Influence of study discontinuation during the run-in period on the estimated efficacy of sacubitril/valsartan in the PARAGON-HF trial.

AIMS The 4822 patients randomized in the PARAGON-HF trial were a subset of 5746 initially eligible patients who entered sequential run-in periods. We identified patient factors associated with study discontinuation during the run-in period and estimated the implications of these discontinuations for the overall study result. METHODS AND RESULTS We utilized multivariable logistic regression models to identify patient factors associated with study discontinuation during the run-in period. The efficacy of sacubitril/valsartan in a broader cohort approximating the full run-in population was estimated by weighting randomized patients according to the inverse probability of run-in completion. A total of 924 (16.1%) subjects failed to complete the run-in period. In multivariable models, non-completion was associated with region other than Central Europe, lower systolic blood pressure, lower serum sodium, lower haemoglobin, lower estimated glomerular filtration rate, higher N-terminal pro-B-type natriuretic peptide, higher New York Heart Association functional class, prior heart failure (HF) hospitalization, and lack of prior use of renin-angiotensin system inhibitors or beta-blocker. In repeat analysis of the effect of randomized treatment in PARAGON-HF giving greater weight to participants resembling those who failed to complete the run-in period, the incidence of HF hospitalizations and cardiovascular death was higher, and sacubitril/valsartan treatment reduced the composite of total HF hospitalizations and cardiovascular death compared with valsartan (rate ratio 0.86; 95% confidence interval 0.74-1.00). CONCLUSION Patients with more advanced HF were at higher risk for non-completion of the run-in period in PARAGON-HF. Re-analysis of study outcomes accounting for the effect of run-in non-completion did not alter the estimated treatment effects of sacubitril/valsartan vs. valsartan.