OBJECTIVES: Obstructive sleep apnea (OSA) is a prevalent sleep disorder that affects approximately 1 billion people worldwide. Positive airway pressure (PAP) therapy is an effective treatment for OSA, yet many women with OSA remain untreated or not adherent to PAP therapy. This analysis used logistic regression modeling to predict 2-year adherence to PAP therapy in a large retrospective sample.

METHODS: De-identified US administrative claims data was linked to diagnostic OSA sleep test data and objective PAP usage data for this analysis. Age and insurer were characterized at the time of the sleep test. Comorbidity status was evaluated in the year prior to the sleep test by assessing ICD-9/10 codes associated with healthcare encounters. A logistic regression model based on predicting non-adherence to PAP therapy over 2 years was developed using baseline covariates.

RESULTS: The model was developed using 5,896 women with OSA. Average age of the cohort was 53.8y and insurance coverage was 81.9% Commercial, 9.3% Medicaid, and 8.8% Medicare Advantage. The average number of comorbidities per person was 3.2, with GERD (29.4%), depression (28.1%), anxiety (26.6%), type 2 diabetes (19.9%), and insomnia (19.5%) as the most prevalent conditions. The model fit (AUC ROC) was satisfactory at 66.3%. Women were more likely to be adherent if they were older (45-54y OR: 0.78, 55-64y OR: 0.59, 65-69y OR: 0.59, ≥70y OR: 0.53), lived in the Midwest region (OR: 0.83), or snored (OR: 0.82). Women were less likely to be adherent if they had Medicaid insurance (OR: 1.79), psychotic disorders (OR: 1.42), insomnia (OR: 1.35), headache (OR: 1.24), or type 2 diabetes (OR: 1.21).

CONCLUSIONS: This analysis from real world data shows the demographic characteristics that are associated with being more or less likely to adhere to PAP therapy. Knowing which factors impact likelihood to adhere may improve patient-centered care for women in sleep medicine.