| HOME | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|---|
| ||||||||||||||||||||||||||||||||
Correspondence: Address correspondence to Marion Danis, MD, Department of Clinical Bioethics, Warren G. Magnuson Clinical Center, National Institutes of Health, Building 10, Rm. 1C118, Bethesda, MD 20892-1156. E-mail: mdanis{at}nih.gov
 |
Abstract |
|---|
 TOP Abstract MethodsResultsDiscussionReferences |
|---|
Key Words: Medicare • Patient participation • Insurance benefits • Health priorities
Information about Medicare enrollees' preferences regarding their insured benefits is sparse. Research has generally focused on measurement of their knowledge about the Medicare program (Murray & Shatto, 1998), satisfaction with the quality of care (Rosenbach, Acamache, & Khanddker, 1995), and identification of factors associated with the likelihood of enrolling or disenrolling in managed care plans. A small amount of information exists regarding the relationship of benefit preferences to selection of supplemental insurance (Vistnes & Banthin, 1997–1998). The value of assessing enrollees' benefit preferences, in the manner demonstrated here, lies in the possibility of incorporating enrollees' views in efforts to set Medicare policy.
|
Methods |
|---|
|
TOPAbstractMethodsResultsDiscussionReferences |
|---|
Study Instruments
Volunteers participated in structured small group exercises using CHAT: Choosing Healthplans All Together, a simulation exercise designed to allow groups of laypersons to construct health plans within the constraint of limited resources, as reported previously (Danis, Biddle, & Goold, 2002). A version of the CHAT exercise specifically based on the actuarial costs of typical senior Medicare enrollees was developed for use in this study.
The first step of the exercise uses a circular game board in which 14 insurance benefit categories are represented (Figure 1). Participants select their insurance package by distributing pegs among the holes on the board. The circular board is designed to avoid any hierarchical presentation of choices that might unduly influence the selection process. Participants can select Basic, Medium, or High options for each benefit category or may forgo a category altogether, thus offering the possibility of numerous variations in the insurance package design. An instruction manual describes the features of each of the benefit categories and their associated options (Table 1). All materials are written to be understood at a sixth-grade reading level.
|
|
Actuarial costs for individual services were estimated on the basis of a breakdown of standard managed care plan costs for major medical categories (inpatient, outpatient, primary care, specialist, radiologic and laboratory tests, other medical services, pharmacy, and mental health); these relative costs were rounded to the nearest $8 so that they can be selected by using the pegs. Costs for additional categories not routinely included in a typical Medicare managed care package, such as dental services and long-term care, were determined as a product of the expected utilization frequency per member and the estimated cost of the service utilization. For example, if a service was expected to be used by 1 out of 1,000 members on a monthly basis and the cost of providing this service was estimated to be $8,000, then the estimated cost of providing this service was estimated to be 1/1,000 x $8,000 or $8.00 PMPM, which in this example would be represented as 1 peg.
The cost of insuring long-term care required arbitrary assumptions because it is currently unavailable to the Medicare population. It was therefore estimated based on expenditures incurred by the Medicaid program for skilled nursing facility (SNF) and intermediate care (IC) facility use by the elderly population. Distributing Medicaid expenses for the elderly population of $29.6 billion/year for SNFs and IC facilities among 38 million Medicare beneficiaries yields an annual cost of $6,630 or $65/month (which would be equivalent to 8 pegs). We assumed the utilization of long-term care through the Medicaid program is dampened by the eligibility requirement to exhaust personal financial assets, and that use of long-term care and actuarial costs might be higher otherwise. We therefore estimated the value for full coverage of long-term care (the Medium option) at 12 pegs ($96 PMPM) and for the coverage of half of long-term care costs (the Basic option) at 6 pegs ($48 PMPM).
The 14 services were assigned a total of 83 holes on the game board. The 50 pegs thus permit participants to select 60% of the services offered in the exercise (Figure 1).
In the second step of the exercise, participants spin a roulette wheel and receive health event cards describing illness scenarios and the associated consequences of coverage choices, including out-of-pocket payment responsibilities, access, and choice of provider or treatment. Players read their events aloud to the group and are asked to comment on their previous choice of coverage.
During the exercise, selection occurs four times to allow participants to make choices and face consequences (a) alone, (b) in groups of three, (c) as an entire group, and (d) once again alone. Players receive event cards following the first two cycles of benefit selection. This sequence of repeated decision cycles is intended to promote group decision making (Duke & Greenblat, 1981), including discussion and deliberation about what participants consider valuable and why. Repeated cycles also allow comparison of individual and group choices. Data were collected on recording sheets at the conclusion of each cycle (for more information, contact the authors).
Self-administered questionnaires were given to participants before and after playing the exercise. The pregame questionnaire included the following: sociodemographic items, health status of the participant and his or her immediate family, experience with chronic illness in the family, self-reported health services use, and out-of-pocket health care expenditures during the prior 12 months. The postgame questionnaire asked whether the participants found the CHAT game understandable, informative, easy to understand, and easy to do on a 4-point scale (1 = very easy, 2 = fairly easy, 3 = fairly hard, and 4 = very hard). In the postgame questionnaire, respondents were also asked how strongly they agreed or disagreed with statements about the exercise, including the following: the way the group reached its decision was fair; my views were considered and taken into account; the way the group reached its decision was equally fair to each member of the group; my own choice of a plan is very different from what the group chose; and I was satisfied with the group's decision.
Participant Protection
This project was approved by the Office of Human Subjects Research at the National Institutes of Health, and the institutional review boards at the University of Michigan, the University of North Carolina at Chapel Hill, and Duke University. During the game, participants were given an alias to preserve their anonymity to the researchers.
Data Analysis
We used descriptive statistics to describe the study participants and the types of coverage and flexibility of restrictions they selected. We considered the selection of Basic, Medium, or High coverage as a measure of the degree of coverage restriction and scored it by using a 3-point scale (1 = Basic, 2 = Medium, and 3 = High). We calculated the overall restrictiveness of the benefit package as the mean of the restrictiveness scores for the benefits the participants selected.
Analyses included chi-square statistics and Fisher's exact test for the analysis of categorical variables and the calculation of means, standard deviations, and Student t tests for continuous variables. We used the Mantel–Haenszel chi-square statistic (Mantel & Haenszel, 1959) to examine linear relationships between pairs of ordinal variables. We used McNemar's chi-square test (1947) to assess the degree of agreement between individual health care coverage choices made during the first and fourth cycles of the game. To use McNemar's test, we recoded coverage choice for each of the 14 service types into a dichotomous indicator (i.e., coverage was either selected or not selected). Results are reported as statistically significant if p <.05. We conducted all analyses by using SAS statistical software, version 6.12 (SAS Institute, Cary, NC).
Participant responses to the pregame questionnaire and the first cycle of the game were considered independent observations and analyzed as appropriate. However, because the postgame questionnaire and the final cycle were conducted after the group exercise, we initially assumed that the responses were not independent between individuals. To test this assumption, we calculated the intraclass correlations by using the group as the class variable. With the exception of two variables, intraclass correlations were less than.10, and those that were larger did not differ significantly from 0. Thus, we rejected the assumption of nonindependence and did not correct for intraclass correlation.
|
Results |
|---|
|
TOPAbstractMethodsResultsDiscussionReferences |
|---|
|
Individual Insurance Preferences
Individual participants chose coverage for hospitalization, pharmacy, primary care, home health, dental, tests, specialty care, and vision care most frequently (Table 3). Half or more the participants opted for coverage above the basic level for all of these services except tests (Table 3). About two thirds of individuals selected long-term coverage both initially and in the final cycle. Participants included an average of 9.0 services with intermediate flexibility (score of 1.60, about midway between basic and medium levels) on a scale from 1 (tightly managed care) to 3 (more flexible care) during their first selection, and 9.2 services with flexibility of 1.56 during their final selection. There was no statistically significant difference between initial and final individual selections in the total number of covered benefits selected or the degree of flexibility of individual choices. However, the selection of particular services did vary in frequency. Those services more frequently selected at the conclusion of the exercise were tests, specialty care, mental health care, and coverage for the uninsured. Services that were less frequently selected by individuals in the last cycle of the game included complementary medicine and last chance coverage. When we examine the degree of concordance between Cycle 1 and Cycle 4 selections by using McNemar's test, only last chance (p =.008), tests (p =.028), specialty care (p =.011), and uninsured coverage (p =.031) revealed significant disagreement. With the exception of last chance coverage, participants were significantly more likely to select each of these service types during the final cycle.
|
Group Preferences
In the group selection cycle (Cycle 3), all 10 groups selected coverage for hospitalization, home health, pharmacy, dental services, primary care, specialty care, tests, and vision care (Table 4). The only service selected at higher than the basic level by more than half the groups was hospitalization. Other categories were selected by some groups but not others. The basic level of long-term care was selected by 7 of 10 groups and not chosen at all by the other three, and 7 of 10 groups chose to include uninsured coverage. All but one group included mental health services. Only one group selected complementary medicine coverage and no groups chose to include coverage for last chance treatments. The 10 groups selected an average of 11.3 types of coverage (SD = 0.2, range = 10–12) with a restrictiveness score of 1.2 on a scale from 1 (tightly managed care) to 3 (more flexible care).
|
The vast majority of individual participants, 87 of 112 individuals (78%), indicated that they would be willing to abide by the health care plan developed by their groups. Seven individuals did not answer this question. Individual sociodemographic characteristics were not associated with the degree of willingness to abide by the group decision. With the exception of mental health (selected by 16% of nonabiders vs. 41% of abiders, Fisher's p =.031), individual benefit choices were not significantly different between those who would and would not abide by the group's decision. However, those individuals who were unwilling to abide by the group choice were more dissatisfied with the process of the group selection. Nonabiders were less likely to agree with the statement "their views were considered by the group" (24% vs. 19%, p =.014).
The group choices (Cycle 3) and final individual choices (Cycle 4) were compared for the 25 participants who indicated that they were unwilling to accept the coverage selections of their groups. We observed significant differences in coverage for the uninsured (McNemar's = 10.29, p =.001) and for "other services" (McNemar's = 5.33, p =.001), with the primary source of disagreement between the group and final cycle choices being that the individuals did not choose coverage but their groups did.
In their evaluation of the process, participants strongly agreed (78%) or agreed somewhat (20%) that the CHAT exercise was enjoyable, strongly agreed (66%) or agreed somewhat (30%) that it was easy to understand, strongly agreed (62%) or agreed somewhat (33%) that it was easy to do, and strongly agreed (65%) or agreed somewhat (32%) that it was informative. Participants with varying amounts of educations gave equal assessments of how enjoyable and informative the exercise was; those with less education were less likely to agree that the game was understandable (p =.03) and easy (p =.01) but more likely to agree that they learned a lot from it (p =.004).
When asked about the group deliberations in the postgame questionnaire, 83% agreed that the way the group reached its decision was fair; 83% agreed that their views were considered and taken into account; and 88% agreed that the way the group reached its decision was equally fair to each member of the group. Although 41% agreed with the statement that their own choice of a plan was very different from what the group chose, 86% agreed that they were satisfied with the group's decision.
|
Discussion |
|---|
|
TOPAbstractMethodsResultsDiscussionReferences |
|---|
We have previously reported Medicare enrollees' choices for insurance benefits using a simpler instrument that assumed an indemnity model of care and did not permit group decision making. In contrast, the exercise reported here is based on a managed care model of insurance and adds a process of group deliberation. In the prior study, enrollees gave highest priority to hospitalization, outpatient care, prescription drugs, eye care, and home care or respite (Danis, Biddle, Henderson, Garrett, & DeVellis, 1997). In the present study, the CHAT exercise makes it possible to offer more complex trade-offs so that, rather than necessarily or exclusively trading one benefit area for another, participants can choose a larger number of services if they are willing to accept more restricted or highly managed care. Furthermore, in keeping with the nature of insurance, which is inherently a group product, the CHAT exercise involves trade-offs between individuals. It is noteworthy that groups selected a greater number of services than individuals selected. This strategy for accommodating varied individual preferences appears to have been effective as indicated by the vast majority of individuals reported being willing to abide by their group's decisions.
Several limitations of the study must be acknowledged. The actuarial estimates for services, particularly pharmacy benefits, which have dramatically risen in cost, were applicable at the time participants were enrolled and must be revised in order to measure current opinion. The study design, which involved the use of prescheduled group sessions, precluded the possibility of a larger and randomly selected sample of Medicare enrollees; thus we cannot generalize broadly to the population of Medicare enrollees. A random sample, for instance, would have included a broader educational level, although the representation of income levels in our study was quite broad. The range of benefits offered in the exercise, which was limited by the need to find accurate actuarial information, was not exhaustive. Recent suggestions for new benefits such as Medicaring (Lynn et al., 1999) or a community-based long-term care option (Leutz, Capitman, & Green, 2001) might be worthwhile additions that participants would have selected. Finally, the extent to which selections made in a decision exercise would be acceptable in reality remains to be examined.
Several of these limitations will be addressed through ongoing research. Through development of an electronic version of the exercise, updated actuarial information will be easily incorporated. The design of a Web-based version of the exercise will facilitate systematic sampling of large number of individuals, which will complement data collected through the small group deliberation exercises. The extent to which choices made during the CHAT exercise are a valid reflection of actual insurance selections will be examined in future research.
The findings we report have several implications. First, they suggest that Medicare enrollees may find solutions for constraining Medicare costs through managed care acceptable if they can have input into the allocation process. This would be a welcome possibility for several reasons. Many managed care plans have found it difficult to continue to offer plans to Medicare enrollees, and those that remain did so by using survival strategies such as cutting services, or increasing enrollees' out-of-packet costs (Benko, 2000). Managed care plans are likely to continue to be important, particularly for Medicare enrollees with inadequate finances, less education, and high levels of social support. They have tended to enroll because of low premiums, enhanced HMO benefits, and pressure from employers providing retiree benefits (Silverman et al., 2000). It is not, however, merely less advantaged enrollees who are likely to find affordable broad coverage important. Although 90% of elderly Medicare beneficiaries were reported to have additional health insurance coverage that provided some supplemental benefits (Rice & Bernstein, 1999), the number of retirees who have supplemental coverage provided by their former employers has been dropping dramatically, having fallen from 40% in 1993 to 23% in 2001 (Iglehart, 2002). This coverage is increasingly unstable. Consequently, structural reform of the Medicare program has to include reconsideration of the basic benefits (Rice & Bernstein, 1999).
The need for Medicare reform is a frequently discussed policy problem, and proposed solutions are strongly contested by health care professionals, politicians, and the public. Proposed solutions have ranged from shifting to a system such as the Federal Employees Health Benefit Program through expansion of contracts with private health plans (Wilensky, 2001) and use of a defined contribution in a premium support model to arguments for leaving the program much as it is with some modification of its benefit package (Vladeck, 2001). Regardless of the direction of future reforms of Medicare, it will require sensitivity to costs and competing priorities in the face of large federal budget deficits and growing demands on the program. Information regarding the preferences of Medicare enrollees, and what, if anything, they would be willing to forgo in exchange, should not rest on speculation. Findings in this study, if reproduced or confirmed in other research, suggest directions Medicare enrollees would endorse.
A second implication is the possibility that Medicare could serve as a basis of expanding insurance coverage for the uninsured. A major obstacle to expanding coverage to the uninsured has been the concern that the need or demand for services may exceed available resources. Many states, for instance, face increasing Medicaid budgets and decreasing revenues and must either restrict eligibility ("ration by people") or cut the types of services covered. Medicare enrollees in this study endorsed the possibility of expanding the program's coverage to the uninsured. Such a proposal is not new (1975), but it has garnered renewed interest as the large number of uninsured Americans has persisted (Cunningham, 1998; Shea, Short, & Powell, 2001). Although the expansion of Medicare raises financial concerns (Scott, 1998), it is important to consider the trade-offs that Medicare enrollees tell us they might consider acceptable in order to make expansion financially feasible.
|
Footnotes |
|---|
1 Department of Clinical Bioethics, Warren G. Magnuson Clinical Center, National Institutes of Health, Bethesda, MD.
2 Department of Health Policy and Administration, University of North Carolina School of Public Health, Chapel Hill.
3 Bioethics Program, University of Michigan Medical School, Ann Arbor.
Decision Editor: Laurence G. Branch, PhD
Received for publication July 12, 2002. Accepted for publication October 30, 2002.
|
References |
|---|
|
TOPAbstractMethodsResultsDiscussionReferences |
|---|
This article has been cited by other articles:
 |
|
 |
|
  M. R. Gold, S. Sofaer, and T. Siegelberg Medicare And Cost-Effectiveness Analysis: Time To Ask The Taxpayers Health Aff., September 1, 2007; 26(5): 1399 - 1406. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
M. Ginsburg, S. D. Goold, and M. Danis (De)constructing 'Basic' Benefits: Citizens Define The Limits Of Coverage Health Aff., November 1, 2006; 25(6): 1648 - 1655. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 S. D. Goold, A. K. Biddle, G. Klipp, C. N. Hall, and M. Danis Choosing Healthplans All Together: A Deliberative Exercise for Allocating Limited Health Care Resources Journal of Health Politics Policy and Law, August 1, 2005; 30(4): 563 - 602. [Abstract] [PDF]
|
|
| ||||||||||||||||||||||||||||||||
| HOME | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|---|
