Research That Matters (January 17 - 20, 2008)

BACKGROUND AND PURPOSE: Previous research has provided evidence that managed care contains service costs through the minimization of inpatient services. Due to concerns for the impact of the HMO model on the seriously mentally ill, the alternative of soft capitation through the carve-out model has recently been promoted. Yet, there have been continuing pressures to capitalize on the greater savings and continuity of care associated with HMOs. This study, therefore, compares service patterns involving acute psychiatric hospitalization, selected outcomes, including rehospitalization, and costs associated with the HMO, carve-out models, and fee-for-service models as they have been implemented in a large northeastern U.S. state between FY1994 and FY2000. The study also aims to test hypotheses that are consistent with economic agency theory about the impact of diverse incentive structures on services. METHODS: This is a longitudinal, quasi-experimental secondary analysis that utilizes unduplicated individual data from a publicly mandated state casemix tracking system. This is supplemented with data from the STF-3C file of the 2000 U.S. Census and from the state’s Hospital Financial Reports. Primary analyses focus on 56,518 individuals who were hospitalized on psychiatric units within 57 hospitals throughout the state. Interrater Kappa and other reliabilities were computed for items such as diagnosis. In addition to standard descriptive statistics, including weighting for differential casemix, the study also employed event history analysis, specifically Cox regression, to compare and model rehospitalization rates between the three programs. RESULTS: The hypotheses of the study were largely supported, that HMOs have the most substantial impacts in minimizing service provision, with the carve-out program having an impact intermediate between the HMO and fee-for-service programs. Lower rates of utilization were associated with lower overall rates of hospitalization, shorter lengths of stay, fewer repeated stays, in addition to diminished geographic access and displacement of psychiatric patients to medical units. The HMOs also saw the greatest rates of AMA discharges. The final model of rehospitalization has an overall predictive accuracy of 59.6%. CONCLUSIONS AND IMPLICATIONS: The differences between the three programs are consistent with economic agency theory which anticipates that particular payment incentives will be associated with either high levels of service provision and costs (in the case of the FFS program); low levels, in the case of the HMOs; or intermediate levels, in the case of the carve-out with soft capitation and a mixed incentive structure. The study also illustrates the importance of examining diverse outcomes, rather than maintaining an exclusive focus on rehospitalization. Financial pressures to minimize inpatient service provision may reduce unnecessary hospitalization, but they may also lead to the displacement of those in need of care to less appropriate settings, among other unanticipated effects. Despite the greater continuity of care and lower costs of HMOs, the findings suggest that this model may not be suited for the SMI who require periodic inpatient care, and therefore, the continued provision of both alternatives may be indicated. Further research is needed that tests the robustness of these findings, for example, through instrumental variables analysis.