CPAP therapy was associated with reduced CVD incidence in younger males with BMI ≥30 (logOR -0.37; 95% CI -0.57 to -0.17), but increased risk in older females with prior CVD.
Observational (n=93,868)
Does CPAP therapy reduce cardiovascular disease incidence in specific subgroups of adults with obstructive sleep apnea?
In silico trial emulation suggests that CPAP therapy may provide cardiovascular benefit to younger, obese men with OSA, but potential cardiovascular harm to older, obese patients with prior CVD.
Effect estimate: logOR -0.37 (95% CI -0.57 to -0.17)
p-value: p=<0.05
Abstract Rationale Randomized controlled trials have failed to show an impact of continuous positive airway pressure (CPAP) therapy on cardiovascular disease (CVD) outcomes in obstructive sleep apnea (OSA). A limitation of these trials may have been the lack of targeted enrollment of patients with OSA who are most likely to experience cardiovascular improvement from CPAP. In this analysis, we sought to establish a set of data-driven eligibility criteria that define the OSA population most likely to derive cardiovascular benefit from CPAP therapy. Methods We performed a structured query language (SQL)-based retrieval in our electronic health record (EHR) to identify adults with OSA (ICD10 G47.33) with or without CPAP use. We then tested a simplified set of eligibility criteria among these patients, including age (60 years old, ≥60, either), sex (female, male, either), body mass index (BMI; 30, ≥30, either), and CVD history (positive, negative, either). For each combination, we created a 1:2 propensity-matched cohort of patients on and off CPAP. To evaluate the effect of CPAP on CVD incidence, we performed logistic regression including an interaction term between CPAP and each set of eligibility criteria, weighted by the inverse propensity score. Significance was defined as a false discovery rate (FDR) 0.05, adjusted using the Benjamini-Hochberg procedure across all combinations. Results We identified 93,868 adults with OSA, of which 4,757 had condition concept names indicating CPAP use. Of the 81 possible sets of eligibility criteria, 16 showed statistically significant differences in CPAP effect on CVD incidence between patients meeting criteria versus those that did not (Figure 1). Patients with the largest beneficial effect of CPAP were generally younger (for example, those meeting the criteria age 60 years, male sex, BMI ≥30kg/m2, +/-CVD; logOR -0.37 95%CI -0.57 - -0.17). Conversely, older patients (≥60 years) with an elevated BMI (≥30kg/m2) and prior history of CVD were more likely to experience cardiovascular harm with CPAP use (for example, those meeting the criteria age ≥60 years, female sex, BMI ≥30kg/m2, +CVD; logOR 0.94 95%CI 0.49 - 1.40). Conclusions We identified specific eligibility criteria associated with CVD benefit, as well as harm, from CPAP among patients with OSA. This analysis demonstrates the utility of performing in silico trial emulation within a large health system-based EHR to optimize trial eligibility criteria. Beyond this proof of concept, we are incorporating increasingly complex criteria, and refining artificial intelligence approaches to more accurately capture OSA diagnosis and CPAP usage in EHR. This abstract is funded by: American Academy of Sleep Medicine Foundation
Cohen et al. (Fri,) conducted a observational in Obstructive Sleep Apnea (n=93,868). Continuous positive airway pressure (CPAP) vs. No CPAP use was evaluated on CVD incidence (logOR -0.37, 95% CI -0.57 to -0.17, p=<0.05). CPAP therapy was associated with reduced CVD incidence in younger males with BMI ≥30 (logOR -0.37; 95% CI -0.57 to -0.17), but increased risk in older females with prior CVD.