Can symptoms, physical status, and OSA-18 questionnaire results identify children with obstructive sleep apnea (OSA) in a clinical setting?
The clinical prediction model created for this study was found to be useful in identifying pediatric patients at high-risk for OSA among those with sleep disturbances.
Explore this issue:October 2016
Background: Because untreated OSA in children is associated with adverse cardiovascular, neurocognitive, and somatic growth consequences, physicians highly prioritize identifying children with OSA. Previous studies have suggested that integrating symptoms, physical examination data, and questionnaire results might be effective in screening children at risk of OSA, but no clinical point-based prediction model has been developed.
Study design: Single institutional, cross-sectional study of 310 children aged 2 to 18 years with symptoms of OSA.
Setting: National Taiwan University Hospital, New Taipei City, Taiwan.
Synopsis: The univariate logistic analytical results indicated that tonsil hypertrophy, obesity, snoring for more than five nights weekly, snoring for more than three months, waking up at night, breathing pauses, weight gain, and mouth breathing were more prevalent, and the adenoid size and the OSA-18 total score were higher in the OSA group. Authors constructed four models, choosing model 4 and creating a point system for estimating OSA risk. Points where allotted as follows: tonsil size (5 points maximum); adenoid size (5 and 20 points for age > 6 years and < 6 years, respectively); obesity (5 points for age > 6 years); and witnessed breathing pauses (5 points). Performance levels of the full and reduced models were compared according to their disease discriminative (AUC) and reclassification abilities (NRI). The AUCs for the main effect full, main effect reduced, interaction effect full, and final reduced models were 0.833, 0.819, 0.833, and 0.832, respectively. Breathing pauses had the highest diagnostic accuracy in predicting pediatric OSA.