Background Our study aimed to develop a predictive model for the risk of obstructive sleep apnea (OSA) in bariatric surgery candidates for utilization during the preoperative evaluation. Methods Relevant clinical data were retrospectively collected for 453 patients who met the inclusion criteria and did not meet the exclusion criteria; the patients were randomized into training and test cohorts. Univariate analysis was performed on the training set. Multiple risk factors associated with OSA were identified using multivariate analysis. These factors were incorporated into a regression model and used to construct a nomogram to predict the risk of OSA. The model was validated with a calibration curve and an operating characteristic curve. The models were verified for discrimination, consistency, and accuracy by calibration and subject operating characteristic curves. Finally, decision curve analysis was used to determine the model’s utility. Results In this study, non-alcoholic fatty liver disease (NAFLD), age, chest circumference (CC), and average SpO2 were found to be independent risk factors for developing OSA in bariatric surgery candidates. The AUC for the training cohort was 0.88 with a sensitivity and specificity of 0.93 (95% CI: 0.84–1.00) and 0.70 (95% CI: 0.64–0.75). The Hosmer–Lemeshow test of the calibration curves for the training and validation sets revealed a P > 0.05 (training cohort: P = 0.955; test cohort: P = 0.440). Conclusions We constructed a prediction model that included NAFLD, age, CC, and mean SpO2, which showed superior predictive performance compared to existing models. This model offers a convenient, cost-effective alternative to PSG, particularly useful in preoperative screening of bariatric surgery patients. In the future, the relationship between NAFLD and OSA needs to be further explored, and the prediction model needs to be externally validated.<Table Float="No" ID="Taba"> <tgroup cols="2"> <colspec align="left" colname="c1" colnum="1" /> <colspec align="left" colname="c2" colnum="2" /> <tbody> <row> <entry nameend="c2" namest="c1"> Key Points 1. OSA increases the risk of postoperative complications in bariatric surgery patients, but there is a lack of tools to effectively predict the risk of OSA in bariatric surgery. 2. NAFLD, age, CC, and average SpO2 were found to be independent risk factors for developing OSA in bariatric surgery candidates. 3. A nomogram was constructed to predict the risk of incidence of OSA in patients undergoing bariatric surgery, offering a practical alternative to PSG. </entry> </row> </tbody> </tgroup> </Table>

A Prediction Model for Obstructive Sleep Apnea in Bariatric Surgery Candidates with Obesity

期刊论文
发表日期：

2025年9月8日

查看全文

Xiaoxiao Zhu,
Bei Xu,
Chengcan Yang,
Shuai Ma,
Guihua Hao,
Fen Gu,
Bing Wang,
Ying Yao

摘要

Background

Our study aimed to develop a predictive model for the risk of obstructive sleep apnea (OSA) in bariatric surgery candidates for utilization during the preoperative evaluation.

Methods

Relevant clinical data were retrospectively collected for 453 patients who met the inclusion criteria and did not meet the exclusion criteria; the patients were randomized into training and test cohorts. Univariate analysis was performed on the training set. Multiple risk factors associated with OSA were identified using multivariate analysis. These factors were incorporated into a regression model and used to construct a nomogram to predict the risk of OSA. The model was validated with a calibration curve and an operating characteristic curve. The models were verified for discrimination, consistency, and accuracy by calibration and subject operating characteristic curves. Finally, decision curve analysis was used to determine the model’s utility.

Results

In this study, non-alcoholic fatty liver disease (NAFLD), age, chest circumference (CC), and average SpO₂ were found to be independent risk factors for developing OSA in bariatric surgery candidates. The AUC for the training cohort was 0.88 with a sensitivity and specificity of 0.93 (95% CI: 0.84–1.00) and 0.70 (95% CI: 0.64–0.75). The Hosmer–Lemeshow test of the calibration curves for the training and validation sets revealed a P > 0.05 (training cohort: P = 0.955; test cohort: P = 0.440).

Conclusions

We constructed a prediction model that included NAFLD, age, CC, and mean SpO₂, which showed superior predictive performance compared to existing models. This model offers a convenient, cost-effective alternative to PSG, particularly useful in preoperative screening of bariatric surgery patients. In the future, the relationship between NAFLD and OSA needs to be further explored, and the prediction model needs to be externally validated.

Key Points

1. OSA increases the risk of postoperative complications in bariatric surgery patients, but there is a lack of tools to effectively predict the risk of OSA in bariatric surgery.

2. NAFLD, age, CC, and average SpO2 were found to be independent risk factors for developing OSA in bariatric surgery candidates.

3. A nomogram was constructed to predict the risk of incidence of OSA in patients undergoing bariatric surgery, offering a practical alternative to PSG.