Lost to Follow‐Up in Pragmatic Trials With Remote Follow‐Up: Insights From CONNECT‐HF and TRANSFORM‐HF

BACKGROUND: Pragmatic trials commonly use remote follow-up to reduce participant burden and costs. However, these approaches may increase the risk of lost to follow-up (LTFU). We identified patient- and site-level factors associated with LTFU in 2 large, pragmatic heart failure (HF) trials. METHODS: We analyzed data from 2 pragmatic US trials (CONNECT-HF [Care Optimization Through Patient and Hospital Engagement Clinical Trial for Heart Failure], TRANSFORM-HF [Torsemide Comparison With Furosemide for Management of Heart Failure]) following HF hospitalization. Both trials employed centralized call center-based follow-up with supplemental data sources. For this analysis, LTFU was defined as the absence of both a completed 6-month assessment and any subsequent clinical follow-up throughout the study (confirmation of vital status alone was insufficient). Multivariable logistic regression models were used to identify factors associated with LTFU in each trial. RESULTS: The LTFU rates were 7.6% (362/4771) in CONNECT-HF and 10.5% (245/2334) in TRANSFORM-HF. In multivariable analyses of both trials, younger age was consistently associated with increased LTFU. Conversely, implantable cardioverter-defibrillator/cardiac resynchronization therapy was consistently associated with lower LTFU (CONNECT-HF: adjusted odds ratio [aOR], 0.56 [95% CI, 0.43-0.75]; TRANSFORM-HF: aOR, 0.59 [95% CI, 0.39-0.89]). In CONNECT-HF, smaller hospital size (aOR per 100 beds, 0.90 [95% CI, 0.83-0.97]) was also associated with higher LTFU. Socioeconomic factors were not significantly associated with LTFU. CONCLUSIONS: In analyses of 2 pragmatic trials of patients hospitalized for HF in the United States, younger patients, those without implantable cardioverter-defibrillator/cardiac resynchronization therapy, and smaller hospital size were associated with higher LTFU. Tailoring retention strategies to both patient and site characteristics should be explored in future work to help mitigate LTFU in pragmatic trials.