OBJECTIVES: to determine the reproducibility and validity of information on medication obtained in a postal questionnaire, with face-to-face interviews providing the gold standard. METHODS: cross-sectional study (Perfil de Utilização de Medicamentos por Aposentados Brasileiros), using two approaches (postal and home visit) to collect information from retirees who were 60 years old or older and beneficiaries of Brazil's national social security system (INSS). Statistics used were: simple kappa (k), adjusted kappa (PABAK), weighted kappa, intraclass correlation coefficients, indicators of sensitivity and specificity, and Luiz et al's graph method. RESULTS: 234 elderly (male = 42%; female = 58%; mean = 71.7 years) answered both approaches. Concordance between the postal and face-to-face interview approaches was excellent (k=0.94) for hypoglycemics; very good (k = 0.83-0.82) for angiotensin-converting enzyme inhibitors and anti-hypertensives; good (k = 0.71) for diuretics; and reasonable (k = 0.47) for non-steroidal anti-inflammatories. Concordance was good (k = 0.61) for information on the total number of medications used. The validity of the postal approach was high, at times total, for the drugs used in treating diabetes (sensitivity and specificity = 100%), followed by anti-hypertensives. The lowest values obtained were for non-steroidal anti-inflammatories (sensitivity = 64%; specificity = 88%). CONCLUSION: the postal approach can be used to obtain accurate information on classes of medications used by a population aged 60 years or more with a social profile similar to that of social security beneficiaries.
Reproducibility; Validity; Concordance studies; Drug use; Health service for the elderly; Medications; Postal questionnaire; Face-to-face interview; Postal survey; Cross-sectional studies
