The National Surgical Quality Improvement Program (NSQIP) dataset (2006 to 2008) was used. Patients undergoing bariatric surgery for morbid obesity (n = 32,889) were divided into training (n = 21,891) and validation (n = 10,998) datasets. Multiple logistic regression analysis was performed on the training dataset. The model fit from the training dataset was maintained and was used to estimate mortality probabilities for all patients in the validation dataset.
Thirty-day mortality was 0.14%. Seven independent predictors of mortality were identified: peripheral vascular disease, dyspnea, previous percutaneous coronary intervention, age, body mass index, chronic corticosteroid use, and type of bariatric surgery. This risk model was subsequently validated. The model performance was very similar between the training and the validation datasets (c-statistics, 0.80 and 0.82, respectively). The high c-statistics indicate excellent predictive performance. The risk model was used to develop an interactive risk calculator.
This risk calculator has excellent predictive ability for mortality after bariatric procedures. It is anticipated that it will aid in surgical decision-making, informed patient consent, and in helping patients and referring physicians to assess the true bariatric surgical risk.