Identification of non-fatal myocardial infarction through hospital discharge data in Western Australia.

The validity of identifying incident cases of non-fatal acute myocardial infarction (AMI) between 1971 and 1982 in Western Australia from routine hospital records was assessed in ages 25-64 years, according to the WHO criteria defined in 1970 and 1983. This was done by reviewing original data sources and by using the Perth Coronary Register of 1971 as an external reference. Events with a coded discharge diagnosis of acute or subacute ischemic heart disease were found to be highly sensitive (97%) for cases of "definite" AMI (WHO 1983 criteria). The specificity of such events was lower (positive predictive value of about 70%) and validation of these events would be necessary for studies requiring high specificity. The sensitivity and specificity of these events for "definite" AMI seemed quite stable over time with similar values being observed in 1971 and 1978. Although the situation for "possible" AMI (non-fatal) is less clear, certainly many more records would need to be reviewed to validate this diagnosis.