Impact of changing diagnostic criteria on incidence, management, and outcome of acute myocardial infarction: retrospective cohort study

Participants, methods, and results

Since 1997, all patients admitted with chest pain to Monklands Hospital, Airdrie, had their troponin T concentrations measured. We identified patients admitted between April 1997 and December 2000 with a principal diagnosis of acute myocardial infarction, according to the old criteria, from routine discharge data, the databases of the coronary care unit and laboratory, and case notes. We used the databases to identify patients admitted for chest pain who had raised troponin T concentrations (≥0.1 ng/ml) in the absence of non-myocardial causes such as renal failure, thromboembolic disease, or myocarditis. The new criteria increased admissions for myocardial infarction by 58%, from 1671 to 2637; this equated to approximately 160 000 additional myocardial infarctions per year in the United Kingdom.

Compared with patients who met the old criteria, the additional 966 patients identified were older (median age 74 v 68 years; P<0.001; Mann Whitney U test) and a higher proportion were women (47% v 38%; P<0.0001, χ² test). Thrombolysis was given to only 13 of the additional patients compared with 672 patients who met the old criteria (1% v 40%; P<0.0001; χ² test). As a result, thrombolysis rates fell from 40% (95% confidence interval 38% to 42%) to 26% (24% to 28%).

Linkage to national admission (Scottish morbidity record) and death data (General Registrar's Office) provided information on survival, readmission for ischaemic heart disease, coronary angiography, and coronary revascularisation. We calculated cumulative probabilities of these outcomes up to one year of follow up using Kaplan-Meier product limit estimates.

The additional patients had higher 30 day mortality (P=0.016, log rank test) (table). The difference in mortality increased over one year (P<0.0001, log rank test). In a Cox proportional hazards model, the difference in survival on univariate analysis (hazard ratio 2.69; 2.32 to 3.12) was slightly attenuated after adjustment for age, sex, and deprivation (2.07; 1.77 to 2.42) and attenuated further after we adjusted for thrombolysis administration (1.87; 1.58 to 2.20). However, the difference remained significant (P<0.001). The univariate differences in readmission (0.71; 0.60 to 0.84; P<0.001), coronary angiography (0.60; 0.47 to 0.76; P<0.001), and revascularisation (0.69; 0.52 to 0.91; P<0.001) were no longer significant after we adjusted for demography.

Comment

The new criteria identified additional patients who were significantly different from those previously classified as having myocardial infarction in terms of demography, eligiblity for thrombolysis, and outcome. Their poorer survival may be, in part, because they are older and ineligibile for thrombolysis because ST segments are not elevated in electrocardiograms. Less frequent revascularisation may be because of poorer survival but may also reflect real differences in practice. Recent evidence suggests these patients might benefit from early revascularisation.4 Thrombolysis rates, 30 day mortality, readmission, and revascularisation are commonly used to assess and compare the quality of care provided by hospitals. Because hospitals vary in both access to troponin assays and adoption of the new diagnostic criteria,5 comparisons between hospitals are misleading. Similarly, longitudinal studies will be unable to differentiate between real changes and artefacts. Our findings have serious implications for auditing, benchmarking, and epidemiology.