Response bias in a health status survey of elderly people.

This paper compares respondents and non-respondents from the community sample of the Saskatchewan Health Status Survey of the Elderly. Response bias was assessed by comparing the demographic characteristics and use of health care services of the two groups. A stratified two-stage area probability sample was drawn from a comprehensive sampling frame. There were 1614 subjects eligible; interviews were completed with 1267 (78.5%). In the very elderly (85+ years) cohort, disproportionately more urban dwellers and more males were interviewed; the sample was otherwise demographically representative of the elderly population. Non-respondents, especially the very elderly, used significantly more medical services than respondents, and had a higher number of hospital admissions. Non-respondents over age 75 experienced significantly longer average lengths of stay. On average, non-respondents used approximately 15% more hospital days. Non-respondents over age 75 appear to be more likely to experience ill health than respondents. Hence, statistics from this survey are conservative estimates of the ill health of the elderly.