This Article Abstract Full Text (PDF) Services Download to citation manager PubMed PubMed Citation

Are Healthier Older Adults Choosing Managed Care?

Correspondence: Address correspondence to Gail A. Jensen, PhD, Institute for Gerontology and Department of Economics, Wayne State University, 87 E. Ferry Street, Detroit, MI 48202. E-mail: gail.jensen{at}wayne.edu

	Abstract

TOP Abstract Empirical Approach Data Findings Discussion References

 
Purpose: We attempt to determine whether older workers and early retirees avoid managed care plans and to explore whether health plan choices are linked to the health status of workers or their spouses. Design and Methods: We use the responses of those born between 1931 and 1941 to the 1994 and 1998 waves of the Health and Retirement Survey. We analyze current workers and early retirees separately, using cross-tabular and multinomial logit techniques. Results: Among older adults with active worker coverage, 60% were enrolled in either a health maintenance organization or preferred provider organization in 1998; 42% of early retirees were enrolled in these plans. Those with a choice of plans were even more likely to be in managed care. When demographic characteristics, time, and differences in the benefits and cost of the various plans offered by an employer are controlled for, health status (measured in a variety of ways) has little bearing on an older worker's choice of health plan. Implications: These findings suggest that older workers choose plans much as younger workers do. Employers are likely to continue to offer managed care as their workers age. The lack of unidirectional findings on health status bodes well for the long-term practicality of managed care under Medicare. Many workers are choosing an insurance type early in their tenure and remaining with that type of plan as they age.

Key Words: Health insurance • Retiree benefits • HMOs

Near-elderly persons make up a rapidly growing segment of the U.S. population. The number of U.S. residents between the ages of 55 and 64 years will increase by nearly 50% by the year 2010. They will comprise 13% of the population by the year 2020 (U.S. Census Bureau, 1995). Thus, older workers will be an increasingly large factor in the insurance decisions made by employers over the next two decades.

Existing studies have indicated that older workers are much more likely to have conventional insurance coverage (Dowd & Feldman, 1985; Ellis, 1985; Feldman, Finch, Dowd, & Cassou, 1989). This suggests that employers may face increasing pressures to continue and even expand their nonmanaged care offerings as their workforces age. However, most of these studies predate the explosion of managed care enrollment that occurred in the 1990s (Jensen, Morrisey, Gaffney, & Liston, 1997). The most recent data indicate that over 90% of workers with employer-sponsored health insurance are now in one or another form of managed care (Gabel et al., 2001). Morrisey and Jensen (2001) reported that 61% of near-elderly workers aged 57–63 and 49% of early retirees in this same age range were enrolled in a managed care plan in 1998.

Older workers may also hold a key to Medicare reform. The more likely approaches to reform tend to rely on managed care in one form or another. The "defined contribution" model essentially provides each Medicare beneficiary with a voucher to purchase private insurance. The "premium support" model envisions competition among Medicare-approved health plans (Reischauer, 1997). It is difficult to imagine how either of these proposals would realistically function without a major role for managed care options. Even incremental Medicare reform is likely to have a larger role for managed care. The Centers for Medicare and Medicaid Services are currently working to refine the payment formulas for Medicare + Choice to provide payments that reflect levels of illness.

There is a large body of literature on the effects of health status on insurance plan choice, particularly focusing on traditional plans versus health maintenance organizations (HMOs). For recent reviews, see Miller and Luft (1994) and Glied (2000). The vast majority of this literature concludes that HMOs, for whatever reason, attract those who use fewer services. Altman, Cutler, and Zeckhauser (2000), for example, concluded that half of the cost differential between traditional and managed care plans can be attributed to a favorable difference in the prevalence of illness within managed care plans. (The other half they attribute to lower negotiated prices.)

Among adults over the age of 65, research has examined the effects of health status on the acquisition of private coverage to supplement Medicare. Whereas earlier work yielded mixed findings, more recent work by Lillard, Rogowski, and Kington (1997) found relatively little evidence to support the contention that adverse selection was an important factor in the purchase of insurance in older populations. Research has also focused on enrollment in Medicare HMOs. Studies have typically used prior utilization as a measure of health status and have found that Medicare HMOs tended to enroll lower utilizers of services (Physician Payment Review Commission, 1996). This was born out in the Commission's own analysis of 1989–1994 Medicare claims data. It found that, in the 6 months prior to enrolling in a Medicare HMO, beneficiaries had claims experience that was only 56% of average. Thus, although health status may not affect the decision of seniors to have coverage in addition to traditional Medicare, it appears to affect the type of Medicare coverage they acquire.

There has been much less research on the near-elderly population and their choice of health plans. Jensen (1992) reported that the lack of insurance coverage in this cohort tends to be associated with the work history of the individual and his or her spouse. Sloan and Conover (1998) examined the 1992 and 1994 waves of the Health and Retirement Survey. They concluded that individuals whose health had deteriorated significantly were no more likely than others to suffer a subsequent loss of active worker coverage. This was largely due to the substitution of retiree coverage, individually purchased coverage, and newly acquired coverage through Medicare or Medicaid. Aside from disability status, other measures of health and functional status did not affect the probability of being insured.

There appears to be no published research on the effects of health status on the plan choices of the near-elderly population. Although recent research has documented the growth over the 1990s of managed care among older adults (Morrisey & Jensen, 2001), we still know little about just who has migrated into managed care. However, studies conducted prior to this recent surge in managed care had found that older adults in HMOs tended to be healthier. Is this true today among older adults? It is possible that since managed care plans came to dominate the market by the mid-1990s (Jensen et al., 1997), they might be drawing a more representative risk mix among older subscribers. In this article we examine this issue.

	Empirical Approach

TOP Abstract Empirical Approach Data Findings Discussion References

 
This article first reviews the extent to which near-elderly persons with employer-sponsored coverage are covered through a managed care plan. It reports on trends in enrollment in HMOs, preferred provider organization plans (PPOs), and traditional fee-for-service (FFS) coverage between 1994 and 1998, and the extent to which older adults are offered a choice of plans. It then reports the characteristics of those within each type of plan. Coverage is examined separately for adults with active worker coverage and adults with retiree coverage.

This descriptive analysis is followed by a series of multinomial logit regressions to examine more carefully the relationship between health status and the type of coverage held by an older adult. The models distinguish between three types of employer plans: HMOs, PPOs, and traditional FFS coverage. Separate models are estimated for those with active worker coverage and retiree health benefits, because preferences and choice sets may be very different for active workers and retirees.

For each cohort we first estimate models for the type of employer coverage held by an older adult, among all survey respondents in employer plans. We then estimate type-of-plan models for the subset of adults who said they chose their plan from a menu of offerings by the firm. This allows us to better account for plan choices and examine the effects of one's own health status as well as spousal health status in plan choice decisions.

The models estimated for those who faced a choice of health plans posit that plan choice is a function of an individual's demographic characteristics, such as age, gender, race, marital status, and household income, the different benefits and cost of the different plans offered, and time (i.e., whether it was 1994 or 1998). The models also include a set of health status measures for the individual and, if the individual is electing family coverage, his or her spouse's health status as well. Whether persons have access to public coverage is also accounted for in these models. Rather than estimating pairwise logit regressions, multinomial logit models have the advantage of using a common error distribution across the entire choice set, thereby ensuring that the probabilities of different types of plans sum to one for each individual.

	Data

TOP Abstract Empirical Approach Data Findings Discussion References

 
Data for the study are drawn from the 1994 and 1998 waves of the Health and Retirement Survey (HRS) sponsored by the National Institute of Aging. In 1992, the HRS began to follow a cohort of men and women born between 1931 and 1941. Individuals had to meet this age criterion, but spouses of any age were included. The participants were reinterviewed in 1994, 1996, and 1998. The first wave interviewed more than 12,000 persons throughout the United States. It contained an oversample of African Americans and residents of Florida. The initial response rate was 81.7%, and the reinterview response rates were 91.8%, 93.1%, and 93.8%, respectively, after an adjustment was made for the approximately 2% of the original respondents who died between each wave (see www.umich.edu/hrswww). We use the 1994 and 1998 waves because, in these years, data on the key variables of interest are available.

Our focus is limited to respondents born between 1931 and 1941 who had employer-sponsored health insurance at the time they were interviewed. There were 5,946 persons in 1994 and 4,251 persons in 1998 who met these criteria. In the 1994 sample there were 4,697 persons with active worker health insurance and 1,267 persons with postretirement health coverage. In the 1998 sample, the corresponding counts were 2,565 and 1,686, respectively. Data from the 1992 wave are not used, because in that year the survey did not record the nature of a respondent's health plan, in particular whether it was an HMO, a PPO, or an FFS type of plan. Data from the 1996 wave are not used because the public-use tapes for that year lacked data on annual household income.

For our descriptive statistics, the data are weighted by use of the HRS sampling weights, so the resulting estimates should be nationally representative for adults with employer coverage in this age range.

Questions concerning health insurance coverage, like most of the questions in the HRS, are for a single point in time and relate to the day on which the survey was answered. The 1998 survey directly asked the nature of employer coverage. The 1994 survey was different. In that year, respondents who had earlier indicated (in 1992) that they had employer coverage were asked if they still had it. If they said "yes," they were then asked about the nature of the coverage, with the questions being more detailed than they were in 1992. Persons who had not had employer coverage in 1992 were asked if they had employer coverage now, and if they said "yes," a parallel set of detailed questions on the nature of coverage was asked. The year 1994 was the first in which HRS respondents were asked about they type of employer health plan they had, and whether they had a choice of plans. The HRS used a 5-point scale to measure self-reported health status of the individual and his or her spouse. The responses ranged from poor to excellent. In addition, in both years the survey asked about disability status, that is, whether one had high blood pressure, diabetes, emotional problems, or other chronic conditions. It also asked if one was a smoker and whether one had been admitted to a hospital within the prior 12 months.

	Findings

TOP Abstract Empirical Approach Data Findings Discussion References

 
Table 1 shows the extent to which near-elderly persons with employment-related coverage have various types of coverage. Among those with coverage through an active worker, 75.5% had coverage in their own name in 1998; the remaining 24.5% had coverage through a spouse who had active worker coverage. Among retirees in 1998, two thirds (66.9%) had retiree coverage in their own name. The 1998 values reflect increases over 1994 and probably reflect increased labor force participation by women over the 1990s.

Enrollment in managed care plans may reflect direct insurance decisions by active workers or retirees. It may also reflect the single plan offered by an employer. Respondents were asked whether the employer had offered them a choice of plans. Nearly 47% of respondents with active worker coverage in 1998 faced a choice of health plans; nearly 53% of retirees did. These percentages reflect an increase from 1994. Those offered a choice of plans were much more likely to be in a managed care plan than were those without a choice. Indeed, among retirees in 1998, 70.6% of those with a choice of plans were in one or another form of managed care, compared with only 46.7% among retirees who did not have a choice of plans. Indeed, traditional FFS coverage was much more common among those near-elderly persons who did not have a choice of plans.

It does not follow, of course, that those without a direct choice of plans are in plans that they dislike. A profit-maximizing firm is expected to balance the preferences of workers with the costs of offering multiple plans (Goldstein & Pauly, 1976; Jensen, 1986). Firms with more homogeneous workforces are more likely to offer a single plan that reflects the preferences of the average or median worker (Moran, Chernew, & Hirth, 2001).

Among the 4,697 older workers with active worker coverage in 1994, 44.3% continued to receive active worker coverage in 1998 and 19.9% retired and began receiving retiree health benefits by 1998. The rest, 35.8%, either moved into public coverage (7%) or individually purchased insurance (9%), became uninsured (11%), or else dropped out of the HRS altogether (9%). Did the adults who retired with retiree health insurance shift to managed care to the same extent as those who remained with active worker coverage between 1994 and 1998? Table 2 addresses this question. Although there was clearly a shift into HMOs between both groups, it was less pronounced among those adults who retired and began receiving retiree health benefits. This is consistent with the pattern observed in Table 1: although it is true that both groups are migrating into managed care, older workers are doing so more rapidly.

One reason for this finding may be the inclusion here of persons who were not offered a choice of plans by their employer. Segregation by type of plan is clearly more likely to occur when individuals are offered a choice of plans by the firm. Consequently, including the former group in the estimation may be masking the relationship between health status and plan enrollment.

In Table 5 we report the results of model estimation when we restrict the sample to individuals who were offered a choice of plans and who selected single coverage (i.e., insurance for themselves only). In Table 6 we report the models for individuals who were offered a choice and who selected family coverage (i.e., for themselves and their spouse). In the family coverage models, we also include explanatory variables measuring the health status of the spouse; presumably, these too may have influenced plan choice because the insurance was to cover both husband and wife. Finally, because the models in Tables 5 and 6 are for individuals who faced a choice of plans, we attempt to control for their out-of-pocket premium costs. If one plan costs more than another, prior research strongly suggests that this will influence plan choice (Dowd & Feldman, 1994–1995). Unfortunately, although the HRS reports the employee-paid portion of the premium for the actual plan chosen, it does not report comparable data for the competing plans on the menu. However, it does report the responses to two questions that were asked of individuals who said they were offered a choice of plans. The first was "Is there another plan that was offered that provides better coverage?" The second was "Is there another plan offered that costs more?" We include both measures as a proxy for the perceived relative value of the plan chosen.

Table 6 reports analogous multinomial logit regressions for those who faced a choice of plans and who elected family coverage. Here, too, individual and spousal health characteristics matter little among couples electing joint coverage. A likelihood ratio test for the joint significance of an individual's own health characteristics reveals that they are not significant as a group in the active worker coverage equation (² = 26.1 with 20 df) and barely significant in the retiree coverage model (² = 28.5 with 20 df). A parallel test for the joint significance of the spouse's health characteristics finds them insignificant in both equations (² = 24.9 and 19.2, respectively).

	Discussion

TOP Abstract Empirical Approach Data Findings Discussion References

 
This study has used the 1994 and 1998 waves of the HRS to explore the link between health status and the nature of the employer-sponsored health insurance held by older workers. Two findings emerge from the analysis. First, older workers dramatically switched into managed care between 1994 and 1998. Among adults born between 1931 and 1941 with active worker coverage, three out of every five (59.8%) were enrolled in either an HMO or PPO by 1998. Among those with retiree health benefits, more than two out of every five (42.2%) had managed care coverage. These findings are not driven by a lack of choice among older workers. Those with a choice of plans were even more likely to be in a managed care plan, and those without a choice were the ones more likely to be enrolled in a traditional FFS plan.

Second, our explorations of self-reported health status suggest that healthier older adults, as well as those who are less healthy, are equally represented within HMOs, PPOs, and FFS plans. Very few measures of own and spouse's health status appear related to an individual's choice of plan, and, even among these, the effects are not unidirectional. The relationships appear much more complex. It may be that managed care plans provide greater perceived value than FFS plans for some chronic conditions. Although not consistent with the research on the choice of Medicare HMOs, the results are within the mainstream of findings on health insurance and older workers. As we already noted, that literature tends to find that work history has a much stronger link to insurance status than health status does per se.

We draw several implications from these findings. First, employers may stay with managed care options as the workforce ages. Older workers are subject to the same trade-offs as younger workers in deciding what sort of coverage to take. For the most part, they appear to be making much the same choices.

Second, the growing use of managed care by older workers and the choice of managed care options by early retirees suggests that workers have become familiar with managed care plans over the past decade and are choosing to continue with the sort of coverage they knew as active workers. This is consistent with research on coverage stability in transitions to retirement by Morrisey and Jensen (2001) that found that the newly retired overwhelmingly stay with the plan type they had as active workers.

This may explain why health status had such a small effect on plan choices in this study. It is not unreasonable that workers made plan choices in middle age or even earlier, and they remained with that plan type over their remaining working career even as they may have changed particular plans. If so, this is consistent with the findings of Feldman and colleagues (1989), who found that workers were very price sensitive within a plan "nest" but much less so when moving from one type of plan to another. Although their work focused exclusively on HMO coverage versus FFS, it is not difficult to imagine that the result might generalize to nests of managed care plan types.

Third, the findings have implications for Medicare. The substantial use of managed care plans by the near-elderly population suggests that Medicare reformers have a wider range of options for consideration than would otherwise be the case. Baby boomers are familiar with managed care and may be receptive to reforms that have a major role for such plan types. Indeed, it would come as no surprise if, within a few years, most new Medicare enrollees find traditional Medicare to be an alien form of coverage.

Finally, Medicare is currently in the process of implementing a more sophisticated risk adjustment system to pay Medicare HMOs. Our findings of a lack of clear health status effects on plan choice among the near-elderly population and particularly among early retirees suggests that the lack of good data on prior health services use or on other measures of health status will not necessarily have important consequences for the program. It may be that Medicare can defer the use of sophisticated risk adjusters until it has a year or two of utilization experience on new beneficiaries.

There are limitations to this study. First, we are particularly concerned about the lack of good data on the out-of-pocket premiums faced by those in the sample. There is a relatively large body of literature showing that employee plan choice is strongly influenced by the price of the offered health plans. We had information on the premium of the plan chosen but no hard data on the prices of the available plans not chosen. Future research has to better incorporate the relevant relative prices. Second, we had only limited information on prior use of health services and on health status. All were self-reported measures, so our analyses may suffer from classic errors in the variables. Although these would not bias our estimates as long as the errors in reporting were random, they do reduce the precision of the estimates. Thus, our estimates of the effects of health status may actually be stronger than reported here. Finally, we had a relatively small sample with which to work and, therefore, relatively few cases of seriously impaired health. This would tend to result in an underestimate of the economic and political significance of health status in the models estimated.

	Footnotes

 
This project was funded by Contract J-9-P-7-0048 from the U.S. Department of Labor, Pension and Welfare Benefits Administration. We thank Xiao Xu for research assistance.

¹ Institute for Gerontology and Department of Economics, Wayne State University, Detroit, MI.

² Lister Hill Center for Health Policy, University of Alabama at Birmingham.

Decision Editor: Laurence G.Branch, PhD

Received for publication June 24, 2002. Accepted for publication October 11, 2002.

	References

TOP Abstract Empirical Approach Data Findings Discussion References

 

• Altman, D., Cutler, D. M., Zeckhauser, R. J., 2000;. Enrollee mix, treatment intensity, and cost in competing indemnity and HMO plans (Working Paper 7832). Cambridge, MA: National Bureau of Economic Research.
• Dowd, B., Feldman, R., 1985;. Biased selection in twin cities health plans. Advances in Health Economics and Health Services Research,. 6:253-271.[Medline]
• Dowd, B., Feldman, R., 1994–1995;. Premium elasticities of health plan choice. Inquiry,. 31:438-444.
• Ellis, R. P., 1985;. The effect of prior-year health expenditures on health plan coverage plan choice. Advances in Health Economics and Health Services Research,. 6:149-170.[Medline]
• Feldman, R., Finch, M., Dowd, B., Cassou, S., 1989;. The demand for employer based health insurance plans. Journal of Human Resources,. 24:115-142.
• Gabel, J. R., Levitt, L., Pickreign, J., Whitmore, H., Holve, E., Rowland, D., et al 2001;. Job-based health insurance in 2001: Inflation hits double digits, managed care retreats. Health Affairs,. 20:(3), 180-186.[Abstract/Free Full Text]
• Glied, S., 2000;. Managed care. In A. J. Culyer & J. P. Newhouse (Eds.), Handbook of health economics (Vol. 1A, pp. 707–754). Amsterdam: Elsevier.
• Goldstein, G. S., Pauly, M. V., 1976;. Group health insurance as a public good. In R. N. Rosett (Ed.), The role of health insurance in the health services sector (pp. 73–109). New York: National Bureau of Economic Research.
• Jensen, G. A., 1986;. Employer choice of health insurance benefits. Unpublished doctoral dissertation, University of Minnesota, Minneapolis.
• Jensen, G. A., 1992;. The dynamics of health insurance among the near elderly. Medical Care,. 30:598-614.[Medline]
• Jensen, G. A., Morrisey, M. A., Gaffney, S., Liston, D. K., 1997;. The new dominance of managed care: Insurance trends in the 1990s. Health Affairs,. 16:(1), 125-136.[Abstract]
• Lillard, L., Rogowski, J., Kington, R., 1997;. Long-term determinants of patterns of health insurance coverage in the Medicare population. Gerontologist. , 37:314-323.[Medline]
• Miller, R. H., Luft, H. S., 1994;. Managed care plan performance since 1980. Journal of the American Medical Association,. 271:1512-1519.[Abstract/Free Full Text]
• Moran, J. R., Chernew, M. E., Hirth, R. A., 2001;. Preference diversity and the breath of employee health insurance. Health Services Research,. 36:911-934.[Medline]
• Morrisey, M. A., Jensen, G. A., 2001;. The near-elderly, early retirees, and managed care. Health Affairs,. 20:(6), 197-206.[Free Full Text]
• Physician Payment Review Commission., 1996;. Risk selection and risk adjustment in Medicare (Annual Report to Congress 1996). Washington, DC: Author.
• Reischauer, R. D., 1997, July;. Medicare: Beyond 2002. In S. H. Altman, U. E. Reinhardt, & D. Shactman (Ed.), Policy options for reforming the Medicare program. Waltham, MA: Council on the Economic Impact of Health System Change.
• Sloan, F. A., Conover, C. J., 1998;. Life transitions and health insurance coverage of the near elderly. Medical Care,. 36:110-125.[Medline]
• U.S. Bureau of the Census., 1995;. Statistical abstract of the United States. Washington, DC: U.S. Government Printing Office.

This Article Abstract Full Text (PDF) Services Download to citation manager PubMed PubMed Citation