High-sensitivity troponin and in-hospital mortality in non-ACS cardiac ICU admissions: a 14-year cohort

Background High-sensitivity cardiac troponin (hs-cTn) elevation is frequent in critically ill patients without acute coronary syndrome (ACS). However, its independent prognostic value and incremental contribution beyond established severity scores in real-world cardiac intensive care unit (CICU) populations remain debated, especially over long periods with potential assay drift. Purpose To evaluate the association between peak hs-cTn and in-hospital mortality in non-ACS CICU admissions and to quantify its incremental discriminative value over SAPS3 severity score using year-standardized troponin. Methods Retrospective single-center cohort including 15,271 CICU admissions (July 2011–February 2025). ACS admissions were excluded based on primary diagnosis code consistent with myocardial infarction (I21–I22) or unstable angina (I20.0). Exposure was defined as the peak troponin concentration during the CICU stay; high-sensitivity troponin assays were used throughout the entire study period at our institution. To address temporal drift, troponin was log-transformed [log(1+hs-cTn)] and standardized within admission year (z-score). The primary endpoint was in-hospital death. Incremental discrimination was assessed by comparing ROC AUC of a base model including SAPS3 points only versus an extended model SAPS3 + year-standardized hs-cTn, using 5-fold stratified cross-validation. Association adjusted for SAPS3 was estimated with logistic regression. Results After excluding ACS (n=1,740), 13,531 non-ACS admissions remained; troponin was available in 9,329 (68.9%). Median age was 67 years (IQR 53–80) and 52.8% were male. In-hospital mortality was 5.6% (520/9,329). Mortality increased across troponin quartiles: Q1 3.6% (83/2,333), Q2 5.3% (123/2,332), Q3 5.8% (136/2,332), Q4 7.6% (178/2,332). After adjustment for SAPS3, hs-cTn remained independently associated with mortality (adjusted OR 1.16 per 1-SD increase, 95% CI 1.08–1.25, p<0.001); Q4 vs Q1 showed higher risk (adjusted OR 1.95, 95% CI 1.44–2.65, p<0.001). Discrimination improved from AUC 0.862 (SAPS3-only) to AUC 0.866 (SAPS3 + hs-cTn), ΔAUC +0.004. Conclusion In a large real-world CICU cohort of non-ACS admissions, peak hs-cTn—standardized by admission year—was independently associated with in-hospital mortality and do not provided a incremental gain in discrimination beyond SAPS3.