|Home | About | Journals | Submit | Contact Us | Français|
Medicare managed care may offer enrollees lower out-of-pocket costs and provide benefits that are not available in the traditional fee-for-service Medicare program. However, managed care plans may also restrict provider choice in an effort to control costs. We compared rates of voluntary disenrollment from Medicare managed care to traditional fee-for-service Medicare among Medicare managed care enrollees with and without a cancer diagnosis.
We identified Medicare managed care enrollees aged 65 years or older who were diagnosed with a first primary breast (n = 28331), colorectal (n = 26494), prostate (n = 29046), or lung (n = 31243) cancer from January 1, 1995, through December 31, 2002, in Surveillance, Epidemiology, and End Results (SEER) cancer registry records linked with Medicare enrollment files. Cancer patients were pair-matched to cancer-free enrollees by age, sex, race, and geographic location. We estimated rates of voluntary disenrollment to fee-for-service Medicare in the 2 years after each cancer patient’s diagnosis, adjusted for plan characteristics and Medicare managed care penetration, by use of Cox proportional hazards regression.
In the 2 years after diagnosis, cancer patients were less likely to disenroll from Medicare managed care than their matched cancer-free peers (for breast cancer, adjusted hazard ratio [HR] for disenrollment = 0.78, 95% confidence interval [CI] = 0.74 to 0.82; for colorectal cancer, HR = 0.84, 95% CI = 0.80 to 0.88; for prostate cancer, HR = 0.86, 95% CI = 0.82 to 0.90; and for lung cancer, HR = 0.81, 95% CI = 0.76 to 0.86). Results were consistent across strata of age, sex, race, SEER registry, and cancer stage.
A new cancer diagnosis between 1995 and 2002 did not precipitate voluntary disenrollment from Medicare managed care to traditional fee-for-service Medicare.
Medicare managed care may offer lower out-of-pocket costs and provide benefits that are not available in traditional fee-for-service Medicare.
Registry study of Medicare managed care enrollees that used Surveillance, Epidemiology, and End Results (SEER) records linked to Medicare enrollment files to obtain rates of voluntary disenrollment to fee-for-service Medicare for cancer patients in the 2 years after a cancer diagnosis and cancer-free enrollees.
In the 2 years after diagnosis, patients diagnosed with breast, colorectal, prostate, or lung cancer were less likely to disenroll from Medicare managed care than their matched cancer-free peers. These results were consistent across strata of age, sex, race, SEER registry, and cancer stage.
If voluntary disenrollment is in fact the way that beneficiaries “vote with their feet,” then these results suggest that those facing a serious illness are satisfied with Medicare managed care.
Because Medicare does not process claims for managed care enrollees, no information was available on the medical or treatment information for cancer-free enrollees. No information was available on enrollees’ socioeconomic status or relationships with specific providers.
From the Editors
Managed care was introduced to the Medicare program with the hope that competition between plans would help to control escalating costs, expand preventive services, and improve coordination of care (1,2). During the 1990s, the number of Medicare beneficiaries participating in managed care increased substantially, from 1.3 million (4% of beneficiaries) in 1990 to 6.3 million (16% of beneficiaries) in 2000 (3). Although the availability of plans and beneficiary participation vary geographically, more than 80% of Medicare beneficiaries now have access to at least one managed care plan (4).
Managed care organizations generally aim to improve disease prevention and management while containing the costs of care. However, the mechanisms that they use to achieve these goals, such as primary care gatekeeping, provider networks, and higher cost sharing for out-of-network services, may restrict enrollees’ access to specific providers and create barriers or disincentives to the use of expensive services (5). These mechanisms may deter enrollment and retention of individuals with chronic or complex diseases, particularly conditions for which access to specialized treatment centers or physicians is perceived as advantageous. Numerous studies have found that Medicare managed care plans tend to attract and retain beneficiaries who are younger and healthier and have lower pre-enrollment medical expenditures than beneficiaries in the traditional Medicare indemnity insurance (ie, “fee-for-service”) program (6–15). Survey results indicate that voluntary disenrollment from a Medicare managed care plan may reflect beneficiaries’ perceptions of poor quality of care, impaired access to services or providers, or overall dissatisfaction with their plans (8,16).
With more than 55% of all cancer diagnoses and 70% of all cancer deaths occurring in people aged 65 years or older (17), the experience of cancer patients in the Medicare program is of particular concern. Cancer patients and survivors typically require regular visits with specialists, coordination of care among multiple providers, and frequent testing to monitor disease. For the past decade, there has been increasing emphasis on the value of provider expertise in treating complex medical conditions such as cancer (18,19). Because managed care plans may constrain enrollees’ choice of providers (20), participation in Medicare managed care may feel especially restrictive to enrollees with cancer. Problems with access have been correlated with beneficiary dissatisfaction (21) and may also prompt disenrollment (22,23). However, Medicare managed care plans often provide benefits that are not available in the traditional fee-for-service segment of Medicare. Perhaps more importantly, because participation in a Medicare managed care plan generally entails lower out-of-pocket health care spending than fee-for-service Medicare (24), enrollees may be reluctant to disenroll when facing a serious illness that requires substantial medical care utilization. Our objective was to compare rates of voluntary disenrollment to fee-for-service Medicare among Medicare managed care enrollees with and without a cancer diagnosis.
Our primary data source was the linked Surveillance, Epidemiology, and End Results (SEER)–Medicare dataset. The SEER program, sponsored by the National Cancer Institute, is a consortium of population-based cancer registries in selected states and metropolitan areas. The areas covered by SEER included approximately 14% of the US population during the 1990s and more than 20% since 2000. The registries collect data on all incident cancers in their coverage area, including clinical and sociodemographic characteristics, with active follow-up for date and cause of death. The SEER registries maintain high standards of data quality, and the program's overall completeness of case ascertainment is 98% (25).
For cancer patients aged 65 years or older residing in SEER areas, Medicare files have been linked to SEER records (25). Medicare is the primary health insurer for 97% of Americans aged 65 years or older. Nearly all beneficiaries receive inpatient hospital, skilled nursing facility, and home health benefits under Medicare Part A. Ninety-five percent of beneficiaries subscribe to Medicare Part B, which covers physician services and outpatient care. The SEER–Medicare dataset contains Medicare enrollment information, including monthly indicators for participation in a Medicare managed care plan, and cancer registry data for beneficiaries with fee-for-service or Medicare managed care coverage. Medicare claims are available for beneficiaries in fee-for-service Medicare but not for managed care enrollees. The National Cancer Institute and Centers for Medicare and Medicaid Services have also made available the Medicare enrollment files for 5% of beneficiaries, randomly sampled from SEER areas, who have never been diagnosed with cancer.
Additional information regarding each beneficiary's Medicare managed care plan and county of residence at the time of diagnosis was obtained from the National Cancer Institute and the Centers for Medicare and Medicaid Services. Plan data provided by the Centers for Medicare and Medicaid Services included plan tax status (for profit vs not for profit), plan type (group/staff model or independent practice association [IPA]), and plan participation in Medicare during the study period (continued, terminated, withdrew, contract not renewed). We obtained annual, county-level estimates of Medicare managed care penetration (ie, the proportion of Medicare beneficiaries enrolled in a managed care plan) from the Area Resource File (26).
We identified all beneficiaries diagnosed with a first primary breast, colorectal, lung, or prostate cancer from January 1, 1995, through December 31, 2002, who were enrolled in a Medicare managed care plan at the time of cancer diagnosis. We included beneficiaries who were at least 65.5 years old and had been enrolled in a risk-based Medicare managed care plan for at least 6 months before diagnosis. Beneficiaries lacking Part A or B of Medicare and those diagnosed with cancer only at death were excluded. We also excluded beneficiaries enrolled in cost-based Medicare managed care plans because these plans enroll fewer than 1% of all Medicare beneficiaries (27) and do not bear the same financial risks for beneficiary care as the risk-based plans (1). In total, we identified 28695 patients who were diagnosed with a first primary breast cancer, 26901 patients with colorectal cancer, 37143 with prostate cancer, and 31783 with lung cancer.
Each beneficiary with cancer was matched to one without cancer, as identified in the Medicare 5% sample, by exact year of birth, sex, race, and SEER region. Race was categorized as white, black, Hispanic, Asian/Pacific Islander, or Native American, as recorded in the Medicare Enrollment Database; beneficiaries with unknown race were excluded. Beneficiaries in the cancer-free cohort were enrolled in a Medicare managed care plan at the time that their matched cancer patient was diagnosed with cancer and for the preceding 6 months. Cancer-free beneficiaries were matched to only one beneficiary with cancer within each cancer site, but they could also be matched to a beneficiary with a different cancer. We successfully matched 99% of beneficiaries with breast cancer, 99% of those with colorectal cancer, 78% of those with prostate cancer, and 98% of those with lung cancer. Compared with the beneficiaries who were successfully matched, unmatched beneficiaries with cancer were slightly older and somewhat less likely to be white. The final sample for analysis included more than 115000 matched pairs (n = 28331 for breast cancer analysis, n = 26494 for colorectal cancer analysis, n = 29046 for prostate cancer analysis, n = 31243 for lung cancer analysis). For the purpose of studying Medicare managed care disenrollment in beneficiaries with less common cancers, we also identified and matched three additional cohorts: enrollees with non-Hodgkin lymphoma (n = 7463), acute leukemia (n = 292), and soft tissue sarcoma (n = 560).
For enrollees with cancer, the primary endpoint of the analysis was voluntary disenrollment from Medicare managed care to traditional fee-for-service Medicare within 2 years of cancer diagnosis. For each cancer-free enrollee, the endpoint was voluntary disenrollment within 2 years of the matched cancer patient's diagnosis. The time origin—month of diagnosis for the beneficiary with cancer—was identical for both members of each matched pair. We had complete follow-up for all beneficiaries in the sample. Observations were censored when one of the following events occurred: 1) death; 2) withdrawal, non-renewal, or termination of the beneficiary's plan from the Medicare program; or 3) admission to hospice. Although beneficiaries are not automatically disenrolled from Medicare managed care at the time of hospice admission, hospice care is reimbursed on a per diem basis outside of the Medicare managed care capitated payment rate (28).
We used Kaplan–Meier methods to estimate the percentage of patients who remained continuously enrolled in Medicare managed care. Hazard ratios (HRs) for voluntary disenrollment, comparing the cancer and matched cancer-free cohorts, were estimated separately for each cancer site in Cox proportional hazards regression models. Plots of the Schoenfeld residuals indicated that the proportional hazards assumption was not violated (29). Models were fit by use of estimating equations and a robust covariance matrix to account for paired data (30). In multivariable models, we adjusted for plan type (group or staff model vs IPA), plan tax status (for profit vs not for profit), county-level Medicare managed care penetration (<20%, 20%–39%, or ≥40%), and year of cancer diagnosis. These factors have previously been associated with voluntary disenrollment from Medicare managed care (15,31). In the adjusted models, we also examined interactions between cancer status and plan characteristics, and we estimated hazard ratios stratified by matching characteristics (age at diagnosis, sex, race, and SEER area). In analyses of cancer patients only, we evaluated the relationship between stage at diagnosis and disenrollment, with stage categorized as in situ, local, regional, or distant according to the SEER historic staging scheme (32). All statistical tests were two-sided, and all analyses were performed in SAS (version 9.1, Cary, NC).
Age at diagnosis, sex, race, geographic location, and the cancer stage distribution of each cancer cohort are shown in Table 1. The frequency distributions of these characteristics are identical in the respective cancer-free cohorts. Plan characteristics and county-level Medicare managed care penetration were similarly distributed among the cancer and cancer-free cohorts (Table 2). Most beneficiaries were enrolled in independent practice association plans, and most were in for-profit plans. At least half of the beneficiaries lived in counties with moderate (20%–39%) Medicare managed care penetration.
The number of voluntary disenrollments to fee-for-service Medicare and censored events for each cancer site and for each matched cancer-free cohort are shown in Table 3. The frequency of disenrollment varied from 5% to 13% among enrollees with cancer and from 13% to 14% among their matched cancer-free peers. In all cohorts, few beneficiaries (<5%) were affected by the withdrawal or termination of plans from the Medicare program. The frequency of death during the follow-up period varied across cancer sites, with the greatest mortality observed in beneficiaries with lung cancer.
Beneficiaries with cancer were less likely to disenroll from Medicare managed care than their matched cancer-free peers (Figure 1). This result was observed in each cancer site and persisted after adjustment for plan characteristics, Medicare managed care penetration, and year of diagnosis (Table 4) (adjusted HR for disenrollment, comparing those with cancer to their matched cancer-free peers = 0.78, 95% confidence interval [CI] = 0.74 to 0.82 in the analysis of enrollees with breast cancer; adjusted HR = 0.84, 95% CI = 0.80 to 0.88 in the analysis of enrollees with colorectal cancer; adjusted HR = 0.86, 95% CI = 0.82 to 0.90 in the analysis of enrollees with prostate cancer; and adjusted HR = 0.81, 95% CI = 0.76 to 0.86 in the analysis of enrollees with lung cancer). Plots of estimated 6-month interval hazards for disenrollment (data not shown) indicated that our findings were consistent throughout the 2-year period evaluated.
We found an increased risk of voluntary disenrollment from Medicare managed care, regardless of cancer status and across all cancer sites, among beneficiaries in independent practice association plans and among those in for-profit plans (Table 4). Across cancer sites, there was a consistent inverse relationship between county-level Medicare managed care penetration and disenrollment, with enrollees living in high-penetration counties having lower disenrollment rates than those in counties with moderate penetration and those living in low-penetration counties having higher disenrollment rates than those in counties with moderate penetration. We also examined interactions between cancer status and plan type and cancer status and plan profit status. Analysis of interaction terms (data not shown) indicated that there were no subgroups of plan type or profit status in which beneficiaries with cancer were more likely to disenroll than their cancer-free peers.
Across strata of each matching characteristic, the relationship between cancer diagnosis and hazard of voluntary disenrollment from Medicare managed care was consistent (data not shown). We observed an association that was inconsistent with our overall finding in only one stratum: patients with breast cancer in the New Jersey SEER registry were more likely to disenroll than their matched cancer-free peers (HR = 1.22, 95% CI = 1.06 to 1.41).
In analyses of cancer patients only, stage at diagnosis had no impact on the likelihood of disenrollment. This finding is illustrated for the colorectal cancer cohort (Figure 2); results for the other cancer sites were similar. When we repeated our primary analysis in patients with less common cancers—non-Hodgkin lymphoma, acute leukemia, and soft-tissue sarcoma—we found no statistically significant differences in disenrollment between beneficiaries with and without cancer (data not shown).
In this population-based analysis of Medicare managed care enrollees, a cancer diagnosis did not precipitate voluntary disenrollment from a Medicare managed care plan to fee-for-service Medicare. In the 2 years following cancer diagnosis, beneficiaries with cancer were less likely to disenroll than their matched cancer-free peers. This result was evident across a spectrum of different cancers and in nearly all strata of age, sex, race, and geographic region. Although characteristics of Medicare managed care plans influenced the likelihood of disenrollment, they did not meaningfully confound or modify the relationship between cancer status and disenrollment.
We examined voluntary disenrollment from Medicare managed care to fee-for-service Medicare presuming that the frequency with which seniors exercise their option to disenroll signals their dissatisfaction with Medicare managed care, which in turn reflects actual or perceived problems with access to care or quality of care in their plans. If voluntary disenrollment from Medicare managed care is, in fact, beneficiaries’ way of “voting with their feet,” then our results suggest that enrollees facing a serious, potentially life-threatening illness are as satisfied with Medicare managed care, if not more so, than their cancer-free peers. This conclusion is consistent with surveys that have compared Medicare beneficiaries in managed-care plans with those in fee-for-service Medicare and found similar levels of overall satisfaction (33).
Our results may also be related to reduced cost sharing for Medicare-covered services and the availability of outpatient prescription drug coverage in Medicare managed care plans. During the study period, Medicare beneficiaries lacking employer-sponsored supplemental insurance faced deductibles and coinsurance payments for most covered health services (including most physician-administered chemotherapy, covered under Part B of Medicare) and the full costs of outpatient prescription drugs that were not covered under Part B. For these beneficiaries, enrolling in a Medicare managed care plan was generally a lower-cost alternative to purchasing an expensive supplemental “Medigap” policy (34,35). Several studies (31,36) have shown that the absence or removal of a prescription drug benefit was associated with a greater likelihood of disenrollment in Medicare managed care plans. Surveys of Medicare beneficiaries have found greater financial access to care and greater satisfaction with costs among Medicare managed care enrollees than beneficiaries with traditional fee-for-service coverage (21,37). Medicare managed care enrollees may also have remained in their plans if they were concerned about obtaining a supplemental insurance policy upon returning to fee-for-service Medicare. For beneficiaries who did not enroll in a managed care plan when they first became eligible for Medicare, resumption of a self-purchased supplemental policy was not guaranteed (38).
An alternative explanation for our findings is that after developing a serious illness, seniors may be less willing to make a change in insurance coverage, even if such a change would facilitate access to a broader choice of providers. A preference for the status quo, as described in the economic literature (39–41), has been observed in health insurance decisions (42) and may be more common with increasing age (43). Substantial gaps in knowledge about the Medicare program in general and Medicare managed care plans in particular (44,45) suggest a mechanism for enrollee inertia: beneficiaries may be especially reluctant to change their insurance enrollment if they do not fully understand their options.
Breast, colorectal, prostate, and lung cancers are often treated by community-based physicians and do not necessarily require services that are available only at specialized centers. Therefore, even Medicare managed care plans that limit access to specific providers may still offer satisfactory care to patients with these common cancers. Positing that beneficiaries with cancers that are rarer or require more complex or less standardized treatment regimens might be more inclined to disenroll from Medicare managed care, we repeated our analyses in cohorts with non-Hodgkin lymphoma, acute leukemia, and soft-tissue sarcoma, and we found no effect of the cancer diagnosis on the likelihood of disenrollment. We also expected that disenrollment would be more common among enrollees diagnosed with advanced-stage cancer, for whom a generally poor prognosis may provoke a sense of urgency and a desire for specialty consultations and investigational therapies. However, we observed no difference in disenrollment by stage at diagnosis. Therefore, despite the array of mechanisms used by Medicare managed care plans to manage care and control costs, cancer patients may not feel so restricted in their choice of providers that they will leave managed care and return to traditional fee-for-service Medicare.
Our results echo those of Riley et al. (46) who evaluated disenrollment in Medicare managed care plans in the late 1980s and found that beneficiaries diagnosed with cancer after enrollment in Medicare managed care were less likely than other beneficiaries to disenroll. With the expansion of the Medicare managed care market since 1990 (3), we were able to evaluate a much larger sample of beneficiaries enrolled in a more recent era. Although the Medicare managed care market has evolved considerably in the past two decades (47,48), the trends that we observed were similar to those reported by Riley et al. (46). Our study also corroborates the low disenrollment rates found by Field et al. (49) in their analysis of enrollees with cancer in health maintenance organizations in Medicare, Medicaid, and commercial plans.
Several limitations of our analysis warrant mention. Because Medicare does not process claims for managed care enrollees, we had no information on the medical conditions and health care utilization of beneficiaries in the cancer-free cohorts. To the degree that health status and the incidence of other diseases are associated with age, sex, race, and geographic location, we would expect the matched cohorts with and without cancer to be similar in this regard. We also had no information on enrollees’ socioeconomic status, which has been shown to influence participation in Medicare managed care (8,10,15,31,50), nor did we have information on beneficiaries’ relationships with specific providers. Our dataset did not permit identification of specific plan factors, such as the type of supplemental benefits offered and cost-sharing requirements, which are also likely to influence Medicare managed care enrollment and disenrollment decisions (31,36,51).
The geographic scope of the study sample may limit the generalizability of our findings. Although SEER areas are relatively representative of the US population that is 65 years or older (25), health insurance markets vary widely and our sample is heavily influenced by the experience of beneficiaries in California, a state with a mature managed care market. The applicability of our findings to other Medicare managed care markets may be limited. In addition, we did not examine the experience of beneficiaries in private Medicare fee-for-service plans, a growing segment of the Medicare program (4).
Our findings have practical implications for Medicare beneficiaries, managed care plans, and policymakers. In an effort to improve continuity of care, stabilize the Medicare managed care market, and discourage beneficiaries from “gaming the system,” the 2003 Medicare Modernization Act eliminated the opportunity to switch plans monthly and limited beneficiaries to an annual plan election followed by a brief period during which they could make one additional plan change (52). Although the new policy, implemented in 2006, resembles the annual open enrollment period commonly offered by employers who provide a choice of commercial insurance plans, the “lock-in” provision in Medicare has prompted concern that seniors receiving dissatisfactory or substandard care might suffer as a result of reduced choice. However, we did not observe an exodus of enrollees subsequent to cancer diagnosis during our study period, before 2006, when monthly opt-out was possible. Thus, the move from monthly to yearly open enrollment appears unlikely to be problematic for most Medicare managed care enrollees, including those with a serious illness.
In addition to limiting the frequency of changes in plan enrollment, the Medicare Modernization Act encourages the expansion of Medicare managed care to include a wider array of plans and a larger number of beneficiaries, prompting concern about the quality of care in Medicare managed care, especially among vulnerable subgroups of the Medicare population (53,54). The adoption of increasingly sophisticated beneficiary-level risk adjustment methods for determining Medicare managed care payments may encourage plans to focus more on the needs of seriously ill beneficiaries and thus enhance services. Our results suggest that Medicare managed care enrollees with cancer are sufficiently satisfied with their care to remain in a managed care plan. Monitoring of access to care and quality of care, within both the fee-for-service and managed care segments of Medicare, remains a priority.
The enrollment and disenrollment decisions of Medicare beneficiaries also have financial implications for the Medicare program. Disproportionate disenrollment of sick enrollees would exacerbate favorable selection within the managed care sector of Medicare and simultaneously would increase the average profit per enrollee for Medicare managed care plans. Our findings indicate that enrollees with cancer do not disenroll from Medicare managed care plans to traditional fee-for-service Medicare at a disproportionate rate. Analysis of enrollees with other serious conditions would complement these results and advance our understanding of the causes and ramifications of Medicare managed care disenrollment.
R03 HS14831 was co-founded by the Agency for Healthcare Research and Quality and the National Cancer Institute; R21 CA98353 was funded by the National Cancer Institute.