MSS852: A CLINICAL NOMOGRAM FOR EARLY IDENTIFICATION OF PATIENTS AT RISK FOR DELAYED HOSPITAL DISCHARGE AFTER BARIATRIC SURGERY
Michael T Olson, MD; Yun Beom Lee, MD; Brian Layton, DO; Pamela Masella, DO; Department of Minimally Invasive and Bariatric Surgery, Brooke Army Medical Center, Fort Sam Houston, Texas, USA
Objectives: Early discharge after bariatric surgery is routine under enhanced recovery protocols, but still many patients remain hospitalized beyond postoperative day (POD) 1. We aimed to identify perioperative factors associated with increased risk for delayed hospital discharge and to develop a clinical nomogram for predicting that risk after bariatric surgery.
Methods: We conducted a retrospective cohort study of patients undergoing bariatric surgery between 01/01/2023 and 01/01/2024 at a military treatment facility. Delayed discharge was defined as length of stay >1 day. Candidate predictors included operative time >150 minutes, ≥1 overnight antiemetic dose, POD 0 oral intake <200 mL, POD 1 hemoglobin drop ≥2 g/dL, and any overnight hydromorphone use. Significant predictors from univariate logistic regression were entered into a multivariable model. A nomogram was constructed from the final model coefficients. Predicted probabilities were calculated for each patient and stratified into low (<0.40), moderate (0.40-0.60), and high-risk groups (>0.60) for delayed discharge based on an optimal threshold of 0.59. Model discrimination was assessed by area under the receiver operating characteristic curve (AUC), and internal validation was performed using bootstrap resampling (1000 iterations).
Results: Sixty four of 122 patients (52.5%) had delayed discharge during the study period. Significant predictors in multivariable analysis included operative time >150 minutes (OR 2.51, 95% CI 1.00–6.58), ≥1 overnight antiemetic dose (OR 2.26, 95% CI 0.97–5.37), POD 0 oral intake <200 mL (OR 3.24, 95% CI 1.42–7.69), POD 1 hemoglobin drop (OR 3.04, 95% CI 0.96–11.03), and any overnight hydromorphone use (OR 4.24, 95% CI 0.72–37.28). The nomogram demonstrated good discrimination (AUC 0.773), with sensitivity 0.61, specificity 0.82, positive predictive value 0.80, and negative predictive value 0.65 (Image 1A). Internal validation using bootstrapping (1000 resamples) showed optimism-corrected C-index 0.739 and calibration slope 0.80, indicating reasonable model performance and modest overfitting. Observed rates of delayed discharge were 28%, 52%, and 79% for low, moderate, and high-risk groups, respectively, demonstrating good alignment between predicted and observed outcomes (Image 1B).
Conclusions: We developed a clinically applicable nomogram to predict risk of hospital stay >1 day after bariatric surgery. Key perioperative predictors include operative time, antiemetic use, oral intake, hemoglobin drop, and opioid requirement which can all be collected before patient bedside evaluation on POD 1. Internal validation suggests the model is robust, supporting its use for prospective patient risk stratification and optimization of discharge planning.
