Unrecognized myocardial infarctions (UMI) are detected by surveillance electrocardiograms (ECGs). In epidemiologic studies, different sets of ECG criteria have been used to define myocardial infarction, possibly contributing to significant differences in prevalence estimates and risk factor associations. We sought to summarize the rationale behind the various UMI-ECG definitions and to suggest an approach to develop uniform criteria.

A comprehensive review of relevant publications from 1966 to 2002 was conducted.

Out of 14 major studies, the occurrence of UMI as a proportion of all infarctions varied from 4% to 44%, with markedly varying ECG criteria. No study directly addressed the rationale behind selection of ECG criteria. Computerized ECG analysis appears superior to visual reading due to better reliability, speed and cost while maintaining a similar predictive validity. Criteria requiring only major Q waves have the highest specificity for ECG-MI and for prediction validity for future coronary heart disease in middle-aged white men. The addition of minor Q waves and ST-T abnormalities improves predictive validity in middle-aged women and elderly men. Minor Q waves can have their diagnostic accuracy improved by several strategies, including use of ST-T wave changes, Washington Code, vector cardiogram, chronic obstructive pulmonary disease–ECG criteria, and serial ECG analysis.

Currently the most cost-effective and valid method for detecting UMI in epidemiologic studies appears to be computerized ECG analysis using major Q waves in middle-aged white men. Issues needing further research include morphologic validation of ECG-UMI criteria and the influence of age, sex, and race on ECG-MI criteria.