The validity of a questionnaire on medicines used in health care practice: comparison of a questionnaire and computerized medical record survey

Background. To assess the effect of sociodemographic characteristics, perceived health, chronic disease, and country of birth on use of medicines and the validity of a questionnaire survey on the use of medicines. Methods. A questionnaire was handed out to all 1776 patients aged 16 years or over who had attended Jordbro Health Center, Haninge municipality, Sweden, during the preceding 6 months (response rate 81%), and the findings were linked to information obtained from the computerized prescriptions of medicines. Three measurements were used: medicine used, prescriptions, and medicines prescribed. Results. The data were analyzed by logistic regression technique in univariate and multivariate forms. Of the respondents 73% reported using some type of medicine, compared to 63% who had been given a prescription by a physician in the Health Center during the past year. The number of medicines used was 3.26 vs. 4.17 prescribed, but after controlling for possible confounders the figures were 3.51 vs. 3.94. Female gender and chronic disease were related to high prevalence of drug use. Old age, female gender, sick-leave or disability pension, and chronic disease were also related to a high prevalence of prescribed medicines, while old age, sick-leave, perceived bad health, and chronic disease were related to the numerous medicines used. Old age, perceived bad health, and chronic disease were related to a large number of prescriptions, while old age and chronic disease were related to a large number of prescribed medicines. Conclusions. Differences were found between the two methods used to analyze data. The prevalence of medicines used was overestimated, but the number of medicines used was underestimated. This indicates that the prevalence rates based on analysis of the questionnaire are more reliable, while the number of used medicines should be interpreted with caution, and the effect of patient-concordance should be considered in future studies.