Using the STOPBANG questionnaire and other pre-test probability tools to predict OSA in younger, thinner patients referred to a sleep medicine clinic

Background: The STOPBANG questionnaire is used to predict the presence of obstructive sleep apnea (OSA). We sought to assess the performance of the STOPBANG questionnaire in younger, thinner patients referred to a sleep medicine clinic. Methods: We applied the STOPBANG questionnaire to patients referred for level I polysomnography (PSG) at our sleep center. We calculated likelihood ratios and area under the receiver operator characteristic (AUROC) curve and performed sensitivity analyses. Results: We performed our analysis on 338 patients referred for PSG. Only 17.2% (n = 58) were above age 50 years, and 30.5 and 6.8% had a BMI above 30 and 35 years, respectively. The mean apnea-hypopnea index (AHI) was 12.9 ± 16.4 and 63.9% had an AHI ≥5. The STOPBANG (threshold ≥3) identified 83.1% of patients as high risk for an AHI ≥5, and sensitivity, specificity, positive (PPV), and negative predictive values (NPV) were 83.8, 18.0, 64.4, and 38.0%, respectively. Positive and negative likelihood ratios were poor at 1.02–1.11 and 0.55–0.90, respectively, across AHI thresholds (AHI ≥5, AHI ≥15 and AHI ≥30), and AUROCs were 0.52 (AHI ≥5) and 0.56 (AHI ≥15). Sensitivity analyses adjusting for insomnia, combat deployment, traumatic brain injury, post-traumatic stress disorder, clinically significant OSA (ESS >10 and/or co-morbid disease), and obesity did not significantly alter STOPBANG performance. Conclusions: In a younger, thinner population with predominantly mild-to-moderate OSA, the STOPBANG Score does not accurately predict the presence of obstructive sleep apnea.