Comparison of diagnostic accuracy between three different rules of interpreting high sensitivity troponin T results

With the introduction of high sensitivity troponin-T (hs-TnT) assay, clinicians face more patients with 'positive' results but without myocardial infarction. Repeated hs-TnT determinations are warranted to improve specificity. The aim of this study was to compare diagnostic accuracy of three different interpretation rules for two hs-TnT results taken 6 h apart. After adjusting for clinical differences, hs-TnT results were recoded according to the three rules. Rule1: hs-TnT >13 ng/L in at least one determination. Rule2: change of >20 % between the two measures. Rule3: change >50 % if baseline hs-TnT 14-53 ng/L and >20 % if baseline >54 ng/L. The sensitivity, specificity and ROC curves were compared. The sensitivity analysis was used to generate post-test probability for any test result. Primary outcome was the evidence of coronary critical stenosis (CCS) on coronary angiography in patients with high-risk chest pain. 183 patients were analyzed (38.3 %) among all patients presenting with chest pain during the study period. CCS was found in 80 (43.7 %) cases. The specificity was 0.62 (0.52-0.71), 0.76 (0.66-0.84) and 0.83 (0.74-0.89) for rules 1, 2 and 3, respectively (P < 0.01). Sensitivity decreased with increasing specificity (P < 0.01). Overall diagnostic accuracy did not differ among the three rules (AUC curves difference P = 0.12). Sensitivity analysis showed a 25 % relative gain in predicting CCS using rule 3 compared to rule 1. Changes between two determinations of hs-TnT 6 h apart effectively improved specificity for CCS presence in high-risk chest pain patients. There was a parallel loss in sensitivity that discouraged any use of such changes as a unique way to interpret the new hs-TnT results