Managed care has become the dominant mode of care delivery

Transcription

1 Commercial Plans In Medicaid Managed Care: Understanding Who Stays And Who Leaves Many of the factors that influence plans exit decisions are within the control of state policymakers and program administrators. by Sharon K. Long and Alshadye Yemane ABSTRACT: Although the rapid increase in Medicaid managed care during the early 1990s attracted commercial plans to the program, by the late 1990s commercial plan participation in Medicaid had begun to decline. This study examines the role of Medicaid policies, plan characteristics, and local health care market conditions in a commercial plan s decision to exit. We find that many of the factors that influence commercial plans decisions to exit Medicaid are within the control of state policymakers and program administrators, including capitation rates, service carve-outs, mandatory enrollment policies, and the number of Medicaid enrollees and areas served by the plan. Managed care has become the dominant mode of care delivery within the Medicaid program, covering nearly 60 percent of all enrollees in That dominance is expected to grow as states expand Medicaid managed care (MMC) to additional geographic areas and populations, including rural areas and disabled people. Much of the growth in Medicaid managed care has been in full-risk capitated managed care programs. In 1999 there were 316 plans in 45 states serving 11.4 million Medicaid enrollees in fully capitated programs, representing approximately 64 percent of all MMC enrollees. 2 Commercial plans, which we define as plans with less than 25 percent of their total enrollment from Medicaid or other public programs, represented 57 percent of these full-risk plans and accounted for 58 percent of all full-risk MMC enrollees. 3 Although commercial plans play a large role in fully capitated MMC programs, published reports indicate that the share of commercial plans participating in Medicaid has been declining since the mid-1990s. At the height of commercialplan participation in 1996, there were 211 such plans in Medicaid managed care. By 1999 that figure was down to 181, with the number of commercial plans exiting Medicaid (34) exceeding the number of plans entering (12) by nearly three to one. 4 Plan exits from Medicaid are of concern because they raise the potential for ma- Sharon Long (SLong@ui.urban.org) is a principal research associate at the Urban Institute in Washington, D.C. Alshadye Yemane is a research associate there July/August 2005 DOI /hlthaff Project HOPEThe People-to-People Health Foundation, Inc.

2 Medicaid Managed Care jor disruptions in care. Disruptions would arise, for example, if enrollees needed to change primary care providers (PCP) because their current providers were not participating in the new plan or if enrollees treatment plans were changed as a result of different utilization management techniques (for example, drug formularies), general approaches to care, or changes in procedures for obtaining services under the new plan. If states are to reduce plan exits from the MMC market, it is important to understand the factors that influence plans decisions to exit. Although a number of qualitative studies have explored plans MMC participation, no empirical studies have addressed this issue. 5 This study provides the first empirical analysis of the factors associated with plan exits from Medicaid managed care. Specifically, we examine how MMC policies, plan characteristics, and local health care market conditions affect a commercial plan s decision to continue in or exit from Medicaid managed care in a county. This analysis can help state officials develop strategies to keep commercial plans in or attract new commercial plans to the Medicaid market. We focus on commercial plans MMC participation because (1) many of the recent exits from Medicaid managed care have been by commercial plans; and (2) we would expect market behavior to differ for Medicaid-dominated and commercial plans, since they are likely to have different motivations for serving the Medicaid population. Although it would be of considerable interest to explore those differences, data on Medicaid-dominated firms are not generally available. Information on commercial firms, which must be licensed by the states in which they operate, can be obtained through existing data sources; no data sources have information on all Medicaid-dominated firms, because not all such firms face licensure requirements. 6 Study Data And Methods Conceptual framework. Although no empirical studies have examined the factors affecting plan exits from the MMC market, a number of empirical studies have looked at plan exits from the Medicare+Choice (M+C) market. 7 Building on that work, we modeled a plan s decision about MMC participation in a county as a function of (1) state Medicaid policies that affect return on investment in the county MMC market, (2) the characteristics of the plan, and (3) the characteristics of the local market. 8 In specifying our model, we drew from the qualitative and quantitative literature examining commercial plans exits from Medicaid managed care and also from Medicare managed care (Exhibit 1). Data sources. Enrollment. We relied on county Medicaid enrollment data for commercial plans from InterStudy, cross-checked against data provided by the states. 9 Basedonthosedata,weidentified154commercialplansthatwereserving Medicaid beneficiaries in 745 counties in 39 states as of 1 January 2000 (Exhibit 2). After we eliminated observations with missing values for the explanatory variables included in our model (summarized in Exhibit 1), the final analysis sample was 1,115 HEALTH AFFAIRS ~ Vo l u m e 2 4, N u m b e r

3 EXHIBIT 1 Variables Related To A Commercial Plan s Decision To Exit Medicaid Managed Care (MMC) Variable Hypothesized effect on plan exits Explanation of hypothesized effects Medicaid payment and other policies Medicaid capitation rate in county Higher capitation rates increase the return to the Medicaid line of business, reducing the probability of an exit Growth in state s Medicaid capitation rates Capitation payments that increase over time are more likely to keep pace with increases in health care costs, reducing the probability of an exit State carves out services under MMC Both Carve-outs could reduce the costs of providing care, making Medicaid more profitable for plans and reducing the probability of an exit (negative); alternatively, carve-outs could increase the costs of providing care by interfering with plans ability to manage care, making Medicaid less profitable and increasing the probability of an exit (positive) Medicaid beneficiaries required to enroll in fully capitated managed care in county Plan characteristics Age of plan (years) Plan size Plan affiliation/sponsorship Affiliated with national firm Affiliated with BCBS Provider-sponsored independent firm Other type of independent firm Plan is for-profit Positive Positive Positive Mandatory enrollment could limit plans ability to benefit from favorable selection by healthier Medicaid beneficiaries, making Medicaid less profitable and increasing the probability of an exit Plans with more years of experience and larger enrollments may be more likely to operate more efficiently, making them better able to control costs, increasing the likelihood that Medicaid will be profitable and reducing the probability of an exit Plans that are more closely connected to the local health care market (such as independent firms) and plans that have an organizational mission to serve vulnerable populations (such as BCBS, provider-sponsored, and not-for-profit firms) will be less likely to exit Plan uses closed managed care model Closed-model plans have more opportunities to control costs, increasing the likelihood that Medicaid will be profitable and reducing the probability of an exit Plan s investment in MMC market Plan s share of county MMC market Share of counties in state in which plan has MMC enrollment Plans that are more heavily invested in Medicaid at the county and/or state level will be less likely to exit plan-county observations in 33 states. The number of plan-county observations varied greatly across the states, ranging from 186 in Minnesota to only 1 in North Dakota and 1 in the District of Columbia. Plan characteristics. InterStudy data were the source for all plan characteristics, including age of plan, affiliation and sponsorship, tax status, Medicaid market share, total enrollment, percentage of counties in a state in which the plan serves Medicaid beneficiaries, and total market share, which was used to calculate the 1086 July/August 2005

4 Medicaid Managed Care EXHIBIT 1 Variables Related To A Commercial Plan s Decision To Exit Medicaid Managed Care (MMC) (cont.) Variable Market characteristics Hypothesized effect on plan exits Explanation of hypothesized effects Returns to private market and Medicare relative to returns to Medicaid Both As the relative returns to commercial and Medicare markets increase, plans may exit Medicaid to invest in those markets (positive); alternatively, higher relative returns in those markets could increase plans ability to cross-subsidize their Medicaid product, making them less likely to exit (negative) Operating profit margin in local area Plans operating in a local market that is profitable may be better able to cross-subsidize their Medicaid product, reducing the probability of an exit Size of overall managed care market in county Competitiveness of HMO market in county Competitiveness of provider market in county Positive Positive Larger populations provide a greater potential for economies of scale in service delivery and, thus, lower costs and a lower probability of an exit More competitive markets make it more difficult for plans to obtain favorable rates for health care services, leading to higher costs and higher probability of an exit County is urban Plans in urban markets will be less likely to exit because the more extensive health care systems in these areas make it easier for plans to manage care and control costs SOURCE: Authors analysis. NOTES: BCBS is Blue Cross Blue Shield. HMO is health maintenance organization. Herfindahl-Hirschman Index (HHI) of managed care market competitiveness. State and local area characteristics. State and local area characteristics come from a variety of sources. We measured the relative returns in the private market and Medicare to the returns under Medicaid managed care as the ratio of the average employer-sponsored insurance premiums in the county to the Medicaid capitation rate in the county, and the ratio of average M+C rates in the county to the EXHIBIT 2 Commercial Plans Participation In Medicaid Managed Care (MMC), 2000 And 2001 Plans Number of plans Number of counties Number of plancounty observations Total participation in 2000 County exits by 2001 Total participation in , ,514 Net change between 2000 and SOURCE: Authors tabluations using InterStudy data. HEALTH AFFAIRS ~ Vo l u m e 2 4, N u m b e r

5 Medicaid capitation rate in the county. 10 County capitation rates for the M+C program came from the Centers for Medicare and Medicaid Services (CMS). 11 County Medicaid capitation rate data came from a 2001 Urban Institute survey of state Medicaid program officials. 12 Information on state carve-outs came from the same Urban Institute survey. We focused on carve-outs for mental health and prescription drug services because these are two of the most costly services that are carved out by states under Medicaid managed care. Mandatory enrollment. To identify counties with mandatory enrollment in fully capitated MMC programs for both poverty-related and Supplemental Security Income (SSI) Medicaid populations, we used data from the CMS s 2000 National Summary of State Medicaid Managed Care Programs and Medicaid Managed Care Enrollment Reports. Other variables. We calculated the HHI of county hospital market competitiveness using market share data from the 2000 American Hospital Association annual survey. We obtained the remaining variables from the 2003 Area Resource File from the Health Resources and Services Administration (HRSA). Analyses. We estimated logit models to predict the probability that a plan s MMC exit in a county in 2001 was a function of the characteristics of the Medicaid program, the plan, and the local health care market in 2000 (Exhibit 1). We report our results as odds ratios, which are interpreted as the odds of exiting from Medicaid managed care. To provide a sense of the relative magnitude of the influence of the different variables on the probability that a plan exits, we simulated the effects of changes in the explanatory variables. For dummy variables, such as whether the plan is a for-profit plan, we simulated the effect of a shift from no to yes (that is, the shift from 0 to 1). For continuous variables, such as Medicaid capitation rates, we simulated the effect of an increase of one-half standard deviation from the mean. Study Results The decline in commercial plan MMC participation that has been reported in earlier work continued in Between 2000 and 2001, fifty-seven plans exited the Medicaid market in one or more counties across thirty-one states. The exiting plans left Medicaid in an average of five counties each, leading to the 282 plancounty MMC exits between 2000 and It is worth noting that only seven of these plans (representing fourteen plan-county exits) exited completely from the Medicaid program in a state. Role of MMC rates and policies. We found that Medicaid capitation rate levels, capitation growth rates, and other Medicaid policies have strong effects on the probability that a plan will exit Medicaid managed care (Exhibit 3). An increase in the Medicaid capitation rate of one-half standard deviation, which is an increase of about $15 per member per month, led to a 3.2-percentage-point drop in the probability of exiting, all else being equal. We also found that an increase in the capitation 1088 July/August 2005

6 Medicaid Managed Care EXHIBIT 3 Odds Ratios And Simulations From Logit Model Of Medicaid Managed Care (MMC) Plan Exits, 2000 And 2001 Change in predicted probability of exit a Variable Mean Odds ratio With a change in value from 0 to 1 With increase of one-half standard deviation (STD) Change in probability STD/2 Medicaid payment and other policies County Medicaid capitation rate ($) Growth in Medicaid capitation rates (standardized) State carves out mental health services States carves out pharmacy services ** 0.53*** 0.49*** 2.35*** SSI population required to enroll in HMO in county Poverty-related population required to enroll in HMO in county ** Plan characteristics Age of plan (years) Plan s total number of enrollees in county (thousands) Plan uses closed managed care model *** 8.5 Plan is affiliated with BCBS Plan is independent firm Plan is provider sponsored Plan is for-profit *** 1.70** Plan s share of county MMC market Share of counties in state in which plan has MMC enrollment *** 0.13*** Market characteristics Private capitation rate/medicaid capitation rate in county M+C capitation rate/medicaid capitation rate in county Median operating profit margin in local area (standardized) ** *** Competitiveness of HMO market in county (HHI) Competitiveness of hospital market in county (HHI) Total number of physicians per 10,000 population in county growth rate of one-half standard deviation, which represents a 1.4-percentage-point increase, reduced the probability of exit by 2.2 percentage points. 13 Service carve-outs and mandatory enrollment in fully capitated managed care also have significant effects on plan exits. Plans in our sample that faced a carveout for mental health were less likely to exit, while plans that faced a prescription drug carve-out were more likely to do so. The requirement that the SSI Medicaid HEALTH AFFAIRS ~ Vo l u m e 2 4, N u m b e r

7 EXHIBIT 3 Odds Ratios And Simulations From Logit Model Of Medicaid Managed Care (MMC) Plan Exits, 2000 And 2001 (cont.) Change in predicted probability of exit a Variable Mean Odds ratio With a change in value from 0 to 1 With increase of one-half standard deviation (STD) Change in probability STD/2 Market characteristics County population in lowest quartile County population in higher quartile County is urban ** 0.79 b 2.36*** SOURCES: Authors analysis using data from InterStudy and other sources. NOTES: Growth in the capitation rate is measured relative to the average growth across the states. Median operating profit margin is measured relative to the median profit margin across the states. We assessed the goodness-of-fit of the logit model by predicting the probability of exit for each plan, assuming that a plan with an exit probability of greater than.50 would actually exit (and a plan with an exit probability of.50 or less would remain in MMC), and then comparing the predicted exit status to actual exit status for each plan. Our model correctly predicted exit status for 81 percent of all observations. Looking at the accuracy of predictions by exit type, 95 percent of plans that actually remained in Medicaid were accurately predicted as not exiting Medicaid, and 35 percent of plans that actually exited were accurately predicted as exiting Medicaid. The lower rate of accurate predictions for exiting plans is an expected result of logistic regressions when trying to predict an event that occurs relatively infrequently. HMO is health maintenance organization. SSI is Supplemental Security Income. M+C is Medicare+ Choice. BCBS is Blue Cross Blue Shield. HHI is Herfindahl-Hirschman Index. N = 1,155 plans. Chi-square (df = 24) is a We estimated changes in the predicted probability of exit for those variables that have a significant effect on the probability of exit in the multivariate analysis. b Not significant at the.10 level. ** (***) Significantly different from zero at the.05 (.01) level, two-tailed test. population enroll in fully capitated managed care plans was associated with a greater likelihood of exit, with mandatory enrollment contributing to a 9.7- percentage-point increase in the probability. The mandatory enrollment of the poverty-related population was also associated with an increased probability of exit, although it was not statistically significant. Medicaid policies requiring beneficiaries to enroll in fully capitated managed care might limit a plan s ability to benefit from the favorable selection that often occurs under other models. 14 Role of plan characteristics. Plans in our sample with a larger share of the county MMC market and plans that served Medicaid enrollees in more counties within the state were less likely to exit from Medicaid managed care in a county. Model type, plan sponsorship, and for-profit status also matter. Plans with a closed model and provider-sponsored plans were less likely to exit, while for-profit plans were more likely to exit, all else being equal. Role of market characteristics. Finally, market characteristics are important. MMC exits were less likely in counties with private capitation rates that were higher relative to Medicaid rates. This suggests that higher rates in the private market might allow plans to subsidize the costs of care in the Medicaid market. Consistent with this, MMC exits were less likely in local areas where plans operating profit margins were higher. In contrast, exits were more likely in counties with a 1090 July/August 2005

8 Medicaid Managed Care lower concentration of physicians, in less populated counties, and in urban counties. The association between physician supply and probability of exiting from Medicaid managed care could suggest that a tighter supply of physicians makes it more difficult for plans to negotiate favorable physician rates. It could also suggest that plans have a harder time recruiting providers to serve their Medicaid clients in such markets. It is not clear how to interpret the finding that plans were more likely to exit from Medicaid managed care in urban counties than in rural counties. We suspect that it could indicate that plans that decided to enter rural MMC markets were more committed to remaining in those markets than in urban markets, all else being equal. This would be true, for example, for plans that viewed the MMC market as a way to gain entry into rural managed care markets more generally. 15 For these plans, MMC participation in rural areas would be part of a larger growth strategy. Discussion And Policy Implications Commercial plans continue to be an important segment of the MMC market, serving more than half of all MMC enrollees. Given the potential adverse effects on continuity of care when plans exit the MMC market, there are clear reasons why states might want to invest in maintaining commercial plans MMC participation. Our findings suggest several factors that such states should consider in formulating their MMC policies. Capitation, growth rates, and carve-outs. First, higher MMC capitation rate levels and growth rates are associated with a lower probability of exiting Medicaid. Second, service carve-outs can encourage plan participation by removing the burden of providing expensive services (as in mental health care) or discourage participation by limiting a plan s ability to manage patient care (as in prescription drug services). Therefore, states might want to examine the extent to which carve-out policies affect plans ability to manage care as part of their MMC design process. Mandatory enrollment. Third, counties with mandated enrollment in fully capitated managed care for SSI beneficiaries are more likely to lose commercial plans than counties that do not have mandatory enrollment. This finding is of particular concern because it suggests that states that have made a commitment to fully capitated managed care are most likely to lose commercial plan participation. It may be that once commercial plans can no longer benefit from the favorable selection that often occurs under other types of managed care, they cannot achieve adequate returns on their Medicaid line of business at the capitation rates provided. The stronger negative effect of mandatory enrollment for the SSI population than the poverty-related population likely reflects the greater costs and risks associated with serving a disabled population. Although capitation rates could, in theory, adjust for differences in the cost of providing care to enrollees with different levels of disability, few states have implemented rate differentials that adjust for health status. Further, plans have reported that MMC rates do not fully com- HEALTH AFFAIRS ~ Vo l u m e 2 4, N u m b e r

9 States that hope to maintain or encourage commercial plan MMC participation need to establish sound capitation rates. pensate for the additional costs of serving their SSI populations. 16 MMC market share and investment. Fourth, plans that have a larger share of the MMC market in a county and plans that have invested in Medicaid managed care in a larger share of the counties in a state are less likely to exit than plans with more limited MMC investment. This is because plans that invest heavily in Medicaid are likely to be better able to spread the fixed costs of participating in Medicaid and to pool risk more effectively than plans with a more marginal investment. Type of plan. Beyond these factors, we also found that provider-sponsored and not-for-profit plans are less likely to exit than other plans. These plans may be more likely to have a sense of social mission or obligation toward serving the Medicaid population and, consequently, be less likely than other plans to exit when facing the same MMC rates and policies. This is consistent with qualitative research on the factors that affect commercial plan participation in MMC. General health care market. Finally, plans in markets with higher operating profit margins and higher commercial rates relative to Medicaid rates are less likely to exit Medicaid managed care. This finding might suggest that commercial plans in the MMC market are focused on their overall bottom line (rather than just the returns to their Medicaid line of business) in deciding whether to remain in the MMC market. Although we are not aware of other empirical studies examining the factors affecting commercial plan MMC exits, there is a growing literature examining plan exits from the M+C market. Our results are generally consistent with that literature. 17 Study limitations. Beyond the measures included in this study, other factors could also affect a plan s decision to continue in or exit from Medicaid managed care in a county. Unfortunately, we do not have access to data for all states and for all plans on the complete set of factors that could play a role in a plan s participation decision. This is particularly problematic at the state level because additional state policies could be affecting plans participation decisions that we are not capturing in our model. Consequently, this analysis should be viewed as a first step in understanding the factors that influence a commercial plan s decision to exit Medicaid managed care. A second limitation is that this study is a cross-sectional analysis. We were able to identify associations between the factors in our model and MMC exits but could not determine whether the exits were actually caused by those factors. Finally, we treated a plan s decision about MMC participation in a county as independent of its participation decision in other counties in the state. In reality, plans may make decisions about participation in a group of adjacent counties or in the state as a whole rather than county by county. Developing a more complex model of plan decision making would be a useful extension of this work July/August 2005

10 Medicaid Managed Care Our findings offer cause for optimism about the stability of commercial plan MMC participation. Recent research has found that commercial plans that remain in Medicaid markets are serving more Medicaid enrollees than in the past. 18 If this trend holds, our study results would suggest that the probability of commercial plan MMC exits should decline in the future. Furthermore, states might consider taking advantage of this pattern by expanding MMC plan enrollment. This could be accomplished by limiting the number of plans awarded contracts within a county or by contracting with plans for groups of counties. Our results also suggest that states that hope to maintain or encourage commercial plan MMC participation need to establish sound capitation rates and havereasonableincreasesinthoseratesovertimesothatpaymentsreflectthe costs of serving the Medicaid populations enrolled in managed care. States must also take care to ensure that other program policies (such as service carve-outs) are not interfering with plans ability to manage care for Medicaid beneficiaries and that plans have the willingness and capacity to enroll sicker populations, particularly if fully capitated managed care will be the only service option for those enrollees. To increase retention rates for commercial plans, states should also play attention to ensuring that plans are able to enroll an adequate number of Medicaid enrollees. Finally, states concerned with keeping commercial plans in their MMC programs might want to monitor changes in the broader managed care market, since the returns in the private market and the overall profitability of managed care in a market appear to have implications for commercial plans decisions to continue in or exit from Medicaid managed care. This research was supported by the Centers for Medicare and Medicaid Services (CMS) under Contract no The study benefited from the insightful comments of Paul Boben and other CMS staff and Terri Coughlin andjohnholahanattheurbaninstitute. HEALTH AFFAIRS ~ Vo l u m e 2 4, N u m b e r

11 NOTES 1. Centers for Medicare and Medicaid Services, Managed Care Trends, , medicaid/managedcare/trends03.pdf (13 April 2005). 2. Ibid.;andS.Felt-Lisk,R.Dodge,andM.McHugh,Trends in Health Plans Serving Medicaid 2000 Data Update (Washington: Kaiser Commission on Medicaid and the Uninsured, 2001). 3. This definition of commercial plans, which has been used in earlier work on plan MMC participation, is based on the old 75/25 rule, which required that health plans serving Medicaid beneficiaries have at least 25 percent of their enrollees be non-medicaid beneficiaries. The rule was eliminated in the Balanced Budget Act (BBA) of S. Felt-Lisk, The Changing Medicaid Managed Care Market: Trends in Commercial Plans Participation (Washington: Kaiser Commission, 1999); and Felt-Lisk et al., Trends in Health Plans Serving Medicaid. 5. See, for example, M.J. McCue et al., Reversal of Fortune: Commercial HMOs in the Medicaid Market, Health Affairs 18, no. 1 (1999): ; T.A. Coughlin, S.K. Long, and J. Holahan, Commercial Health Plan Participation in Medicaid Managed Care: An Examination of Six Markets, Inquiry 38, no. 1 (2001): 2234; and D.A. Draper, R.E Hurley, and A.C. Short, Medicaid Managed Care: The Last Bastion of the HMO? Health Affairs 23, no. 2 (2004): Although we do not have information on all Medicaid-dominated plans in our database, we compared the exit behavior of the Medicaid-dominated plans in our data with that of commercial plans. We found significant differences in the factors associated with MMC exits for the two types of plans. 7. Mostrecently,M.P.Glavinetal., AnExaminationofFactorsintheWithdrawalofManagedCarePlans from the Medicare+Choice Program, Inquiry 39, no. 4(200203): ; and T. Lake and R. Brown, Medicare+Choice Withdrawals: Understanding Key Factors (Menlo Park, Calif.: Henry J. Kaiser Family Foundation, 2001). 8. This model assumes that the plan makes the decision on whether to continue participating in Medicaid managed care in a particular county. It should be noted that in some cases, states make a decision to not renew a contract for a plan in a particular market. We cannot distinguish between exits that are a result of a decision by the plan and exits that are by state choice. 9. For information on the InterStudy data, see InterStudy Competitive Edge, CompetitiveEdge.asp (12 May 2005). 10. Data for the average premium for employer coverage came from Medical Expenditure Panel Survey (MEPS)andCountyBusinessPatterns(CBP)data.MEPSprovidessingleandfamilypremiumdataby state and firm-size category. CBP provides the share of workers in a given firm-size category. The measure of county-level premiums is constructed by multiplying the share of workers in each firm-size category in the county with the corresponding premium data for that firm-size category in the state and summing up the values across the firm-size categories for the county. 11. CMS, Announcement of Calendar Year (CY) 2000 Medicare+Choice Payment Rates, 1 March 1999, (13 April 2005). 12. J. Holahan and S. Suzuki, Medicaid Managed Care Payment Methods and Capitation Rates in 2001, Health Affairs 22, no. 1 (2003): Exhibit 3 reports one-half standard deviation of the standardized growth rate (which is constructed as each state s growth rate divided by the mean growth rate across all states). For ease of interpretation, we report the change in terms of percentage points in the text. 14. See, for example, A. Leibowitz, J.L. Buchanan, and J. Mann, A Randomized Trial to Evaluate the Effectiveness of a Medicaid HMO, Journal of Health Economics 11, no. 3 (1992): I. Moscovice, M. Casey, and S. Krein, Expanding Rural Managed Care: Enrollment Patterns and Prospects, Health Affairs 17, no. 1 (1998): Coughlin et al., Commercial Health Plan Participation. 17. Glavin et al., An Examination of Factors ; and Lake and Brown, Medicare+Choice Withdrawals. 18. Felt-Lisk et al., Trends in Health Plans Serving Medicaid July/August 2005