Background Enhanced recovery after surgery (ERAS) pathways have reduced hospital stays after bariatric surgery, yet a subset of patients still require prolonged hospitalization. We aimed to identify predictors of delayed discharge and develop a validated clinical nomogram to estimate the likelihood of > 1 day postoperative stay. Methods We performed a retrospective cohort study of consecutive adults undergoing primary or revisional minimally invasive sleeve gastrectomy or Roux-en-Y gastric bypass at a single military treatment facility from 01/01/2022 to 12/31/2024. Demographic, comorbidity, operative, and perioperative data were abstracted. The 2023 cohort served as the development set for univariable and multivariable logistic regression to identify predictors of delayed discharge (> 1 day). A clinical nomogram was constructed from independent predictors and internally validated via bootstrap resampling (200 iterations). Temporal validation was performed on 2022 and 2024 cohorts. Results Among 281 patients (mean age 47.2 ± 11.3 years; mean BMI 40.5 ± 6.0 kg/m2; 26.7% male), 141 (50.2%) experienced delayed discharge. Independent predictors included operative time > 150 min (OR 3.00, 95% CI 1.14–8.09), overnight hydromorphone use (OR 3.78, 95% CI 1.40–11.0), ≥ 1 overnight antiemetic dose (OR 2.55, 95% CI 1.04–6.27), postoperative day (POD) 0 oral intake < 200 mL (OR 2.43, 95% CI 1.01–6.01), and POD 1 hemoglobin decrease ≥ 2 g/dL (OR 4.16, 95% CI 1.25–15.3). The final five-variable model demonstrated strong discrimination (AUC 0.77; bias-corrected C-index 0.74) and calibration (Hosmer–Lemeshow p = 0.17). Temporal validation confirmed robust performance (AUC 0.77–0.87). In sensitivity analysis, model discrimination remained high for both primary (AUC 0.79) and revisional cases (AUC 0.88). A web-based Shiny risk calculator was developed for bedside use (<a href="https://michaeltolson.shinyapps.io/bariatric-delayed-discharge-2023/">https://michaeltolson.shinyapps.io/bariatric-delayed-discharge-2023/</a>). Conclusions A five-variable nomogram accurately predicts delayed discharge following bariatric surgery and demonstrated strong temporal validation. This tool may aid individualized discharge planning.

Development and temporal validation of a clinical nomogram to predict delayed discharge after bariatric surgery

期刊论文
出版模式：

开放获取
发表日期：

2026年3月6日

查看全文

Michael T. Olson,
Yun Beom Lee,
Pamela Masella,
Brian D. Layton

摘要

Background

Enhanced recovery after surgery (ERAS) pathways have reduced hospital stays after bariatric surgery, yet a subset of patients still require prolonged hospitalization. We aimed to identify predictors of delayed discharge and develop a validated clinical nomogram to estimate the likelihood of > 1 day postoperative stay.

Methods

We performed a retrospective cohort study of consecutive adults undergoing primary or revisional minimally invasive sleeve gastrectomy or Roux-en-Y gastric bypass at a single military treatment facility from 01/01/2022 to 12/31/2024. Demographic, comorbidity, operative, and perioperative data were abstracted. The 2023 cohort served as the development set for univariable and multivariable logistic regression to identify predictors of delayed discharge (> 1 day). A clinical nomogram was constructed from independent predictors and internally validated via bootstrap resampling (200 iterations). Temporal validation was performed on 2022 and 2024 cohorts.

Results

Among 281 patients (mean age 47.2 ± 11.3 years; mean BMI 40.5 ± 6.0 kg/m²; 26.7% male), 141 (50.2%) experienced delayed discharge. Independent predictors included operative time > 150 min (OR 3.00, 95% CI 1.14–8.09), overnight hydromorphone use (OR 3.78, 95% CI 1.40–11.0), ≥ 1 overnight antiemetic dose (OR 2.55, 95% CI 1.04–6.27), postoperative day (POD) 0 oral intake < 200 mL (OR 2.43, 95% CI 1.01–6.01), and POD 1 hemoglobin decrease ≥ 2 g/dL (OR 4.16, 95% CI 1.25–15.3). The final five-variable model demonstrated strong discrimination (AUC 0.77; bias-corrected C-index 0.74) and calibration (Hosmer–Lemeshow p = 0.17). Temporal validation confirmed robust performance (AUC 0.77–0.87). In sensitivity analysis, model discrimination remained high for both primary (AUC 0.79) and revisional cases (AUC 0.88). A web-based Shiny risk calculator was developed for bedside use (https://michaeltolson.shinyapps.io/bariatric-delayed-discharge-2023/).

Conclusions

A five-variable nomogram accurately predicts delayed discharge following bariatric surgery and demonstrated strong temporal validation. This tool may aid individualized discharge planning.