Application of a Right-Sided Colectomy Anastomotic Leak Score to a Multi-institutional Cohort

An anastomotic leak (AL) calculator was previously developed and internally validated using National Surgical Quality Improvement Program (NSQIP) data for right-sided resections to assist with intraoperative decision-making regarding proximal stoma diversion. This study aimed to externally validate the existing right-sided AL score on institutional-level NSQIP data across three hospitals.

Patients undergoing right-sided colon resections with ileocolic anastomosis between 2011 and 2020 were identified within a multi-institutional NSQIP database that had not been used to develop the original AL risk score. Additional data were obtained through retrospective chart review. AL risk scores were calculated for these patients using the previously developed score. The AL risk score was then evaluated by calibration and discrimination analysis.

A total of 503 nondiverted patients were identified, with an overall 30-day AL rate of 5.6%. The C-statistic was 0.59, compared to 0.64 in the original development cohort. In seven of ten deciles, the actual AL rate exceeded the predicted rate. Stratifying patients into deciles 1-7 versus deciles 8-10 revealed a significant difference in the AL rate (4.9%, 95% confidence interval: 3.1%-7.4% versus 14.3%, 95% confidence interval: 4.6%-33.3%, P = 0.04).

Although the observed AL rates were higher than that predicted by the previously developed AL risk score, the score effectively stratified patients into lower and higher risk categories. Further investigation is needed to identify institutional-specific AL risk factors that may not have been considered in the current scoring system.