Background: The definition of acute myocardial infarction (AMI) requires a rise and/or fall in troponin with 1 or more results ≥99th percentile of the reference range. How much troponin must change has not been specified. We ascertained whether dynamic changes (δ) in high-sensitivity troponin T (hs-TnT) improved diagnostic and prognostic test performance in the emergency department.
Methods: We recruited 939 patients with symptoms suggestive of acute coronary syndrome (without ST elevation). hs-cTnT was measured at 0 h and 2 h after presentation. End-points were admission diagnosis of AMI and 1-year adverse events (composite of death, AMI, revascularization).
Results: Diagnostic specificity of 0-2-h hs-cTnT for AMI (incurred by 200 patients) improved from 79.8% (78.8%-80.5%) by using the 99th percentile alone to 94.2% (92.9%-95.4%) when we also included a δ ≥20%, but diagnostic sensitivity decreased from 94.5% (90.7%-96.9%) to 49.5% (44.6%-53.9%). With the inclusion of those patients with a δ ≥20% when 0-2-h hs-cTnT was <99th percentile, in addition to any with concentrations ≥99th percentile, diagnostic sensitivity increased to 97.5% (94.4%-98.9%). hs-cTnT ≥99th percentile predicted adverse events (incurred by 111 patients), adjusted hazard ratio 1.9 (1.2-2.8), whereas a δ ≥20% did not, hazard ratio 1.1 (0.7-1.7).
Conclusions: Diagnostic specificity of hs-cTnT improved with the use of a δ ≥20% in those patients with concentrations ≥99th percentile, but at a cost of a large reduction in sensitivity. Diagnostic sensitivity improved with the use of a δ ≥20% in patients with 0-2-h concentrations <99th percentile. Both approaches may be required for optimum rule-in and rule-out strategies, respectively. The δ criteria seem to be less useful for medium-term risk stratification.