[Theory and practice of sampling studies: exemplified by the MONICA survey]

In parallel with the systematic registration of myocardial infarction, the MONICA-Project attempts to investigate at three different times the prevalence of risk factors for cardiovascular disease in the population. This article presents the sampling plan of the first MONICA survey in the cantons of Vaud, Fribourg and Tessin. The sampling procedure was at two levels: first, a sample of communes stratified according to community size was chosen, and secondly, within these communities, individuals were selected from the population registries. The prerequisites for an efficient sampling plan are discussed on a theoretical level. In addition, the practical constraints (budget, organizational problems, population registry files) are presented. Finally, all steps of the sampling procedure are described including the difficulties encountered. The discussion attempts a critical evaluation of the whole MONICA sampling procedure whose results are largely satisfactory.