0626 Determining the Effect of PAP on Cardiovascular Risk in the Real World: A Target Trial Emulation of the SAVE Trial Using State-Wide Medicare Claims

Abstract Introduction Positive Airway Pressure (PAP) is the treatment of choice for severe or symptomatic obstructive sleep apnea (OSA). However, randomized controlled trials (RCTs) failed to demonstrate that PAP prevents major adverse cardiovascular events (MACE). Conversely, observational studies showed significant effects. It is unclear whether differences are due to design considerations or patient selection. We proposed a target trial emulation (TTE) of the Sleep Apnea Cardiovascular Endpoints (SAVE) study, the largest RCT assessing the effect of PAP on MACE risk. We hypothesize that a SAVE trial emulation using observational data would result in a non-significant effect. Methods We used the TTE framework to formalize the analysis of state-wide Medicare claims across nine states in the Central U.S. from 2011-2020 towards addressing the question proposed by the SAVE trial. We explicitly described all components of our target trial and operationalized them using observational claims data, making any necessary adjustments. Evidence of PAP initiation was based on claims and MACE was defined as a composite of myocardial infarction, heart failure, stroke, or coronary revascularization. We used prescription time distribution matching to assign index dates and propensity score matching to control for confounding using baseline covariates when assessing causal contrasts. Cox proportional hazards models estimated treatment effect sizes. Results After applying the SAVE eligibility criteria, 168,415 beneficiaries were included (mean [SD] age 75.1 [6.3] years; 40% women), and 22,372 (13.3%) had evidence of initiating PAP. Our estimated intention-to-treat effect (hazard ratio [HR]) of PAP on MACE risk was 1.21 (95%CI: 0.70-2.09), suggesting that in this emulated trial, PAP was not associated with lower MACE risk, consistent with estimates from SAVE (1.10 [0.91-0.32]). Conclusion Emulation of the SAVE trial using Medicare claims data did not show an association between PAP and lower MACE risk. Results suggest that differences between RCTs and observational studies are likely to reflect patient selection (e.g., non-sleepy, secondary prevention, age) rather than biases attributed to observational studies (e.g., healthy adherer bias). This study reinforces the use of TTE to guide the design of robust observational studies assessing the effect of OSA therapies on cardiovascular risk. Support (if any) AHA (20CDA35310360); PCORI (RI-CRN-2020-003-IC); NCATS (UL1TR002366).