0512 Predictors of Obstructive Sleep Apnea Severity Misclassification due to Changing Hypopnea Scoring Criteria: A Clinical Cohort Study

Abstract Introduction The apnea-hypopnea index (AHI) continues to be the key metric for assessing obstructive sleep apnea (OSA) diagnosis and severity. However, variability in hypopnea scoring criteria can lead to substantial differences in AHI calculation and OSA severity classification, potentially impacting clinical decision-making, treatment eligibility, and health outcomes. Despite its clinical relevance, predictors of misclassification and the magnitude of AHI differences between these scoring approaches remain poorly characterized. Here, we characterize and identify clinical and demographic predictors of OSA severity misclassification and differences in AHI when applying the two different hypopnea scoring criteria. Methods We conducted a retrospective cohort study of adult patients who underwent diagnostic sleep testing at the University of Kansas Sleep Disorders Center between 2023-2025. Included modalities were in-lab polysomnography (PSG), home sleep tests (HST), and peripheral arterial tonometry (PAT) studies. Each study was scored using the 3% desaturation or arousal (3%a) criterion and the 4% desaturation criterion. Misclassification rates across OSA severity categories (none, mild, moderate, severe) were calculated. Associations between misclassification and demographic (age, sex, race, ethnicity) and clinical factors, comorbidities and study type were assessed using chi-squared tests. Differences in AHI between scoring criteria were analyzed using multivariate linear models. Results We included 8,001 individuals (mean [SD] age 54.8 [16.5] years; 50.1% women; 77.7% White; 90.6% non-Hispanic). Overall, 41.8% of patients were classified into a more severe OSA category using the 3% rule compared to the 4% rule. Misclassification was most frequent among PAT studies. Stratified analyses by study type showed that older age (PSG and PAT), male sex (PSG), higher BMI (PSG), lower ESS (PAT), cardiovascular/cerebrovascular disease (PAT), and current smoking (PAT) were significantly associated with misclassification. Multivariate models indicated greater AHI differences for PSG (β=2.7 events/h) and PAT (β=7.7 events/h) versus HST (p< 0.01). Older age (β=0.08 events/h), male sex (β=0.45 events/h), and Black race (β=0.93 events/h) were also associated with larger AHI differences. Conclusion OSA severity change in classification is common when applying the 3%a rule versus the 4% rule, particularly among PAT studies, older individuals, and men. These findings underscore the need for harmonized scoring standards to reduce diagnostic variability. Support (if any) NCATS (UL1TR002366).