Related Experiment Video
Updated: May 31, 2026
07:54
Drug-Induced Sleep Endoscopy (DISE) with Target Controlled Infusion (TCI) and Bispectral Analysis in Obstructive Sleep Apnea
Published on: December 6, 2016
Predicting sleep apnea in bariatric surgery patients
Ronette L Kolotkin1, Michael J LaMonte, James M Walker
1Obesity and Quality of Life Consulting, Durham, North Carolina 27705, USA. rkolotkin@qualityoflifeconsulting.com
Summary
Obstructive sleep apnea (OSA) prediction models for bariatric surgery patients showed lower accuracy than previously reported. An alternative model improved specificity but still fell short of original findings.
Area of Science:
- Bariatric Surgery
- Sleep Medicine
- Medical Diagnostics
Background:
- Obstructive sleep apnea (OSA) is common and serious in obese individuals undergoing bariatric surgery.
- Existing prediction models for OSA in this population have varying accuracy.
- The Dixon model, using common variables, showed high sensitivity and specificity in prior studies.
Purpose of the Study:
- To evaluate the performance of the Dixon OSA prediction model in gastric bypass patients.
- To identify alternative predictors and develop a new model for OSA detection in this cohort.
Main Methods:
- Overnight limited polysomnography was used to measure the apnea-hypopnea index (AHI) in 310 gastric bypass patients.
- The Dixon prediction model was applied to the study sample.
- An alternative prediction model was developed and validated.
Main Results:
- 44.2% of patients had moderate-to-severe OSA (AHI ≥ 15/h).
- The Dixon model's sensitivity (75%) and specificity (57%) were lower than reported.
- An alternate model with 10 predictors improved sensitivity (77%) and specificity (77%) but still underperformed original findings.
Conclusions:
- Neither the Dixon model nor the newly developed model achieved the previously reported sensitivity and specificity for OSA prediction in gastric bypass patients.
- Predictive models for OSA in bariatric surgery patients require further refinement.
- Overlapping predictors exist between models, but performance varies significantly across patient groups.
