Risk prediction score in laparoscopic colorectal surgery training: experience from the English National Training Program

Abstract

Objective: The overall aim was to develop and validate a risk prediction score for laparoscopic colorectal surgery training cases.

Background: Published risk prediction scores are not transferable between hospitals because they are derived from a single institution's data and are not designed for use in training situations.

Methods: Cases from the prospectively collected database of the National Training Programme in Laparoscopic Colorectal Surgery, between July 2008 and July 2012, were analyzed. Independent risk factors for conversion were identified by the logistic regression. Converting the odds ratios into integers created a risk prediction score for conversion. The clinical impact of this score was investigated by comparing postoperative complications and the level of trainer input in high- and low-risk cases. To study whether adverse outcomes in predicted high-risk cases occur outside the National Training Programme in Laparoscopic Colorectal Surgery, 2 external data sets were examined.

Results: A total of 2341 cases carried out in 42 hospitals were analyzed. Significant risk factors for conversion were body mass index, American Society of Anesthesiology classification, male sex, prior abdominal surgery, and resection type. At a risk score of more than 6, complication rates increased, including mortality (2.9% vs 0.5%, P < 0.001), anastomotic leak (4.3% vs 1.4%, P = 0.002), and a higher level of trainer input (32.2% vs 19.9% of cases, P < 0.001). Analysis of 786 external cases showed that high-risk cases had higher conversion (18.8% vs 7.1%, P < 0.001), overall complication (36.4% vs 15.0%, P < 0.001), and leak rates (4.0% vs 1.3%, P = 0.015).

Conclusions: A risk predication score to facilitate case selection in laparoscopic colorectal surgery training was developed and validated.