Abstract: Background: Acute heart failure (AHF) patients with high troponin levels have a worse prognosis. High-sensitive troponin T (hs-TnT) has been used as a tool to stratify prognosis in many scales but always as a qualitative and not as a quantitative variable.
Objectives: The main objective of this study was to determine the best hs-TnT cut-off for prediction of 30-day all-cause mortality.
Methods: The EAHFE registry, a prospective follow-up cohort of patients with AHF, was analysed. We performed a propensity score analysis of the optimal hs-TnT cut-off point previously determined by receiver operating characteristic (ROC) curve analysis.
Results: Of the 13 791 patients in the EAHFE cohort, we analysed 3190 patients in whom hs-TnT determination was available. The area under the ROC curve for 30-day all-cause mortality was 0.70 (CI95% 0.68 to 0.71; P < .001), establishing an optimal cut-off of hs-TnT of 35 ng/L. The sensitivity and specificity of this cut-off were 76.2 and 55.5%, respectively, with a negative predictive value (NPV) of 95.3%. A propensity score was made with 34 variables showing differences based on the cut-off of 35 ng/L for hs-TnT. In the analysis of the population obtained with the propensity score, patients with hs-TnT > 35 ng/L showed a greater 30-day all-cause mortality, with a HR of 2.95 (CI95% 1.83-4.75; P < .001). External validation reported similar results.
Conclusions: An hs-TnT value of 35 ng/L is an adequate cut-off to evaluate the prediction of 30-day all-cause mortality with a NPV of 95.3%.