High night-to-night variability in childhood obstructive sleep apnea severity

Overnight in-laboratory polysomnography is the gold standard test for childhood obstructive sleep apnea (OSA). Night-to-night variability is a well-documented phenomenon in adult literature but is poorly characterized in children. We evaluated the test-retest reliability of polysomnography in children with moderate-severe OSA.

This is a secondary analysis of two pediatric trials evaluating OSA therapies for children aged 4-18 years old with moderate-severe OSA. The baseline diagnostic polysomnography data were compared to repeat polysomnography data in trial.

Twenty-seven children were included (median age 8 years, 41% females, body mass index z-score 1.4). The median inter-test interval was 48 days (range: 14-78 days). There were 7/27 children (26%) with moderate-severe OSA whose second polysomnography showed mild or no OSA. The median (interquartile range) obstructive apnea-hypopnea index on the first night was 9.4 (7.4-14.0) events/hour and 14.2 (5.4-21.6) events/hour on the second night, with poor correlation (intraclass correlation coefficient; ICC 0.27; 95% CI -0.04, 0.51; p = 0.080). There was poor correlation for the oxygen desaturation index (8.9 versus 10.9 events/hour; ICC 0.10; 95% CI -0.26, 0.53; p = 0.31), and poor correlation for hypoxic burden (31.9 versus 48.7 %min/hour; ICC 0.45; 95% CI 0.10, 0.70; p = 0.0072).

Polysomnograms demonstrate poor test-retest reliability in children with moderate-severe OSA. There should be caution relying solely on a single night polysomnogram in children with moderate-severe OSA.