This article is more than 5 years old. Information may no longer be current.
Digoxin in HF: The debate continues
Click Here to Manage Email Alerts
Despite more than 200 years of use, digoxin remains controversial in the management of HF. This past year, multiple observational studies came to divergent conclusions regarding the benefit or harm of digoxin in patients with HF.
A cardiac glycoside, digoxin exerts its inotropic effect by blocking the Na+/K+-adenosine triphosphatase with a resultant increase in intracellular levels of Ca2+. At therapeutic serum concentrations, digoxin increases vagal tone and decreases resting heart rate. Increased parasympathetic tone can also modulate the renin-angiotensin-aldosterone system, providing modest neurohormonal attenuation. Accordingly, its use in HF and atrial fibrillation is firmly grounded in theoretical benefit.
Landmark trial, current use
The landmark DIG trial further established the role of digoxin in the reduction of HF-related hospitalizations and improved symptoms. In this study, no mortality benefit was observed, but digoxin was associated with a 28% lower risk for hospitalization for worsening HF compared with a placebo group. Largely based on these findings, along with several smaller trials, digoxin is recommended in the 2010 Heart Failure Society of America and 2013 American College of Cardiology Foundation/American Heart Association HF guidelines to decrease hospitalizations in patients with HF who are otherwise on optimal medication therapy.
Felix K. Yam
However, not everyone is convinced of the benefits of digoxin. Some have criticized the inability of the DIG trial to conclude any overall mortality benefit. Other issues include post-hoc analyses that demonstrated an association between higher serum digoxin concentrations and increased mortality.
Digoxin use also declined with the emergence of beta-blockers used in combination with ACE inhibitors and angiotensin receptor blockers as key neurohormonal modulators, as well as the addition of mineralocorticoid receptor antagonists such as spironolactone or eplerenone. Despite its diminishing role among contemporary HF therapies, digoxin is still widely used among prescribers who struggle to keep their HF patients out of the hospital. Registry data from the Worcester Heart Failure Study showed that 60% of patients with acute HF with either new-onset or existing AF were treated with digoxin. In a recent trial that evaluated the effectiveness of serelaxin (Novartis) for acute HF, 20% of patients had received digoxin therapy at baseline.
Studies question safety
Two recent studies, however, call to question the safety of digoxin in HF with or without AF. In a post-hoc analysis of data from the landmark AFFIRM trial, Whitbeck and colleagues concluded that digoxin use was associated with a significant increase in all-cause mortality in patients with or without HF. In the subgroup analysis, patients with HF who received digoxin (n=1,076) had a 41% increase in all-cause mortality (estimated HR=1.41; 95% CI 1.09-1.84). Researchers for the study also reported a trend toward increased mortality related to CV and arrhythmic deaths; however, these findings did not reach statistical significance.
Soon after, another post-hoc analysis by Gheorghiade and colleagues, using the same AFFIRM trial data, concluded that digoxin was not associated with increased mortality in patients with AF with or without HF (HR=1.06; 95% CI 0.83-1.37). Although it was not a prespecified endpoint, digoxin use was also not associated with risk for all-cause hospitalization.
Two more post-hoc analyses using the same data resulted in opposite conclusions. These studies raise significant concerns in the interpretation and validity of post-hoc analyses.
Each study used propensity score matching to adjust for differences in baseline characteristics. The two studies differed mainly in how each defined digoxin exposure groups. In the Whitbeck and colleagues study, digoxin exposure was assessed as a time-varying covariate, meaning the same patient can contribute data in either group depending on digoxin exposure at any given time, compared with the Gheorghiade and colleagues study in which digoxin groups were established at study baseline, excluding patients with prior use within 6 months and missing data at the time of randomization. There are merits and disadvantages to both methods, which ultimately led to divergent conclusions. As discussed by Sabina A. Murphy, MPH, who wrote an editorial accompanying both papers in the European Journal of Cardiology, use of a time-dependent covariate can be problematic because of indication bias in which digoxin gets added for worsening HF symptoms and does not contribute directly to mortality. Murphy also mentioned differences between the two analyses in the sample extracted from the AFFIRM trial with potential for selection bias in the Gheorghiade analysis because of exclusions due to missing data at a single time point. There were also differences in the propensity analysis methods.
Debate reignited
Before the dust settled on this controversy, two other observational studies were published that reignited the digoxin debate. First, Freeman and colleagues identified a cohort of 2,891 Kaiser Permanente Northern California patients with newly diagnosed HF between 2006 and 2008. They were stratified based on incident digoxin use and followed for a median of 2.5 years. Incident digoxin use was associated with higher rates of death in the group assigned digoxin compared with the group that did not receive digoxin (14.2 vs. 11.3 per 100 person-years; HR=1.72; 95% CI 1.25-2.36). Digoxin users in this study were less likely to also receive loop diuretics, ACE inhibitors or beta-blockers and more likely to have AF or chronic lung disease. After adjustment for these factors, an association toward increased mortality remained. The researchers concluded that in a “contemporary” cohort of patients with systolic HF, digoxin was associated with higher mortality and there was no difference in risk for HF hospitalization.
This study raises legitimate concerns for digoxin use in HF, considering it was performed in a cohort of patients who represented contemporary practice in contrast to those patients enrolled in the DIG and AFFIRM trials. A limitation to this observational analysis, however, is the different length of follow-up time observed in digoxin users compared with nonusers. In the nonusers, there were 6,548 person-years of follow-up available vs. only 450 person-years of follow-up for digoxin users. Similar to the Whitbeck and colleagues post-hoc analysis of the AFFIRM data, Freeman and colleagues also used a time-varying exposure approach for assigning adverse events to digoxin use. In the Kaiser Permanente study, indication bias also could have occurred with patients switched to digoxin potentially experiencing worsening disease with intermediate events such as new-onset AF. This study controlled for time-varying medication exposure, laboratory test results and comorbidities, but the researchers also recognized that residual confounding cannot be ruled out.
Finally, to confuse matters more, the latest observational study from the Alabama Heart Failure Project indicated that incident digoxin use in a propensity score-matched cohort study of Medicare beneficiaries hospitalized with acute HF was associated with lower 30-day all-cause hospital readmission. This finding was observed only in patients with HF with reduced ejection fraction, not in those with preserved ejection fraction. This difference in all-cause hospital readmission was still present at 1 year. Also at 1 year, the researchers reported a significant reduction in all-cause mortality (HR=0.83; 95% CI 0.71-0.88). Interestingly, there was no difference in HF-specific readmissions among digoxin users vs. nonusers. These data were also obtained from patients admitted from 1998 to 2001, with only about 30% prescribed concomitant beta-blockers and less than 75% prescribed ACE inhibitors or angiotensin receptor blockers, which limits the generalizability of this study.
Future unclear
Although the observational nature of these studies do little to end the digoxin debate, they serve as strong hypothesis-generators. The studies also raise awareness to understanding different statistical approaches used in adjusting for confounding. Confounding by indication, however, remains a legitimate hurdle to our interpretation of the findings. Additionally, favorable digoxin studies in HF appear to be against a background of lower-than-expected beta-blocker and ACE inhibitor or angiotensin receptor blocker use. Unfortunately, little funding is likely to be allocated for a full-fledged prospective, randomized digoxin trial in HF patients with “contemporary” neurohormonal blockade.
Concurrent to the latest evidence and along with declining digoxin use, the price of generic digoxin has increased dramatically. There are now anecdotal reports in the news media of patients stopping digoxin, requesting changes in therapy and insurance companies requiring prior authorization for digoxin approval. The price increase was attributed to declining manufacturers along with numerous other market drivers. Whether the latest price increase will further impact the prescribing of digoxin remains to be seen.
In the meantime, we should consider the totality of evidence and be cautious in our interpretation and generalization of results from these observational analyses. At best, the magnitude of benefit appears to be small relative to ACE inhibitors or beta-blockers and, at worst, the magnitude of harm may be slightly increased in an already sick population. Agree or disagree, digoxin appears to be sticking around for the foreseeable future and the debate will likely continue.
Disclosure: Yam reports no relevant financial disclosures.