What’s leak got to do with it? Association of mask leak and positive airway pressure adherence from the homepap study

Positive airway pressure (PAP) therapy is the gold standard treatment for obstructive sleep apnea (OSA), yet adherence remains suboptimal. Mask leak is a common barrier, but current leak metrics don’t distinguish intentional from unintentional leak. We evaluated a novel “Real Leak” measure and compared its relationship to adherence with conventional leak metrics. We conducted a secondary analysis of the HomePAP trial, which randomized adults at high risk for OSA to home sleep apnea testing or in-laboratory polysomnography, followed by PAP initiation. Real Leak was calculated by subtracting mask-specific intentional leak from device-reported Average Leak, essentially representing unintentional leak over 1 month. PAP adherence was defined as ≥ 4 h/night on ≥ 70% of nights at 1 and 3 months. Pearson correlations and multivariable linear regression adjusted for age, sex, BMI, race, education, Epworth Sleepiness Scale, and apnea-hypopnea index. Data were available for 139 and 124 participants at 1 and 3 months, respectively. At 3 months, adherence was 46.2%. Real Leak and Average Leak were highly correlated and inversely associated with adherence (rho[95%CI]: 1-month 0.73 [0.61,0.81], 3-month 0.90 [0.85,0.93], p < 0.001) at both timepoints. Each 1 L/min increase in Real Leak corresponded to a 0.72% decrease in adherence days, compared to 0.55% in Average Leak (p = 0.008/0.018) (7.2% and 5.5% per 10 L/min of change, respectively, p = 0.008/0.018). Race and AHI (p = 0.02/p < 0.001, respectively) were independent adherence predictors. Both Real Leak and Average Leak negatively predicted PAP adherence, with no significant performance difference. Nonetheless, standardizing leak terminology and incorporating intentional leak adjustments may improve clarity and decision-making. Mask leak is a common barrier to Positive airway pressure (PAP) adherence, but existing leak metrics don’t distinguish intentional from unintentional leak, limiting their clinical utility. The American Academy of Sleep Medicine emphasizes minimizing leak during PAP titration but doesn’t provide clear guidance on which leak metric to prioritize. We proposed a novel “Real Leak” measure as a potentially more clinically meaningful indicator of leak-related adherence barriers. In the HomePAP trial, Real Leak and Average Leak were highly correlated and similarly predicted reduced adherence, with no significant performance difference. We propose standardizing leak definitions for clinical practice and research and suggest that Real Leak - essentially unintentional leak over 1 month - is a conceptually clearer and more meaningful metric.