Abstract Rationale The apnea-hypopnea index (AHI), the current metric for obstructive sleep apnea (OSA) severity, does not reliably predict clinical improvement or changes in key neurobehavioral outcomes, after treatment. We hypothesized that ventilatory burden (VB), an automated breath-by-breath measure representing the proportion of overnight breaths with 50% normalized amplitude, independent of hypoxemia or arousal, would correlate with worse executive function and attention scores and serve as a stronger predictor of these outcomes than AHI. Methods Data was obtained from the Pediatric Adenotonsillectomy Trial for Snoring (PATS) study via National Sleep Research Resource, a multicenter randomized clinical trial comparing early adenotonsillectomy (Arm 1) versus watchful waiting (Arm 2) in children with mild sleep-disordered breathing (SDB). Primary outcomes included 12-month changes in Behavior Rating Inventory of Executive Function Global Executive Composite (GEC) T score and Go/No-Go (GNG) test d-prime score. Polysomnography data with cannula flow signal quality 95% (316 of 726) was included. A total of 303 studies (Arm 1 = 149, Arm 2 = 154) met inclusion criteria. Mean VB and percent change were calculated for each arm and timepoint. Paired t-tests assessed within-arm changes. Linear regression evaluated associations between VB, AHI, and neurobehavioral measures, with R² values computed for each model. Results There was a significant difference in AHI across the two timepoints for both arms. Arm 1demonstrated a statistically significant decrease in AHI (p = 0.0124) and marginal decrease in VB (p = 0.1698). Arm 2 demonstrated an increase in AHI (p = 0.5953) and increase in VB (p = 0.1297). There was a significant difference in GEC and GNG scores across the two timepoints for both arms. Arm 1 demonstrated a significant difference in GEC scores (p = 0.02393) and GNG scores (p = 0.05714). Arm 2 demonstrated a significant difference in GEC scores (p = 0.1896) but not in GNG scores (p = 0.5065). There was a significant correlation between delta VB and delta GNG for Arm 2 (r=-0.404, p = 0.001). There was no significant correlation between delta VB and delta GEC or between delta AHI and delta GEC or delta GNG scores (Figure 1). Conclusion Among children with mild SDB, VB was a stronger predictor of attention performance (GNG) than AHI in those managed with watchful waiting. The lack of correlation between VB and neurobehavioral outcomes after early adenotonsillectomy aligns with prior findings showing limited neurocognitive benefit of early surgery in mild disease. These results highlight VB as a physiologically relevant, outcome-linked metric that may better capture the neurobehavioral impact of SDB than AHI. This abstract is funded by: None
Kenney et al. (2026) studied this question.