567 Validity of Sleep Related Breathing Disorder Scale and Anthropometrics in Obese/Overweight Children with Obstructive Sleep Apnea

Prevalence of OSA, a multifactorial disorder associated with significant morbidity, has increased due to the pediatric obesity epidemic. A key existing gap is lack of a valid OSA screening tool in overweight/obese pediatric patients incorporating anthropometrics. Our aim is to examine the validity of inclusion of anthropometrics to the existing SRBD scale to detect moderate (apnea hypopnea index-AHI≥5) or severe OSA(AHI≥10). Methods Consecutive obese/overweight(n=89) patients(body mass index-BMI percentile>85th for age/gender) in obesity management clinic with SRBD scale, polysomnogram(PSG) and anthropometrics (neck circumference (NC),waist circumferences(WC), height), systolic and diastolic blood pressure(BP) percentiles are included. Receiver operating characteristic(ROC) analysis with AHI as the outcome variable, sensitivity, specificity, positive(PPV), negative predictive values(NPV) for an SRBD cutoff score of 8 and SRBD score found using Youden’s index in ROC and 95% confidence intervals using the exact binomial method are presented. Prediction model, interaction and discrimination (Outcome:AHI; Independent: age, sex, WC, NC, SRBD) were analyzed. Results Study population characteristics: age 12.6±3.4years, 55% female, 62% non-white and AHI=13.0±20.7,AHI>5=65.1% and AHI>10=37.1%. No significant differences were noted in item endorsement or SRBD total score using either AHI≥5 or AHI≥10 (all P>0.10). The area under the ROC curve for SRBD detecting AHI≥5 and 10 was 0.491(95%CI=0.352–0.630) and 0.559(95%CI=0.439–0.679). SRBD≥8 had sensitivity 0.759(0.628–0.861), specificity 0.387(0.218–0.578), PPV 0.698(0.570–0.808), NPV 0.462(0.266–0.666), for AHI≥5 and for AHI≥10, 0.848(0.681–0.949), 0.375(0.249–0.515), 0.444(0.319–0.575), 0.808(0.606–0.934) respectively. The SRBD cutoff score for Youden’s index was 7 for both AHI cutoffs of 5 and 10 and produced similar results to using SRBD cutoff score of 8. The prediction models including age, sex and WC (NC was not significant) had optimism-corrected c-statistics of 0.724 and 0.627 for AHI≥5 and 10, respectively. Adding SRBD total score to the models actually reduced these values to 0.702 and 0.614. Conclusion SRBD alone has fair sensitivity, but poor specificity for significant OSA in overweight/obese. The addition of anthropometrics to SRBD decreased discrimination of OSA in prediction models. Anthropometrics may differ in pre pubertal and post pubertal phenotypes of OSA and may or may not aid in increasing predictability of OSA with SRBD. Support (if any):