suggest that skilled, subacute care in long-term care facilities is increasingly a substitute for hospital care.

Thus, overall, trends in the industry make it difficult to accurately develop address lists for health care facilities, and the internal diversity in living situations in facilities labeled “hospitals” or “nursing homes” exacerbate the problem. Address listing efforts that rely on the name of an institution, or the classification derived from state licensing regulations, may be out of step with the specific unit-by-unit duration of stay and level of care or oversight provided in the facility. For example, the unit occupied by “a person with Alzheimer’s [disease] who lives in a residential care setting that provides 24 hour, 7 day a week oversight could be counted as residing in a non-family housing unit, an institutional setting, or a [group quarters] non-institutional setting depending upon the name of the place (assisted living, nursing home or group home, respectively)” in which the unit is nested (Drabek, 2005:1).

Noting this limitation of facility-based analysis, McCormick and Chulis (2003:143) pursued another route by analyzing the Medicare Current Beneficiary Survey, whose sample of respondents is drawn from the Medicare Enrollment Files (person-based), and thus can be contacted wherever they are found. The authors estimate that 30 percent of Medicare recipients living in long-term care facilities lived in elder group residential arrangements in 2001, a near doubling from the 16 percent estimated in 1996; the percentage living in Medicare- or Medicaid-certified nursing homes (institutional) dropped from 70 percent in 1996 to 59 percent in 2001 (McCormick and Chulis, 2003:144).