| Summary: | 碩士 === 國立臺灣大學 === 流行病學與預防醫學研究所 === 102 === Background: Obstructive sleep apnea syndrome (OSAS) is an upper airway disorder. Over-night polysomnography is the “gold standard” for the diagnosis of pediatric OSAS. Information from objective and subjective measures for children with OSAS helps clinicians in decision making.
Purpose: To assess diagnostic abilities of objective measures, subjective measures, and combined objective and subjective measures in detecting pediatric obstructive sleep apnea syndrome, and to compare performance difference and clinical utilities between objective measures, subjective measures, and combined objective and subjective measures for detection of pediatric OSAS.
Study Design: Cross-sectional study.
Methods: Children aged 2-18 years were recruited. Children were assessed objectively for tonsil size, adenoid size, and obesity; tonsils were graded by otolaryngologist using the scheme by Brodsky et al.; adenoid size was measured based on a lateral cephalometric radiographs (Fujioka method); obesity was determined by a measure of body mass index percentile of each child. Subjective measures for symptoms were recorded using a standard sheet. Objective measures significantly correlated with OSAS were put into the objective model, whereas subjective measures into the subjective model. Accordingly, objective and subjective measures significantly correlated with OSAS were served as the combined model. Diagnosis of OSAS was made by polysomnography. Diagnostic performances of models in detecting OSAS were analyzed by model fit, discrimination (C-index), calibration (Hosmer-Lemeshow test), and reclassification. The model was internal validated using the leave-one-out cross-validation, bootstrapping method, and k-fold cross-validation.
Results: In total, 222 children were enrolled. Objective model included tonsil hypertrophy, adenoid hypertrophy, and obesity, whereas subjective model included snoring frequency, snoring duration, awaken, and breathing pause. The chi-square test was significant in the objective model, subjective model, and the combined model (P < 0.001). The C-index was 0.84 for the combined model, which was significantly differed from that in the objective model (0.78, P = 0.0032) and the subjective model (0.72, P = 0.0001). The Hosmer-Lemeshow test showed adequate fit (P > 0.05) for all models. Compared to objective model or subjective model, the combined model correctly reclassified 10.3% (P = 0.044) and 21.9% (P = 0.003) of all subjects. Internal validation of the combined model showed fair model performance and no obvious over-fitting.
Conclusions: Overall performance of combined objective and subjective measures, as compared with objective measures or subjective measures alone, offer incremental utility in detecting OSAS. This finding provides the rationale to combine both objective and subjective measures in developing a screen tool for pediatric OSAS.
|
|---|
