Research with older people: Comparison of a general practice and electoral roll sample with the Australian national health survey

Objectives. To establish the utility of general practice and electoral roll sampling frames in research involving older age groups. Methods. Demographic characteristics, medication use, self-rated health and walking behaviour of research based samples derived from general practice patients, and members of the electoral roll, were compared with participants in the National Health Survey (NHS) residing in Melbourne, Victoria, Australia. Participants. The GP sample comprised 267 older patients, (64% patient participation rate), randomly selected from 42 general practices, (52% GP participation rate). The electoral roll sample comprised I000 older people sampled from electoral roll data (70% participation rate). These two samples were compared with duta on those over age 65years from the 199.5 NHS sample, a multi-stage area sample (93.5 % completion rate). Results. The distribution of demographic characteristics in the GP group was similar to that of the NHS group whereas the sample based on the electoral roll contained more men. The GP sample rated their health in the middle of the range and the electoral roll sample rated their health more positively compared with the NHS group. Medication use differed in both groups compared with the NHS sample. Conclusion. Health related characteristics of both the general practice and the electoral roll samples differed from the "gold standard" NHS sample. Interpretation of research findings from similar samples would benefit from consideration of potential effects of selection and response bias. General practice sampling frames may provide a useful alternative for population based research.