Inconsistent Outcome Reporting in Heart Failure Randomized Controlled Trials: Inconsistent Outcome Reporting in HF RCTs

Randomized controlled trials (RCTs) may report outcomes different from those prespecified on trial registration websites, protocols, and statistical analysis plans (SAPs). This study sought to investigate prevalence and characteristics of heart failure (HF) RCTs that report outcomes different from those prespecified.MEDLINE via PubMed was searched to include phase II-IV HF RCTs in nine high-impact journals from 2010 to 2020. Outcomes reported in trial publications were compared with pre-specified outcomes in protocols, registration websites, and SAPs. Chi-squared or Fisher exact test were used to analyze correlations between trial characteristics and inconsistencies. Among 216 trials, 32 inconsistencies were observed in 28 trials (13.0%). Among 32 inconsistencies, 2 (6.3%) pertained to omission of prespecified primary outcomes, 4 (12.5%) pertained to omission of prespecified secondary outcomes, 2 (6.3%) pertained to changing prespecified primary outcomes to secondary outcomes, and 2 (6.3%) pertained to changing pre-specified secondary outcomes to primary outcomes. Three inconsistencies (9.4%) pertained to addition of new primary outcomes, 17 (53.1%) pertained to addition of new secondary outcomes, and 2 (6.3%,) pertained to change in timing of assessment of primary outcomes. Majority of the inconsistencies favored statistically significant findings., Seventy-eight (36.1%) were registered retrospectively. Single-center recruitment was associated with outcome inconsistencies (β=-0.14; 95% CI, -0.22 - -0.01,p=0.035).More than one in ten trials reported outcomes inconsistent with those specified in trial registration websites, SAPs, and protocols. An action plan is warranted to minimize selective reporting and improve transparency.