Validation of a health administrative definition of obstructive sleep apnea in children in Ontario, Canada

Obstructive sleep apnea (OSA) affects approximately 5% of children and requires polysomnography (PSG) for diagnosis. As such, population-based longitudinal studies of OSA in children are scarce. We aimed to validate pediatric case definitions of OSA using provincial health administrative datasets and establish a methodology for future longitudinal studies in this population. We performed a multicenter validation study, linking Ontario health administrative data with clinical data for children aged 0-18 years who underwent PSG between 2009-2016 at two tertiary care children's hospitals. We used various administrative case definitions to identify those with the highest sensitivity, specificity, positive and negative likelihood ratios for capturing children with moderate-severe OSA (apnea-hypopnea index ≥5), stratified by age groups (<10 and age ≥ 10 years). The reference cohort included 1,254 children who underwent PSG, with 317 moderate-severe OSA. The overall mean age was 7.7 ± 4.8 years, with 64.8% < 10 years of age and 44% female. The best-performing case definitions included combinations of PSG and adenotonsillectomy or initiation of positive airway pressure therapy within 0-18 months post PSG, or PSG and a diagnostic code for OSA within 6 months pre and 0-18 months post PSG. In children <10 years, these definitions exhibited high specificity (0.80-0.88), with moderate sensitivity (0.62-0.79), positive (3.82-5.25) and negative likelihood ratios (0.27-0.43). However, the performance in children ≥10 was less robust with high specificity (0.85-0.92), moderate positive likelihood ratios (3.48-4.19) and low sensitivity (0.28-0.59), and negative likelihood ratios (0.48-0.78). We have identified case definitions within Ontario health administrative data that have good specificity, and moderate sensitivity, positive and negative likelihood ratios for identifying moderate-severe OSA in children <10 years of age. These can be used in future studies to understand the natural history, predictors and outcomes of young children with OSA, while refined case definitions may still be needed for children ≥10.