Evaluating the causal effect of medication adherence on adverse clinical outcomes in the target trial framework

Abstract Objectives We sought to evaluate the causal effect of adherence to Sacubitril/Valsartan (SAC/VAL) on adverse clinical outcomes in patients with heart failure (HF) with reduced ejection fraction, employing methods designed to minimize biases observed in earlier studies. Methods Medicare Fee-For-Service claims data was utilized to emulate a target trial comparing two strategies of adherence to SAC/VAL: (a) continued adherence, (b) continued non-adherence. Patients with proportion of days covered ≥80% during an adherence assessment period of 60 days post-initiation were categorized as the continued adherence group, while those with a proportion of days covered <80% were categorized as the continued non-adherence group. Adherence was measured monthly, and patients were censored upon deviating from their strategy. HF hospitalization or all-cause mortality over 1 year was assessed. Marginal structural models with inverse probability treatment weighting to estimate hazard ratios (HRs), while accounting for both baseline and time-varying confounders were employed. A bias calibration analysis using osteoporotic fractures as a negative control outcome accounted for residual bias. Key findings A total of 42 915 individuals were identified in the continued adherence group and 19 782 in the continued non-adherence group. The unadjusted HR for continued adherence vs. continued non-adherence for the composite outcome was 0.52(95% CI, 0.50–0.55). After adjusting for baseline and time-varying covariates, the HR was 0.59(95% CI, 0.56–0.62). The bias-calibrated HR was 0.77(95% CI, 0.71–0.84), indicating the effect is not fully explained by residual confounding. Conclusions Among Medicare beneficiaries with HF newly initiated on SAC/VAL, optimal medication adherence was associated with a substantially lower risk of HF hospitalization or all-cause death. For assessing the causal relationship between medication adherence and clinical outcomes, the target trial framework provides a principled approach to avoid common biases.