Prediction analysis for delayed discharge following gastric bypass surgery.

BACKGROUND: Growing rates of obesity and an increased demand for bariatric surgery have led to shorter post-operative hospital stays. While next-day discharge is becoming increasingly common, limited data exists to identify patients requiring longer stays. This study aims to develop a prediction model using preoperative factors to predict the successful discharge within 24hours (single-day discharge) following Roux-en-Y gastric bypass.

METHODS: A retrospective chart review was performed. Patient age, gender, preoperative weight, body mass index (BMI), medication use, comorbidities and discharge data were recorded. Cross-validation (CV) was applied to assess the generalizability of the model and CV error was used to identify the best predictor set. Results A total of 525 patients were included. The final model predicted prolonged hospital stay with a prediction error of 0.1545. The best predictive variables for delayed discharge included were asthma, preoperative weight, gastroesophageal reflux disease (GERD), gabapentin use, preoperative BMI, preoperative anticoagulant use, chronic pain, obstructive sleep apnea (OSA), dyslipidemia, chronic kidney disease (CKD) and type 2 diabetes mellitus (T2DM). Conclusions A prediction model including preoperative BMI, obesity-related comorbidities and medication use was generated. This model has the potential to support hospital resource management and triaging patients suitable for short-stay and increase the efficiency of resource allocation for gastric bypass.