Managed care today, for both Medicare beneficiaries and others, bears increasingly less resemblance to the managed care plans of the early 1990s (Lesser, Ginsburg, and Devers 2003). Utilization management strategies such as gatekeeping and preauthorization are on the decline. Many health maintenance organizations (HMOs) are starting to look more like preferred provider organizations (PPOs) because of reductions in selective contracting. And cost sharing is increasingly shifting to patients. With Medicare reform looming, accompanied by heated debates about a larger role for managed care and competition, there is a heightened need for data on the performance of managed care in its current form. But managed care's rapid evolution has made it a moving target, particularly challenging to study in a timely fashion. Researchers often lack up-to-date data, so that studies frequently draw conclusions that have considerably less relevance when they are published than they would have had were managed care a more stable element in the health insurance market. But getting the data to keep up with managed care is difficult to achieve. Thus, at a minimum, the data used should contain elements pertinent to the contemporary status of managed care. Whether data sources are old or new, to have findings that are useful to policymakers, researchers must interpret their findings in the context of the changes managed care has undergone. Several papers in this issue of Health Services Research touch on different aspects of managed care. What they have in common is research involving data that predate recent changes in managed care, yet each provides valuable lessons for managed care's latest evolution.
Despite the controversy over increasingly incorporating managed care and competition in Medicare, there is a growing recognition that managed care—in any of a variety of incarnations—will coexist with fee-for-service (FFS) Medicare (Gold 2003). This coexistence has the potential for influencing delivery and cost of care for FFS patients, known as spillover effects. While there is no overall consistent pattern, studies over the past three decades have shown some benefits from managed care in the FFS sector, including reductions in hospital spending, increased rates of cancer screening, and modest reductions in mortality (Baker 2003; Mukamel, Zwanziger and Tomaszewsk 2001). More recently, research has addressed spillover effects on FFS Medicare beneficiaries with specific medical conditions, focusing on process of care and health care outcomes. For example, Heidenreich et al. (2002) used data from the Cooperative Cardiovascular Project (CCP) to examine the relationship between managed care penetration and process of care and health outcomes for patients with acute myocardial infarction. They demonstrated a modest but statistically significant and positive association between HMO market share and use of beta-blockers and aspirin among FFS patients.
The paper by Bundorf and colleagues in this issue of HSR refines our understanding of spillover effects by examining managed care competition in addition to market share (Bundorf et al. 2003). Competition is an infrequently considered aspect of market dynamics in studies of managed care spillover, in contrast to the relatively extensive literature on spillover and market share (Baker 2003). Like Heidenreich, Bundorf et al. used CCP data to study spillover effects in patients with acute myocardial infarction. They found that high managed care competition (greater than 10 HMOs per market area), controlled for managed care market share, was associated with a small but significantly increased rate of revascularization and angiography procedures (measured as a combined endpoint). Additionally, annual costs for FFS patients in high-competition markets were 11 percent greater than for FFS patients in low-competition markets. Competition per se was not associated with improved mortality, and the appropriateness of the increased utilization of services remains uncertain because the period of follow-up observation was relatively brief and quality of life outcomes were not studied. In any event, these findings stand in stark contrast to the assumed societal benefit of competition between plans that is frequently articulated by policymakers.
Bundorf and colleagues' analysis remains timely despite changes in managed care because they have deconstructed managed care into elements that remain relevant in today's market. Indeed, part of what makes their findings provocative is the dramatic flux in the Medicare managed care markets. In recent years markets have undergone considerable consolidation, through withdrawals and mergers (Feldman, Wholey, and Christianson 1999). At the same time, enrollment in M+C programs have declined nearly 50 percent since 1998 (Gluck 2001). Though published data are lacking on the current status of competition, it is likely that competition among Medicare HMOs has declined substantially. Moreover, a major source of change in Medicare managed care programs has been low reimbursement rates, which also affect competition. And as another paper in this issue of HSR suggests, a new risk-adjustment system designed to improve reimbursement for care of sicker patients may not achieve its desired goals (Beeuwkes Buntin, McClellan, and Newhouse 2003). Thus, fewer markets today and in the future may have the capacity to produce the market share and competition associated spillover effects observed in Bundorf's study, which used data collected from 1994 to 1996. This paper highlights the need for more research on the influence of competition on spillover effects given that changes in penetration and in competition may or may not occur in parallel.
While Bundorf et al. examined some of the individual elements of market dynamics, two other papers in this issue disentangle some of the elements of managed care plans themselves. Studies of managed care's impact on health services use, spending, and health outcomes for managed care enrollees or for FFS patients tend to aggregate plans with little regard to plan structure. Plans often differ in their use of gatekeepers and preauthorization requirements, defined provider networks, selective and risk contracting, and premium and cost-sharing structures, among others. Recognizing the lack of detail on plan structure as a potentially important limitation in previous health services research, Tye and colleagues characterize managed care plans by their individual components in an attempt to create a new typology for managed care (Tye, Liang, and Haas 2003). They then examine the association between different plan characteristics and rates of mammography use among managed care enrollees. Using the 1996 Medical Expenditures Panel Survey (MEPS)—which they identify as the only nationally representative survey to include detailed data on health plan structure—the authors found higher rates of mammography use in MCOs with defined provider networks than plans without such networks. They also report higher mammography rates among enrollees in plans with gatekeepers. Utilization was not associated with other elements of MCO structure, including in-network restrictions, copayment, deductible or coinsurance rates, or breadth of benefit coverage. In an accompanying paper, Phillips and colleagues also report a positive association between gatekeeper status and use of mammography, as well as for clinical breast exam, and cervical cancer screening using the Pap test (Phillips et al. 2003). In contrast, they did not find a significant association between gatekeepers and prostate cancer screening, a more controversial intervention. This latter finding suggested that differential use of services under gatekeeping is associated with evidence-based practices.
As with the paper on spillover effects, these findings must be viewed in the context of the changing managed care environment. Data from the Community Tracking Study suggest that major declines in both narrowly defined provider networks and gatekeeping have occurred in recent years (Lesser, Ginsburg, and Devers 2003; Mays, Hurley and Grossman 2003) and there are few indications that they will regain popularity in the near future. Indeed, gatekeeping alone has received mixed reviews for reducing costs and limiting moral hazard (Clancy and Franks 1997; Martin et al. 1989; Escarce et al. 2001; Kapur et al. 2000). Thus, gatekeeping and defined provider networks may play a more limited role for managed care organizations going forward. But the more important take-home point from these studies may actually be that health services researchers, and policymakers for that matter, need to pay closer attention to the details of managed care organizations. Managed care plans have pursued different avenues for cost-containment and adoption of quality improvement strategies, and have variably shifted their emphasis on consumerism, all of which may produce a more heterogeneous market (Draper et al. 2002; Gold 2003; Mays, Hurley, and Grossman 2003). Focusing on “managed care” or “HMO” coverage may thus conceal important effects. In this regard, even though they use data from 1996, the work of Tye and Phillips is particularly relevant today because of their focus on the individual components of managed care plans. Having the detailed data of the MEPS allowed them to focus on characteristics, rather than labels.
It is likely that the impact of different plan characteristics on services utilization, expenditures, and health outcomes depends on the specific outcome under study. But regardless of the association, investigators conducting research where managed care is a major focus need to give serious consideration to collecting and using more detailed data on plan structure. Indeed, it is sobering that among major surveys only the Medical Expenditures Panel Survey contains such information. Significant advances in the field of managed care research could be made by adding such detailed information to other major, richly detailed surveys, including the Medicare Current Beneficiary Survey or the Health and Retirement Study.
Managed care is likely to stay with us for some time, both for Medicare patients and others. Understanding its effects on health services use and health outcomes will continue to be a challenge for health services researchers as plans grow increasingly difficult to categorize. Indeed, the extent of change in managed care in recent years suggests that interplan variation may sometimes exceed the difference between FFS and managed care, thus calling into question whether the whole construct of “managed care” makes any sense at all.