High burden of moderate-to-severe obstructive sleep apnea: comparative accuracy of screening tools against polysomnography

Abstract

Background and objectives

Timely diagnosis of obstructive sleep apnea (OSA) is essential, but polysomnography (PSG) is resource intensive. Simple screening tools may aid triage, yet their comparative accuracy in high-risk populations remains unclear. This study aimed to assess OSA burden and evaluate the diagnostic performance of three tools (STOP-BANG, Epworth Sleepiness Scale (ESS), and NoSAS) against PSG, along with a Combined Screening Model (CSM) integrating these tools.

Methods

In this cross-sectional study, 140 patients referred for overnight PSG were screened using STOP-BANG, ESS, and NoSAS. Sensitivity, specificity, predictive values, and area under the receiver operating characteristic curve (AUC) were calculated using an apnea–hypopnea index (AHI) ≥5 events/h and oxygen desaturation index (ODI) ≥5 events/h as diagnostic thresholds. A CSM was developed to assess the combined predictive value.

Results

OSA was confirmed in 91% of patients, with 74% having moderate-to-severe disease. NoSAS had the highest AUC (0.70), followed by STOP-BANG (0.67), CSM (0.65), and ESS (0.61), with no significant differences. CSM showed the greatest sensitivity (71%), while NoSAS had the highest specificity (64%).

Conclusion

OSA was highly prevalent and often severe in this referral population. NoSAS performed slightly better individually, and combined models showed potential for supporting early triage. However, PSG remains essential for definitive diagnosis and management.