Reliability of reported family history of myocardial infarction

Abstract

Objective - To assess the reliability of reported family histories of myocardial infarction. Design - A case-control study in which reported histories of first degree relatives were validated from death certificates, general practitioners' records, and hospital notes. Setting - Participants enrolled in the Belfast centre of the World Health Organisation's study monitoring trends and determinants in cardiovascular disease (MONICA). Subjects - 200 men who survived myocardial infarction and 200 age matched controls drawn randomly from the population. Main outcome measures - The sensitivity, specificity, positive predictive value, and proportion of overall agreement with validated records of reported family histories of myocardial infarction, given at least one reported relative or at least one verified relative being affected. Results - 349 of the 400 probands provided detailed family histories, reporting on 2812 first degree relatives. The overall sensitivity, specificity, and positive predictive value of reported histories were 67.3%, 96.5%, and 70.5% for cases and 68.5%, 97.7%, and 73.8% for controls. The κ coefficients were modest: 0.65 for cases and 0.68 for controls. The odds ratios for myocardial infarction, given at least one affected relative, were not substantially inflated by recall bias. Some recall bias was evident for the probands' reports of their siblings' histories of myocardial infarction, the odds ratio for a reported history being 1.67 (95% confidence interval 1.09 to 2.57) and for the validated history 1.54 (1.01 to 2.37). Conclusions - Although the relative risk of disease is correctly estimated, the predictive accuracy of a casual family history of myocardial infarction may limit the effectiveness of targeted screening programmes. They may, however, complement other strategies based on genetic testing.