C54-13 Elevated Troponin as a Predictor of Short-term and Long-term Mortality in Sepsis: A Bayesian Random-effects Meta-analysis

Abstract Background Sepsis remains a leading cause of mortality worldwide, with early risk stratification critical for clinical decision-making. Elevated troponin, a marker of myocardial injury, is frequently observed in sepsis but its prognostic value across short- and long-term outcomes is uncertain due to study heterogeneity. We conducted a Bayesian meta-analysis to synthesize evidence on the association between elevated troponin and mortality in sepsis patients. Methods We conducted a Bayesian random-effects meta-analysis to assess the association between elevated troponin levels and mortality in patients with sepsis. Studies reporting adjusted hazard ratios (HRs) with corresponding standard errors for short-term (28-day) and long-term mortality were included. Effect sizes were analyzed on the natural logarithm (log HR) scale to ensure normality assumptions. A vague normal prior (0,10) was assigned to the pooled effect size (μ). Between-study heterogeneity (τ) was modeled using a half-normal prior with scale parameter = 1. Sensitivity analyses were performed by varying the prior distribution for τ, to evaluate the robustness of the pooled estimates and heterogeneity parameters. Posterior inference was based on Markov chain Monte Carlo (MCMC) sampling. Summary statistics included the posterior median and 95% credible intervals (CrI) for the pooled HR (exponentiated μ), prediction intervals for future studies, median between-study variance (τ), and relative heterogeneity (I²). Results In this Bayesian meta-analysis, elevated troponin levels were significantly associated with increased mortality in sepsis patients. For short-term (28-day) mortality, nine studies were included. The pooled hazard ratio (HR) was 1.66 (95% CrI: 1.29 to 2.14), with posterior mean log HR = 0.51 (95% CrI: -0.06 to 1.08), indicating a 66% higher risk of death. The prediction interval for a future study was 0.94 to 3.39, reflecting uncertainty due to heterogeneity. Heterogeneity was moderate (posterior median I² = 26.9%; τ = 0.17). For long-term mortality, 15 studies contributed data. The pooled HR was 2.08 (95% CrI: 1.51 to 2.93), with posterior mean log HR = 0.73 (95% CrI: -0.28 to 1.79), suggesting more than doubled risk. The prediction interval was 0.76 to 5.98. Heterogeneity was high (I² = 84.1%; τ = 0.44), attributable to differences in follow-up duration, troponin assays, and patient populations. Conclusions Elevated troponin is a robust predictor of both short-term (HR 1.66) and long-term (HR 2.08) mortality in sepsis, with stronger effects over extended follow-up despite high heterogeneity. These findings support routine troponin assessment for risk stratification in sepsis management. This abstract is funded by: None