EGS SO13 - Evaluation of Different Surgical Risk Prediction Models in Predicting Morbidity and Mortality in Giant Hiatus and Diaphragmatic Hernia Repairs

Abstract Background Emergency giant hiatus hernia and diaphragmatic hernia repairs are associated with high rates of mortality and morbidity; different studies reported variable outcomes. Accurate risk assessment empowers patients to provide informed consent and also serves as a valuable tool for identifying individuals who may require enhanced postoperative care, such as respiratory support and critical care admission. Various risk prediction tools have been developed to assist clinicians in making decisions to aid in stratifying the risk of surgical management. We aimed to assess the effectiveness of different risk prediction models in predicting mortality and morbidity in this patient cohort. Method Retrospective analysis performed on cases between May, 2010 and January, 2021 in a tertiary adult upper gastro-intestinal centre. The outcomes from the operations were assessed and compared with the predicted morbidity and mortality for different risk prediction models. Including the Surgical Outcome Risk Tool (SORT), the National Emergency Laparotomy Audit (NELA), parsimonious NELA and the American College of Surgeons National Surgical Quality Improvement Programme (ACS-NSQIP). The models were compared using the area under the curve (AUC). Morbidity models included Portsmouth Physiological and Operative Severity Score (P-POSSUM) and the ACS-NSQIP. Morbidity was evaluated by calculating the comprehensive complication index (CCI). Results A total of 108 patients met the criteria and were included in the analysis. Among these 50.9% were females. The median age was 69 (IQR 59-78). The 30-day mortality rate was 6.93% (7 patients), six patients died during the same admission, and one patient died during a re-admission. ACS-NQSIP had the highest predictive power for mortality (AUC = 0.6344), in comparison to NELA (AUC=0.5665), parsimonious NELA (AUC=0.4866) and SORT (AUC = 0.4342). Both ACS-NSQIP and P-POSSUM showed moderate correlation to CCI (rho = 0.489, p<0.001 and 0.446, p<0.001 respectively). Conclusion ACS-NSQIP is a better predictor of both mortality and morbidity in emergency giant hiatus and diaphragmatic hernia repairs when compared to NELA, parsimonious NELA, P-POSSUM and SORT. Interestingly, parsimonious NELA had a lower prediction rate when compared with the older NELA model. ACS-NSQIP may have a role in pre-assessment and consenting of emergency hiatus and diaphragmatic hernia repairs. Multi-centre prospective studies could be used to validate these findings.