Digitoxin's reported 18% risk reduction (HR 0.82) for mortality or worsening heart failure is statistically fragile, as post hoc analysis reveals the DIGIT-HF trial had only 60% effective power.
Does digitoxin reduce the composite of all-cause mortality or hospitalization for worsening heart-failure in patients with HFrEF?
Although the DIGIT-HF trial reported a statistically significant 18% risk reduction with digitoxin in HFrEF, the trial was severely underpowered, raising concerns about the reliability and generalizability of its findings.
Absolute Event Rate: 0% vs 0%
To Editor, Bavendiek et al. reopened a long-standing debate over the role of cardiac glycosides in heart failure with reduced ejection fraction (HFrEF) in the digitoxin in patients with heart failure (DIGIT-HF) trial.1 The DIGIT-HF trial evaluated whether digitoxin, a cardiac glycoside that is hepatically metabolized and hence more stable and safer in chronic kidney disease, could reduce the composite of all-cause mortality or hospitalization for worsening heart-failure (HF).2 The trial reported a modest but statistically significant risk reduction (hazard ratio HR 0.82; 95% confidence interval CI 0.69–0.98; P = 0.03), reviving interest in digitalis therapy use as part of the guideline-directed medical therapy (GDMT). However, DIGIT-HF was underpowered, raising concerns about the reliability and clinical generalizability of its findings. The trial’s original sample size calculation projected 2190 patients and 734 events to achieve 80% power for detecting the primary composite outcome at 26% with digitoxin versus 31% with the placebo group.3 Due to slow recruitment, the sample size was revised to 1653 patients and 716 events without performing an interim analysis. Yet the final analysis included only 1212 patients and 506 events after waiving the interim analysis, representing barely 70% of the revised sample size. Using the observed event rate and effect size reported in the DIGIT-HF for the primary outcome (39.5% vs. 44.1%; HR 0.82; 95% CI 0.69–0.98; P = 0.03), our post hoc power analysis using STATA 18 estimated an effective power of 60%, which is below the conventional threshold for adequately powered trials.4 Consequently, the robustness of the findings is debatable, which is reflected in the cumulative incidence curves for the primary endpoint Figure 1a in DIGIT-HF.1 The curves started to overlap after approximately 60 months, suggesting the unsustainability of treatment effect. Indeed, to maintain 80% power at the observed hazard ratio (0.82), DIGIT-HF would have required a total enrollment of 1746 participants. DIGIT-HF represents the growing challenge of conducting long-term HF trials during the rapidly evolving GDMT. When DIGIT-HF was conceived, sacubitril/valsartan and SGLT2 inhibitors were not yet established as standard therapy in HFrEF. Their subsequent adoption in 2016 and 2019, respectively, during recruitment certainly reduced event rates, which might have diminished the statistical efficiency of the trial. The DIGIT-HF investigators transparently reported their recruitment and amendment challenges, which is commendable. Yet DIGIT-HF underscores a pressing need for adaptive trial frameworks, conditional power monitoring, and sample size re-estimation methods to allow recalibration of statistical power midtrial without compromising blinding or integrity.4 Adequate power is an ethical obligation. Underpowered studies risk exposing patients to interventions without sufficient potential for conclusive insight. When sample size amendments or early closures are unavoidable, transparent post hoc power reporting should become standard to ensure proper interpretation of study findings. Although encouraging, the apparent 18% risk reduction in DIGIT-HF may not survive replication under adequately powered conditions. CONCLUSION Digitoxin’s long half-life, hepatic metabolism, and reduced renal clearance were hypothesized to offer a safer profile in modern HFrEF, yet DIGIT-HF could not conclusively validate that premise. The underpowered primary outcome of all-cause mortality or hospitalization for worsening HF, along with the observed attenuation of benefit over time, reflects statistical fragility. DIGIT-HF reopens a vital discussion about the intersection of trial methodology, evolving standards of care, and enduring pharmacologic principles. While the study was commendably executed, its underpowered nature tempers the confidence of translating its findings into practice. Financial support and sponsorship Nil. Conflicts of interest There are no conflicts of interest.
Rahhal et al. (Wed,) reported a other. Digitoxin's reported 18% risk reduction (HR 0.82) for mortality or worsening heart failure is statistically fragile, as post hoc analysis reveals the DIGIT-HF trial had only 60% effective power.