Author + information
- Todd F. Dardas, MD, MS and
- Wayne C. Levy, MD∗ ()
- Department of Internal Medicine, Division of Cardiology, University of Washington, Seattle, Washington
- ↵∗Reprint requests and correspondence:
Dr. Wayne C. Levy, University of Washington, Department of Internal Medicine, Division of Cardiology, Box 356422, 1959 NE Pacific Street, Seattle, Washington 98195.
The digitalis glycosides (digoxin) have been ensconced by history, efficacy, therapeutic necessity, and lack of definitive evidence as treatments for both heart failure (HF) and atrial fibrillation (AF). There is a reluctance to abandon digoxin because of its perceived ability to additively improve symptoms and reduce hospitalizations in HF and to control heart rate in AF when other agents have failed or have intolerable effect profiles (1). Simultaneously, there is trepidation toward routine use in HF because of the perception of a narrow therapeutic window and the absence of well-conducted studies demonstrating an additive effect in a background of modern pharmacological and device therapy. The use of digoxin in AF as a single or initial agent is not recommended, but it is sometimes necessary for rate control when other medications have failed or hypotension is encountered. Evidence is emerging that digoxin use in patients with AF (without HF) may be associated with harm (2–5). The call for a randomized controlled trial in these patient populations may be warranted, especially because of the null or negative effects reported from observational studies. As with all observational studies, the true association of a risk factor or treatment with outcomes can only be approximated as bias and confounders are often present. In the case of digoxin, patients with more advanced HF, interval hospitalization, or more recalcitrant AF are more likely to be prescribed digoxin, which is likely to associate digoxin with worse clinical outcomes.
In this issue of the Journal, Allen et al. (6) present an analysis of a large registry of patients with AF, some of whom had HF, and some of whom were treated with digoxin. The investigators took advantage of the wealth of data collected in the registry to measure the association of digoxin with adverse outcomes stratified by HF status. The investigators rigorously evaluated digoxin’s effects by employing state-of-the-art analyses designed to mitigate the confounders and some of the known issues with propensity matching, which is a surrogate for randomization. In the present study, there was no increase in mortality among HF patients treated with digoxin (prevalent use of digoxin adjusted hazard ratio: 1.04; p = NS; incident use of digoxin adjusted hazard ratio: 1.05; p = NS). This result remains true to the mortality endpoint from the DIG (Digitalis Investigation Group) trial (6). Prevalent digoxin use and incident digoxin use were similar and included 1.0 in their respective confidence intervals, which is an indication that patients with worsening HF were not additionally harmed by digoxin itself. The current study differed from the DIG trial and found no association between digoxin treatment and hospitalization rates in patients with HF. The original DIG trial excluded AF at entry, and the current findings suggest that AF nullifies the effect of digoxin in reducing hospitalizations for HF patients. The absence of benefit or harm associated with digoxin in this population of AF patients with HF adds further credence to the call for a better understanding of digoxin’s role in modern HF therapy.
Among patients with AF and without HF, the investigators reported an increase in deaths, cardiovascular hospitalizations, and the combination of death and cardiovascular hospitalization in those patients with incident digoxin use. This effect could result from unmeasured confounders or variables absent from the propensity model, such as incident hospitalizations, which would also correlate with a higher incidence of events regardless of digoxin therapy. However, because of the multiple adjustments and propensity matching, a truly harmful effect of digoxin might be present in this population. Harm from digoxin is consistent with recent reports from the TREAT-AF (Retrospective Evaluation and Assessment of Therapies in Atrial Fibrillation) and ROCKET-AF (Rivaroxaban Once Daily Oral Direct Factor Xa Inhibition Compared with Vitamin K Antagonism for Prevention of Stroke and Embolism Trial in Atrial Fibrillation) studies (2,3). Currently, no study demonstrating a benefit of digoxin in this population has been reported.
Limitations of the present study include the potential for absence of important factors that influence digoxin use from the propensity model, such as incident hospitalizations, imputation of ejection fraction data in 11% of patients, and lack of information regarding serum digoxin levels or dosing. Overall, Allen et al. (6) report perhaps the most rigorously adjusted analysis addressing the effects of digoxin.
Does the investigators’ work and those of others establish or create equipoise for a study of digoxin in HF or AF? For patients with HF with reduced ejection fraction who are in either sinus rhythm or AF, there remains an unmet need for therapies that further ameliorate patient reported symptoms and reduce hospitalizations. Because of the current paper and the work of others, equipoise exists for randomization to either digoxin or placebo in a background of guideline-directed medical therapy. Equipoise is best established for defining digoxin’s role as superior to guideline-directed medical therapy alone for reducing hospitalizations or improving symptoms. A hospitalization endpoint would be acceptable to clinicians and is reliably measurable. An absolute reduction in hospitalization of 5% could be supported as a minimally important treatment difference. Effects on symptoms would be more difficult to quantify because of dropouts, but symptoms can certainly be objectively measured via exercise testing (6-min walk) or subjectively measured via validated HF symptom inventories. Past analyses would also suggest that a novel trial contain explicit dosing parameters (digoxin levels <0.8 ng/ml) and have the power to detect a large difference in response or safety profile between the sexes. The equipoise for mortality is less certain because of the DIG trial results and observational studies that suggest no trend toward improved survival in modern samples, including the current report from the ORBIT-AF (Outcomes Registry for Better Informed Treatment of Atrial Fibrillation) registry. Even the original DIG investigators suggested that digoxin’s effect on mortality is likely “moderate” (7).
The use of digoxin in AF without HF with a reduced ejection fraction or HF symptoms deserves a separate evaluation of equipoise because of the results of the present analysis, other observational studies, and subgroup analyses of large AF trials. The argument for routine use of digoxin as an initial therapy or single agent in AF is weak, although its use among those with hypotension or following failure of other atrioventricular nodal blockers is occasionally necessary. However, designing a trial that requires treatment failure or the constellation of parameters favoring digoxin use (e.g., hypotension, reduced left ventricular function) before randomization would be exceedingly difficult. In addition, there is no alternate treatment to which patients could be randomized. Randomization to placebo or atrioventricular node ablation and a pacemaker at this phase of treatment would seem unethical, and inconsistent with clinical practice and guidelines. Accordingly, the population of AF patients for whom equipoise exists at this time is unclear. If one believes in the harm of digoxin measured in observational trials, proving this effect may seem warranted because of the large population at risk for harm, but the same belief in the trend toward harm, as seen with incident digoxin use in this study, may preclude the ability to randomize patients. In the absence of equipoise or a reasonable trial design, digoxin will most likely continue to fill a niche as a secondary therapy for AF, as currently recommended.
The analysis by Allen et al. (7) supports the case for a novel study to investigate the efficacy of digoxin in HF patients for the purpose of reducing hospitalizations and improving symptoms. There remains a need for digoxin use among selected patients with AF, while the design and equipoise for a clinical trial in AF are unclear. Thus, providers may need to accept 200 years of accumulated experience with digoxin to guide its use as AF therapy.
↵∗ Editorials published in the Journal of the American College of Cardiology reflect the views of the authors and do not necessarily represent the views of JACC or the American College of Cardiology.
Dr. Dardas has received research funding from the ACCF/Daiichi Sankyo Career Development Award. Dr. Levy has served as a consultant for Novartis, GE Healthcare, Biotronik, and Pharmin; and has received research funding from HeartWare, Amgen, Novartis, Resmed, and Medtronic.
- American College of Cardiology Foundation
- Washam J.B.,
- Stevens S.R.,
- Lokhnygina Y.,
- et al.
- Turakhia M.P.,
- Santangeli P.,
- Winkelmayer W.C.,
- et al.
- January C.T.,
- Wann L.S.,
- Alpert J.S.,
- et al.
- Camm A.J.,
- Kirchhof P.,
- Lip G.Y.,
- et al.
- Allen L.A.,
- Fonarow G.C.,
- Simon D.N.,
- et al.