Review of the DIG trial

The Effect of Digoxin on Mortality and Morbidity in Patients with Heart Failure

N Engl J Med 1997;336:525-33

Background: Digoxin is a natural drug that comes from the Foxglove plant (Digitalis purpurea). It has been used for the treatment of congestive heart failure for over 200 years. At the time this trial was undertaken, it was given to the overwhelming majority of patients with severe congestive heart failure. The percentage of patients on digoxin was over 90% in all of the seminal heart failure trials we have reviewed thus far. However, despite its frequent use, it was unknown whether the drug improved major morbidity or mortality, and new drug classes were emerging with positive effects on hard outcomes. The primary aim of the Digitalis Investigation Group was to design a large, pragmatic trial to test the long-term effect of digoxin versus placebo on all-cause mortality as well as hospitalization for heart failure.

Patients: The study enrolled 6,800 patients with congestive heart failure and an EF of </=45% in the main trial and enrolled another 988 patients with heart failure and an EF >45% in an ancillary trial. The patient population in the trial was intended to be diverse with no upper age limit. If patients were already on an ACE inhibitor, and had been stable for at least 2 weeks, they could be randomized immediately. For patients not on an ACE inhibitor, they were prescribed the drug and reevaluated in 2 weeks, at which time they could be randomized, if stable. Patients could be randomized into the trial whether they were taking digoxin or not and there was no washout phase.

Important exclusion criteria included an MI, cardiac surgery or percutaneous coronary intervention within 4 weeks; unstable or refractory angina for less than 1 month; 2nd or 3rd degree AV block or sick sinus syndrome without a pacemaker; atrial fibrillation or atrial flutter; cor pulmonale; acute myocarditis; amyloid cardiomyopathy; hypertrophic cardiomyopathy; complex congenital heart disease; current treatment with IV inotropes; potassium <3.2 or >5.5 mmol/L; need for cardiac surgery or percutaneous coronary intervention in near future; patients on heart transplant list; severe kidney (Cr >3.0 mg/dL) or liver disease; any non-cardiac disease that shortens life-expectancy to <3 years (i.e., most cancers).

Baseline Characteristics: The mean age of patients was 63.5 years and approximately 78% were male. Nearly 30% of patients were above the age of 70. The average EF was 29% and the median duration of CHF was 17 months. Most patients had mild to moderate heart failure, with 13% in NYHA Class I, 53% in Class II, 31% in Class III, and only 2% in Class IV. Ischemic cardiomyopathy was the main cause of heart failure (71%); idiopathic causes and hypertensive heart disease accounted for 15% and 9%, respectively. Most patients were on an ACE inhibitor (95%) and diuretic (82%). Nitrates or other vasodilators were used in 44%. Prior to enrollment, 44% of patients used digoxin.

Trial Procedures: This was designed to be a pragmatic trial. Entry conditions required only a 2 week period of clinical stability, which was not rigorously defined. A recent hospitalization for heart failure was not an exclusion; however, patients could not be on IV inotropic agents. Patients could be on digoxin at the time of screening and if so, were randomized without a washout period. For patients not already on an ACE inhibitor, an ACE inhibitor was to be prescribed with reassessment in 2 weeks and randomization if stable.

Randomization was stratified by study center and EF (</=45% or >45%). The recommended initial dose of the study drug (digoxin or placebo) was determined with an algorithm based on the patient’s age, sex, weight and renal function. Local investigators were permitted to modify the dose on the basis of other factors. All patients returned for follow-up visits 4 weeks and 16 weeks after randomization and every 4 months thereafter.

Monitoring of serum digoxin levels for dose titration and toxicity via use of local laboratories was not encouraged in the trial. High serum digoxin levels without clinical signs of toxicity do not prove toxicity and conversely, a portion of patients with true toxicity have serum digoxin levels in the normal range. If a serum digoxin level was felt to be essential for clinical management, local investigators sent blood samples to a central laboratory and the results were reported as either: subtherapeutic, therapeutic or possibly toxic, or probably toxic within 48 hours. A subset of patients in the trial had serum digoxin levels drawn at 4 weeks and 12 months. This data was not used for the individual management of patients in the trial but rather to assess the safety and efficacy of the dosing algorithm used.

If during the trial, patients had worsening symptoms of heart failure, it was recommended that other therapy for heart failure be used in an optimal fashion. If patients remained symptomatic despite efforts to optimize other forms of treatment, open-label treatment with digoxin was allowed and the study drug was discontinued.

Endpoints: The primary outcome was all-cause mortality. The secondary outcomes included mortality from cardiovascular causes, death from worsening heart failure, hospitalization for worsening heart failure, and hospitalization for other causes, including suspected digoxin toxicity. In the ancillary trial, the primary endpoint was a composite of death or hospitalization due to worsening heart failure.

It was estimated that 6,700 patients would be needed to detect a 15% relative reduction in mortality based on an estimated 3-year mortality rate of 27% in the placebo group. It was anticipated that 3 years would be required to recruit the anticipated number of participants and follow-up would be for a minimum of 2 years and maximum of 5 years. In a methods paper, which proceeded publication of the main results, a range of sample size estimates are provided.

Results: The mean duration of follow-up was 37 months or 3.1 years. At randomization, the median daily dose of the assigned study drug was 0.250 mg in both groups. Among the 1485 patients in the digoxin group for whom blood samples were obtained, the mean steady-state serum digoxin level was 0.86 ng/ml at the 1-month visit and 0.80 ng/ml at the 12-month visit. At 1-month, 88% of patients in the digoxin group had serum digoxin levels within the therapeutic range of 0.5 to 2.0 ng/ml. At 1-year, 86% of patients in the digoxin group were taking the study drug and 83% of patients in the placebo group were taking placebo.

At the final study visit, 71% of surviving patients in the digoxin group were taking the study drug, and an additional 10% were taking open-label digoxin. In the placebo group, 68% of the surviving patients were taking placebo and 16% were taking open-label digoxin. Open-label digoxin was used at some point in the trial by 14% of patients in the digoxin group versus 22% of patients in the placebo group (P<0.001). The primary reason for discontinuing the study drug were the use of open-label digoxin to treat worsening heart failure.

There was no significant difference in overall mortality, which was 34.8% in the digoxin group and 35.1% in the placebo group. There was also no significant difference in cardiac mortality; however, there was a trend toward a lower risk of mortality due to worsening heart failure in the digoxin group (11.6% vs 13.2%; RR 0.88; 95% CI 0.77 to 1.01).

Significantly fewer patients were hospitalized for worsening heart failure symptoms in the digoxin group (26.8% vs 34.7%; RR 0.72; 95% CI 0.66 to 0.79). Overall, there were 626 fewer hospitalizations for worsening heart failure in the digoxin group.

The risk of death from any cause or hospitalization for worsening heart failure was significantly reduced in the digoxin group (RR 0.85; 95% CI 0.79 to 0.91) as was the endpoint of death from worsening heart failure or hospitalization due to worsening heart failure (RR 0.75; 95% CI 0.69 to 0.82). These benefits were seen soon after randomization, and they persisted throughout the trial.

Significantly fewer patients were hospitalized for any cause in the digoxin group (64.3% vs 67.1%; RR 0.92; 95% CI 0.87 to 0.98) and there were 6% fewer hospitalizations for any cause in the digoxin group (6356 vs 6777) and 10% fewer hospitalizations for cardiovascular causes (4106 vs 4570). There were fewer hospitalizations per patient for any cause in the digoxin group (P=0.01) as well as fewer hospitalizations per patient for cardiovascular causes (P<0.001).

Significantly more patients were hospitalized with suspected digoxin toxicity in the digoxin group compared to placebo (2.0% vs 0.9%; RR 2.17; 95% CI 1.42 to 3.32). The proportion of patients hospitalized for non-cardiovascular reasons was similar in both groups.

Based on planned pre-specified subgroup analysis there was no significant treatment effect heterogeneity noted on a relative scale. However, with respect to the combined endpoint of mortality from worsening heart failure or hospitalization from worsening heart failure, the absolute benefit of digoxin was greater among patients at the highest risk. The absolute risk differences for patients with an EF <25%, a cardiothoracic ratio >0.55, and a NYHA class III or IV was approximately -11%, which translates to a NNT of less than 10 patients to prevent 1 from experiencing the combined endpoint.

In the ancillary trial involving patients with an EF >45%, digoxin did not significantly reduce the primary endpoint of all-cause death or hospitalization for worsening heart failure, but this was likely due to low power as the treatment effect was consistent with the result in the main trial (RR 0.82; 95% CI 0.63 to 1.07).

Conclusions: In this large, pragmatic clinical trial, digoxin did not reduce all-cause mortality compared to placebo, but it did significantly reduce other important endpoints including all-cause hospitalizations and hospitalizations for worsening heart failure as well as combined endpoints including mortality and heart failure hospitalizations. Moreover, the benefits of digoxin were more evident on an absolute scale for patients at the highest levels of risk.

This trial has important strengths including its large size and pragmatic design. It did not have a run-in period, serum digoxin levels were not used for monitoring purposes, and the entry criteria were relatively lax, which increases its external validity.

A unique feature of the trial that compromised its internal validity and may have made it more difficult to show a benefit for digoxin was the allowance for open-label digoxin use in patients with symptomatic heart failure who could not be treated effectively with other measures. The potential benefit of digoxin on mortality may have been masked by such crossover. This is supported by the fact that approximately 8% more patients in the placebo group used open-label digoxin during the trial and the primary purpose for its use was to treat worsening heart failure. This is not generally considered standard practice for trials of new drugs. In fact, in the trial of carvedilol in severe heart failure that we will review in the coming weeks, investigators were explicitly instructed not to institute open-label treatment with a beta-blocker.

In conclusion, digoxin did not reduce mortality but did significantly reduce other important measures of morbidity in a diverse population of heart failure patients and some of the benefits may have been masked by allowance of crossover to open-label digoxin in patients with worsening heart failure. In our opinion, this trial supports a strong effect of digoxin on heart failure morbidity with high external validity, and it does not exclude the possibility that digoxin could reduce death in a trial employing more traditional design methods for establishing proof of efficacy.