Risk of complication at the time of laparoscopic hysterectomy: a prediction model built from the National Surgical Quality Improvement Program database

Abstract

Background: Although laparoscopic hysterectomy is well established as a favorable mode of hysterectomy owing to decreased perioperative complications, there is still room for improvement in quality of care. Previous studies have described laparoscopic hysterectomy risk, but there is currently no tool for predicting risk of complication at the time of laparoscopic hysterectomy.

Objective: This study aimed to create a prediction model for complications at the time of laparoscopic hysterectomy for benign conditions.

Study design: This is a retrospective cohort study that included patients who underwent laparoscopic hysterectomy for benign indications between 2014 and 2017 in US hospitals contributing to the American College of Surgeons - National Surgical Quality Improvement Program database. Data about patient baseline characteristics, perioperative complications (intraoperative complications, readmission, reoperation, need for transfusion, operative time greater than 4 hours, or postoperative medical complication), and uterine weight at the time of pathologic examination were collected retrospectively. Postoperative uterine weight was used as a proxy for preoperative uterine weight estimate. The sample was randomly divided into 2 patient populations, one for deriving the model and the other to validate the model.

Results: A total of 33,123 women met the inclusion criteria. The rate of composite complication was 14.1%. Complication rates were similar in the derivation and validation cohorts (14.1% [2306 of 14,051] vs 13.9% [2289 of 14,107], P=.7207). The logistic regression risk prediction tool for hysterectomy complication identified 7 variables predictive of complication: history of laparotomy (21% increased odds of complication), age (2% increased odds of complication per year of life), body mass index (0.2% increased odds of complication per each unit increase in body mass index), parity (7% increased odds of complication per delivery), race (when compared with white women, black women had 34% increased odds and women of other races had 18% increased odds of complication), and American Society of Anesthesiologists score (when compared with American Society of Anesthesiologists 1, American Society of Anesthesiologists 2 had 31% increased odds, American Society of Anesthesiologists 3 had 62% increased odds, and American Society of Anesthesiologists 4 had 172% increased odds of complication). Predicted preoperative uterine weight also had a statistically significant nonlinear relationship with odds of complication. The c-statistics for the derivation and validation cohorts were 0.62 and 0.62, respectively. The model is well calibrated for women at all levels of risk.

Conclusion: The laparoscopic hysterectomy complication predictor model is a tool for predicting complications in patients planning to undergo hysterectomy.

Keywords: NSQIP; laparoscopic hysterectomy; risk prediction.