RT Journal Article SR Electronic T1 Diagnosis of past history of myocardial infarction in epidemiological studies: an alternative based on the Caerphilly and Speedwell surveys. JF Journal of Epidemiology and Community Health JO J Epidemiol Community Health FD BMJ Publishing Group Ltd SP 116 OP 120 DO 10.1136/jech.42.2.116 VO 42 IS 2 A1 Yarnell, J W A1 Sweetnam, P M A1 Baker, I A A1 Bainton, D YR 1988 UL http://jech.bmj.com/content/42/2/116.abstract AB In epidemiological studies the diagnosis of a past history of myocardial infarction is made from the answer to a single question: "Have you ever had a severe pain across the front of your chest lasting for half an hour or more?" Two additional questions, which form an optional part of the London School of Hygiene and Tropical Medicine chest pain questionnaire, were used in two large community studies, with other information to determine the likely accuracy of the diagnosis ("Did you see a doctor about this pain?" If so, "What did he say it was?") The prevalence of possible myocardial infarction from the use of the single question was significantly higher among men from South Wales than among men from Speedwell, Bristol (10.1% and 6.9% respectively); in contrast, positive responses to the additional questions reduced the prevalence in the two populations to 5.8% and 4.9% respectively. These latter figures are very similar to those of self-reported coronary thrombosis in the two populations. Among subjects with positive responses to the additional questions the prevalence of ECG ischaemia was about 50%; in contrast, the prevalence of ECG ischaemia among those positive only to the severe chest pain question was very similar to that among those with no history of chest pain (12%). Preliminary mortality data show a similar classification of level of risk. These findings indicate that the false positive error rate for possible myocardial infarction could be significantly reduced by the use of two additional questions which form an optional part of the London School of Hygiene chest pain questionnaire but are rarely used. However, the present findings relate to populations with uniform levels of adequately accessible medical care; comparisons between populations with different levels of medical care will require cautious interpretation.