Misclassification in a questionnaire survey of varicose veins.J Clin Epidemiol. 1995 Sep; 48(9):1175-8.JC
The objective of this study was to assess the determinants and the effect of misclassification in a questionnaire survey of varicose veins. A random sample of 166 individuals aged 50 years in a larger prevalence study (n = 6874) was examined by a surgeon to validate the self-reporting of varicose veins. Varicose veins were assessed by a mailed questionnaire completed by the study subjects themselves and validated in a clinical examination by a surgeon. The sensitivity, specificity, positive and negative predictive values, kappa coefficient, and Chamberlain indices of the self-assessed diagnosis in reference to the diagnosis by surgeon were estimated overall and in categories of determinants. The overall sensitivity and specificity estimates were 0.92 and 0.93, respectively. The specificity appeared to be poorer among subjects who reported positive family history of varicose veins (0.83) than among those with negative family history (0.98). With respect to some other determinants of varicose veins such as gender, height and weight, occupational status and position at work, there was at best weak but nevertheless inconclusive evidence on possible differential misclassification. Sensitivity appeared to be relatively good for epidemiologic purposes but one could have hoped for better specificity. The correction for misclassification resulted in estimated prevalence of 19% (uncorrected 24%) in men and 43% (uncorrected 45%) in women. A questionnaire survey of varicose veins showed a relatively good validity. The effect of misclassification in the total 5-year-old study population was small in women but moderately large in men. Familial predisposition to varicosis was associated with a lower accuracy of self-assessed diagnosis.