Exploring nightly variability in sleep apnea and changes in severity classification with multi night sleep testing in a population based cohort of young children

Background Obstructive sleep apnea (OSA) is typically evaluated with single-night sleep studies. In adults, night-to-night variability (NtNV) is well documented, but less so in children. The current study evaluated NtNV in apnea events for misclassification in disease severity in young children to investigate if multi-night sleep-recordings may be clinically relevant to improve diagnostic accuracy and affect treatment decisions. Methods Parents/guardians of healthy 4–9-year-old children participating in a study to evaluate prevalence of OSA were asked to record the child's sleep for up to 5-nights. This analysis is from children who completed three consecutive night sleep recordings. Results Average age of participants was 6.1 years ( n = 200, 58.0%-girls/42.0%-boys), prevalence of overweight/obesity was 25% and moderate-severe-OSA (AHI 3% ≥ 5) based on single-night testing protocol was 29.0%. Using the highest AHI 3% value from the three recordings, the first-night correctly identified 57.0% of participants and adding a second night identified 86.0% of participants, with 14% of participants not identified until the third night of sleep testing. Bland-Altman plots (BA plots ) demonstrated non-statistically significant mean-differences and 95% upper and lower-limits (LOA 95% ) close in absolute terms night-one/night-two; −0.024 events/hour-of-sleep LOA 95% [−5.062, 5.013], 8.0% of observations outside the LOA 95% ( p = 0.894) and night-two/night-three; −0.009 events/hour of sleep, LOA 95% [−5.227, 5.209] and 6.0% of observations outside LOA 95% , ( p = 0.9620). The intra-class correlation coefficient (ICC) for AHI 3% for all 3 nights was 0.45 [0.37, 0.54], p < 0.001, for night-one/night-two (0.44; p < 0.001), and night-two/night-three (0.44; p < 0.001). Cohens Kappa (0.304–0.462) and Brennan-Prediger Kappa (0.403–0.490), respectively indicated low-to-moderate interrater agreement, with quadratic-weighted Cohen's Kappa for the four ordinal severity categories of 0.37 and 0.48 for the two night-pairs, demonstrating NtNV, which is consistent with the variability demonstrated by the BA plots . Conclusion Observed NtNV in apnea events affected severity categorization, suggesting that multi-night sleep testing may have clinical value and improve accuracy of sleep apnea evaluation in young children. Clinical trial registration URL: clinicaltrials.gov/study/NCT05479201 , identifier NCT0547920 .