TPW 4.02 Correlation Between Observed Mortality and NELA, ACS NSQIP and SORT Predicted 30-Day Mortality in Romania: A Retrospective Multicentre Observational Cohort Study

Abstract

Emergency laparotomy (EL) is associated with substantial postoperative mortality, with reported 30-day mortality (30-dm) rates of 8–18% in unselected adults. Although 30-dm risk models are widely used, their validation and implementation in Eastern European healthcare systems remain limited. This study evaluated the correlation between observed 30-dm and that predicted by the National-Emergency-Laparotomy-Audit (NELA), the American-College-of-Surgeons-National-Surgical-Quality-Improvement-Program (ACS-NSQIP), and the Surgical-Outcome-Risk-Tool (SORT) in Romania. This was a retrospective, multicentre cohort study spanning 2 years and included all adults ≥18 years undergoing EL at 4 acute care hospitals in North Eastern Romania, including one tertiary centre. The primary outcome was defined as the correlation between actual 30-dm and NELA, ACS-NSQIP or SORT predicted 30-dm. To strengthen transparency and reproducibility, the study adhered to the TRIPOD checklist. Statistical analysis was performed using GraphPad Prism v10.6.1. 793 cases were analysed. Median age was 69 (57,77) years, with M/F ratio of 1.27, ASA grade 3 (3,4), Clinical-Frailty-Score 5 (4,6), and Charlson Comorbidity Index 5 (3,7). Actual 30-dm was 21.68%. Predicted and correlated mortality estimates were: NELA 31.4% (16.6,56.1), R²=0.38, AUC=0.8968, p<0.0001; ACS-NSQIP unadjusted 8.00% (1.5,24.8), R²=0.49, AUC=0.9226, p<0.0001; ACS-NSQIP risk somewhat higher 19.0% (9.4,29.6), R²=0.43, AUC=0.8992, p<0.0001; ACS-NSQIP risk significantly higher 26.9% (12.7,34.9), R²=0.32, AUC=0.8474, p<0.0001; and SORT 10.1% (3.8,26.1), R²=0.21, AUC=0.8139, p<0.0001. Data are presented as median±IQR. All models correlated with 30-day mortality, yet validation showed strong discrimination alongside markedly variable calibration.