Objective: We validated the diagnoses of self-reported acute myocardial infarction (AMI) treated in hospital.
Study design and setting: The agreement between myocardial infarction reported in a postal questionnaire in 1998 and data from the Augsburg Coronary Event Register were assessed in a representative sample of German men and women (n = 9,176) aged 25 to 74 years at baseline examination.
Results: Of the 9,176 persons, 207 men and women reported an incident AMI treated in hospital during the follow-up period. Of these, 148 persons fulfilled the criteria for verified AMI (positive predictive value 71.5%). Among the 8,969 respondents who reported no AMI, three persons had an AMI (negative predictive value 100%). The sensitivity was 98.0%, and the specificity 99.3%. Much of the false-positive reporting was related to cardiac hospitalizations, predominantly for coronary heart disease (42%).
Conclusion: A postal questionnaire seems to be a useful method to identify hospitalizations for incident nonfatal AMI cases in epidemiologic cohort studies. Because the proportion of false negatives is low, medical record reviews for case ascertainment can be limited to the group of positive responders.