We used MAX 2006 Person Summary (PS) files to develop our measures of LTC system performance. MAX PS files contain demographic and enrollment information for each Medicaid enrollee, as well as information on total Medicaid expenditures for services used during the calendar year, by service type. They also contain information on use of and spending on 1915(c) waiver services -- an important vehicle that states use to provide expanded HCBS to select populations.
We defined HCBS to include services covered under Section 1915(c) waivers and personal care, residential care, home health care, adult day care, and private duty nursing services that are mandatory or provided at state option outside of waiver programs. Institutional care includes nursing home care, ICFS/IID care, inpatient psychiatric services for people under age 21, and psychiatric hospital services for those 65 and older. The MAX PS files cannot be used to differentiate between people using institutional care for long periods and those using Medicaid institutional care for acute events. This study's operational definition of ILTC thus broadly includes all care received in institutions, whether or not a person is using them for LTC.5 Moreover, as earlier noted, this analysis was unable to differentiate between Medicaid beneficiaries living in and associated Medicaid spending on certain residential care facilities certified as "small" ICFs/IID (those with no more than 16 residents) and similar settings and their residents in other states licensed as "group homes" (and therefore qualifying for Medicaid HCBS reimbursement only for services and not room and board costs also covered in ICFs/IID).
Our analyses were limited to Medicaid enrollees eligible on the basis of disability or age and who were eligible for full Medicaid benefits in 2006. We excluded Program of All-Inclusive Care for the Elderly (PACE) or other managed LTC enrollees because information on their use of services (HCBS or institutional care) often is missing or unreliable in MAX. We also excluded from the analysis 11 states with MAX fee-for-service (FFS) data that are potentially unrepresentative or unreliable, including Arizona, Maine, Massachusetts, Michigan, Minnesota, Montana, New Hampshire, Oregon, Pennsylvania, Rhode Island, and Texas. Finally, we could not differentiate enrollees with physical disabilities from those with ID/DD in the District of Columbia, Washington, and Wisconsin, and excluded these states from our subgroup analyses. A more detailed discussion of the MAX data, analyzed measures, and methods used is in Appendix B. Appendix C lists state-specific data anomalies.
The analysis of state constraints and policies related to LTC provision relied on a wide range of publicly available data sources. When available, we used data from 2005, 2006, or previous years to capture policies in place and state characteristics at the time that services were being used in 2006. We also used the 2007 American Community Survey's (ACS's) income and disability data to construct estimates of the number of people over 65 or with disabilities potentially eligible for Medicaid in each state.6 Medicaid programs vary substantially in terms of the populations they cover. We used ACS-based measures of the size of potential Medicaid-eligible populations (assuming national eligibility criteria) to determine the extent to which cross-state differences in LTC utilization and spending result from state coverage policies.