Validation of the modified berlin questionnaire for the diagnosis of obstructive sleep apnea in patients with COVID-19 infection

Obstructive sleep apnea (OSA) is a condition characterized by intermittent cessation of breathing and hypoxia during sleep. Less is known regarding the actual prevalence of OSA in COVID-19 patients, given that objective sleep studies with polysomnography are not accurate in the face of an active contagious respiratory infection. The Berlin Questionnaire (BQ) is a widely used survey to predict OSA, which includes 10 questions in three categories (snoring severity, daytime sleepiness, presence of hypertension and/or obesity). Considering the confounding effect of obesity and hypertension on the clinical course of COVID-19, we have recently developed a modified BQ (mBQ) for addressing the clinical outcomes in patients with high-risk vs low-risk OSA. The primary purpose of the current study was to validate this questionnaire. All cases with a history of COVID-19 infection between March 10 and June 22, 2020, and who completed the mBQ in our first study, were invited to participate. Participants refilled the questionnaires, and an attended polysomnography was conducted. The mBQ was validated based on the polysomnography results. A cut-off value of an apnea-hypopnea-index 15 events/hour was used for the OSA diagnosis. Out of 50, 22 participants (44%) were categorized as having high risk of OSA based on the mBQ. In all, 27 cases (54%) were categorized as positive on snoring intensity and/or snoring frequency, 7 cases (14%) on witnessed apneas, and 28 cases (56%) on the tiredness in the morning and/or during the whole day. According to the PSG results, 19 patients with high risk OSA (86.4%) and 2 patients with low risk (7.1%) were found to have AHI ≥15 events/h. Thus, the mBQ had a sensitivity of 90.5%, a specificity of 89.6%, a positive predictive value of 86.4%, a negative predictive value of 92.9%, and an accuracy of 90%. The area under the curve was 0.89 (95% CI 0.78 – 0.99) confirming a very good performance of the mBQ in screening for OSA for the AHI threshold 15 events/h. The Cronbach's alpha value was 0.70 indicating an acceptable internal reliability whereas the test-retest reliability was low (α = 0.44). The principal component analysis supported the three factor structures. The convergent validity was confirmed by a moderate agreement between the mBQ and the STOP-Bang. A strong discriminant validity was confirmed by the weak correlation between the mBQ and the Narcissistic Personality Inventory scores. A relatively good correlation between the total scores of mBQ and BQ supported the concurrent validity of the mBQ. To conclude, the mBQ is a practical, feasible, and cost-effective method to predict OSA in adults with a history of COVID-19 infection. It requires short administration time, and has a good level of diagnostic sensitivity, specificity and accuracy, suggesting that it can be used as a screening tool during COVID-19 pandemic. Moreover, the mBQ can be used in clinical management of cases with high-risk OSA in hospitals with long waiting lists for PSG, and to eliminate unnecessary PSG investigations for adults with low-risk OSA.