Background Planning of clinical services for acute myocardial infarction (MI) and studies of trends in disease incidence is usually based on hospital discharge diagnostic coding data, from which hospital episode statistics (HES) are derived. However, the validity of these data is uncertain. In the first ever population-based study of all incident and recurrent acute MI without restriction by age, we attempted to validate routinely collected hospital discharge diagnosis coding of hospitalised patients with acute MI.
Methods The Oxford Vascular Study (OXVASC) is a prospective population-based study of all incident and recurrent acute vascular events in a population of 90 542 in Oxfordshire, UK. Multiple sources of case-ascertainment are used, including daily identification and assessment of all potentially eligible hospital admissions. All patients in OXVASC with a diagnosis of acute MI using the European Society of Cardiology and American College of Cardiology definitions, and hospitalised in the Oxford Radcliffe Hospitals group were included in this substudy. Cases ascertained in OXVASC were cross-referenced with hospital discharge codes, with both the primary and secondary diagnosis fields being matched with the relevant hospital admission. All primary and secondary ICD 10 discharge codes were pooled for analysis.
Results Of 820 acute MI identified in the first 5 years of OXVASC (2002–7), only 438 (53%) had a hospital discharge diagnostic code of acute MI (ICD 10 121–122). A further 205 (25%) cases had other non-specific ICD 10 codes (I20, I23, I24, I25), indicating the presence of some form of coronary heart disease (CHD). There was no mention of CHD (ICD 10 120–I25) in 175 (21.3%) cases. The accuracy of discharge coding decreased with age (fig), and only 302 (49%) patients with acute MI aged over 65 years were given a hospital discharge diagnosis of acute MI. Of the 820 events included in the study, 114 (17.6%) occurred in patients already in hospital for other reasons (eg, elective surgery, other acute medical admission, etc), of which only 34 (23.6%) were given a discharge diagnosis of acute MI.
Conclusions Hospital discharge coding misses a substantial proportion of patients hospitalised with acute MI, particularly at older ages and following inhospital events. This lack of sensitivity will undermine most NHS service planning and research studies based on HES data and other similar routinely collected data on acute MI.