Risk stratification for myocardial injury and mortality in acute carbon monoxide poisoning: a multivariable predictive model

Background: Myocardial injury following carbon monoxide (CO) poisoning is associated with increased mortality, yet early predictors remain poorly characterized. This study aimed to develop predictive models for early risk stratification using readily available clinical data. Methods: In a retrospective analysis of 714 patients with acute CO poisoning (2019-2024), we evaluated clinical and laboratory variables to identify predictors of myocardial injury (defined as cTnI ≥ 0.05 ng/mL) and 90-d mortality. Multivariable logistic regression was used to identify independent predictors, and model performance was assessed via ROC analysis. Results: Myocardial injury occurred in 132 patients (18.5%). Patients with injury were older (median age: 65 vs. 54 years, p < 0.001), had higher lactate levels (2.5 vs. 1.8 mmol/L, p < 0.001), higher carboxyhemoglobin concentrations (26.65% vs. 21.75%, p < 0.001), and more frequent hypocapnia (50.8% vs. 24.7%, p < 0.001). In the multivariable model adjusting for clinically relevant and univariately significant variables, age (per year increase; aOR = 1.02, 95% CI: 1.01-1.03), hypocapnia (aOR = 2.23, 95% CI: 1.23-4.07), and severe neurological impairment (aOR = 3.86, 95% CI: 2.12-7.05) were independently associated with myocardial injury. For 90-day mortality, independent predictors were age (per year increase; aOR = 1.14, 95% CI: 1.08-1.21) and severe neurological impairment (aOR = 7.53, 95% CI: 2.50-22.67). The models demonstrated good predictive accuracy for myocardial injury (AUC = 0.750, 95% CI: 0.703-0.797) and excellent predictive ability for mortality (AUC = 0.895, 95% CI: 0.846-0.944). Conclusions: The clinical risk model incorporating advancing age, severe neurological impairment, and hypocapnia enables risk stratification for myocardial injury, a critical intermediate marker of mortality in acute CO poisoning.