To Switch or Be Switched: Examining Changes in Drug Plan Enrollment among Medicare Part D Low-Income Subsidy Enrollees

Appendix 1: Study Methodology

The analysis for this study is based on a 5-percent random sample of Medicare beneficiaries, obtained from the Centers for Medicare & Medicaid Services (CMS), for each year from 2006 to 2010. The sample dataset includes information on the characteristics of beneficiaries from the Master Beneficiary Summary File and Part D plan information from the Part D Plan Characteristics Files. Certain identifying information for both beneficiaries and drug plans, including the plan name, are encrypted. These standard CMS datasets were supplemented by a customized CMS dataset with supplemental data on the beneficiary enrollment decisions. The latter file was a preliminary version of a file being developed by CMS for use by researchers; the production version of the file has been recreated with the correct reassignment codes for 2007, but was not available in time for this study. The additional information in the customized dataset was essential to studying plan switching by LIS beneficiaries, some of whom are enrolled in new plans as a result of reassignments by CMS.

Defining LIS Status

Defining a sample of LIS beneficiaries has additional complexities because there are multiple versions of the Low-Income Subsidy and different ways to qualify. Medicare beneficiaries who qualify for full Medicaid benefits (dual eligibles), those enrolled in Medicare Savings Programs (e.g., Qualified Medicare Beneficiary (QMB), the Specified Low-Income Medicare Beneficiary (SLMB), Qualifying Individual (QI), or Qualified Disabled and Working Individuals (QDWI) programs), and those receiving Supplemental Security Income (SSI) are deemed eligible for the LIS and are not required to apply. Using data for December 2009, 87 percent of all LIS beneficiaries qualified in one of these ways (Table A1). Those with deemed coverage are all eligible for premium-free coverage provided they enroll in a benchmark plan. Some (11 percent of all LIS beneficiaries) are exempt from all cost sharing based on residence in a nursing facility or (starting in 2012) those receiving Medicaid home and community based services. Others with deemed coverage receive lower or higher copay levels, based on whether their incomes are below or above 100 percent of the federal poverty level (FPL). Other beneficiaries may apply for LIS status based on meeting income and asset requirements. About 10 percent of all LIS beneficiaries qualified via the application route for full LIS status, based on incomes up to 135 percent of the FPL. Applicants with incomes between 135 percent and 150 percent (who also meet asset requirements) qualify for partial LIS status (about 4 percent of all LIS beneficiaries), which provides different levels of premium support depending on incomes and limit cost sharing to 15 percent of a drug’s cost. For this study, we include only those partial LIS beneficiaries who qualify for full premium support. In total, the sample criteria include about 97 percent of all LIS beneficiaries.

Table A1: Part D Low Income Subsidy Program Enrollment Status, as of December 2009
Code Status Premium Cost Sharing Percent of All LIS Enrollees Included in Sample
01 Deemed; institutional residence $0 None 10.7% Yes
02 Deemed, <=100% FPL $0 Low 55.0% Yes
03 Deemed, >100% FPL $0 High 20.9% Yes
04 LIS, not deemed, <135% FPL $0 High 9.8% Yes
05 Partial LIS, >135% FPL $0 15% 0.9% Yes
06 Partial LIS, >135% FPL 25% 15% 1.0% No
07 Partial LIS, >135% FPL 50% 15% 0.9% No
08 Partial LIS, >135% FPL 75% 15% 0.7% No
NOTE: LIS is low income subsidy. FPL is federal poverty level.

We also require that beneficiaries maintain full LIS premium support over the time period for any particular sample. Most beneficiaries (96.3 percent) had no change in their LIS status (full LIS, partial LIS, or not LIS) over a two-year period from January 2009 to December 2010. Of those who made some change, 3.3 percent had only a single change in status, while 0.4 percent had two or more changes. To understand the nature of the exclusions, we determined that in the same two-year period (2009-2010), only 1.6 percent of those with LIS status in January 2009 lost that status at least once in the next 23 months. Thus, a larger share of those without continuous LIS status consisted of beneficiaries who gained LIS status after January 2009, whether through applying or gaining Medicaid eligibility. Elsewhere we have discussed the fact that LIS eligibility assessments and reassessments may affect participation rates, since eligibility from one year to the next is not always automatic. Of those who were LIS in December 2009, our data show that 1.9 percent lost LIS status as of January 2010 (most redeterminations are due on the first of the year). One-third (31 percent) of that group then regained their LIS status by the end of 2010. It is possible that these LIS beneficiaries, excluded because of changes in their LIS status, may change plans more often than those with continuous LIS status.

Defining Two-Year and Five-Year Samples of LIS Beneficiaries

From the one-year samples for the five years that constituted the overall dataset, we matched beneficiaries across years to build a series of samples of beneficiaries who were enrolled in the Part D program for each of four two-year periods (2006-07, 2007-08, 2008-09, and 2009-10) (Table A2). To be included in a two-year sample, beneficiaries (1) must have been alive and enrolled in a non-employer Part D plan1 continuously throughout the two-year period (or from July 2006 in the case of the 2006-07 file);2 (2) must have been eligible for full-premium LIS support (codes 01-05 in Table A1) in all months for which they were enrolled in a drug plan; and (3) must have been enrolled in a PDP during the entire time they were enrolled in Part D.

Table A2: Number of Part D Enrollees in Two-Year Samples, by Exclusion Criteria
  2006-07 2007-08 2008-09 2009-10
Total in 5% sample 2,431,158 2,475,122 2,519,634 2,565,873
Part D enrollment, December of Year 1 1,307,391 1,380,897 1,444,534 1,500,204
Part D enrollment, January of Year 2 1,354,075 1,424,688 1,479,202 1,521,299
Total LIS enrollees continuously enrolled in Part D and with full-premium LIS status (including both PDP and MA-PD enrollment) 424,099 408,442 421,140 432,989
LIS beneficiaries in PDPs and not employer plans = final two-year samples 345,824 324,986 332,901 335,057
Projected total enrollment for this sample 6.9 million 6.5 million 6.7 million 6.7 million

We also report in Table A2 the projected number of total LIS PDP enrollees who meet all the criteria for this analysis. These numbers are modestly lower than the actual numbers of LIS PDP enrollees due to the exclusions delineated above. Depending on the year, about 16 percent of LIS PDP enrollees are excluded from our analysis.

We built another sample of beneficiaries who were enrolled in the program for the entire 2006-2010 period (Table A3). To be included in the five-year sample, beneficiaries (1) must have been alive and enrolled in a non-employer Part D plan continuously from July 2006 through December 2010; (2) must have been eligible for full-premium LIS support during all these months; and (3) must have been enrolled in PDPs throughout all months of this time period. Of the sample that meets the first two criteria (continuous Part D enrollment with full-premium LIS support), 76 percent were in PDPs for five years (thus qualifying for the analysis sample), whereas 10 percent were in MA-PD plans for the entire five years, and 14 percent spent some time enrolled in each type of plan.

Table A3: Number of Part D Enrollees in the Five-Year Sample
(July 2006-December 2010)
Description of Sample Number
Total LIS enrollees continuously enrolled in Part D and with full-premium LIS status (includes both PDP and MA-PD enrollment) 313,422
Total LIS enrollees continuously enrolled only in PDPs and not employer plans with full-premium LIS status = final five-year sample 239,221
Projected total enrollment for this sample 4.8 million

For comparisons, we use a sample of those who are non-LIS beneficiaries for the entire analysis period, the same sample as in our earlier report.3 For the five-year file, the N is 586,897 for the full sample and 313,418 for the PDP-only sample. For the most part, non-LIS results are drawn from our previous report on switching by non-LIS Part D enrollees.

Defining Part D Enrollment Status

The Master Beneficiary Summary File identifies the plan in which a beneficiary is enrolled in any given month, but does not distinguish when the enrollment results from a beneficiary election as opposed to an assignment or reassignment by CMS. The supplemental file we received from CMS was created to capture from internal transaction files, for each month, the beneficiary enrollment code for the most recent enrollment transaction. The key enrollment codes in this file whether a beneficiary’s plan for a given months was the result of a beneficiary election (the beneficiary made an active choice of the plan), assignment by CMS (usually applied to the first time an LIS beneficiary was assigned to a plan), or reassignment by CMS (usually applied when an LIS beneficiary who reassigned to a new plan).4

Specifically, the enrollment status variable supplied by CMS defines, for each month during the analysis period, the “enrollment type” of the most recent change in enrollment (Table A4). The code is updated each time a new enrollment occurs. In January 2010, a “B” code means that a beneficiary selected a plan on her own either in the most enrollment period or the last time she changed plans. An “A” code means that the beneficiary remains in the same plan to which she was initially auto-enrolled by CMS. A “C” code captures facilitated enrollment, which have the same result as auto-enrollments. Facilitated enrollments are generally used for non-deemed LIS beneficiaries who are given the opportunity to enroll on their own before being assigned to a plan by CMS (deemed LIS beneficiaries are assigned to a plan, but retain the ability to switch to a plan on their own). These “A” and “C” codes are retained until the beneficiary is reassigned or selects a plan on her own, in which case her code would change to another code. An “H” code means that the beneficiary was reassigned by CMS to a different plan. Other codes identify less common situations, such as “E” or “F” codes (when the plan submits the auto-enrollment, generally in situations where one plan sponsor has multiple benchmark plans) or “D” codes (when a beneficiary is placed into a different plan as a result of a CMS-approved crosswalk, for example, when a sponsor consolidates existing plans). For most analytic purposes, the codes A, C, E, and F are combined as “A” for an initial CMS plan assignment. Where possible, beneficiaries with “D” codes (6 percent of the sample) are reassigned to the previous code, because a rollover election can occur to beneficiaries who elected their previous plan or were assigned to it. The few beneficiaries with “G” and “I” codes are combined as other and are generally excluded from tables that rely on enrollment status.

Table A4: Low Income Subsidy Enrollment Status Codes
Code Type of Enrollment Percent of All LIS Enrolleesi
A Auto-enrolled by CMS (generally applies to dual eligibles) 11.3% 15.7%
C Facilitated enrollment by CMS (generally applies to other LIS) 4.4%
E Plan-submitted auto-enrollment <0.1%
F Plan-submitted facilitated enrollment <0.1%
H Re-assigned by CMS 25.4% 25.4%
B Beneficiary election 52.3% 52.3%
D Rollover election: plan change as a result of a crosswalk between plans offered by the same plan sponsor 6.2% 6.2%
I Other plan-submitted enrollment 0.3% 0.3%
G Enrollment by the point-of-service vendor that serves LIS beneficiaries who lack a plan <0.1%
  TOTAL 100% 100%
NOTE: iPercent of LIS enrollees as of December 2009.

Defining Plan SwitchesThere is one significant limitation in the preliminary supplemental file available for this study. A substantial number of the reassignments by CMS that were effective in January 2007 were incorrectly coded in the transaction files as beneficiary elections. Thus, a beneficiary who was reassigned to a new plan effective January 2007 was coded “B” instead of “H.” Because codes are retained until the next transaction, the legacy of these incorrect codes continues through the end of the study period (December 2010), although its effect gradually diminishes over time. In the sample, about two-thirds of those eligible for reassignment and in a new plan in January 2007 are coded as if they selected the new plan, whereas in 2008 the similar share was 5 percent. Where noted in this report, certain results (especially for January 2007 or January 2008) are affected by this incorrect coding. CMS has recreated this file with the correct codes, but the new version was not available for this analysis.

For the LIS Part D enrollees in each two-year sample or the five-year sample, we define a year-end plan switch based on enrollment in a different plan in January of the second year, compared to December of the first year (plan elections and reassignments made at any time during the annual enrollment period are effective on January 1). Plan switches are further coded as either a switch by beneficiary choice or a switch by CMS assignment based on the enrollment status code for January. In a small number of cases, the type of switch cannot be determined.

We do not count year-end plan switches that are involuntary under the following two circumstances: (1) a plan enrollee whose plan (as designated by the contract ID and plan ID combination) changes, but where the old plan and the new plan are matched (“crosswalked”), and who accepts the automatic transfer to the crosswalked plan;5 and (2) any plan enrollee whose plan exits the program without any crosswalked plan and who therefore must select a new plan to remain in Part D.6

We also consider plan switches that occur outside the annual enrollment period, which include switches effective in any month other than January. For LIS beneficiaries, beneficiary elections may generally be made in any month. As with year-end switches, these switches are classified as either switches by beneficiary choice or switches by CMS assignment. The latter occur in several circumstances, such as when a beneficiary moves to a different region. Even beneficiaries who previously selected their own plans will be assigned to a plan by CMS if they do not select a plan in the new region. In addition, beneficiaries in a plan that exits the market in midyear are reassigned to a new plan if they do not select one. In 2010, one plan was excluded from the program, resulting in the reassignment of its enrollees. Tables in this report that show rates of plan switching outside the annual enrollment period exclude those reassigned to new plans, whereas tables showing total plan changes over five years include both types of switches outside the annual enrollment period.

Other Methods Considerations

Because the switching analysis presented here is based on individual decisions, the sample of beneficiaries for the analysis of each two-year file excludes newly eligible beneficiaries, beneficiaries who died during the relevant time period, and others who did not participate in Part D in both years. In addition, this analysis excludes from the estimate of switching rates various types of involuntary switches. As a result, the results reported here cannot be used to project the total net enrollment change from one year to the next. The net enrollment change is a product of how many people switch plans voluntarily, those who switch involuntarily as a result of plan exits, new enrollees to Part D, and those who have died or disenrolled from Part D.

The encryption of plan identifiers imposes some limitations on our study. Some plan sponsors operate under multiple contract numbers, often because they acquire the contract number from another plan sponsor through a corporate merger or acquisition. As a result, we cannot reliably distinguish between switches within plans offered by the same sponsor and those across plan sponsors. In addition, the use of encrypted plan identifiers makes it difficult to analyze plan characteristics that are not reported in the plan characteristics files.

In addition, the study is limited to the plan and beneficiary characteristics in the Medicare Beneficiary Summary File. We did not obtain for this study the prescription drug events (claims) for beneficiaries or claims from Medicare Parts A and B. Thus, we cannot calculate beneficiary risk scores or use measures of drug spending or other Medicare spending beyond the summary measures available in the Master Beneficiary Summary File.

Appendix 2: Benchmark Plans

LIS beneficiaries may choose to enroll in any plan, but premium-free coverage is only available from a designated set of “benchmark” plans that change annually.7 Each year, the benchmark is calculated separately for each of the 34 PDP regions based on the enrollment-weighted average premium bid submitted by stand-alone PDPs and MA-PDs or is set at the premium of the lowest-cost stand-alone PDP if no plan has a premium below the average. This average is a weighted average of premiums for PDPs and most MA-PD plans for the basic Part D benefit. Thus premiums corresponding to enhanced benefits are excluded, as are the value of any Part A/B rebate used by MA-PD plans. The weight for each PDP and MA-PD plans is based on the number of LIS beneficiaries enrolled in each plan.

In certain years, Part D sponsors have been permitted to waive a de minimis amount of the premium above the LIS benchmark for an LIS enrollee ($2 in 2007, $1 in 2008, and $2 in 2011 and thereafter) in order to retain enrollees in their premium-free status from the prior year. The de minimis policy was instituted by CMS on a demonstration basis for 2007 and 2008; it was made permanent by the Affordable Care Act, effective in 2011 (CMS has designated the amount as $2 each year). Plans may volunteer to waive the portion of their monthly premium that is a de minimis amount above the benchmark for eligible individuals, and thus CMS will not reassign enrollees away from these plans.

The availability of premium-free coverage for LIS Part D enrollees fluctuates from year to year, potentially exposing Part D LIS enrollees to higher costs unless they change to another zero-premium (“benchmark) drug plan.8 The total number of benchmark plans offered peaked at 640 in 2007 and dropped as low as 307 in 2010. Only 65 plans, however, retained benchmark status over the entire period from 2006 to 2010. Table A5 identifies on a national basis the number of benchmark plans available each year, as well as the how many of those plans lost benchmark status for the following year.

Table A5: Availability of Part D LIS Benchmark Plans and Changes over Time, 2006-2010
Annual enrollment period 2006-07 2007-08 2008-09 2009-10 2010
Number of benchmark plans in Year 1 409 640* 495** 308 307
Net change in benchmark plans, Year 1 to Year 2 +231* -145** -187 -1 NA
Number of plans losing benchmark status, Year 1 to Year 2 90 240 218 93 NA
Percent of plans losing benchmark status, Year 1 to Year 2 22% 38% 44% 30% NA
Number of enrollees in plans losing benchmark status, Year 1 to Year 2 1.6 mil 3.7 mil 2.5 mil 1.5 mil NA
Percent of enrollees in plans losing benchmark status, Year 1 to Year 2 20.0% 46.1% 31.8% 18.4% NA
Percent of enrollees in plans losing benchmark status, Year 1 to Year 2 (adjusted for other plans from sponsor) 11.7% 28.9% 25.6% 18.4% NA
NOTE: NA = Not applicable for this study. mil is million.
* Amount includes 157 plans qualifying in 2007 through the de minimis policy. Otherwise, there would have been 483 benchmark plans in 2007, up 74 from 2006.
** Amount includes 53 plans qualifying in 2008 through the de minimis policy. Otherwise, there would have been 442 benchmark plans in 2007, down 198 from 2007.

The year-to-year variation has been substantial. Between 2007 and 2008, the number of benchmark plans rose by 50 percent, but at the same time, over one-third of the benchmark plans in 2007 lost that status for 2008. A similar number of plans lost their benchmark status between 2008 and 2009, but the total number of benchmark plans decreased. The high volatility in these two years probably represents several factors, including strategic decisions by some large plan sponsors around the structure of their plan offerings and the level of premiums charged by different plans. In later years, various rules and guidelines from CMS have reduced somewhat the volatility in the market for benchmark plans.

The volatility in the availability of benchmark plans has differed by plan sponsor and by region. For example, benchmark plan offerings by UnitedHealth, the program’s largest plan sponsor, dropped from 34 regions to 10 regions and rose back to 25 regions between 2007 and 2009. Humana, the second largest plan sponsor, had benchmark plans in all 34 regions in 2007, but had no benchmark plans in 2009 and only three in 2010.

All LIS enrollees in the plans losing benchmark status are subject to paying a premium the next year, but some of them are eligible for reassignment to a new benchmark plan during the annual enrollment period. Table A5 shows how many LIS beneficiaries are in plans losing benchmark status across the period studied for this report. The largest number of beneficiaries affected came in 2008, when nearly half of all LIS beneficiaries enrolled in PDPs were affected. One complicating factor has been the reorganization by many plan sponsors of their overall array of plan offerings (for some, partly in response to acquisitions of other plan sponsors). Thus, some benchmark plans were effectively replaced by other offerings of the same sponsor.9 The share of LIS beneficiaries affected in 2008 is reduced considerably (from 46 percent to 29 percent) if we exclude situations where another benchmark plan option is available from the same sponsor. However, these transfers between plans may not occur automatically.

Key Findings

KFF Headquarters: 185 Berry St., Suite 2000, San Francisco, CA 94107 | Phone 650-854-9400
Washington Offices and Barbara Jordan Conference Center: 1330 G Street, NW, Washington, DC 20005 | Phone 202-347-5270 | Email Alerts: | |

The independent source for health policy research, polling, and news, KFF is a nonprofit organization based in San Francisco, California.