, most (14 of 20) of the counties with the lowest per capita spending are in western states (OR, CO, NM, HI, MT, WA). The 20 highest-spending counties included substantially larger numbers of Medicare beneficiaries than the 20 lowest-spending counties (totaling 4.6 million versus 735,000 in 2013).

The highest-spending counties differ from the lowest-spending counties on several dimensions, including beneficiary health status, income, race and ethnicity, and measures of provider supply (Figure 3; Appendix 2: Table 2). For example, 42.5 percent of beneficiaries in the highest-spending counties have five or more chronic conditions, almost double the rate among beneficiaries in the lowest-spending counties (23.4%), and higher than the national average (34.0%). The average poverty rate among people ages 65 and older is nearly two times greater in the 20 highest-spending counties (14.7%) than in the 20 lowest-spending counties (7.8%), and higher than the national average (9.8%). Relatively high poverty rates in the highest-spending counties are reflected in the larger share of beneficiaries dually eligible for Medicare and Medicaid: 34.9 percent in the 20 counties with the highest Medicare per capita spending compared to 16.5 percent in the 20 counties with the lowest per capita spending. A much larger share of beneficiaries in the highest-spending counties are black (19.1%) than in the lowest-spending counties (1.0%) or nationally (9.8%). Similarly, Hispanic beneficiaries account for 17.8% percent of beneficiaries in the highest-spending counties, but just 7.1 percent of beneficiaries in the lowest-spending counties and 6.0 percent nationally.

Counties with higher unadjusted Medicare per capita spending in 2013 typically had a larger supply of various types of providers than the lowest-spending counties (Appendix 2: Table 2). Notably, the 20 highest-spending counties had 33.7 doctors per 10,000 county residents in 2012 (the most recent year available), 9.7 more than the national average and 9.3 more than in the lowest-spending counties. The highest-spending counties also had more hospital beds and home health agencies per 10,000 county residents compared to both the national average and the lowest-spending counties, but a smaller number of hospices and ambulatory surgical centers.

The 20 highest- and lowest-spending counties differed substantially in spending per capita for specific Medicare-covered services in 2013 (Figure 4; Appendix 2: Table 3). By far the most important service category in terms of explaining spending differences between the highest- and lowest-spending counties is hospital inpatient care—which is perhaps not surprising, given that inpatient spending is among the most costly types of Medicare-covered service both on a per-capita and a per-user basis. Average spending per capita on hospital inpatient care in 2013 was more than twice as high in the highest-spending counties than in the lowest-spending counties ($4,914 versus $2,335). The difference in hospital inpatient spending is due to a combination of a larger share of the beneficiary population using hospital inpatient services, higher prices paid, and greater quantity and intensity of services received by inpatient users in the 20 highest-spending counties. For example, in the highest-spending counties, an average of 19.2 percent of traditional Medicare beneficiaries had a hospital inpatient stay, compared to the national average of 17.5 percent and 13.2 percent in the 20 lowest-spending counties. Hospital inpatient use was also higher in the 20 highest-spending counties, averaging 2,081 days per 1,000 beneficiaries, compared to 1,530 nationally and 988 in the 20 lowest-spending counties.

The 20 highest-spending counties in 2013 also had substantially higher spending and use for post-acute care (skilled nursing facility (SNF) and home health care services) relative to the national average and the 20 lowest-spending counties. SNF spending per capita in 2013 averaged $1,117 in the 20 highest-spending counties, 44 percent higher than the national average ($774) and 140 percent higher than in the 20 lowest-spending counties ($466). The per capita spending differences were even larger for home health services ($962, $489, and $192, respectively). The percent of traditional Medicare beneficiaries using SNF in the 20 highest-spending counties in 2013 (5.7%) was higher than national average (5.1%) and higher than in the lowest-spending counties (3.3%). The differences in use rates for home health are even more striking: 13.8 percent of beneficiaries used home health services in the 20 highest-spending counties in 2013, compared to a national average of 9.4 percent, and just 4.7 percent in the 20 lowest-spending counties. The number of SNF covered days per 1,000 beneficiaries was significantly higher in the 20 highest-spending counties (2,429) than the national average (1,887) or the average for 20 lowest-spending counties (1,093). Home health visits per 1,000 beneficiaries averaged 6,207 in the highest-spending counties—six times more than in the lowest-spending counties (1,019) and twice the national average (3,062).

Taken together, these findings suggest that higher unadjusted county-level Medicare per capita spending is partly driven by having a traditional Medicare beneficiary population that is poorer and sicker than average and that uses hospital inpatient services and post-acute care at higher rates and with greater intensity than beneficiaries in lower-spending counties. Counties with relatively high Medicare per capita spending also have a larger supply of certain health care providers than lower-spending counties, which may be related to having a sicker beneficiary population.

How does the amount of variation, and the characteristics of high- versus low-spending counties, differ if rankings are based on unadjusted Medicare spending versus Medicare spending adjusted for differences in prices and beneficiary health status?

Ranking counties based on their actual (unadjusted) Medicare per capita spending reveals the counties where Medicare spends the most and the least, but these spending amounts reflect both the prices that Medicare pays for services at the local level and the health status of beneficiaries living in each county. When we adjust for these price and health-risk differentials and compare the average adjusted per capita spending amounts between the 20 highest-spending counties and the 20 lowest-spending counties, we find that the gap between the averages narrows substantially—from a 96 percent difference ($13,139 versus $6,726) to a 22 percent difference ($9,344 versus $7,640)—but does not disappear (Figure 5). This reduction in the gap between the average spending amounts based on adjusted per capita spending makes sense, since the adjustment mitigates two of the factors (price and health risk) that contribute to county-level spending variation.

Another way of exploring the effect of price and health-risk adjustments is to examine the ranking of counties based on adjusted Medicare per capita spending, which produces a different set of counties at the top and bottom of the rankings than ranking based on unadjusted per capita spending. Based on this ranking, 19 of the 20 counties with the highest adjusted Medicare per capita spending are located in southern states (TX, LA, FL, OK, AL), with 9 counties located in Texas alone, and only one county located in a non-southern state (OH) (Figure 6; Appendix 2: Table 4). In contrast, a majority (17 out of 20) of the lowest spending counties based on price and risk-adjusted per capita spending are located in western states (CA, CO, HI, NM, OR), including 9 counties in California alone; the remaining 3 counties are in AK and NY.

Based on the county rankings by price and health-risk adjusted spending, we find that some demographic differences between the 20 counties at the top and bottom of the ranking, but those differences are not as large as the differences between the counties ranked by unadjusted per capita spending. The 20 highest adjusted spending counties have a somewhat sicker beneficiary population (HCC score of 1.11 versus 0.96), and a larger share of black (8.9% versus 6.8%) and Hispanic (15.8% versus 10.6%) beneficiaries, but a smaller share of beneficiaries eligible for both Medicare and Medicaid (25.3% versus 28.9%) (Appendix 2: Table 5).

In terms of provider supply, the 20 counties with the highest adjusted per capita Medicare spending had 26.5 percent fewer physicians per 10,000 residents than the 20 counties with the lowest adjusted spending, but a larger supply of hospital beds and ambulatory surgical centers, as well as more post-acute providers, including SNF beds, home health agencies, and hospices. Higher adjusted Medicare per capita spending in the top 20 adjusted spending counties could reflect what the Dartmouth researchers refer to as “practice patterns” related to having a larger supply of certain types of providers, or it could reflect higher levels of demand related to having somewhat sicker beneficiary populations.

How much variation exists across counties in the rate of growth in Medicare per capita spending between 2007 and 2013, and what are the characteristics of the counties with the highest and lowest per capita spending growth rates?

The annual rate of growth in Medicare per capita spending between 2007 and 2013 averaged 2.2 percent nationally, and ranged from -0.9 percent, on average, among the 20 counties with the lowest spending growth rate (i.e., a decline in nominal per capita spending) to 4.6 percent, on average, among the 20 counties with the highest spending growth rate (Figure 7).

Between 2007 and 2013, five counties experienced negative average annual growth in per capita spending: Miami-Dade County, FL (-1.84%), Hidalgo County, TX (-1.70%); Wilson County, TN (-0.35%), Orange County, NC (-0.17%), and Walker County, AL (-0.08%). Fifteen of the 20 counties with the lowest spending growth rates are in southern states, while the 20 counties with the highest spending growth rates are more geographically dispersed (Figure 8; Appendix 2: Table 6). None of the 20 counties with the highest spending growth rates between 2007 and 2013 were among the 20 counties with the highest per capita spending amounts in 2013; similarly, none of the 20 counties with the lowest spending growth rates between 2007 and 2013 were among the 20 counties with the lowest per capita spending amounts in 2013.

Unlike the 20 highest-spending and lowest-spending counties, where a comparison reveals several differences in beneficiary characteristics in 2013, there are few notable differences in health status or demographics between the counties with the highest spending growth rates and lowest spending growth rates (Appendix 2: Table 7). The 20 counties with the highest spending growth rates had somewhat fewer Medicare beneficiaries in 2013 than counties with the lowest spending growth rates (totaling 658,000 versus 994,000), but these two sets of counties were similar in terms of average HCC scores, the share of black beneficiaries, and the share of beneficiaries who were eligible for both Medicare and Medicaid in 2013. The only demographic characteristic that differed between counties with the highest and lowest spending growth rates was that Hispanic beneficiaries represented a much larger share of beneficiaries in the lowest spending growth counties, on average, in 2013 (32.4% versus 4.4%); this is likely due to the fact that Miami-Dade County and a handful of counties in Texas with much higher-than-average shares of Hispanic beneficiaries were among the 20 counties with the lowest Medicare per capita spending growth rates across these years.

We also did not observe major changes in beneficiary demographics between 2007 and 2013 that would suggest that such changes played a large role in the rate of Medicare per capita spending growth in these counties. We did observe a difference in the average annual rate of growth in the overall poverty rate among county residents of all ages, which increased at an average annual rate of 4.4 percent in the 20 counties with the highest spending growth rates, compared to 2.3 percent in the 20 counties with the lowest spending growth rates. Average annual growth in the share of black, Hispanic, and Medicare-Medicaid eligible beneficiaries was slightly higher in the highest spending growth counties than in the lowest spending growth counties, while Medicare Advantage penetration increased at the same average annual rate in both sets of counties.

We examined changes in the rate of growth in measures of provider supply to assess whether counties with higher rates of Medicare per capita spending growth also had faster growth in provider supply compared to counties with lower spending growth rates (Appendix 2: Table 7). Contrary to expectations, we found no discernable relationship. For example, in the 20 counties with the fastest growth in Medicare per capita spending, the average number of physicians per 10,000 county residents grew slightly at an average annual growth rate of 0.4% (from 21.0 in 2007 to 21.5 in 2012, the latest year available), and was unchanged in the 20 slowest-growing counties (27.4 in both 2007 and 2012). For another example, we observed a decline over time in the number of hospital beds and skilled nursing facility beds per 10,000 county residents in both the 20 counties with the highest spending growth rates and the 20 counties with the lowest spending growth rates.

Although we did not observe major distinctions between high-spending growth and low-spending growth counties in terms of changes in beneficiary demographics and provider supply measures, we did observe notable differences between these groups of counties in the rate of change in service spending and use between 2007 and 2013. Three service categories account for much of the difference in spending growth between the counties with the highest and lowest spending growth rates: hospital inpatient, home health, and durable medical equipment (DME) (Figure 9; Appendix 2: Table 8).

In the 20 highest spending growth counties, average hospital spending per capita increased by 4.0 percent between 2007 and 2013, while it decreased by 0.4 percent in the 20 lowest spending growth counties. Readmission rates increased by 0.2 percent, on average, in the highest spending growth counties, but decreased by 0.8 percent in lowest spending growth counties. The difference in growth in spending on home health services was even more striking: increasing at an average annual rate of 5.7 percent in the highest spending growth counties, but decreasing at an average annual rate of 4.4 percent in the lowest spending growth counties. DME spending per capita fell in both the highest and lowest spending growth counties, but the average annual decrease was much smaller in the former set of counties than the latter (-1.2% versus -13.6%).

Compared to the 20 lowest spending growth counties, the 20 counties with the highest spending growth experienced higher average annual growth in the share of beneficiaries using home health services (3.3% versus 1.0%), and a smaller average annual decrease in the share of beneficiaries using hospital inpatient services (-1.3% versus -2.9%). Conversely, the 20 lowest spending growth counties experienced a reduction in the rate of growth in users of DME services, compared to a flat rate of growth in the highest spending growth counties (-1.8% versus 0.0%).

Did the amount of geographic variation in Medicare per capita spending increase or decrease from 2007 to 2013?

Our analysis shows that, in general, counties with relatively high Medicare per capita spending in 2007 tended to experience relatively low spending growth between 2007 and 2013, and vice versa (Figure 10). But although county-level 2007 Medicare per capita spending is negatively related to spending growth over time, high levels of Medicare per capita spending by county have tended to persist over time. Of the 20 counties with the highest per capita spending in 2013, all but two were among the 20 highest-spending counties in 2007 (Appendix 2: Table 9).

To further explore the question of whether geographic variation narrowed or widened over these years, we measured the beneficiary-weighted county-level coefficient of variation for each year from 2007 through 2013 for two spending measures: unadjusted per capita spending, and price and health-risk adjusted per capita spending, including all counties (not just counties with 10,000 or more beneficiaries). We found that the coefficient of variation for unadjusted per capita spending increased slightly from 2007 (0.141) to 2009 (0.143), and then decreased each year thereafter, falling to 0.125 in 2013, a 13 percent decline (Figure 11; Appendix 2: Table 10). The coefficient of variation in adjusted per capita spending followed a similar trend, rising initially and then falling by 15 percent from 2009 to 2013 (from 0.078 to 0.066). We also examined the coefficient of variation by type of service, and found that declines in geographic variation were particularly pronounced for three service categories: home health, durable medical equipment, and hospice. This convergence, or narrowing of variation, in county-level Medicare per capita spending in recent years represents the continuation of a trend since the 1970s toward a reduction in geographic variation in Medicare per capita spending.¹¹ 

Previous research on geographic variation in Medicare spending suggested that wide variation in per capita spending was driven by differences in the supply of providers combined with an inappropriate profit orientation in some areas, and that the gap between high-spending and low-spending areas could be narrowed by policies that encourage providers in high-spending areas to behave more like providers in low-spending areas. More recent research has shown that the areas of the country with relatively high unadjusted Medicare per capita spending are marked by a confluence of poor health, high rates of poverty, and high prices. Thus, changing provider practice patterns may help to curtail spending growth and reduce variation in spending across counties but will not eliminate the abiding socioeconomic, demographic, and health disparities between the highest- and lowest-spending counties.

At the same time, our analysis shows that differences in county-level spending remain even after adjusting for differences in prices and beneficiary health status that affect Medicare per capita spending. In light of our finding that counties with the highest price- and health-risk adjusted per capita spending have a larger supply of certain types of providers, including post-acute care providers, the question remains whether higher spending in these areas is driven by medical practice styles or by demand for care from a relatively sicker beneficiary population, or some combination of both. Further research is needed to understand this relationship.

In recent years, policies have been implemented within Medicare that may have helped to curb some of the more notable excesses in high-spending areas of the country, and new efforts are underway in the program to change how providers deliver care and how Medicare pays for it. These activities include: an increased focus on program integrity in Medicare, with a special emphasis on high-fraud regions and services; a new competitive bidding program for durable medical equipment; a hospital readmission reduction program; the Medicare Shared Savings Program; and bundled payments for episodes of care. It is beyond the scope of this paper to quantify the extent to which these policies have affected geographic variation in Medicare per beneficiary spending, but our findings suggest that they may have played some role in the narrowing of variation.

The Affordable Care Act included a number of provisions designed to encourage greater efficiency in the delivery of care for Medicare beneficiaries by modifying incentives for providers to reduce excess costs and improve quality of care. Some of these initiatives may help to constrain the growth in per capita spending in all areas and further reduce the variation between the highest- and lowest-spending areas, but some regional differences are likely to persist due to profound differences in the health and socioeconomic characteristics of the Medicare population across the county, along with differences in the prices Medicare pays for services.

The primary data source for this analysis is the February 2015 update of the Medicare Geographic Variation Public Use File (GV PUF) from the Centers for Medicare & Medicaid Services (CMS).¹²  The GV PUF includes data for the U.S. overall and for three different geographic levels: state, county, and hospital referral region (HRR). We analyzed data at the county level because that is the most-granular level available. Our analysis focuses on the years 2007 and 2013, the earliest and latest years, respectively, in the 2015 GV PUF.

The GV PUF reports spending and beneficiary characteristics only among Medicare beneficiaries enrolled in both Parts A and B (i.e., not just one or the other), and in the traditional Medicare program (i.e., excluding beneficiaries enrolled in a private Medicare Advantage plan). Because of increases in Medicare Advantage enrollment, the share of Medicare beneficiaries included in the GV PUF study population declined from 71 percent in 2007 (33.0 million out of 46.7 million) to 62 percent in 2013 (34.3 million out of 55.2 million).

The GV PUF include three types of spending measures: “actual costs” (unadjusted Medicare payments, excluding beneficiary cost sharing and third-party payments), “standardized costs” (that is, a simulated measure of costs calculated by CMS using a single national price schedule that reflects the quantity and intensity of services but does not include market- or provider-level price adjustments), and “standardized risk-adjusted costs” (that is, standardized costs adjusted for differences in health risk by dividing by the HCC (hierarchical condition categories) score). (In this paper, we refer to “standardized costs” as price-adjusted spending, and “standardized risk-adjusted costs” as price- and risk-adjusted spending.) In addition, for each combination of county, year, and service category we created a set of Medicare price indexes, equal to the ratio of actual costs over standardized costs. In a county with prices equal to the national average, the price index equals 1.00. These price indexes reflect the market-level and provider-specific adjustments that are applied in each county in Medicare’s price-setting formulas, such as differences in local wages and certain provider characteristics such as add-on payments for teaching hospitals. In our discussion of spending by service category, we focus on spending per capita since this measure is the product of the percent of beneficiaries using each service, the Medicare price index for that type of service, and the standardized (i.e., price-adjusted) costs per user.

The county-level GV PUF include 3,136 counties, but many of those counties contain relatively few Medicare beneficiaries. Medicare per capita spending in those small counties can vary widely from county to county and from year to year due to random beneficiary-level variation. To avoid mistakenly identifying counties as high- or low-spending based on a few individual outliers, our analysis included only the 736 counties with an average of 10,000 or more traditional Medicare beneficiaries from 2007 through 2013. These 736 counties included 25.9 million traditional Medicare beneficiaries in 2013, which represents 76 percent of the GV PUF population in that year. The national averages discussed in our analysis include all counties, regardless of the number of beneficiaries in any given year.

We merged the county-level GV PUF with county-level measures of population: the poverty rate among county residents of all ages and the supply of health care providers from the Area Health Resources Files (AHRF) produced by the Health Resources and Services Administration (HRSA).¹³  For some measures of population demographics and supply, the most recent data available in the AHRF was prior to 2013. To calculate county-level supply measures, we divided the number of providers (e.g., active physicians) by the total number of county residents (i.e., not just Medicare beneficiaries). To calculate national average measures of supply, we first summed supply and population nationwide, and then calculated supply per capita.

We analyzed county-level data on the share of traditional Medicare beneficiaries with multiple chronic conditions, using a 5 percent sample of claims from the 2013 CMS Chronic Conditions Data Warehouse (CCW). We calculated the percent of beneficiaries living in metropolitan areas, based on our analysis of the AHRF. We also analyzed county-level data from the American Community Survey (ACS) 2008-2012 five-year pooled data from the 2013-2014 AHRF on the share of county residents ages 65 and older living in poverty. We matched variables from these separate data files to the GV PUF using the five-digit Federal Information Processing Standard (FIPS) codes, which uniquely identify counties and county-level equivalent areas in the U.S.

We ranked the 736 counties in our analysis using three different spending measures: 1) unadjusted Medicare per capita spending in 2013, to show how counties rank on their actual Medicare per spending levels, regardless of price and health risk differences; 2) price- and health-risk adjusted Medicare per capita spending in 2013; and 3) the annual rate of growth from 2007 to 2013 in unadjusted Medicare per capita spending.¹⁴  Our measure of county-level price- and health-risk adjusted spending equals price-adjusted spending divided by the mean HCC score.

For each of these three spending measures, we identified the 20 counties at the top and bottom of the rankings, which yielded six sets of 20 counties. We tested whether our results were sensitive to the selection of 20 counties in each group by replicating our analysis for groups of 50 counties. The results were not appreciably different. For each of those six sets of counties, we calculated beneficiary-weighted averages of spending and utilization measures and demographics, which gives greater weight to counties with larger numbers of traditional Medicare beneficiaries.

Our analysis compares these groups of counties in terms of Medicare beneficiary characteristics, health care provider supply measures, spending and utilization measures, and county-level poverty:

• Beneficiary characteristics include percent black, percent Hispanic, percent eligible for both Medicare and Medicaid, and health risk scores (HCCs) from the GV PUF; the percent of traditional Medicare beneficiaries with five or more chronic conditions, based on our analysis of a five percent sample of Medicare claims from the 2013 CCW; and the percent of county residents ages 65 and over living in poverty, based on our analysis of ACS 2008-2012 five-year pooled data from the 2013-2014 AHRF, and the percent of beneficiaries living in metropolitan areas, based on our analysis of the AHRF.
• Measures of health care provider supply are reported per 10,000 county residents and include the number of physicians, primary care physicians as a percent of all physicians, hospital beds, skilled nursing facility beds, home health agencies, ambulatory surgical centers, and hospices. Supply measures are from our analysis of the AHRF for various years.
• Measures of spending and utilization from the GV PUF include the percent of beneficiaries using specific Medicare-covered services (hospital inpatient, outpatient, evaluation & management, procedures, skilled nursing facility, home health, durable medical equipment, and hospice), spending per capita and per user, and event counts (days, visits, procedures) per 1,000 beneficiaries.
• County-level poverty from the AHRF is the poverty rate among county residents of all ages.

We used the “coefficient of variation” (COV) as a summary measure of the amount of geographic variation in county-level Medicare per capita spending to examine whether geographic variation increased or decreased between 2007 and 2013. The COV equals the beneficiary-weighted standard deviation in Medicare per capita spending divided by the beneficiary-weighted mean. We included all 3,136 counties in the COV calculation. The COV is useful for measuring trends in spending variation, while adjusting for inflation.

Our analyses use the entire population of U.S. counties and the entire population of Medicare fee-for-service beneficiaries within those counties, rather than a random sample. Although conventional tests of statistical significance are not necessary in this situation, we present results from statistical significance testing for the comparisons of county-level averages in the appendix tables. For these tests, we used the TTEST procedure in SAS, weighted by the number of beneficiaries and assuming unequal variance.

1. A. Gawande, “The Cost Conundrum: What a Texas Town can Teach Us about Health Care,” New Yorker (June 2009); available at http://www.newyorker.com/magazine/2009/06/01/the-cost-conundrum. ↩︎
2. Hospital referral regions (HRRs) are regional markets for tertiary medical care including at least one hospital performing major cardiovascular procedures and neurosurgery. See “Appendix on the Geography of Health Care in the United States” in Center for the Evaluative Clinical Sciences Dartmouth Medical School, The Quality of Medical Care in the United States: A Report on the Medicare Program, 1999; available at http://www.dartmouthatlas.org/atlases/99Atlas.pdf. ↩︎
3. E.S. Fisher, D.E. Wennberg, T.A. Stukel, D.J. Gottlieb, F.L. Lucas, E.L. Pinder, “The Implications of Regional Variations in Medicare Spending. Part 2: Health Outcomes and Satisfaction with Care,” Annals of Internal Medicine 138, no. 4 (February 18 2003): 288-298. ↩︎
4. J. Skinner, E.S. Fisher, Reflections on Geographic Variations in U.S. Health Care, May 12 2010; available at http://www.dartmouthatlas.org/downloads/press/Skinner_Fisher_DA_05_10.pdf. ↩︎
5. J.D. Reschovsky, J. Hadley, P.S. Romano, “Geographic Variation in Fee-for-Service Medicare Beneficiaries’ Medical Costs Is Largely Explained by Disease Burden,” Medical Care Research and Review OnlineFirst (May 28 2013); available at http://mcr.sagepub.com/content/early/2013/04/26/1077558713487771.abstract. ↩︎
6. S. Zuckerman, T. Waidmann, R. Berenson, J. Hadley, “Clarifying Sources of Geographic Differences in Medicare Spending,” New England Journal of Medicine 363, no. 1 (July 1 2010): 54-62. ↩︎
7. L. Sheiner, “Why the Geographic Variation in Health Care Spending Cannot Tell Us Much about the Efficiency or Quality of Our Health Care System,” Brookings Papers on Economic Activity, Vol. 2014, No. 2, 2014, pp. 1-72. ↩︎
8. J.P. Newhouse, A.M. Garber, R.P. Graham, M.A. McCoy, M. Mancher, A. Kibria, et al., editors. Variation in Health Care Spending: Target Decision Making, Not Geography. Washington (DC): National Academies Press, 2013. ↩︎
9. Centers for Medicare & Medicaid Services, Medicare Data for the Geographic Variation Public Use File: A Methodological Overview, February 2015; available at http://www.cms.gov/Research-Statistics-Data-and-Systems/Statistics-Trends-and-Reports/Medicare-Geographic-Variation/Downloads/Geo_Var_PUF_Methods_Paper.pdf. ↩︎
10. The top 20 counties are different when ranked by total, rather than per capita, Medicare spending in 2013. Nine counties appear on both lists, but some counties—most likely by virtue of the number of traditional Medicare beneficiaries residing in the county (e.g., Cook County, Illinois; Maricopa County, Arizona; and San Diego County, California)—appear on the list of 20 highest-spending counties by total spending but not by per capita spending. ↩︎
11. See Figure 4 in Congressional Budget Office, Geographic Variation in Health Care Spending, February 2008, Pub. No. 2978; available at http://www.cbo.gov/ftpdocs/89xx/doc8972/02-15-GeogHealth.pdf. ↩︎
12. Centers for Medicare & Medicaid Services, Medicare Data for the Geographic Variation Public Use File: A Methodological Overview, February 2015; available at http://www.cms.gov/Research-Statistics-Data-and-Systems/Statistics-Trends-and-Reports/Medicare-Geographic-Variation/Downloads/Geo_Var_PUF_Methods_Paper.pdf. ↩︎
13. Health Resources and Services Administration, Technical and User Documentation for the 2013-2014 County Area Health Resources Files, 2015; available at http://datawarehouse.hrsa.gov/DataDownload/ARF/AHRF_USER_TECH_2013-2014.zip. ↩︎
14. We calculated the annual growth rate in x from 2007 to 2013 as   ↩︎

Since 2006, Medicare beneficiaries have had access through Medicare Part D to prescription drug coverage offered by private plans, either stand-alone prescription drug plans (PDPs) or Medicare Advantage prescription drug plans (MA-PD plans). Now in its tenth year, Part D has evolved due to changes in the private plan marketplace and the laws and regulations that govern the program. This report presents findings from an analysis of the Medicare Part D marketplace in 2015 and changes in features of the drug benefit offered by Part D plans since 2006. Key findings are summarized below.

Key Findings

Part D Enrollment and Plan Availability

In 2015, more than 39 million Medicare beneficiaries are enrolled in Medicare drug plans. Since 2006, the share of Medicare beneficiaries enrolled in a Part D plan has increased from 53 percent to 72 percent of all eligible Medicare beneficiaries.

• The majority (61 percent) of Part D enrollees are in PDPs, but enrollment in MA-PD plans is growing more rapidly. MA-PD plan enrollment accounts for two-thirds of the net increase in Part D enrollment from 2014 to 2015, and the share of MA-PD enrollment has grown over the past decade from 28 percent to 39 percent.
• In 2015, 6.6 million Medicare beneficiaries are enrolled in an employer-only Part D plan designed solely for retirees of a former employer. Enrollment in employer-only Part D plans has quadrupled since 2006.

The average Part D enrollee had a choice of 30 PDPs and 15 MA-PD plans in 2015.

• The average number of PDPs offered to enrollees has dropped from a high of 55 plans in 2007 to 30 plans in 2015. Between 2014 and 2015, the average number of PDPs offered to enrollees dropped from 35 to 30 plans, while the number of MA-PD plans per enrollee increased slightly from 14 to 15 plans.

In 2015, three Part D sponsors account for half of all Part D PDP and MA-PD enrollees.

• UnitedHealth, Humana, and CVS Health have enrolled half of all participants in Part D. This level of market concentration is relatively unchanged since 2006. UnitedHealth and Humana have topped enrollment since the program began, while enrollment in CVS Health has grown through acquisition of other plan sponsors. UnitedHealth, by itself, has maintained the top position for all ten years of the program, and in 2015 provides coverage to more than one in five PDP and MA-PD enrollees.

Part D Premiums

Average monthly PDP premiums have been essentially flat since 2010; MA-PD premiums have risen modestly in the past few years.

• PDP enrollees pay $37.02 per month, on average, in 2015, whereas enrollees in MA-PD plans pay only $17.29.The combined average for PDP and MA-PD plan enrollees is $30.02 in 2015.

Premiums for some of the most popular plans increased in 2015, while for others, premiums fell.

• Humana’s Walmart Rx PDP, which was new in 2014, raised its premiums by 24 percent (an average increase of about $3 per month) in 2015. By contrast, the SilverScript Choice PDP lowered its premium by 21 percent (an average decrease of about $6 per month) in 2015.

PDP premiums vary widely across plans and across regions.

• Premiums vary for plans with equivalent benefits, ranging from $12.60 to $101.40 per month for plans offering the basic Part D benefit.
• One source of variation is geographic differences. Average PDP premium for plans with the basic benefit are $16.97 in New Mexico and $36.90 in the Idaho/Utah region.

Part D Benefit Design and Cost Sharing

In 2015, most plans use five cost-sharing tiers: preferred and non-preferred tiers for generic drugs, preferred and non-preferred tiers for brand drugs, and a tier for specialty drugs.

• The vast majority of all Part D enrollees (80 percent of PDP enrollees and 91 percent of MA-PD enrollees) are in plans that use five cost-sharing tiers, a design that gained popularity starting in 2012.

Cost sharing for brand-name drugs has been relatively stable in recent years, but has risen substantially since the start of Part D; MA-PD plan enrollees generally pay somewhat higher cost sharing than PDP enrollees.

• Cost sharing for brands increased between 2006 and 2015 by about 36 percent for beneficiaries enrolled in PDPs and by nearly 70 percent for those in MA-PD plans. In the same period, cost sharing for generics decreased.
• On average, MA-PD plan enrollees pay somewhat higher cost sharing for their drugs than PDP enrollees, particularly for brand-name drugs. For example, median cost sharing for preferred and non-preferred brands in MA-PD plans is $45 and $95, respectively, compared to $38 and $80 in PDPs.
• Copayments for brand-name drugs are higher than those typically charged by large employer plans, while copayments for generics are lower.
• In 2015, many PDP enrollees are in plans that charge coinsurance instead of copayments: 28 percent of enrollees face copayments for preferred brand drugs and 63 percent for non-preferred brand drugs.

Nearly all Part D plans use specialty tiers for high-cost drugs and charge coinsurance of 25 percent to 33 percent during the benefit’s initial coverage period.

• Nearly half of all PDP enrollees and nearly three quarters of all MA-PD plan enrollees are in plans that charge the maximum 33 percent for these high-cost drugs.

The use of tiered pharmacy networks has grown rapidly in recent years and is now the norm in PDPs.

• The share of Part D stand-alone drug plans with tiered pharmacy networks grew from 7 percent in 2011 to 87 percent in 2015. Enrollees in these plans pay lower cost sharing if they use pharmacies offering preferred cost sharing and higher cost sharing if they use other pharmacies.

Low-Income Subsidy Program

• About 11.7 million Part D enrollees (30 percent) receive extra help through the Part D Low-Income Subsidy (LIS), a majority of whom (8 million) are enrolled in stand-alone PDPs. The subsidy reduces cost sharing and pays their drug plan premiums, as long as they enroll in PDPs designated as benchmark plans.
• In 2015, 15 percent of LIS beneficiaries enrolled in PDPs (1.2 million) are paying monthly premiums, and of this group, three-fourths are paying $10 or more per month. In addition, 381,000 LIS beneficiaries enrolled in MA-PD plans are paying premiums in 2015. CMS does not reassign these beneficiaries to a zero-premium PDP because they have actively selected the plan they are in.
• On average, LIS beneficiaries paying premiums for their PDPs pay $18.90 per month, well above the average in previous years.
• Part D Plan Performance Ratings

Nearly half of PDP enrollees are in plans with the highest star ratings (4 stars or more).

• Part D plan ratings in 2015 are up considerably from 2014 levels. The share of PDP enrollees in plans with at least four stars out of a maximum five stars rose from 5 percent in 2014 to 48 percent in 2015.

Discussion

Now in its tenth year of operation, the Part D program has experienced relative stability in recent years. The program has had consistently high levels of plan participation, offering dozens of plan choices for beneficiaries in each region and broad access to generic and brand-name drugs. While the Part D program has matured since 2006, the marketplace also changes every year. Plans can and do enter and drop out of the market annually, and enrollees can and do experience changes in premiums, cost sharing for their medications, which drugs are covered by their plan, and which pharmacies they can use without paying higher cost sharing.

Beneath the surface, there are trends that could pose cost and access challenges for Part D enrollees. At a time of heightened public concern about the cost of prescription drugs, median cost sharing for brand-name drugs has increased over the years and more plans impose coinsurance for brand tiers. Although premiums have been flat in recent years, the high cost of newly approved drugs and others in the pipeline could change that trend. In the absence of an absolute limit on out-of-pocket costs, Part D enrollees with high drug use could face growing out-of-pocket costs.

Since 2006, Medicare beneficiaries have had access to prescription drug coverage offered by private plans, either stand-alone prescription drug plans (PDPs) or Medicare Advantage prescription drug plans (MA-PD plans). These Medicare drug plans (also referred to as Part D plans) receive payments from the government to provide Medicare-subsidized drug coverage to enrolled beneficiaries. Part D plans are required to offer a defined standard benefit or one that is equal in value (Exhibit I.1). They may also offer an enhanced benefit. Medicare drug plans must meet defined requirements, but may vary in terms of premiums, benefit design, gap coverage, formularies, and utilization management rules.

In 2015, more than 39 million Medicare beneficiaries are enrolled in Medicare drug plans, including 24 million in PDPs and 15 million in MA-PD plans.^{1 , 2}  About 12 million Part D enrollees are receiving extra help through the Part D Low-Income Subsidy (LIS) program to pay their drug plan premiums and cost sharing. Part D has evolved since its inception in 2006 due to changes in the private plan marketplace and the regulations that govern the program. The 2010 Affordable Care Act (ACA) is bringing significant improvements to the program, primarily phasing out the coverage gap, or “doughnut hole,” in the drug benefit.³  In addition to a 50 percent manufacturer discount on the price of brand-name drugs in the gap, the law further reduces cost sharing for brand-name and generic drugs in the gap over time, reducing cost sharing to the level that applies before the gap and eliminating the coverage gap in 2020. In addition, the Centers for Medicare & Medicaid Services (CMS) has implemented other statutory and regulatory changes that have resulted in some consolidation of Part D plan offerings, along with a degree of greater standardization.

This report presents findings from an analysis of the Medicare Part D marketplace in 2015, the program’s tenth year, and changes in various features of the drug benefit since 2006.⁴  It presents key findings in five different areas:

• Enrollment and plan availability;
• Premiums;
• The design of Part D benefits, including cost sharing, specialty tiers, formularies, utilization management, the coverage gap, and tiered pharmacy networks;
• The Low-Income Subsidy (LIS) program for low-income beneficiaries; and
• Plan performance ratings.

The findings are based on data from CMS for all plans participating in Part D. More detail about the methods used in this analysis is provided on page 53.

Beneficiary Participation in Part D

More than 39 million Medicare beneficiaries are enrolled in a Part D plan, either a PDP or MA-PD in 2015, representing 72 percent of all eligible Medicare beneficiaries. This is an increase of nearly 2 million beneficiaries since 2014 and of nearly 17 million beneficiaries since 2006 when only 53 percent of eligible beneficiaries were enrolled in Part D plans

1

. In 2006, a larger share of Medicare beneficiaries received drug coverage from other sources such as former employers, so that a total of 90 percent of beneficiaries had coverage for their drugs from some source.⁵  In 2012 (the most recent year for which data are available), 88 percent had drug coverage at least comparable to Part D coverage and 12 percent of Medicare beneficiaries had no drug coverage whatsoever.⁶ 

More than half of the ten-year increase comes from additional enrollment in MA-PD plans, which is likely a mix of enrollees new to Part D or to Medicare altogether and those who switched from traditional Medicare supplemented by PDP coverage to MA-PD plans.⁷  The rest of the increase is a mix of higher PDP enrollment and more enrollment in employer-only Part D plans. About two-thirds of the increase from 2014 to 2015 is from enrollment gains for MA-PD plans.

The 4.8 percent enrollment increase from 2014 to 2015 is lower than some earlier years. Growth was higher in 2007 and 2008 as the program ramped up and higher again in 2012 and 2013 as a result of increased enrollment of retirees in employer-only Part D plans. Over ten years, the average annual rate of increase is 6.4 percent.

The share of Medicare beneficiaries with Part D coverage varies by state, ranging from 38 percent (Alaska) to 77 percent (Michigan)

2

. States with the highest shares of Part D enrollment are California, Michigan, New York, and Ohio, each with 75 percent of Medicare beneficiaries in Part D. Alaska, Maryland, Wyoming, and the District of Columbia have fewer than 60 percent of their residents on Medicare in Part D plans. Some of these states have high shares of federal employment; federal retirees get drug coverage outside Part D through the Federal Employees Health Benefits Program.

Nationally, about 61 percent of Part D enrollees are in PDPs; the remaining 39 percent are in MA-PD plans, with considerable variation by state 

2

, Appendix Table 1, Appendix Table 2. PDP enrollment accounted for 72 percent of total enrollment in 2006, but this share has been declining over time as MA-PD plan enrollment has grown more rapidly than PDP enrollment in recent years. From 2006 to 2015, non-employer MA-PD plan enrollment grew by 10.5 percent annually, whereas non-employer PDP enrollment grew by only 2.4 percent annually. The more rapid growth in MA-PD enrollment from 28 percent to 39 percent of Part D enrollees reflects the broader trend of greater enrollment in Medicare Advantage.

PDPs account for 100 percent of Part D enrollees in Alaska and more than 90 percent of enrollees in six other states with low populations (Delaware, New Hampshire, North Dakota, South Dakota, Vermont, and Wyoming). By contrast, MA-PD plans account for half or more of Part D enrollees in five states: Arizona, California, Florida, Hawaii, and Oregon.

Between 2011 and 2013, total enrollment in employer-only Part D plans doubled from 2.9 million to 5.9 million beneficiaries, rising more modestly to 6.6 million in 2015

4

, Appendix Table 2). Total enrollment in employer-only Part D plans in 2015 is more than four times the level in 2006. The biggest increase was between 2012 and 2013, when enrollment in these plans grew 63 percent. The growth rate slowed to less than 1 percent in 2015.

The major impetus for this growth was a provision in the ACA that eliminated the tax deductibility of the 28 percent Retiree Drug Subsidy (RDS), effective in 2013.⁸  This subsidy, paid to employers who provide creditable prescription drug coverage to Medicare beneficiaries, was included in the original Part D legislation to encourage employers to maintain existing drug coverage for their retirees. In 2006, 7.2 million Medicare beneficiaries were covered in retiree health plans that received the RDS (with an average subsidy payment of $527 per person in 2006, rising to a projected $625 in 2015). With the changed tax status of the RDS, enrollment in subsidized retiree plans dropped to 2.7 million in 2014, and the Medicare Trustees project a drop to 2.2 million in 2015 and 0.9 million by 2019.⁹ 

Most employers that no longer elected to receive the subsidy after the change in tax treatment have shifted their retirees to employer-only Part D plans. Most of the new enrollment was in employer-only PDPs. In 2013, enrollment was up 99 percent in employer-only PDPs and just 7 percent in employer-only MA-PD plans. In 2014 and 2015, as growth in employer plans slowed overall, there was a modest shift toward MA-PD plan enrollment.

Continuity of Part D Plan Offerings

About one-fifth of the PDPs that participated in Part D in its first year are still in the market today. There are 315 PDPs operating under the same contract and plan identification number in 2015 as in 2006. About one-fourth of these PDPs operate under the same plan name, while others have similar names. For example, what was originally the AARP MedicareRx Plan, sponsored by UnitedHealth, is now called AARP MedicareRx Preferred. Others have changed as a result of acquisitions; for example, five PDPs offered by Sterling in 2006 are now known as WellCare Simple PDPs. About one-fourth of these continuously operating PDPs have changed their benefit type; in most cases, PDPs originally offering the basic benefit now operate as enhanced benefit PDPs (a change in status no longer allowed by CMS).

These continuously operating PDPs accounted for about 44 percent of all PDP enrollees in 2006, and 55 percent of total enrollment in 2015. Most of the remaining PDP enrollees are in other plans offered by sponsors that have participated in Part D since 2006. Of the six plan sponsors that entered the Part D market after the program’s first year, none have attracted a significant market share. The most successful new entrant has been Envision RxPlus, which held a 2.5 percent share of PDP enrollment in 2013, declining to 2 percent in 2015.

Firm Participation

In 2015, the ten largest sponsors of Part D plans account for more than three-fourths of all enrollees, three firms account for half of all enrollees, and UnitedHealth alone accounts for more than one in five Part D enrollees (including 21 percent of PDP enrollees and 20 percent of MA-PD plan enrollees)

5

.¹⁰  This pattern of a few plan sponsors having a substantial share of Part D enrollment has held over the program’s first ten years. The ten largest Part D plan sponsors in 2015 have enrolled 30.9 million beneficiaries in either a stand-alone PDP or an MA-PD plan

Patterns are similar for the PDP and MA-PD plan markets, viewed separately

7

. In the PDP market alone, the top three sponsors (UnitedHealth, CVS Health, and Humana) account for nearly 60 percent of enrollment in 2015—slightly more than the 54 percent share held by the top three sponsors (UnitedHealth, Humana, and Member Health) in 2006. Among MA-PD plans, UnitedHealth and Humana each have 20 percent of the market in 2015, followed by Kaiser Permanente, with 9 percent; these are the same three sponsors that held the most MA-PD market share in 2006.

The growth pattern among the largest PDP sponsors illustrates some of the different strategies used since the start of the program

8

. UnitedHealth has been the largest PDP sponsor since 2006, and Humana has generally been the second largest. CVS Health has used an acquisition strategy to move solidly into the third position. Aetna and Cigna have moved higher up in recent years, also through acquisitions. It is noteworthy that both acquiring firms in recently proposed mergers (Aetna and Anthem) lost enrollment in the PDP market from 2014 to 2015, whereas both firms to be acquired (Humana and Cigna) gained enrollment.

Seven of the top ten firms in 2015 sponsor both stand-alone PDPs and MA-PD plans. Kaiser Permanente is the only sponsor among the top ten that offers only MA-PD plans, and Anthem is the only other firm with more MA-PD enrollees than PDP enrollees. CVS Health and Express Scripts offer only PDPs. Other than Kaiser Permanente and Anthem, at least 60 percent of each of the top firms’ enrollment is in PDPs.

Enrollment growth since 2006 for CVS Health, Express Scripts, Aetna, and Cigna is due largely to acquisitions of other plan sponsors. CVS Health has used an acquisitions strategy to become the third largest sponsor in the Part D marketplace. The parent company now includes 5 of the 18 firms that had the most enrollees in 2006. Cigna and Aetna have grown their Part D market shares through similar acquisitions strategies. Express Scripts has grown both through its recent acquisition of Medco, but also through the overall increase in enrollment in employer-only plans since the RDS tax status change. Four plan sponsors dominate the employer-only segment of the Part D market, collectively accounting for about two-thirds of all enrollees in employer-only Part D plans: Express Scripts (34 percent), CVS Health (16 percent), UnitedHealth (10 percent), and Kaiser Permanente (7 percent).

National plan sponsors have dominated the Part D PDP market from the start. Firms serving all or most regions hold 94 percent of the PDP market in 2015, up from 89 percent from 2006. Most other sponsors with plans in no more than a few regions are local Blue Cross Blue Shield (BCBS) plans. Among the PDPs serving no more than a few regions, the only non-Blue plan with a PDP market share of greater than 1 percent in a region is the WPS Insurance Company, a not-for-profit plan sponsor with 7 percent of the market in Wisconsin. BCBS plans with large market share in their regions are the coalition of BCBS plans in the seven-state upper Midwest region (23 percent), the Health Care Service Corporation affiliates in Illinois (21 percent) and Oklahoma (14 percent), Arkansas BCBS (14 percent), the coalition of BCBS plans in southern New England (11 percent), and Blue Cross Blue Shield of Kansas (10 percent). Five regions have no PDP sponsored by a BCBS plan.

UnitedHealth and Humana have been the two largest Part D plan sponsors from the start of the program, but their combined share of enrollment has dropped from 45 percent in 2006 to 39 percent in 2015. UnitedHealth, due in part to its successful marketing relationship with AARP, has maintained its top position for all ten years of the program and has seen its enrollment grow by about 46 percent since 2006. Humana has maintained a strong Part D presence, due in part to offering the lowest PDP premiums in 2006 and retaining many of those enrollees over time despite premium increases for its older plans. While higher-than-average premium increases and a loss of LIS benchmark status in most regions contributed to a drop in Humana’s Part D enrollment between 2006 and 2010, Humana’s introduction of new lower-premium PDPs in 2011 and 2014 reversed this decline, contributing to a net enrollment gain of 62 percent in Humana’s Part D plans between 2006 and 2015.

Market Concentration

At the national level, the Part D market among PDPs is not concentrated. As measured by the Herfindahl-Hirschman index, a statistical measure of market competition, the index value of 1,362 falls a little below the threshold for a concentrated market.¹¹ 

Within regions, however, the Part D (at the firm level), is moderately concentrated with an average index value across regions of 1,896. Overall, 30 of 34 regions qualify as moderately concentrated, while 2 are highly concentrated and 2 are not concentrated. The two regions classified as highly concentrated are Florida and Nevada.

If non-LIS and LIS beneficiaries are treated as separate markets in each region, both are more concentrated at the firm level than at the level of overall enrollment. In 2015, the LIS population reaches the level considered moderately concentrated in 25 of 34 regions and highly concentrated in another 3 regions. Comparable numbers for the non-LIS population are 20 moderately concentrated regions and 14 highly concentrated regions. The non-LIS market is considerably more concentrated than the LIS market, probably because the process for assigning LIS beneficiaries to benchmark plans reduces concentration. The most concentrated regions among non-LIS beneficiaries are Alaska, Arizona, Florida, Hawaii, New Jersey, and New York.

Proposed plan acquisitions of Humana by Aetna and Cigna by Anthem would increase market share among two of the top ten Part D plan sponsors. Should the proposed acquisition of Humana by Aetna be approved by federal regulators, the combined firm will increase its market share from 18 percent to 24 percent and surpass UnitedHealth to become the largest sponsor of Part D plans (both for PDPs and MA-PD plans). The share of enrollment for the top three firms would increase from 50 percent to 56 percent. The share of the top three firms for PDP enrollment would increase from 58 percent to 64 percent.

By contrast, the MA-PD plan market is less concentrated at the national level, but much more concentrated at the local level.¹²  Only three firms in the MA-PD market, accounting for about half of MA-PD plan enrollment (excluding special needs plans), offer plans in at least half the states. The vast majority of MA-PD plan sponsors are local or regional, offering plans in just one or two states. These firms collectively account for about 30 percent of MA-PD enrollment.

Market Share Among Part D Plans

There has been some turnover among the top Part D plans from 2014 to 2015, with one plan moving into the top ten PDPs or MA-PD plans by enrollment

9

. Humana’s Walmart Rx PDP, new in 2014, moved up to the fifth position, while Kaiser Permanente’s Senior Advantage dropped from the ninth to the eleventh spot. In addition to Humana Walmart Rx PDP, four other top plans gained significant numbers 0f enrollees from 2014 to 2015: Cigna-HealthSpring Rx Secure PDP, SilverScript Choice PDP, AARP MedicareRx Saver Plus PDP, and Humana Gold Plus HMO. The first two of these gained enrollees through consolidations of other plans offered in 2014 by the same sponsor. Three of the top plans in 2015 (Humana’s Preferred Rx PDP and Walmart Rx PDP and UnitedHealth’s AARP MedicareRx Saver Plus PDP) are recent entries to the market, featuring low initial premiums and tiered pharmacy networks.

Only three of the top ten PDPs or MA-PDs by enrollment in 2015 were among the top ten in 2006. They are UnitedHealth’s AARP MedicareRx Preferred PDP, Humana’s Enhanced PDP, and CVS Health’s SilverScript Choice PDP (renamed from SilverScript Basic in 2015)

9

. Within many plan sponsors’ offerings, there have been significant changes in enrollment, partly due to sponsors adding, dropping, or consolidating plans. Six of the original top ten plans from 2006 exited the market due to consolidations by plan sponsors, some after being acquired by another sponsor.

Enrollment gains at the plan level between 2014 and 2015 were experienced by both plans raising premiums and plans with significant premium reductions. Two PDPs gaining large numbers of enrollees (Humana Walmart Rx PDP and UnitedHealth’s AARP MedicareRx Saver Plus PDP) raised premiums by more than 20 percent, but still had relatively low premiums compared to the competition. In particular, the Humana Walmart Rx PDP, which was introduced in 2014 and had the lowest premium in most regions in 2015 despite a premium increase, nearly doubled its enrollment (95 percent) between April 2014 and April 2015 and became the fourth largest PDP in 2015. By contrast, Humana’s more expensive plan (Humana Enhanced PDP) experienced a 8 percent loss in enrollment between 2014 and 2015, Over the program’s first decade, Humana has refreshed its offerings by introducing new low-premium plans while raising premiums for its older plans. Two other PDPs with large enrollment gains from 2014 to 2015 had significant premium reductions (SilverScript Choice PDP and Cigna-HealthSpring Rx Secure-Xtra PDP). It is unclear whether the gains were beneficiaries new to Medicare, transfers from other PDPs offered by the same sponsor, or switchers from plans offered by other sponsors.

Low premiums were associated with enrollment growth at times other than the annual open enrollment period. Humana has been most successful in this regard, benefiting from its strategy of offering PDPs with low premiums to capture a large share of the market. The firm’s Walmart Rx PDP enrolled 229,000 new members outside the annual enrollment period (February to December 2014) for a 35 percent increase in enrollment and another 164,000 new members (12 percent increase) between February and September 2015. Because most enrollees have no option to switch plans in these months, it is likely that most of these new enrollees are newly eligible Medicare beneficiaries opting for the least expensive plan. UnitedHealth’s AARP MedicareRx Saver Plus PDP and the WellCare Classic PDP, also offering some of the lowest PDP premiums during 2014, picked up substantial numbers of enrollees during 2014 and the UnitedHealth plan had further gains during 2015.

Enrollment shifts among the top plans and plan sponsors also have been brought about by automatic re-assignment of LIS beneficiaries. If a plan loses its designation as a benchmark plan (available to LIS beneficiaries for zero premium), CMS reassigns certain beneficiaries to a benchmark plan offered by the same sponsor if one is available; otherwise they are switched at random to a benchmark plan offered by another sponsor. In 2015, for example, Aetna and CVS Health each picked up nearly 90,000 enrollees through these types of reassignments. By contrast, three PDPs not designated as benchmark plans experienced disenrollment of LIS enrollees and thus had among the largest decreases in total enrollment from 2014 to 2015.

The most popular plans vary considerably by region. UnitedHealth and CVS Health each have the largest PDPs in nearly half of all regions in 2015.¹³  UnitedHealth’s AARP MedicareRx Preferred PDP is the largest PDP in 16 regions, and CVS Health’s SilverScript Choice PDP is the largest in 14 regions. Humana Preferred Rx PDP holds the lead in Colorado and Nevada. Humana Walmart Rx PDP and Cigna-HealthSpring Secure PDP have the largest shares of enrollment in the Idaho/Utah region and Hawaii, respectively.

The most popular plans also differ for non-LIS and LIS beneficiaries. AARP MedicareRx Preferred PDP has enrolled 27 percent of all non-LIS beneficiaries nationally and has the most non-LIS enrollees in 29 of 34 PDP regions

10

. The next most popular non-LIS plan, Humana’s Walmart Rx PDP, with 12 percent of non-LIS enrollees, has the most non-LIS enrollees in three regions, and local Blue Cross Blue Shield PDPs have the largest share of non-LIS enrollment in Arkansas and the upper Midwest region.

CVS Health’s SilverScript Choice PDP dominates the LIS market with more than one-fourth of national LIS enrollment (28 percent) and the highest share of LIS enrollees in 25 PDP regions. PDPs sponsored by Humana, Cigna, and WellCare have the most LIS enrollees in the other 9 PDP regions. Unlike the situation in recent years, four of the five PDPs with the most LIS enrollees have an enrollee mix with at least one-third non-LIS enrollees.

Plan Availability

Choice remains plentiful in Part D; in 2015, the average Part D enrollee had a choice of 30 PDPs and 15 MA-PD plans. The average number of PDPs per region has come down from a high of 56 in 2007 and 35 in 2014 to 30 in 2015 (weighted by regional enrollment)

11

. At least 24 PDPs are offered in every region this year (excluding the territories). In 2015, virtually all beneficiaries have at least one Medicare Advantage option with drug coverage as well, and the average beneficiary has 15 options for Medicare Advantage drug plan enrollment, down from 26 in 2009 and up from 14 in 2014.¹⁴ 

The number of PDPs offered was down by 14 percent from 2014 to 2015. There are 1,001 PDPs in 2015, well below the number of PDPs offered between 2006 and 2010. While the number of PDPs rose sharply between 2006 and 2007, the number decreased each year since (other than 2014) as a result of both marketplace and policy factors

12

. Over its first ten years, the Part D market has witnessed several mergers between sponsoring organizations and consolidation of plan offerings by sponsors. In 2010, CMS issued regulations aimed at discouraging duplicative plan offerings and plans with low enrollment. For example, many sponsors now offer just two plan options (one basic and one enhanced) instead of the three options they had offered in previous years.

The drop in PDP offerings from 2014 to 2015 reflects plan terminations and a few offsetting new offerings. The only near-national plan sponsor leaving Part D, HealthMarkets, had never attracted significant enrollment in its two years. Three plan sponsors with a small regional presence also terminated contracts. In addition, Anthem terminated its MedicareRx Rewards PDPs in 24 regions. These PDPs entered the program in 2006, but enrollment had fallen steadily in recent years. The largest number of terminations (168 PDPs) reflected plan consolidations by five national plan sponsors and two local plan sponsors. Some consolidations resulted from earlier mergers among sponsors, and others were in response to CMS guidance on duplicative plan offerings. Partially offsetting the plan terminations were 70 new plans, nearly all from three national or near-national plan sponsors (EnvisionRx, Symphonix Health, and United American) with relatively few enrollees. There are two entirely new contracts offering one PDP in Colorado and two in Nevada. Together all the new plans attracted only 170,000 enrollees (about 1,000 enrollees for the new contracts). Over half (93,000) were in one new plan (sponsored by Blue Cross Blue Shield in the upper Midwest region); the rest averaged just over 1,000 enrollees per plan. Some enrollees in new PDPs appear to have shifted from others offered by the same sponsor.

Current CMS policies suggest that the number of PDPs might decline again in future years. In the call letter issued in early 2015 spelling out the terms of plan participation for the 2016 contract year, CMS reiterated the agency’s authority not to renew plans with low enrollment.¹⁵  Currently, 257 PDPs (26 percent of all PDPs in 2015) have fewer than 1,000 enrollees, the level at which CMS urges sponsors to consider plan withdrawal or consolidation; 63 of these PDPs have fewer than 100 enrollees each.¹⁶  The low-enrollment PDPs include the Symphonix Premier Rx and Symphonix Rite Aid Premier Rx PDPs in 25 regions, which were offered for the first time in 2014.

In addition to its policy on low-enrollment plans, CMS continues to maintain a policy that PDPs offered by the same sponsor must be meaningfully different from the sponsor’s other offerings. This policy encourages plan sponsors to reduce their PDP offerings, thereby simplifying the choice environment in Part D.

In 2015, 1,605 Medicare Advantage drug plans are offered, essentially the same number as the year before. The number of MA-PD plans increased by about 50 percent between 2006 and 2009, from 1,333 plans to 1,991 plans.¹⁷  However, the availability of MA-PD plans has fallen since then; the 1,605 MA-PD plans offered in 2014 is about 19 percent lower than at the peak. The number of MA-PD plans available to the average Medicare Part D enrollee has remained nearly constant at 15 plans since 2011.

National Premium Trends

Since 2006, the average PDP premium, weighted by enrollment, has increased by 43 percent, but the 2015 average is 2 percent lower than in 2014 and 4 percent lower than at the peak in 2013. The weighted average monthly premium paid by beneficiaries for stand-alone Part D coverage has increased since the start of the program, from $25.93 in 2006 to $37.02 in 2015

1

.^18 ,19  Premiums have been essentially flat since 2010, down slightly from 2010 to 2015. A key factor driving slow premium growth in recent years is the availability of generic versions of many drugs used for common chronic conditions, which helps to limit growth in total plan costs and hence premiums.²⁰ 

The average monthly Part D premium, including both PDPs and MA-PDs, has increased from around $23 per month in 2006 to $30 per month in 2015, but has been essentially flat since 2010

1

.

Average monthly MA-PD plan premium are lower than average premiums for PDPs, but have been increasing since 2011 (compared to the flat premium trend for PDPs). The average 2015 monthly premium amount attributable to drug benefits in MA-PD plans is $17.29, up 18 percent from $14.70 in 2014 and up 30 percent from $13.30 in 2013. MA-PD premiums on average are higher in 2015 than in any year since the program began.²¹ 

At the same time, the MA-PD average monthly premium is about $20 below the PDP average monthly premium, because many MA-PD plans use a portion of rebates from the Medicare Advantage payment system to reduce or eliminate their premiums. In 2015, CMS calculated that the average MA-PD premium prior to rebates was about $3 per month lower than those for PDPs (a smaller differential than in previous years); thus, the average plan applied a rebate amount of about $17 to lower the premium in 2015.²²  The increase in the MA-PD premium from 2013 to 2015 may reflect changes in the Medicare Advantage payment rules, which may have lowered these rebates. Nearly half (42 percent) of all MA-PD plans, with 52 percent of MA-PD enrollees, charge no premium for their drug benefit.

PDP premiums vary widely. Nationwide, the least expensive PDP has a $12.60 monthly premium, while the most expensive PDP has a $171.90 premium, a 14-fold difference. Although the difference can be explained partly by the relative generosity of the benefits offered or the relative efficiency across plans, these factors seem unlikely to explain the full difference. Even among plans with equivalent benefits (those offering the basic Part D benefit), premiums vary from $12.60 to $101.40 per month. As illustrated in the lowest and highest-premium PDPs in the Florida region, benefit differences are modest relative to the large premium differences

2

. Although enrollees in the highest-premium enhanced plan have some coverage in the gap and no deductible and flat cost sharing for brand drugs, they face higher cost sharing for generic and specialty drugs than the lowest-premium enhanced PDP in Florida. Those enrolled in the highest-premium basic PDP have a higher deductible than those in the lowest-premium basic PDP and are subject to coinsurance for brand drugs.

Plan-Level Premium Trends

Seven of the eight PDPs with the highest enrollment charged higher average premiums in 2015 compared to 2014, whereas one lowered its average premium. More generally, the modest decrease in the average premium for all Part D enrollees hides larger changes at the plan level

3

. For example, the plan with the highest enrollment, UnitedHealth’s AARP MedicareRx Preferred PDP, increased the monthly premium by 16 percent compared to 2014 (from $43.43 to $50.19). All three plans sponsored by Humana raised premiums by 11 percent or more. By contrast, the Silverscript Choice PDP, offered by CVS Health, lowered its average premium by 21 percent (from $29.47 to $23.13).

Some older, established plans have raised premiums more rapidly than the national average, while newer plans tend to set premiums low in order to build enrollment but may raise them rapidly in subsequent years. As a result, beneficiaries who stay in the same plan tend to pay more over time, as earlier research finds relatively few enrollees switch plans voluntarily in a given year.²³  Established plans tend to retain enrollees as they age, when they typically use more drugs, whereas newer plans attract younger enrollees who are likely to have lower drug use and also more likely to shop based on premiums when they first enter the market. Premiums for some new plans have increased rapidly within a year or two of entering the market. For some plan sponsors, this strategy may be a conscious attempt to attract younger enrollees in newer, less expensive plans while still retaining their existing enrollees in older, more expensive plans.

Most plans that have been in the program since 2006 have increased premiums by more than the national average. Overall, of 315 PDPs that have operated under the same contract and plan numbers from 2006 to 2015 (despite some corporate acquisitions and name changes), about 40 percent have monthly premiums in 2015 that are at least double the premium in 2006, and two-thirds have raised premiums over this period by more than the national average increase. For example, the average monthly premium for Humana’s Enhanced PDP in 2015 is more than 3.5 times its 2006 average ($52.86 versus $14.73)

4

. Over the same period, the premium for UnitedHealth’s AARP Medicare Rx Preferred PDP nearly doubled ($50.19 versus $26.31). By contrast, one-fourth of these continuously operating PDPs have a lower premium in 2015 than in 2006. Silverscript Choice PDP had a 18-percent lower premium in 2015 compared to 2006 ($23.15 versus $28.32).

Some of the program’s newer plans have experienced large percentage premium increases. The average premium for Humana’s Preferred Rx PDP is higher by 79 percent ($14.80 to $26.42) in four years after entering the program. UnitedHealth’s AARP MedicareRx Saver Plus PDP has experienced a net 87-percent premium increase ($15.00 to $28.09) in three years.

Overall, enrollment increased (by 13 percent) in plans that lowered premiums, while enrollment was generally flat (down by 1 percent) in plans that raised premiums. PDPs that lowered premiums by at least 20 percent had an average enrollment gain of 34 percent. By contrast, PDPs that raised premiums by at least 30 percent experienced a 15 percent average drop in enrollment. Patterns were not consistent, however, as PDPs that raised premiums 20 percent to 30 percent gained enrollment, and those that lowered premiums up to 10 percent lost enrollment. Some of the enrollment gain came from random assignment of LIS enrollees, but enrollment by non-LIS beneficiaries was considerably higher as well. Our analysis of plan switching from 2006 to 2010 found that 87 percent of beneficiaries in any particular annual enrollment period did not change plans.²⁴  But generally consistent with patterns in 2015, those whose premiums were increasing by $10 or more were more likely to change to plans with lower premiums; 21 percent of those with a $10 to $20 premium increase and 28 percent of those with a premium increase of $20 or more changed plans.

As with PDPs, average premiums vary considerably by MA-PD plan sponsor. Plans offered by UnitedHealth, with 20 percent of the MA-PD market, have a weighted average premium of $11.44 for the drug benefit (in addition to a Part C premium of $5.02 that covers the medical benefits normally provided by traditional Medicare). The other leader in this market segment is Humana, also with 20 percent of MA-PD enrollees, which has a modestly higher average premium of $14.37 (plus $17.94 for Part C). The next two largest MA-PD sponsors are Kaiser Permanente, with a 9 percent market share and a $4.79 average premium (plus $35.85 for Part C), and Aetna, with a 6 percent market share and a $7.13 average premium (plus $16.62 for Part C).²⁵ 

Geographic Variations in Premiums

Average premiums are considerably higher in certain regions than in others in 2015. Beneficiaries enrolled in a basic PDP pay an average of $36.89 in the Idaho/Utah, more than double the average ($16.97) in New Mexico

6

.²⁶  Regional differences in premiums have generally persisted from year to year and continued to grow wider in 2015. New Mexico and Arizona have been among the regions with the lowest average premiums since the program began, while the Idaho/Utah region has been among the most expensive regions.

At the same time, some regions have seen significant changes in their average PDP premiums relative to other regions. The average PDP premium in New York, for example, was below the national average from 2006 to 2010, and then increased to be above average each year since then. Regional differences in the average PDP premium were smaller in the program’s first two years, before plan sponsors could look at actual claims experience for guidance in setting premium levels.

Although persistent regional differences in premiums are driven in part by underlying regional differences in drug utilization, further explanations are not readily apparent. Plan risk scores, averaged by region for 2013 (the most recent year available) are correlated modestly (r=0.24) with that year’s average premium for basic PDPs, providing some support for the notion that utilization is a factor. In 2013, there was also a modest correlation (r=0.33) between the number of basic PDPs in each region with the average premium for those PDPs, counter to the notion that more competing plans leads to lower plan bids. In 2015, however, number of plans was not correlated with premiums, but was modestly correlated (r=-0.25) with a measure of concentration; regions that were more concentrated had lower premiums—again running counter to expectations. More research is needed to understand the patterns of geographic variation in Part D premiums.

Geographic differences in premiums are greater for some plan sponsors than others; some sponsors charge as much as three times more for the identical basic PDP from one region to another. Twelve plan sponsors offer a basic PDP in at least 30 of the 34 PDP regions. For seven of these national or near-national PDPs, premiums for the identical plan design are at least two times greater in one region than in another

6

. The largest absolute premium difference is for the Cigna-HealthSpring PDP, which charges beneficiaries $20.50 in Arizona and $83.30 in Florida for the same coverage—a difference of $62.80 (a wider difference than in 2014). By contrast, the Humana Preferred Rx PDP has a difference of only $13.30 between its lowest and highest regions ($20.20 in New Mexico and $33.50 in New York), but that difference is also wider than in 2014. For six of these national or near-national PDPs, the highest premiums are in states (Florida, Missouri) where premiums overall are below the national average.

Within each region, some plan sponsors charge several times more than competing sponsors for their basic PDPs

7

. In the region that includes Delaware, Maryland, and Washington, DC, the highest premium for a basic PDP is $34.30 for United American’s Select PDP, which is only one-third higher than the $25.30 premium for Aetna’s Medicare Rx Saver PDP. This region had the largest range of premiums in 2014, but several of the high-premium plans exited the market including the BlueRx Standard PDP with a 2014 premium of $111.40. The largest premium range among basic PDPs is in Florida, where the Express Scripts Medicare-Value PDP charges $91.40, seven times the lowest premium in its region ($17.60 for Aetna Medicare Rx Saver PDP). By law, all basic PDPs provide a benefit with the same actuarial value. Different utilization patterns by plan enrollees (adverse selection, beyond what can be compensated for by the risk-adjustment system used by CMS) may be a key factor driving the larger premium differences.

Premium Variations by Plan Type

Beneficiaries selecting PDPs with an enhanced benefit package pay higher premiums on average for their Part D coverage, even for the part attributable to the basic benefit package. The weighted average monthly premium for PDPs with enhanced benefits is $47.81, compared to $28.28 for PDPs offering the basic benefit package

8

. Thus, enrollees pay nearly 70 percent more to get enhanced benefits.

Plan sponsors mostly add value in their enhanced plans by lowering deductibles and sometimes adding coverage in the gap. Most enhanced plans lower or eliminate plan deductibles; 71 percent of enhanced PDPs have no deductible, compared to 7 percent of basic PDPs (the share of enhanced PDPs without a deductible is down from 89 percent in 2014). If eliminating the deductible were the only difference, beneficiaries would be paying an additional premium of $19.53 per month ($47.81 versus $28.28) or $234 annually to eliminate a $320 deductible. Some enhanced plans also expand the coverage of drugs during the coverage gap beyond the amount included in the basic benefit (48 percent of enhanced PDPs). Plans also may use lower cost sharing as part of an enhanced benefit, but as shown in the section below on cost sharing, enhanced PDPs on average charged higher cost sharing in 2015.

Starting with PDPs offered in 2011, CMS has required sponsors to ensure that benefits in enhanced PDPs are meaningfully different than the basic benefits and have a measurable added value. This policy has led to a larger spread between premiums for basic PDPs and enhanced PDPs than in previous years. In 2015, an enhanced PDP must have cost-sharing differences that result in at least $20 lower monthly out-of-pocket costs than the corresponding basic PDP—an amount that is essentially offset by the $19.53 average monthly premium difference between basic plans and enhanced PDPs.

Some PDP sponsors offer two enhanced plans, a less generous first option and a more generous second option; average monthly premiums for the more generous enhanced PDPs offered by these sponsors are much higher than the first option’s premium ($65.15 versus $28.78). As part of its policy on meaningful differences, CMS allows sponsors to offer a second enhanced PDP only if expected out-of-pocket cost sharing amounts are lower (by $25 per month) than for the first enhanced PDP and the second enhanced PDP has coverage for at least some brand-name drugs in the coverage gap. The $36 difference in premiums exceeds the required difference in out-of-pocket costs.

Although higher premiums partly reflect the cost of offering enhanced benefits, the portion of the premium that corresponds to the basic benefit ($38.60 on average for enhanced PDPs) is higher than the premium for basic PDPs ($28.28)

8

. For some sponsors, the difference is much greater. Risk selection may be a factor in these higher premiums, although in 2013 enrollees in enhanced PDPs actually had lower risk scores (0.94) than those in basic PDPs (1.05), mostly attributable to enrollees in the less generous enhanced PDPs. But it may be that the enhanced plans have attracted beneficiaries with higher drug needs beyond differences captured by risk adjustment. Furthermore, even though program rules seem to disallow it, plan sponsors may cross-subsidize some of the supplemental portion of bids for enhanced plans with the basic component in order to maximize the government subsidy.²⁷ 

Some plan sponsors offer enhanced PDPs that have the minimum level of enhanced coverage required by the meaningful difference tests and are offered at low premiums with the apparent goal of attracting beneficiaries with low expected drug costs. In 2015, for example, one of Humana’s enhanced PDPs (Walmart Rx) is offered at an average premium of $15.67 per month, whereas its basic PDP (Preferred Rx) is $26.42. For several other plan sponsors, the portion of the premium attributable to a plan’s basic benefits (thus excluding the value of any enhanced benefit) is lower for their enhanced PDPs compared to their basic PDPs. A key reason for lower premiums in these enhanced plans is favorable risk selection that occurs because they are attractive to non-LIS beneficiaries who are using few drugs and because there are relatively few LIS beneficiaries enrolled in enhanced plans. In 2013, the average risk score for enrollees in the less generous enhanced PDPs was 0.93, compared to 1.04 in the more generous enhanced PDPs.

Plan Benefit Design

Most Part D plans do not offer the defined standard benefit (with a $320 deductible in 2015 and 25 percent coinsurance); all PDPs and most MA-PD plans have a tiered cost-sharing structure with incentives for enrollees to use less expensive generic and preferred brand-name drugs. For the first time in 2015, no PDPs offer the defined standard benefit without formulary tiers. Among MA-PD plans, 2 percent offer the defined standard benefit with 1 percent of enrollment.²⁸ 

A majority of Part D enrollees are in enhanced plans. In 2015, 53 percent of all Part D enrollees are in enhanced plans, designed to offer some benefits beyond the basic benefits defined in law. Nearly half (45 percent) of PDP enrollees and 72 percent of MA-PD enrollees are in enhanced plans.

Use of a deductible by stand-alone PDPs is considerably higher in 2015 than in the first few years of the program. About 58 percent of PDPs charge a deductible this year, compared to a high of 60 percent in 2010. Most PDPs with a deductible use the standard deductible allowed by law ($320 in 2015). Nearly half of PDP enrollees are in plans with no deductible (48 percent of enrollees), compared to 56 percent of PDP enrollees in 2006

1

. A smaller number of MA-PD plans (37 percent) than PDPs have a deductible in 2015, but use of a deductible is up substantially from 2014 (14 percent). More than half of the MA-PD plans with a deductible set the level lower than the maximum of $320. Overall, 51 percent of all Part D enrollees are enrolled in plans with no deductible, and 33 percent are in plans with the maximum deductible. In contrast to Part D, only a small share of workers with employer sponsored drug coverage (12 percent) face a separate drug deductible, averaging $231 in 2015.²⁹ 

In 2015, the vast majority of all Part D enrollees (80 percent of PDP enrollees and 91 percent of MA-PD plan enrollees) are in plans that use five cost-sharing tiers: preferred and non-preferred tiers for generic drugs, preferred and non-preferred tiers for brand drugs, and a tier for high-cost specialty drugs (see section below, ‘Specialty Tiers’)

2

. Overall, 84 percent of the combined enrollment in PDPs and MA-PDs are in plans with five tiers. In 2015, 90 percent of PDPs and 87 percent of MA-PD plans used five-tier designs. Most of the other Part D enrollees are in plans with four tiers: one generic tier, two brand tiers, and a specialty tier.³⁰  By contrast, enrollees in 2006 were in PDPs with a greater variety of tier structures, including significant numbers in plans with three-tier and four-tier arrangements. Four-tier arrangements were most common until 2012 when plans began shifting toward the five-tier cost-sharing design.

Part D plans use more cost-sharing tiers than private-sector employer plans. Whereas all PDP enrollees and nearly all MA-PD enrollees are in plans with four or more tiers, only 23 percent of covered workers in employer-sponsored plans are in plans with that many tiers, while 13 percent are in plans with two tiers or no tiered cost sharing.³¹  The trend among employer plans has been to use more tiers, and some have gone to five tiers, but this design is used much less often than in Part D.

Part D Cost-Sharing Amounts

While cost sharing has been relatively stable in recent years, the median cost sharing for a 30-day supply of non-preferred brand-name drugs in stand-alone PDPs has increased by 45 percent since 2006, from $55 to $80, while cost sharing for preferred brand drugs increased by 36 percent, from $28 to $38

3

. Median cost sharing in PDPs for both brand tiers was modestly lower in 2015 compared to 2014. Cost-sharing amounts for brand-name drugs vary widely across Part D plans in 2015, as they have in previous years. For preferred brand tiers, PDPs charge copayments as low as $19 and as high as $45. About 80 percent of spending on brand drugs for Part D enrollees was for drugs on preferred brand tiers,³²  so these are the cost-sharing levels encountered most often. For non-preferred brand tiers, copayments range from $35 to $95. These ranges are less than in some previous years because of CMS guidance that sets maximum allowable copayment levels.

Median cost sharing for preferred generic drugs in PDPs (or for generic drugs among plans with a single generic tier) is $1 in 2015, lower than in any year since the program began. As a result of lower generic copays, the spread between brand and generic tiers widened modestly from 2011 to 2015. For PDPs with two generic tiers (most PDPs and PDP enrollment), the median cost sharing is $1 for the preferred generic tier and $4 for the non-preferred tier. These copay amounts are lower than in any previous year; in fact, the copay for non-preferred generics is lower than the copay for the single generic tier as recently as 2011. By contrast, however, one PDP charges a $28 copayment for its non-preferred generic tier in 2015, well above any other PDP.

Despite the trend to higher cost sharing, beneficiaries’ average total out-of-pocket costs have not increased. According to MedPAC, for the average Part D enrollee not receiving the LIS, monthly out-of-pocket spending declined from $59 to $47 from 2007 to 2012 (the most recent year available).³³  The impact of partially closing the coverage gap and more use of generic drugs has generally balanced the impact of higher cost sharing for brands.

In 2015, median cost-sharing amounts are generally higher in MA-PD plans than in PDPs in all tiers. For example, the median cost sharing for preferred brands in MA-PD plans is $7 more than the median in PDPs ($45 versus $38) and $2 more for preferred generic drugs ($3 versus $1)

4

. The comparisons for UnitedHealth, the sponsor with largest share of both PDP and MA-PD enrollment, illustrate the pattern. For preferred and non-preferred generic drugs, UnitedHealth’s median cost sharing, weighted by enrollment, is $2 and $5 in its PDPs and $2 and $8 in its MA-PD plans, respectively. The differences for brand drugs for UnitedHealth’s plans mirror the national differences. It is unclear why cost sharing is higher for MA-PD plans, especially since more of them offer enhanced benefits. Further work is needed to assess variations in copayments by Part D plan type, including for example, the extent to which these differences persist across all plan sponsors and within different geographic areas. Another question is whether there are differences in tier placement of specific drugs and whether some plans cover more drugs than others on specific tiers that may be factors in explaining differences in cost-sharing amounts between types of Part D plans.

Combined across PDPs and MA-PD plans in 2015, median cost-sharing amounts for generic drugs (weighted by enrollment) are $2 for preferred generics and $6 for non-preferred generics. For brand drugs, median copayments are $40 and $90 for preferred and non-preferred brands.

Copayments in the form of a flat dollar payment amount remain the most common type of cost sharing; however, the share of PDPs using percentage-based coinsurance for non-specialty brand-name drug tiers has increased since 2006. In 2015, 68 percent of PDPs with a tier for non-preferred brand drugs charge a coinsurance rate for drugs on that tier, a large jump from 37 percent in 2014. Nearly two-thirds (63 percent) of PDP enrollees are in plans with coinsurance for this tier. Of these plans, nearly all have a mixed pricing design. Typically they use a flat copayment for their generic drug tiers, and many also use a flat copayment for preferred brand drugs; 27 percent of PDPs (28 percent of enrollees) use coinsurance for preferred brand tiers, about the same as in 2014. The use of percentage coinsurance for drugs remains uncommon among MA-PD plans (13 percent for the non-preferred brand tier). In 2006, only a small share of PDPs with tiered cost sharing used coinsurance for their brand tiers (12 percent of enrollees for preferred brands and 14 percent of enrollees for non-preferred brands), but in addition nearly one-fourth of PDP enrollees were in standard plans with 25-percent coinsurance. Overall, 43 percent and 18 percent of Part D enrollees have coinsurance for non-preferred brands and preferred brands, respectively.

For plans that use percentage coinsurance instead of dollar copayments, the actual amount an enrollee pays depends on the retail price of the drug. The median coinsurance percentage for PDPs in 2015 for the preferred brand tier is 20 percent. For a brand-name drug with a typical cost of $250, the coinsurance would be $50, more than the median copay amount. For drugs on the non-preferred brand tier, the median coinsurance rate is 40 percent, a substantial share of the drug’s cost and more than the median copay for this tier. In fact, 132 PDPs (up from 29 in 2014) require beneficiaries to pay half the cost of drugs on the non-preferred brand tier, the maximum allowed under CMS guidance.

Cost sharing is generally higher in enhanced PDPs than in basic PDPs. Analysis of enhanced PDPs in earlier years sometimes revealed only small benefit differences compared to the same sponsor’s basic PDPs.³⁴  In 2015, cost sharing in each tier is generally higher for enhanced PDPs than in basic PDPs. For generic drugs, median cost-sharing amounts (weighted by enrollment) are $3 and $5 (preferred and non-preferred tiers) for enrollees in enhanced PDPs, compared to $1 and $4 for those in basic PDPs. For brand drugs, they are $40 and $80 (preferred and non-preferred) versus $35 and $60. Enrollees in enhanced PDPs generally have no deductible, which may constitute the added actuarial value that characterizes enhanced plans. As noted in Section 2, average monthly premiums are about $20 higher for enhanced PDPs, raising questions about the value of these plans for Part D enrollees.

Medicare Part D plans generally charge more than private-sector employer plans do for preferred and non-preferred brand drugs, but less for generics. At the median, PDPs charge $38 per month for a preferred brand in 2015, higher than the median $30 charged by employer plans in 2015,

3

.³⁵  Cost-sharing differences are even greater for non-preferred brands ($80 for PDPs versus $50 for employer plans). On the other hand, only about one-fourth of workers are in employer plans that use coinsurance for brand tiers. Employer plans charge much higher copays for generic drugs than PDPs charge ($10 versus $1). Thus the spreads between cost sharing for brands and generics and between preferred and non-preferred brand drugs are greater in Medicare Part D plans. Compared to commercial health plans, the typical structure of cost sharing in Part D offers a greater incentive for plan enrollees to choose generics or preferred brand drugs. Furthermore, employer plans are much less likely to use two generic tiers.

Part D enrollees in some plans have the potential for savings if they obtain prescriptions by mail order; in other plans cost sharing is lower at retail pharmacies. Only 40 percent of PDPs (with about half of PDP enrollees) discount the cost sharing for a 90-day supply of a preferred brand drug obtained by mail order versus the same 90-day supply at retail. Among these PDPs, cost sharing is typically about 10 percent lower. About one-tenth of PDP (4 percent of enrollees) actually charge more; in most cases the cost sharing is equivalent to that charged for pharmacies not offering preferred cost sharing (see section on tiered pharmacy networks). Part D allows plans to offer mail order, but requires a level playing field in that at least one retail pharmacy must be able to dispense prescriptions with 90-day supplies. According to a recent study, mail order represented about 8 percent of prescriptions for 300 top drugs dispensed in Part D in 2010 and about 14 percent of spending on those drugs. The study found that Part D plan sponsors do not achieve savings when patients use mail order, because total mail order pharmacies may charge more for drug than retail pharmacies in Part D.³⁶ 

About one-fifth of Part D PDP enrollees are in plans that offer discounted cost sharing for a 90-day supply at retail pharmacies. In 2015, 11 percent of PDPs with 19 percent of PDP enrollees have discounted cost sharing for a 90-day supply of a drug on a preferred brand tier. Most PDPs offering this discount charge the equivalent of 2.5 monthly copays for a 90-day supply, amounting to about a 17 percent discount. A handful of PDPs charge two monthly copays. All other PDPs have no discount for a 90-day supply.

Specialty Tiers

Nearly all Part D plans use a specialty tier for high-cost medications in 2015. In 2015, all PDP enrollees and 99 percent of MA-PD plan enrollees are in plans with a specialty tier. Across the two enrollee populations, 99.5 percent are in plans with a specialty tier. Specialty tiers are used by Medicare drug plans for relatively expensive drugs (at least $600 per month in 2015—a threshold that has been unchanged since 2008).

Use of specialty drugs has increased rapidly in recent years (46 percent in 2014 and 15 percent in 2013), driven by growth in the average unit cost, rather than utilization.³⁷  The introduction of new drugs to treat hepatitis C was a large driver of the increase in 2014. Although the trend is up, specialty drugs represented only 16 percent of Part D drug spending in 2014, up from 11 percent in 2014.³⁸  Only 2 percent of Part D enrollees used a specialty drug in 2014.

About half of PDP enrollees and most MA-PD plan enrollees are in plans with a 33 percent coinsurance rate for specialty tier drugs. While CMS limits the coinsurance rate for drugs placed on a specialty tier to 25 percent, plans are allowed to impose higher cost sharing (up to 33 percent) for specialty tier drugs if offset by a lower deductible.³⁹  In 2015, 48 percent of PDP enrollees with a specialty tier and 73 percent of MA-PD plan enrollees in a plan with a specialty tier are in plans charging the maximum 33 percent coinsurance for specialty drugs in the initial coverage period

5

. By contrast, in 2006 only 13 percent of beneficiaries in PDPs and 25 percent of enrollees in MA-PD plans with specialty tiers faced a 33 percent coinsurance rate. Overall, 57 percent of PDP and MA-PD plan enrollees are liable for 33 percent coinsurance for specialty tier drugs in 2015—up from 16 percent in 2006.

Placing a drug on the specialty tier or on a non-preferred brand tier with high coinsurance can have significant cost implications for plan enrollees, at least before they reach the catastrophic coverage phase of the Part D benefit. A specialty drug priced at the $600 threshold will cost the beneficiary between $150 and $200 per month during the initial coverage period prior to the coverage gap. But monthly cost sharing for other common specialty drugs, such as Copaxone (for multiple sclerosis), Enbrel (for rheumatoid arthritis), Gleevec (for certain cancers), and Truvada (for HIV) can range from $300 to $2,000, before a beneficiary reaches the coverage gap or qualifies for catastrophic coverage. The cost for the first month of Sovaldi, a newly approved drug for hepatitis C, can exceed $5,000.⁴⁰  For beneficiaries who exceed the catastrophic coverage threshold, the cost sharing is lowered to 5 percent of the drug cost for the remainder of the year.

Tiered Pharmacy Networks

In 2015, 87 percent of all PDPs—representing 81 percent of all enrollees—have a tiered pharmacy network. By contrast, only 7 percent of PDPs (6 percent of enrollees) had a tiered pharmacy network in 2011

6

. Enrollees in these plans pay lower cost sharing for their prescriptions if they use pharmacies that offer preferred cost sharing.⁴¹  This approach to plan design started in 2011 with the market entry of co-branded PDPs featuring relationships with specific pharmacy chains, such as the Humana Walmart-Preferred Rx PDP (new in 2011) and the Aetna CVS/Pharmacy PDP (new in 2012). Many other plan sponsors designated a tiered network in 2013 or 2014, mostly without a co-branded relationship. The idea behind these arrangements is that Part D plans are able to negotiate discounted prices at certain pharmacies in exchange for higher volume of sales. The lower cost sharing creates an incentive for enrollees to use the designated pharmacies. CMS guidelines allow plans to assume use of preferred cost sharing in calculating whether cost sharing is actuarially equivalent to the defined standard benefit.

Tiered pharmacy networks have been employed much less often in Medicare Advantage. In 2015, only 27 percent of MA-PD plans (21 percent of enrollees) used tiered pharmacy networks

7

. This lower rate may seem surprising given that Medicare Advantage plans generally limit networks for other types of providers. The faster adoption of this approach among PDPs compared to MA-PD plans may reflect the efforts of PDP sponsors to match practices of their competitors who were early adopters of this change.

Proportionally, the discounts in cost sharing for filling a monthly prescription in a pharmacy using preferred cost sharing is greater for generic drugs. The median discount (weighted by enrollment) is $3 for preferred generics and $5 for non-preferred generics, which represents a discount of more than half for these tiers. For brand drugs, the differential is much more modest in percentage terms but greater in dollars. The discount is $5 for the preferred tier and $10 for the non-preferred tier (5 percentage points and 13 percentage points, respectively, for plans that use coinsurance instead of copays). There are no differentials for the specialty tier.

For example, in the AARP MedicareRx Saver Plus PDP sponsored by UnitedHealth, the copayment for a preferred brand drug is $20 in a designated pharmacy and $45 in another network pharmacy ($35 versus $60 for other brand drugs). Copayments in the Humana Walmart Rx PDP at a designated pharmacy are $1 for drugs on the preferred generic tier and $4 for drugs on the non-preferred generic tier, compared to $10 and $33, respectively, at other network pharmacies. Coinsurance differences for this Humana PDP are 20 percent versus 25 percent for a preferred brand drug and 35 percent versus 50 percent for a non-preferred brand drug.

Although the difference in cost for filling a single prescription is modest, the financial consequences for non-LIS Part D plan enrollees if they do not use pharmacies offering preferred cost sharing can add up for beneficiaries taking multiple brand-name drugs.⁴²  An enrollee in the AARP MedicareRx Saver Plus PDP who fills two brand drugs and one generic drug per month might pay an extra $624 over the year if she does not use one of the designated pharmacies; an enrollee in Humana Walmart Rx PDP might pay an extra $228.

In 2014 (the most recent available data), PDPs with tiered pharmacy networks designated only about 24 percent of their network pharmacies as offering preferred cost sharing.⁴³  But the size of the preferred cost sharing networks varied enormously from 1 percent to 99 percent of plans’ overall networks. Overall, access to pharmacies is high. Most PDPs contract with at least 95 percent of all available pharmacies in their full pharmacy network.⁴⁴  But generally plan sponsors do not offer the preferred terms to all pharmacies in their networks.

Limited information is available on the share of plan enrollees who fill prescriptions at pharmacies offering preferred cost sharing. A CMS analysis of 2012 claims data found that the share of retail claims in pharmacies offering preferred cost sharing ranged from 19 percent to 79 percent across 13 plans. For 7 of the 13 plans, the share of claims in pharmacies offering preferred cost sharing was 37 percent or less.⁴⁵ 

For some PDPs, access to pharmacies with preferred cost sharing is limited. For some plans, there are few or no designated pharmacies within a reasonable travel distance.⁴⁶  Medicare law requires that retail pharmacy networks as an entirety meet standards whereby certain shares of beneficiaries must have access to a network pharmacy close to their residence, with separate tests for urban, suburban, and rural areas.⁴⁷  For example, 90 percent of beneficiaries in urban areas must have a network pharmacy within two miles of their home (90 percent within 5 miles for suburban areas and 70 percent within 15 miles in rural areas). CMS ensures that all plans meet these standards for their full pharmacy networks. But CMS found that 46 percent of PDPs or MA-PD plans would have failed if the test was applied just to the pharmacies offering preferred cost sharing.⁴⁸  Plans were more likely to fail in urban area, whereas most plans would have met the standard in suburban and rural areas. If the urban access standard were widened from two miles to 4.5 miles, 90 percent of plans would have met the standard.

Plans sponsored by Humana illustrate the trend. In 2014, Humana’s PDPs on average designated only 10 percent of their network pharmacies as preferred, a lower rate than other national plan sponsors.⁴⁹  Because Humana’s PDPs rely solely on Walmart pharmacies, access is lacking in urban areas where Walmart has no market presence.

In the call letter issued in April 2015, CMS indicated that it will monitor pharmacy networks and encourage plans to take action if they have too little meaningful access to the pharmacies that offer preferred cost sharing.⁵⁰  Effective in 2016, the Medicare program intends to improve transparency by publishing information on plans’ access levels for the designated pharmacies. The agency will also require a sponsor to disclose in marketing materials if its network is an outlier on access standards. But the agency has not established specific access standards for tiered pharmacy networks offering preferred cost sharing, beyond the statutory standards that already apply to the overall networks.

Formularies and Utilization Management

In 2015, the average PDP enrollee is in a plan where the formulary lists 83 percent of all eligible drugs, the same as in 2013 and 2014 but slightly below the average in prior years. The scope of formulary coverage, however, continues to vary widely across PDPs in 2015. Some plans list all drugs from the CMS drug reference file on their formularies, while other PDPs list as few as 66 percent of these drugs.⁵¹  Even the most limited formularies, however, exceed the formulary requirements established under law and CMS program guidance.⁵²  The seven largest PDPs range in formulary coverage from 74 percent to 89 percent of drugs in the reference file.⁵³  The average MA-PD plan enrollee is in a plan with slightly more drugs on formulary (87 percent) than PDPs. Beneficiaries retain the option of requesting an exception to have the plan cover an off-formulary drug, or they can obtain the drug by paying the full purchase price out of pocket.

Since 2007, PDPs have applied utilization management (UM) restrictions to an increasing share of on-formulary drugs, increasing from 18 percent in 2007 to 39 percent in 2015

8

. Even if a drug is listed on a plan’s formulary, utilization management rules, including step therapy, prior authorization, and quality limits, may restrict a beneficiary’s access to the drug.⁵⁴  In 2015, more drugs are subject to prior authorization than to other UM tools. On average across all PDPs (weighted for enrollment), prior authorization is applied to 23 percent of formulary drugs, about three times the use in 2007. Quantity limits (e.g., limiting a prescription to 30 pills for 30 days) are applied to 18 percent of drugs in 2015, whereas only 1 percent of drugs are subject to step therapy. MA-PD plans tend to apply UM restrictions to about the same share of drugs as PDPs.

The Coverage Gap

In 2015, most PDPs (74 percent) offer no gap coverage beyond what is required by law; PDPs offering extra gap coverage cost more and have attracted fewer enrollees

9

. In 2015, beneficiaries reaching the gap pay 45 percent of the full price for brand-name drugs in the gap (after a manufacturer price discount of 50 percent; plans pay the remaining 5 percent), and 65 percent of the cost for generics (plans pay the remaining 35 percent). Under current law, beneficiaries will face average cost sharing of only 25 percent for all drugs in the gap by 2020—the same as in the initial coverage period—effectively eliminating the coverage gap.

Plans offering gap coverage typically cover only a subset of drugs when enrollees are in the coverage gap. In past years, CMS has differentiated gap coverage by shares of drugs covered in the gap (few, some, many, or all). The agency terminated these descriptions starting in 2015 on the basis that they are no longer relevant as a result of phasing out of the coverage gap. In our analysis for past years, we classified plans labeled by CMS as covering few brands or few generics (defined as less than 10 percent of drugs in a particular category) as having “little or no coverage.” Because those distinctions are no longer available, some plans labeled as having gap coverage may have additional coverage for only a minimal share of drugs.

In 2015, 90 percent of all PDP enrollees are in plans without additional gap coverage beyond what is required by law

9

. Overall, however, only 49 percent of PDP enrollees are potentially exposed to the gap in coverage if their spending exceeds the initial coverage limit. This lower percentage reflects the fact that LIS enrollees pay the same modest cost-sharing amounts in the gap as in the initial coverage period. In 2015, the vast majority of non-LIS Part D enrollees (84 percent) are enrolled in PDPs with no gap coverage beyond what is required by the ACA.

A larger share of MA-PD plans (44 percent) than PDPs (26 percent) offer additional gap coverage in 2015, and a much larger share of MA-PD plan enrollees than PDP enrollees are in such plans. Nearly half (45 percent) of MA-PD plan enrollees have at least some additional gap coverage beyond what the ACA requires, well above the 10 percent for PDP enrollees

9

. The higher level of additional gap coverage among enrollees in MA-PD plans occurs largely because Medicare Advantage plans are able to use payments received from the government for providing benefits covered under Parts A and B to reduce cost sharing and premiums under Part D.⁵⁵  Furthermore, because Medicare Advantage plans cover hospital and physician services and other Medicare benefits, they have stronger incentives than PDPs to offer at least some gap coverage to forestall the negative health and cost consequences that could arise if enrollees do not take their medications when they reach the gap. Despite these incentives, a majority of MA-PD plans offer no additional gap coverage. Overall, 77 percent of enrollees in PDPs or MA-PD plans are in plans with no gap coverage.

In earlier years, most Part D enrollees with gap coverage (beyond that required by law) were in plans that covered only some generic drugs in the gap. For example, in 2014, only about 3 percent of PDP enrollees and less than 1 percent of MA-PD plan enrollees had any significant gap coverage for brand-name drugs beyond the coverage that all plans must provide. Furthermore, most plans with gap coverage for generics included only a share of all generic drugs in that coverage. Although CMS no longer provides this breakdown, it is likely that the generosity of gap coverage beyond with the ACA requires plans to offer remains limited.

Enrollees in stand-alone Part D plans tend to pay substantially higher premiums for plans with gap coverage (beyond that which is required by law) compared to those without such coverage. On average, the weighted monthly premium for a stand-alone PDP offering additional gap coverage for at least some drugs is $71.00, about $38 per month above that for plans offering no gap coverage, despite the limited added coverage provided by these plans

10

.