Abstract
While randomized controlled trials (RCTs) guide clinical practice, their completion may also influence real-world treatment patterns. We investigated whether the outcomes of trial emulation differ before and after the publication of the WARCEF (Warfarin versus Aspirin in Reduced Cardiac Ejection Fraction) randomized controlled trial. We emulated the WARCEF trial using EHR data from the Mayo Clinic Platform, comparing Warfarin and Aspirin in patients with heart failure and reduced ejection fraction (HFrEF). Analyses were stratified by the WARCEF completion date (July 2014), using both intention-totreat (ITT) and per-protocol (PP) frameworks. For the ITT analysis, the cohort size before 2014 consisted of 4,579 patients on Aspirin and 93 on Warfarin, while after 2014, the cohort included 13,599 patients on Aspirin and 493 on Warfarin. For the PP analysis, the cohort size before 2014 consisted of 4,314 patients on Aspirin and 86 on Warfarin, while after 2014, the cohort included 3,373 patients on Aspirin and 99 on Warfarin. No significant treatment difference was observed before July 2014, 1.3961 (95% CI: 0.696 - 2.802, p = 0.3477) and 1.1572 (95% CI: 0.559 - 2.397, p = 0.6945) for ITT and PP analytic approaches respectively, consistent with the findings of the WARCEF trial. However, after trial completion, Warfarin was associated with increased risk, 2.039 (95% CI: 1.412 - 2.945, p < 0.001) and 3.940 (95% CI: 1.376 - 11.286, p = 0.0107) for ITT and PP analysis respectively. These changes suggest that the WARCEF results and subsequent guideline updates influenced prescribing patterns and patient selection, thereby altering the observed treatment effects. Trial completion and guideline shifts can materially affect real-world emulation outcomes. Ignoring temporal changes in clinical practice may obscure causal inference and overstate or understate treatment effects in emulation studies. The publication of the WARCEF trial results may have influenced clinical decision-making, leading physicians to prescribe Warfarin or Aspirin differently based on trial findings. This shift in treatment patterns can introduce selection bias in real-world data, as patient characteristics associated with each treatment may no longer be randomly distributed after the trial. For accurate trial emulation, it is essential to ensure that the patient cohort aligns precisely with the trial environment, particularly with respect to the trial's completion date.