Selection bias and utilization in the Minnesota Senior Health Options demonstration

Medicare and Medicaid dually eligible beneficiaries (DEBs) are among the frailest, most vulnerable and most costly populations in the United States. DEBs often are disadvantaged by the complicated Medicare and Medicaid programs and the fragmented care systems. The Minnesota Senior Health Options (MSHO) demonstration pools Medicare and Medicaid funding to managed care organizations to provide integrated, coordinated and flexible primary, acute, long term care and social services to DEBs. A prior evaluation of MSHO found the program was more effective in reducing inpatient hospital utilization, emergency room and physician visits for nursing home residents than for community residents compared with control groups during its first three years of operation. However, administrative data limitations, such as missing inpatient diagnosis and functional status, raise concerns about possible selection bias from those unobserved variables and their effects on the findings of the MSHO evaluation study. This thesis uses the quasi-experimental study design of the MSHO evaluation study, but applies different statistical and econometric models for panel and longitudinal data. The panel data models find minor favorable MSHO selection on unobserved chronic health status for the nursing home and community nursing home certifiable groups and minor adverse MSHO selection on unobserved chronic health status for the community non-nursing home certifiable group. The longitudinal data models find little significant selection bias on observed variables. Overall, the various models applied in this thesis do not find that selection bias on unobserved variables was important enough to change the general conclusions of the MSHO evaluation study. This thesis finds that the MSHO effects, which lowered some utilization measures, did not significantly increase with longer MSHO enrollment time during the MSHO evaluation period. One possible reason for the lack of greater effects may be that the present MSHO care coordination is not sufficiently proactive in providing intensive care to the frail DEBs by incorporating innovative geriatric care models and changing the behavior of the primary care physicians of DEBs effectively.