Background

Undiagnosed OSA increases the risk of perioperative complications in bariatric patients. We aimed to determine the ideal STOP-Bang score cutoff that maximizes accurate OSA diagnosis in a cost-effective manner and to evaluate alternative screening tools.

Methods

Bariatric surgery patients from January 2018 to September 2023 were identified from the Metabolic and Bariatric Surgery Accreditation and Quality Improvement Program (MBSAQIP) database. Additional variables for patients with a STOP-Bang score of >=4 were collected from the electronic medical record. The Berlin Score was retrospectively calculated.

Results

Overall, 167 (34.5%) of 484 patients who underwent bariatric surgery had a STOP-Bang score >=4. Thirty-five patients (21%) experienced a delay in insurance submission, averaging 41.5 days. The ROC curve for STOP-Bang scores >=4 had an AUC of 78.5% predicting OSA and 83.7% for OSA requiring treatment, compared to Berlin Scores AUC of 80.7% and 88.6%, respectively. A STOP-Bang score of 4 had a sensitivity of 55.6% and specificity of 36.8%, compared to STOP-Bang score 5 with 29.3% and 66.2%, respectively. Nineteen patients (19.4%) with a score of 4 started OSA treatment versus 15 (28.8%) with a score of 5. A Berlin score of 3 had a sensitivity of 47.5% and specificity of 69.1%, with 30 patients (44.1%) starting OSA treatment.

Conclusions

The Berlin questionnaire outperforms STOP-Bang in predicting OSA and OSA requiring treatment. Raising the polysomnogram referral score from STOP-Bang >=4 to >=5 or utilizing a Berlin score of >=3, may alleviate unnecessary resource burden, reduce costs, and expedite insurance authorization for bariatric surgery.