Diagnostic properties modified OSA-18 questionnaire in children with severe obstructive sleep apnea

• Main Authors: Panchanok Kaewkul, Wish Banhiran, Kitirat Ungkanont, Archwin Tanphaichitr, Cheerasook Chongkolwatana, Vannipa Vathanophas
• Other Authors: Faculty of Medicine, Siriraj Hospital, Mahidol University
• Format: Article
• Published: 2019
• Subjects: Medicine
• Online Access: https://repository.li.mahidol.ac.th/handle/123456789/46768
• Institution: Mahidol University

© 2018, Medical Association of Thailand. All rights reserved. Objective: To determine the diagnostic properties of OSA-18 and its modified version for detection of severe obstructive sleep apnea [OSA] in snoring children. Materials and Methods: The present cross-sectional study was conducted in 123 patients (82 boys and 41 girls), aged younger than 12 years, who had snoring problems and performed polysomnography [PSG] at Siriraj Hospital. Those with incomplete questionnaires and inadequate PSG data were excluded. The patients were divided into two groups, non-severe OSA (apnea-hypopnea index [AHI] lower than 10) and severe OSA (AHI of 10 and above). The scores of OSA-18 questionnaires were compared between the two groups. Five most important questions (one from each domain) were selected to develop a modified shorter version of OSA-18. Results: There was no statistically significant difference in total scores of OSA-18 between severe OSA and non-severe OSA groups. The optimal cut-off score (65) was selected from receiver operating characteristic [ROC]. The original OSA-18 had the specificity of 76%, the sensitivity of 40%, positive predictive value [PPV] of 40%, negative predictive value [NPV] of 76%, accuracy of 66%, and area under the curve [AUC] of 0.59. Meanwhile, at the optimal cut-off score (21), the modified OSA-18 had the specificity of 92%, the sensitivity of 34.3%, PPV of 63.2%, NPV of 77.9%, the accuracy of 76%, and AUC of 0.67, which was better than its original. Conclusion: The modified shorter version of OSA-18 questionnaire with the optimal cut-off score of 21 is more useful, based on its high specificity, to enable physicians to quickly identify patients who require urgent treatment. Nevertheless, further studies of this newer version in different populations is recommended.