0876 AFTER THE REFERRAL: POLYSOMNOGRAPHY COMPLETION RATES IN CHILDREN REFERRED FOR OBSTRUCTIVE SLEEP APNEA

Abstract

Introduction: Despite guidelines for evidence-based diagnosis of pediatric obstructive sleep apnea (OSA), studies in primary care settings have found low rates of polysomnography (PSG) and other referrals for children at-risk for OSA. For children who are referred, little is known about the rate of PSG completion. The objective of the current study is to identify the rate of completed PSG amongst children referred for suspected OSA. Methods: We employed a computer decision support system (Child Health Improvement through Computer Automation; CHICA) to prompt primary care providers (PCPs) for all patients ages 1-11 with a positive screen for snoring. We examined the electronic health records for all children referred for PSG through this process in five primary care clinics between August 2015 and July 2016. Clinics are located in Indianapolis, Indiana, and serve ethnically diverse children from families with a lower income (83.4% with Medicaid insurance). Results: PCPs referred 124 snoring children for PSG. Referred children represented 5.9% of snoring children for whom the PCPs received an electronic medical record prompt to consider OSA. Of those referred, only 54.8% completed the PSG (despite scheduling availability), 25.0% were scheduled but cancelled/missed the appointment, 16.9% were contacted but never scheduled, 2.4% initiated but did not complete the PSG, and one patient (0.9%) was currently scheduled at the time of analysis. OSA was identified in 64.7% of children completing the PSG, with a range of severity: mild (2.0 4.99) = 50%; moderate (5 9.99) = 15.9%; and severe (AHI>10) = 34.1%. Conclusion: Only a small proportion of snoring children were referred for PSG. Amongst those referred, the PSG completion rate was moderate. Many children who ultimately did complete the PSG were found to have OSA. Taken together, these findings suggest that many children with possible OSA remain unidentified, even in a system with automated screening and computer decision support.