Sampling issues in nursing home research.

It has been suggested that two common methods of sampling nursing home populations, cross-sectional sampling and discharge sampling, result in samples with different characteristics and lengths of stay. Comparison of these samples to a sample of nursing home admissions has not been studied. This study compares characteristics and lengths of stay among cross-sectional, discharge, and admission samples. All current residents of three nursing homes in February 1987 made up the cross-sectional sample, all admissions in the following year made up the admission sample, and all discharges in the same year made up the discharge sample. The results of comparing these three sampling techniques show that the most striking differences occur between the cross-sectional sample and the admission sample. Persons in the cross-sectional sample tended to have longer nursing home stays as well as less social support and more behavioral and functional problems than persons in the admission sample, who tended to have shorter stays and more acute medical problems. The discharge sample was more similar to the admission sample than it was to the cross-sectional sample; however, some differences were found between the discharge and admission samples. Based on the differences found among the three samples, appropriate uses for each sample are discussed.