Detection of pediatric obstructive sleep apnea syndrome: history or anatomical findings?

OBJECTIVE: To assess how history and/or anatomical findings differ in diagnosing pediatric obstructive sleep apnea (OSA).

METHODS: Children aged 2-18 years were recruited and assessed for anatomical (ie, tonsil size, adenoid size, and obesity) and historical findings (ie, symptoms) using a standard sheet. History and anatomical findings, as well as those measures significantly correlated with OSA, were identified to establish the historical, anatomical, and the combined model. OSA was diagnosed by polysomnography. The effectiveness of those models in detecting OSA was analyzed by model fit, discrimination (C-index), calibration (Hosmer-Lemeshow test), and reclassification properties.

RESULTS: A total of 222 children were enrolled. The anatomical model included tonsil hypertrophy, adenoid hypertrophy, and obesity, whereas the historical model included snoring frequency, snoring duration, awakening, and breathing pause. The C-index was 0.84 for the combined model, which significantly differed from that in the anatomical (0.78, p = 0.003) and historical models (0.72, p < 0.001). The Hosmer-Lemeshow test revealed an adequate fit for all of the models. Additionally, the combined model more accurately reclassified 10.3% (p = 0.044) and 21.9% (p = 0.003) of all of the subjects than either the anatomical or historical model. Internal validation of the combined model by the bootstrapping method showed a fair model performance.

CONCLUSION: Overall performance of combined anatomical and historical findings offers incremental utility in detecting OSA. Results of this study suggest integrating both history and anatomical findings for a screening scheme of pediatric OSA.