Coding of acute myocardial infarction. Clinical and policy implications

To evaluate the appropriateness of diagnostic coding of acute myocardial infarction across teaching and nonteaching hospitals.

Retrospective review of a random sample of medical records to find evidence of the occurrence and active treatment of acute myocardial infarction during the admission.

Five tertiary teaching, five other teaching, and five nonteaching hospitals in metropolitan Boston.

Random sample of hospital admissions assigned a discharge diagnosis of acute myocardial infarction between October 1984 and September 1985.

Of the 1003 cases reviewed, 260 did not meet the clinical criteria for acute myocardial infarction. At tertiary hospitals, 175 (41.7%) failed to qualify, compared with 25 (9.1%) at nonteaching facilities. In a large fraction of the disqualified cases, the patients had been admitted to exclude the diagnosis of acute myocardial infarction; although explicitly "ruled out," an acute myocardial infarction code was assigned. Sixty-six cases from teaching hospitals did not qualify because the patient had been admitted only for coronary angiography after an uneventful postmyocardial infarction course. Almost one-third of these patients had had their infarcts from 5 to 8 weeks before the angiography admission.

Cases with an inappropriate discharge diagnosis of acute myocardial infarction may be concentrated in teaching hospitals. This finding could have implications for Medicare's diagnosis-related group payment system and governmental and other research efforts that use these data for such purposes as drawing inferences about the quality of hospital care.