A novel model for early prediction of in hospital mortality in seawater drowning: The SNOP score

Background Drowning is a leading cause of preventable mortality worldwide; however, early in-hospital risk stratification remains limited. Although tools such as the Szpilman score assist in early severity assessment, they may not fully capture the evolving clinical status after admission. This study aimed to develop a simplified and objective model based on readily available parameters to predict in-hospital mortality following seawater drowning. Methods This retrospective study was conducted at a referral emergency department (ED) in northern Turkey between July 1, 2011, and December 31, 2024. Of 190 patients initially included, 166 with complete clinical and laboratory data were analyzed. Data were obtained from institutional and national health information systems. Clinical, physiological, and biochemical variables were assessed. Predictors of in-hospital mortality were identified using receiver operating characteristic (ROC) analysis and multivariable logistic regression. Variables with near-perfect discrimination (e.g., GCS, pH, Szpilman score) were excluded to avoid overfitting. Results Among the 166 patients, 34 (20.5%) died during hospitalization. CPR and endotracheal intubation rates were significantly higher among non-survivors (CPR: 97.1% vs. 0%; intubation: 97.1% vs. 2.3%; both p<0.001). Non survivors also presented with lower GCS (median 3 vs. 15), lower arterial pH, and higher Szpilman scores (all p<0.001). ROC analysis identified four potential predictors with AUC values between 0.90 and 0.95—pCO₂, lactate, SpO₂, and sodium—all showing significant discriminatory capacity (p<0.001). These variables were entered into a binary logistic regression model, from which serum sodium (OR=2.110; 95% CI: 1.310–3.401; p=0.002) and SpO₂ (OR=0.902; 95% CI: 0.847–0.961; p=0.001) emerged as independent predictors. These formed the basis of the SNOP score (Saturation and Natremia-based Outcome Predictor), a two-parameter logistic model demonstrating excellent performance: AUC=0.996, sensitivity=99.0%, specificity=96.2%, and overall accuracy=98.4%. Conclusion: The SNOP score is a simple, ED-specific tool for early prediction of in-hospital mortality in seawater drowning. It complements existing assessment systems by incorporating objective, admission-based parameters. Prospective multicenter validation is warranted to confirm its clinical applicability and support broader implementation.