Validity of a postal questionnaire with regard to the prevalence of myocardial infarction in a general population sample.

In order to evaluate the possibility of finding persons who have suffered a myocardial infarction (MI) by postal questionnaire, a self-administered questionnaire was sent to a random sample of 4400 men aged 45-64 years, drawn from the general population. The response rate was 95%. 176 men indicated that they had been hospitalized for MI, out of which 124 cases could be verified from medical records. Of the remaining men, 33 had evidence of cardiovascular disease (CVD) in their records but no MI, and 19 men had no evidence of CVD. The sensitivity (estimated from a subsample) was 100% and the specificity 98.7%. The predictive value was 100% for a negative response and 70.5% for a positive response. The 33 positive responders whose MI could not be verified but who had evidence of CVD had characteristics fairly similar to the responders with verified MIs. However, the 19 positive responders whose MI could not be verified and who had no evidence of CVD had characteristics that were dissimilar from the MI group as well as from the negative responders. The questionnaire thus identified all the MI cases. The need for validation can be limited to the relatively small group of positive responders.