CLINICAL INDICATORS DERIVED FROM THE PATIENT ASSESSMENT INSTRUMENT IN THE LONG-STAY RESIDENTS OF 69 VA NURSING-HOMES

Objectives: Previous work has shown that clinical indicators reflecting occurrence of bedsores, behavioral disturbances, and deterioration of activities of daily living (ADLs) can be calculated for the long-stay residents of Veterans Affairs (VA) nursing homes from the standard biannual patient Assessment Instrument (PAI). The present study aimed to construct national curves for these indicators, against which each facility could in the future compare its own values; and to determine the correlations between the indicator values and selected nursing home characteristics. Methods: Eight indicators were calculated for the long-stay (more than six months) residents of the 69 VA nursing homes housing 50 or more such patients from the 1992 PAI data. The indicators were: prevalence of bedsores; incidence of bedsores; prevalence of physically aggressive behavior; incidence of aggressive behavior; and frequencies of six-month losses of eating, mobility, transfer, and toileting functions by the initially independent residents. Results: There was 2 two- to sixfold difference between the most favorable quartile and the least favorable quartile for the eight indicators. Significant correlations across institutions were found between the prevalence and incidence of bedsores, the prevalence and incidence of aggressive behavior, and the frequencies of declines in the four ADLs by the initially independent residents. One or several of the indicators were significantly superior in nursing homes with these Characteristics: 2 smaller size, a slower resident turnover rate, a smaller proportion of residents with nonorganic psychoses, a lower ratio of short-stay to long-stay residents, and a lower ratio of independent to dependent long-stay residents. Conclusions: These data provide national standards against which each VA nursing home can compare its PAI-derived clinical indicator values. The outcomes measured by these indicators appear to be influenced both by casemix and by environmental factors.