The impact of serum concentration‐guided digoxin therapy on mortality of heart failure patients: A long‐term follow‐up, propensity‐matched cohort study

Abstract Background Recently published studies suggested that digoxin may increase mortality in heart failure with reduced ejection fraction (HFrEF). However, in the vast majority of former trials serum digoxin concentration (SDC) was not measured and therapy was not SDC‐guided. Aim To assess the impact of SDC‐guided digoxin therapy on mortality in HFrEF patients. Methods Data of 580 HFrEF patients were retrospectively analyzed. In patients on digoxin, SDC was measured every 3 months and digoxin dosage was SDC‐guided (target SDC: 0.5‐0.9 ng/mL). All‐cause mortality of digoxin users and nonusers was compared after propensity score matching (PSM). Results After 7.1 ± 4.7 years follow‐up period (FUP) all‐cause mortality of digoxin users (n = 180) was significantly higher than nonusers (n = 297) (propensity‐adjusted HR = 1.430; 95% CI = 1.134‐1.804; P = .003). Patients having SDC of 0.9 to 1.1 ng/mL (n = 60) or > 1.1 ng/mL (n = 44) at any time during the FUP had an increased risk of all‐cause mortality (HR = 1.750; 95% CI = 1.257‐2.436, P = .001 and HR = 1.687; 95% CI = 1.153‐2.466, P = .007), while patients having a maximal SDC < 0.9 ng/mL (n = 76) had similar mortality risk (HR = 1.139; 95% CI = 0.827‐1.570, P = .426), compared to digoxin nonusers. Conclusions According to our propensity‐matched analysis, SDC‐guided digoxin therapy was associated with increased all‐cause mortality in optimally treated HFrEF patients, especially with SDC ≥0.9 ng/mL. These results reinforce the expert opinion that digoxin in HFrEF can only be used among carefully selected patients with close SDC monitoring.


| BACKGROUND
Digoxin is one of the oldest drugs in the cardiology armamentarium.
The main indication for its application is heart failure (HF) and atrial fibrillation (AF). Despite the extensive use of digoxin, only one randomized placebo-controlled clinical trial (RCT) has examined the effect of digoxin on morbidity and mortality. 1 In the DIG (Digitalis Investigation Group) trial, the use of digoxin did not modify all-cause mortality, although a significant reduction in hospitalization due to worsening HF was observed in HF patients with sinus rhythm (SR). Of note, there has been no RCT assessing mortality and morbidity in patients with AF.
Since the DIG trial observational studies, [2][3][4][5] post-hoc analyses of RCTs, [6][7][8][9] and meta-analyses [10][11][12] have been published with the goal of analyzing the mortality effect of digoxin in HF and/or AF. Most of these nonrandomized publications indicated a potential increase in mortality associated with digoxin. However, in most of these studies serum digoxin concentration (SDC) was not measured. The most recent meta-analysis including more than 825 000 patients, confirmed that digoxin use is associated with increased mortality in patients with AF and HF. 13 Notably, only 10 of the 37 studies reported data on daily digoxin dose and/or data on SDC. 1,3,[14][15][16][17][18][19][20][21] In the aforementioned publications remains the concern that the mortality-increasing effect of digoxin may be related to the lack of the control of SDC and consequently elevated SDCs. Furthermore, due to the potentially incomplete adjustment of all the potentially influencing confounders the observed digoxin-associated mortality increase might be due to the more frequent use of this drug in sicker patients. 22 According to the recent guidelines for HF, 23 the use of digoxin has decreased significantly, however digoxin still affects large patient populations. 24 Given the lack of studies that have evaluated the effect of SDCguided digoxin therapy, we aimed to assess the effect of digoxin on long-term survival in a HF with reduced ejection fraction (HFrEF) patient population, where digoxin dose was regularly adjusted based on SDC. Patients were considered to suffer from HFrEF if the left ventricular ejection fraction (LVEF) was <40%. LVEF was measured by echocardiography using the biplane Simpson method. Patients were classified as digoxin users if digoxin was administered at the time of the initiation of HFOC care and digoxin therapy was applied without interruption during the follow-up period (FUP). Patients who received digoxin at the time of referrals, but digoxin therapy was discontinued afterward during the FUP were excluded from the study. Patients were considered to be new digoxin users if digoxin was initiated at the first visit at the HFOC. Patients who did not receive digoxin at baseline, but digoxin treatment was introduced during the FUP were excluded from the study. Patients were considered to be nondigoxin users if digoxin was not used and not started at baseline. Digoxin initial dosing was calculated with a standardized method. 25 Afterwards SDC was measured every 3 months and the dose was adjusted according to it. The goal therapeutic range of SDC was 0.5 to 0.9 ng/mL. 26 SDC samples were usually taken after 4 to 6 hours of oral administration. During follow-up we made every effort to apply guideline-recommended therapy to every patient.
The study was approved by the institutional review board and complies with the ethical guidelines of the Declaration of Helsinki.

| Study end points
The outcome measure of this study was time to all-cause mortality.
This parameter was compared between digoxin users and nonusers across the whole patient population and after propensity score matching (PSM). Digoxin users were also divided into three groups based on the maximal SDC measured during follow-up (maxSDC < 0.9 ng/mL, 0.9 ≤ maxSDC < 1.1 ng/mL, maxSDC ≥ 1.1 ng/mL) and survival was compared among these subgroups of the propensity-adjusted population. Furthermore, the effect of SDC-guided digoxin therapy on allcause mortality was assessed in new digoxin users and in patients with AF and SR also in the propensity-adjusted population. Mortality data were obtained from the database of the National Health Insurance Fund of Hungary.   3 | RESULTS

| Patients characteristics
The baseline characteristics of the patients of the total cohort and patients after PSM with and without digoxin treatment are presented in Table 1. From the total cohort, 185 patients received digoxin at the time of first visit to the HFOC. As expected, digoxin users suffered more often from AF than nondigoxin users (41.1% vs 21.3%) and had more decreased ejection fraction (26.4 ± 6.5% vs 28.0 ± 6.6%). Ischemic etiology was more frequent among nondigoxin users. There was also a significant difference between the two groups regarding baseline device use: significantly more digoxin-treated patients had a previously implanted ICD or CRT-P/D compared to nonusers (13.0% vs 7.6%).
Regarding drug treatment applied at baseline, the minority of patients  Figure S1), although patients with digoxin exposure had higher incidence of AF (39.4% vs 27.9%, P = .009).  (Table 2).

| Subgroup analyses in the propensity matched population
Those patients who had a maxSDC of between 0.9 and 1.  Figure S2a). This phenomenon was not statistically significant among those having AF at baseline (HR: 1.106, CI: 0.756-1.619, P = .604) (Suppl. Figure S2b).
When we analyzed the effect of digoxin among the 123 new digoxin users compared to digoxin nonusers, we found that digoxin resulted in a significant increased risk of all-cause mortality (HR: 1.371, 95% CI: 1.062-1.770, P = .016) (Suppl. Figure S3).
When digoxin level was analyzed as a continuous variable, SDC was associated with a 14% higher adjusted hazard of death for each 0.5 ng/mL increase (P = .0073).

| Main findings
In this real-life, community-based cohort of optimally treated HFrEF patients, we found that SDC-guided digoxin therapy was associated with increased all-cause mortality, especially with SDC ≥ 0.9 ng/mL.

| Serum-concentration-guided digoxin therapy
The narrow therapeutic window for the use of digitalis glycosides is well known. However, most publications that demonstrated an elevated mortality risk associated with digoxin did not report data about daily digoxin dose and/or serum levels. Even in the studies that reported such information, serum digoxin measurements were not performed in a systematic fashion. In the DIG trial SDC was measured only at 4 weeks and 1 year after the start of the study, while digoxin toxicity was followed only by signs and symptoms at 4 months, and every 4 months thereafter. 1 In a study by Freeman et al. comprising 2891 newly diagnosed HFrEF patients, 17 SDC was measured at all in 70% of patients and was measured just once in 27% of patients. Consequently, the lack of regular SDC control and/or higher SDC may have contributed to the adverse mortality effect of digoxin observed in these trials.
Our retrospective study demonstrates that even with an extremely close monitoring strategy, which were performed systematically in every patient, it was only possible to maintain SDC below 0.9 ng/mL in 42% of patients during the entire follow-up. This may be partly due to the pharmacokinetics of digoxin (it eliminates mainly through the kidneys), and the fact that the renal function of HFrEF patients is typically impaired. It therefore appears to be reasonable to use digitoxin instead of digoxin in HFrEF because of its hepatic elimination. Evidence regarding the effects of digitoxin on morbidity and mortality or data about its safe therapeutic range is even more limited.
In a single-center study of 1020 ICD recipients, treatment with digoxin or digitoxin were associated with similarly increased mortality compared to digitalis nonusers. 27 The ongoing DIGIT-HF trial will hopefully be able to clarify the place of digitoxin in therapy for HFrEF. 28 This trial is investigating the hypothesis that digitoxin-at serum concentrations in the lower therapeutic range-reduces mortality and morbidity in patients with HFrEF with or without AF.

| Correlation of SDCs and mortality
A post-hoc analysis of the DIG trial has raised the concern that high SDC (≥1.2 ng/mL) could lead to an increase in all-cause and cardiovascular mortality, and favorable digoxin effects are only expected in patients with SDC between 0.5 and 0.8 ng/mL. 26 In the recently published post-hoc analysis of the ARISTOTLE trial 19  had SDC levels below 0.9 ng/mL, while only 42% of our patient population had maxSCD < 0.9 ng/mL.
In contrast to the DIG study, we could not identify a favorable mortality effect in patients with maxSDC < 0.9 ng/mL. This may be explained by the fact that there were significant differences between our patient population and those cohorts (eg, we included patients with AF also, in contrast to the DIG trial). Moreover, digoxin users had more advanced HF with lower LVEF in our cohort, and the proportion of patients with hypertension or diabetes was higher compared to the DIG trial. It should be also noted that the morbidity-and mortalityreducing drug and device therapies were applied in higher proportion and dose in our patients than they were used in the DIG trial, which also could have modified the possible deleterious effects of digoxin.

| Effect of digoxin in patients with AF and SR
The results of studies that evaluated the effect of digoxin on mortality

| Limitations
However, in our nonrandomized patient cohort analysis we aimed to minimize potential confounding factors by carefully adjusting our data along important patient characteristics potentially responsible for worse outcomes using two different statistical methods (ie, adjusted multivariate Cox regression and PSM), residual bias cannot be excluded, as this was pointed by Aguirre Dávila et al in a recently published post-hoc analysis of the DIG trial. 22 The observed neutral effect of digoxin in the subgroup of patients with SDC < 0.9 ng/mL on mortality should be interpreted carefully, hence this group represents a small number of patients and has limited statistical power.
The data collection process for our patient cohort started in 2007. Since then, there have been changes in the guideline recommendations regarding the pharmacological and device treatment of HFrEF. These changes may have modified the mortality effect of digoxin.
Our single-center patient population consisted of only Caucasians. Accordingly, the results of the study do not necessarily apply to patients outside this group.

| CONCLUSIONS
In the current, retrospective, single-center study, serum-concentration-guided digoxin therapy was associated with increased all-cause mortality in optimally treated HFrEF patients, especially with SDC ≥ 0.9 ng/mL. It can be also concluded that the safe use of digoxin, which does not result in unfavorable outcomes in HFrEF is hardly feasible.

ACKNOWLEDGMENTS
The statistical expertise of Eva Herrmann, Head of Department of Biostatistics, J.W. Goethe University is greatly appreciated by the authors.