Pharmacotherapy in individuals post-aortic valve replacement (TAVI or SAVR) for severe aortic stenosis: a national target trial emulation study of 57,676 individuals
L Pasea, N Aziminia, D E Newby, M R Dweck, G McCann, A Banerjee, A Singh, T A TreibelAbstract
Background
Aortic stenosis (AS) is a common valve disease, with surgical or transcatheter aortic valve replacement (AVR) being the only treatment of proven benefit. Randomised controlled trials (RCT) have shown that sodium–glucose co-transporter 2 inhibitors (SGLT2i) and mineralocorticoid receptor antagonists (MRA) reduce the risk of heart failure (HF) and cardiovascular (CV) death in HF by ~20%. However, these trials excluded individuals with AS undergoing AVR.
Purpose
Using routinely collected linked electronic health records and the target trial emulation (TTE) framework, we aimed to estimate the treatment effects of four HF therapy groups (Renin-Angiotensin-System inhibitors (RAASi); Beta-Blockers (BB); MRA; SGLT2i) in individuals with AS following surgical or transcatheter AVR.
Methods
We used linked primary care, hospital admissions, national cardiac disease and death registry data in England between April 2018 and March 2025 through the ‘Data Science Centre’s Cardiovascular Disease COVID-UK/COVID-IMPACT Consortium’. TTE is an approach to mimic hypothetical RCT using observational data, aligning eligibility, treatment assignment, follow-up and analysis to reduce bias and improve causal inference. We defined exposure as at least one prescription of a drug recorded in primary care data within 90 days post-AVR. The primary outcome was HF hospitalisation or CV mortality and secondary outcomes were all-cause mortality, CV mortality and hospitalisation. Inverse probability of treatment weighting was used to adjust for baseline confounders (demographics, medical history, AVR intervention, clinical biomarkers and prior prescribed medication), to balance the characteristics of those who were and were not prescribed medications. Weighted Cox proportional hazards regression models were used to estimate risk.
Results
We identified 57,676 AS individuals who underwent surgical (26,846, 46.5%) or transcatheter (30,830, 53.5%) AVR followed up for a median of 3 years (IQR: 1.6, 4.9). At 90 days post-AVR, 3,863 (6.7%) individuals had been prescribed an SGLT2i, 6,268 (10.9%) an MRA, 29,159 (50.6%) an RAASi and 37,495 (65.0%) a BB. We estimated lower risk of HF hospitalisation or CV mortality events in individuals prescribed RAASi (HR: 0.93 95% CI:0.89-0.98), BB (HR: 0.86 95% CI: 0.79-0.92) and SGLT2i (HR: 0.78 95% CI: 0.68-0.90), whilst MRA were associated with increased risk (Figure 1). These findings were largely consistent across endpoints and prespecified subgroup analyses (Figure 2).
Conclusion
In this national TTE study of 57,676 individuals with AS who underwent surgical or transcatheter AVR, pharmacotherapy with SGLT2i, RAASi and/or BB post-AVR significantly reduced the risk of CV mortality or HF hospitalisation. Although we used methods to reduce confounding inherent in observational data, residual confounding from unmeasured factors may remain. Efficacy of these drugs post-AVR needs to be confirmed in prospective RCT.