[Changing the order of questions on drug use in the Swiss Health Survey--does it effect prevalence assessment?].

A split-sample study investigated whether changes in the order of questions about the use of several medicines and remedies affected the estimation of prevalence of their use. In October 1996 the two versions of the questionnaires used for the First and Second Swiss Health Surveys were assigned respectively to two random samples (N1 = 537; N2 = 457). The general aim of the study was to investigate the comparability of the two questionnaire versions with regard to future analyses of trends. As assumed by the inclusion/exclusion model it was postulated that questionnaire changes in the order in which items on medicines and remedies were placed produced a contrast effect. A contrast effect would be a reduction in the estimation of the prevalence of use of those groups of medicines that followed remedies. The study found that question order produced no, or only small, effects on estimation of prevalence of use for purposes of future trend analysis. For eight groups of medicines the different questionnaire versions resulted in no significant differences (p > 0.05) in the prevalence of use. Effects on estimation of prevalence, however, must be expected according to whether a skip question about the use of medicines in general precedes the questions on specific medicines. As a tendency, the version without a skip question yielded a higher prevalence of daily use of tonics and fortifiers (4.8% versus 7.4%; p < 0.10). The difference (9.9% with a skip question versus 15.1% without) was significant (p < 0.05) for daily use of vitamins.