Updating the predictive models for mortality and morbidity after low anterior resection based on the National Clinical Database.

INTRODUCTION: We previously developed risk models for mortality and morbidity after low anterior resection using a nationwide Japanese database. However, the milieu of low anterior resection in Japan has undergone drastic changes since then. This study aimed to construct risk models for 6 short-term postoperative outcomes after low anterior resection, i.e., in-hospital mortality, 30-day mortality, anastomotic leakage, surgical site infection except for anastomotic leakage, overall postoperative complication rate, and 30-day reoperation rate.

METHODS: This study enrolled 120,912 patients registered with the National Clinical Database, who underwent low anterior resection between 2014 and 2019. Multiple logistic regression analyses were performed to generate predictive models of mortality and morbidity using preoperative information, including the TNM stage.

RESULTS: We developed new risk prediction models for the overall postoperative complication and 30-day reoperation rates for low anterior resection, which were absent from the previous version. The concordance indices for each endpoint were: 0.82 for in-hospital mortality, 0.79 for 30-day mortality, 0.64 for anastomotic leakage, 0.62 for surgical site infection besides anastomotic leakage, 0.63 for complications, and 0.62 for reoperation. The concordance indices of all four models included in the previous version showed improvement.

CONCLUSION: This study successfully updated the risk calculators for predicting mortality and morbidity after low anterior resection using a model based on vast nationwide Japanese data.