Digoxin: beneficial or harmful?

Heart failure (HF) and atrial fibrillation (AF) are highly prevalent, and the combination of AF and HF is steadily increasing.1–6 Digoxin has been used in HF for 200 years. Current European Society of Cardiology (ESC) HF guidelines give digoxin a IIb B recommendation, and American College of Cardiology (ACC)/American Heart Association (AHA) HF guidelines a IIa B recommendation, on top of an angiotensin-converting enzyme inhibitor (ACE-I), angiotensin receptor blocker (ARB) or angiotensin-receptor neprilysin inhibitor (ARNi), a beta-blocker, and a mineralcorticoid receptor antagonist (MRA) for reducing the risk of hospitalization in HF with reduced ejection fraction (HFrEF) and sinus rhythm.5,7 Digoxin is also used for rate control in AF, with a IB recommendation from ESC AF guidelines for control of heart rate in AF patients regardless of left ventricular ejection fraction.4 However, whether digoxin is beneficial or harmful in HF and/or AF is still debated. The main evidence supporting guideline recommendations on digoxin in HF is the significant reduction in the risk of HF hospitalization and of HF hospitalization or HF mortality (but not mortality alone) in digoxin-treated patients enrolled in the DIG trial published >20 years ago.8 However, the characteristics of (i.e. age, gender, co-morbidities, etc.) and treatments (i.e. MRAs, beta-blockers, etc.) used in HF patients today are completely different from those in the 1990s, leading to questions of whether we can consider the findings of the DIG trial still valid and applicable as a basis for recommending digoxin in guidelines.5,7–10 Additionally, currently there is no head-to-head randomized clinical trial of digoxin in AF patients. The concerns regarding digoxin use in HF and AF are not just theoretical, but are supported by several recent studies. In the Val-HeFT trial enrolling patients with HFrEF, digoxin use was associated with a 28% increase in risk of all-cause mortality and a 35% increase in risk of HF hospitalization after adjustments for several potential confounders.11 In a post-hoc analysis of the AFFIRM trial, digoxin treatment was associated with a 41% increased risk of all-cause mortality, with also trends towards increased cardiovascular mortality and arrhythmic death, in patients with HF and AF, and with a 41, 35, and 61% increased risk of all-cause, cardiovascular, and arrhythmic mortality, respectively, in the overall AF population.12 However, in another post-hoc analysis of AFFIRM, no association between mortality and digoxin use was reported in the overall AF cohort and in the subgroup of patients with concomitant HF.13 Statistical issues and in particular different handling of time-dependent use of digoxin could explain the discrepancies among these studies, emphasizing how sensitive observational studies are to variations in methodology.14 Furthermore, in the ROCKET-AF trial, AF patients receiving digoxin reported a 17% increased risk of mortality.15 Finally, a recent meta-analysis of both randomized and observational data on digoxin reported increased mortality risk in patients with HF and/or AF receiving digoxin.16 New insights into this topic come from Erath et al.17 who performed a retrospective analysis of 1028 patients receiving an implantable cardioverter defibrillator (ICD) or a cardiac resynchronization device (CRT-D) between 1996 and 2010 for the primary or secondary prevention of sudden cardiac death. A total of 43% of patients enrolled were receiving digitalis glycosides either for the treatment of HF or for the control of ventricular rate in AF or for both conditions. Patients receiving digitalis glycosides reported a 65% increase in risk of all-cause mortality after adjustments for potential confounders. More patients treated with digitalis glycosides died because of both cardiac arrhythmic and cardiac non-arrhythmic causes compared with those not receiving the drug, with no differences for non-cardiac death. Digitalis-treated patients reported more ICD shocks compared with controls. Notably, the majority of mean plasma digitalis concentrations tended to be in the low range, suggesting that toxicity may not be a major determinant of the increased mortality in patients using digitalis glycosides. However, in a similar setting, in MADIT-CRT, enrolling 1820 patients with HFrEF (EF <30%), New York Heart Association functional class I–II, and prolonged QRS who received an ICD or CRT-D, digoxin was not associated with increased risk of HF/death, HF alone, and death alone, but was associated with an increased risk of high-rate ventricular tachycardia/fibrillation.18 The study of Erath et al.17 contributes to increase further the concerns about digitalis reported to be associated with increased mortality even in the presence of an ICD and in patients with normal blood digitalis levels. However, given the observational nature of this study, the term ‘effect’ in the title may be more appropriately considered as an ‘association’. While the statement that ‘digitalis should be used with great caution’ is certainly appropriate, it should not be taken as convincing evidence that these or any other data have proven harm from digoxin. Several analyses are now published on this topic, often showing contrasting results. Like Erath et al., they are all observational studies or post-hoc analyses of randomized clinical trials, with important methodological limitations and potential biases. In fact, these studies allow identification of an association between a treatment and an outcome, but not the direction of the association, meaning that the increase in mortality in patients receiving digitalis might be due to the higher use of this drug in sicker patients, and statistical adjustments may be not sufficient to rule out the influence of known confounders and, additionally, cannot deal with unknown confounders. Thus, although alarming, the evidence of increased mortality associated with digitalis in observational and post-hoc studies cannot be used to decide whether digitalis may still have a place in HF or AF, but points out the urgency of randomized clinical trials to address this question. L.H.L. is supported by a clinical research grant from the Swedish Research Council. G.S. is supported by grants from the Swedish Research Council and the Swedish Heart and Lung Foundation to L.H.L., and from the Italian Society of Cardiology in collaboration with MSD Italia-Merck Sharp & Dohme Corporation. Conflict of interest: G.S. declares none related to the present work. L.H.L. declares none related to the present work. Unrelated to the present work: speaker’s or consulting honoraria from St Jude, Novartis, Bayer, ViforPharma, and HeartWare, and research grants to the author’s institution from Boston Scientific, Medtronic, and AstraZeneca.