Predicting Obstructive Sleep Apnea in Patients with Insomnia: A Comparative Study with Four Screening Instruments

Obstructive sleep apnea (OSA) and insomnia are very prevalent disorders, especially in sleep-lab setting, and insomnia may be the presenting complaint of OSA. Here, we aimed to validate No-Apnea as screening tool for OSA in patients with self-reported insomnia complaints and to compare its performance with other models.

This cross-sectional study involved evaluation of No-Apnea as well as STOP-Bang, NoSAS and Epworth Sleepiness Scale (ESS) in subjects with insomnia being evaluated with full in-lab polysomnography. Discrimination was assessed by area under the curve (AUC), while predictive parameters were calculated by contingency tables. OSA severity was classified based on the apnea/hypopnea index: ≥ 5.0/h as any OSA (OSA_≥5), ≥ 15.0/h as moderate/severe OSA (OSA_≥15), and ≥ 30.0/h as severe OSA (OSA_≥30).

Overall, 2591 patients with a clinical diagnosis of insomnia were included. Diagnosis of OSA_≥5, OSA_≥15, and OSA_≥30 was of 76.3%, 53.1%, and 32.6%, respectively. At all levels of OSA severity, No-Apnea had sensitivity ranging from 84.5 to 94.1% and specificity ranging from 58.2 to 35.1%. For screening of OSA_≥5, OSA_≥15, and OSA_≥30, discriminatory ability (AUC) of No-Apnea was: 0.790 [95% confidence interval (CI) 0.770-0.810], 0.758 (95% CI 0.740-0.777), and 0.753 (95% CI 0.734-0.772), respectively. Based on AUCs, No-Apnea, STOP-Bang, and NoSAS performed similar at all levels of OSA severity. The ESS did not present satisfactory discrimination as OSA screening model.

In a large sample of patients with insomnia, No-Apnea, STOP-Bang, and NoSAS, but not ESS, enable satisfactory and similar discrimination at all levels of OSA severity.