Effect of Chronic Digoxin Use on Mortality and Heart Failure Hospitalization in Pulmonary Arterial Hypertension

Abstract

Background

Digoxin acutely increases cardiac output in patients with pulmonary arterial hypertension (PAH) and right ventricular failure; however, the effects of chronic digoxin use in PAH are unclear.

Methods and Results

Data from the Minnesota Pulmonary Hypertension Repository were used. The primary analysis used likelihood of digoxin prescription. The primary end point was a composite of all‐cause mortality or heart failure (HF) hospitalization. Secondary end points included all‐cause mortality, HF hospitalization, and transplant‐free survival. Multivariable Cox proportional hazards analyses determined the hazard ratios (HR) and 95% CIs for the primary and secondary end points. Among 205 patients with PAH in the repository, 32.7% (n=67) were on digoxin. Digoxin was more often prescribed to patients with severe PAH and right ventricular failure. After propensity score‐matching, 49 patients were digoxin users, and 70 patients were nonusers; of these 31 (63.3%) in the digoxin group and 41 (58.6%) in nondigoxin group met the primary end point during a median follow‐up time of 2.1 (0.6–5.0) years. Digoxin users had a higher combined all‐cause mortality or HF hospitalization (HR, 1.82 [95% CI, 1.11–2.99]), all‐cause mortality (HR, 1.92 [95% CI, 1.06–3.49]), HF hospitalization (HR, 1.89 [95% CI, 1.07–3.35]), and worse transplant‐free survival (HR, 2.00 [95% CI, 1.12–3.58]) even after adjusting for patient characteristics and severity of PAH and right ventricular failure.

Conclusions

In this retrospective, nonrandomized cohort, digoxin treatment was associated with greater all‐cause mortality and HF hospitalization, even after multivariate correction. Future randomized controlled trials should assess the safety and efficacy of chronic digoxin use in PAH.

Nonstandard Abbreviations and Acronyms

CO

cardiac output

PAH

pulmonary arterial hypertension

WHO

World Health Organization

Clinical Perspective.

What Is New?

• Chronic digoxin use is more prevalent in patients with severe pulmonary arterial hypertension and right heart failure.

• In a retrospective, nonrandomized cohort, chronic digoxin therapy is associated with increased risk of combined all‐cause mortality or heart failure hospitalization, all‐cause mortality, and heart failure hospitalization, and lower transplant‐free survival, even after adjusting for the severity of pulmonary vascular disease and right heart failure.

What Are the Clinical Implications?

• The safety and efficacy of chronic digoxin use in pulmonary arterial hypertension requires further investigation, but this single‐center study showed no benefit of digoxin use.

Pulmonary arterial hypertension (PAH) is a fatal disease that affects the pulmonary vasculature through vasoconstriction and adverse vascular remodeling that obliterates and stiffens the pulmonary vasculature resulting in increased pulmonary artery pressures, pulmonary vascular resistance, and pulmonary vascular stiffness. ¹ , ² , ³ , ⁴ The resulting increase in right ventricular (RV) afterload eventually leads to RV dysfunction and failure, which remains the strongest predictor of death. ¹ , ² , ³ , ⁵ , ⁶ , ⁷

While currently approved PAH therapies reduce RV afterload, they do not directly target the RV dysfunction, and mortality still remains high in intermediate‐ and high‐risk patients in the modern era. ⁸ , ⁹ , ¹⁰ , ¹¹ The main treatments for RV dysfunction and right heart failure (HF) in patients with PAH are supportive therapies, including diuretics and digoxin. Digoxin, a cardiac glycoside, provides ionotropic support for patients with ventricular dysfunction, but the long‐term benefits of digoxin use in HF remain controversial. ¹² , ¹³ , ¹⁴ Chronic digoxin reduces HF hospitalization in patients with LV systolic dysfunction in sinus rhythm; however the evidence for its utility in isolated RV dysfunction due to PAH is scarce. ¹¹ , ¹⁵ , ¹⁶ Although 1 study shows digoxin acutely increases cardiac output (CO) and reduces circulating norepinephrine levels, ¹⁷ the effect of long‐term digoxin therapy in patients with PAH remains unclear. Furthermore, digoxin use is not without risk as there is an increased risk of mortality among patients using digoxin in an atrial fibrillation population. ¹⁸

Hence, we aimed to describe patients with PAH treated with chronic digoxin therapy and to examine the association of digoxin use with all‐cause mortality and HF‐related hospitalization in patients with PAH using the Minnesota Pulmonary Hypertension Repository.

METHODS

The data that support the findings of this study are available from the corresponding author upon reasonable request.

Study Population

The Minnesota Pulmonary Hypertension Repository has been previously described. ¹⁹ , ²⁰ In brief, it is a single‐center registry that enrolls all consecutive patients treated for pulmonary hypertension at the University of Minnesota Pulmonary Hypertension Clinic since March 2014. Patients who were diagnosed before March 2014 were entered retrospectively. All data are collected by chart review and entered using an internet‐based electronic data‐capture system. Informed consent was obtained from every participant in the repository. The University of Minnesota institutional review boards approved the Minnesota Pulmonary Hypertension Repository.

For this analysis, we evaluated all adult patients (aged $\ge$ 18 years at the time of enrollment) diagnosed with World Health Organization (WHO) Group I PAH who were enrolled between March 2014 and September 16, 2020. In the Minnesota Pulmonary Hypertension Repository, the diagnosis of PAH required the following: (1) a mean pulmonary artery pressure $\ge$25 mm Hg at rest up until 2019 and $\ge$20 mm Hg at rest since 2019, with a pulmonary capillary wedge pressure of <15 mm Hg, and a pulmonary vascular resistance >3 wood units; and (2) the exclusion of other WHO categories of pulmonary hypertension by clinical evaluation and objective tests, including pulmonary function tests and ventilation‐perfusion scan. Patients with obstructive lung disease diagnosed by reduced expiratory flow rates (forced expiratory volume in 1 second/forced vital capacity <75% predicted), more than mild interstitial lung disease diagnosed by reduced total lung capacity <60%, and chronic thromboembolic pulmonary disease were excluded from the study. Additionally, patients who were already on chronic digoxin therapy for other medical indications before the diagnosis of PAH were excluded.

Clinical Covariates

The Minnesota Pulmonary Hypertension Repository collects baseline demographics, clinical characteristics, echocardiographic variables, and hemodynamics from the initial diagnostic right heart catheterization at the time of enrollment. For this study, we analyzed the following baseline demographic and clinical characteristics: age, sex, race, body mass index, etiology of PAH, WHO functional class, 6‐minute walk distance, serum NT‐proBNP (N‐terminal pro–brain natriuretic peptide) levels, glomerular filtration rate, history of atrial fibrillation, echocardiographic variables, and hemodynamics from right heart catheterization. Echocardiographic variables include LV ejection fraction, RV enlargement, qualitative assessment of RV function, tricuspid annular plane systolic excursion, tissue Doppler velocity of the lateral tricuspid annulus (S′), and RV fractional area change. Hemodynamic measurements include mean right atrial pressure (mRAP), mean pulmonary artery pressure, pulmonary capillary wedge pressure, mixed venous oxygen saturation, cardiac index, CO, and pulmonary vascular resistance. Baseline PAH medications at the time of enrollment were also collected.

Long‐Term PAH Management

Patients with PAH who responded to acute vasodilator challenge were treated with calcium channel blockers. All patients who did not respond to acute vasodilator challenge at the time of diagnosis were treated with monotherapy or combination therapy using phosphodiesterase‐5‐inhibitors, endothelin receptor antagonists, and prostacyclin analogue, or prostacyclin receptor agonists based on the severity of the disease as recommended. ¹⁶ Initially, patients with low‐ and intermediate‐risk characteristics were treated with monotherapy or sequential combination therapy, but after 2015, a prospective dual therapy approach using a phosphodiesterase‐5‐inhibitors and an endothelin receptor antagonists was used as clinically indicated. Patients with high‐risk characteristics were treated from diagnosis with initial combination therapy, including a parenteral prostacyclin. Digoxin and diuretics were used to treat right HF as needed at the discretion of the treating physician. Digoxin was dosed based on a nomogram and levels were checked only if clinically indicated. Patients were followed every 3 to 6 months regularly in the outpatient setting, and more frequently if needed.

Vital Statistics

Vital statistics were updated every 6 months through the Minnesota death index. In patients who were not identified as deceased by the Minnesota death index, it was possible to establish vital statistics by chart review. Date of lung transplantation and HF hospitalization were collected by chart review.

Outcomes

The primary outcome was a composite of all‐cause mortality or HF hospitalization. The secondary outcomes include: (1) all‐cause mortality, (2) HF hospitalization, and (3) composite of all‐cause mortality or lung transplantation (transplant‐free survival).

Statistical Analysis

Summary statistics of baseline patient characteristics for the total cohort and matched cohort were calculated and categorized by patient digoxin use status. Continuous variables were reported as mean±SD or median (interquartile range), and categorical variables were presented as frequency and percentage unless indicated. Unpaired t‐tests and Wilcoxon rank sum tests were used to test differences in means and medians, respectively, for continuous variables. Chi‐square tests for independence were used in categorical variables, substituted by Fisher exact tests where appropriate. Covariate balance before and after matching was assessed by standardized mean differences (SMD) and SMD graphical representations (Figure S1). A propensity score for digoxin use was estimated using a logistic regression model with the following covariates: age at index date; sex; diagnosis date; body mass index; PAH etiology; WHO functional class; history of hypertension, coronary artery disease, and diabetes; use of prostacyclin, warfarin, phosphodiesterase 5 inhibitors, endothelin receptor antagonists, calcium channel blockers, and supplemental oxygen; mRAP; mean pulmonary artery pressure; pulmonary capillary wedge pressure; CO; and PVR. The digoxin users were matched on propensity score (maximum caliper=0.05) to nondigoxin users using the greedy matching algorithm (https://bioinformaticstools.mayo.edu/research/gmatch/), allowing up to 2 nondigoxin users for each digoxin user. For patients treated with digoxin, the index date (the date of entry into the survival analysis) was defined as the date of first digoxin prescription. For patients never treated with digoxin, the index date was defined as 75 days after the date of PAH diagnosis. We chose to adjust the index date by 75 days in the nondigoxin users, because it was the median number of days from PAH diagnosis to digoxin initiation among the digoxin users. Kaplan–Meier plots with log‐rank tests compared unadjusted survival rates in the matched cohort using this same index date scheme. Multivariable Cox proportional hazards regression compared the risk of various clinical end points by digoxin use status after assessing the proportional hazards assumption. Models were adjusted for age at index date, sex, index date, and propensity score. The proportional hazards assumption was assessed by testing interactions between the treatment and time within the Cox regression models and by qualitative assessment of the unadjusted Kaplan–Meier plots. No meaningful violation of this assumption was found. We also conducted a sensitivity analysis, restricting to patients with RV dysfunction, defined as having a cardiac index <2.5 L/min per m² or a mRAP >8 mm Hg at baseline. Sensitivity to incomplete matching was examined by assessing outcomes in the unmatched cohort and in a cohort matched with a maximum caliper of 0.4. Figure S1 illustrates the balance of covariates with the various approaches. Statistical significance was determined at an alpha level of 0.05. Statistical analyses were performed with SAS, version 9.4 (SAS Institute Inc) and R version 4.2.1.

RESULTS

Baseline Characteristics

A total of 266 patients with PAH were initially identified in the Minnesota Pulmonary Hypertension Repository. Sixty‐one patients with missing data for variables used in the propensity score were excluded, resulting in a final unmatched cohort of 205 patients (Figure 1). Of the 205 patients in the unmatched cohort, 67 (32.7%) initiated digoxin therapy and 138 (67.3%) were never prescribed digoxin. Table S1 compares the baseline demographics, clinical characteristics, echocardiographic data, and hemodynamics of the unmatched cohort based on digoxin use. Additionally, baseline clinical and hemodynamic characteristics comparing matched subjects and unmatched subjects by treatment status are shown in Table S2.

Figure 1. Study flow diagram.

PAH indicates pulmonary arterial hypertension; and WHO, World Health Organization.

After propensity score matching using the Greedy algorithm, which permitted 1 to 2 nondigoxin users per digoxin user, there were a total of 119 patients (49 in digoxin group and 70 in nondigoxin group) included in the outcomes analyses. At baseline, patients on digoxin therapy had higher serum NT‐proBNP (2539 pg/mL versus 531 pg/mL; P=0.001, SMD=0.586), a greater prevalence of RV enlargement (91.1% versus 71.9%; P=0.014, SMD=0.511) on echocardiogram, and lower mixed venous oxygen saturation (62.0% versus 65.4%; P=0.039, SMD=0.415) (Table 1). There were no statistically significant differences in the other baseline characteristics, although some had moderately large SMDs at ≈0.3.

Table 1.

Baseline Patient Characteristics by Digoxin Use in the Matched Cohort

Characteristic	Matched Cohort			SMD
	Never on digoxin	Digoxin	P value
	n=70*	n=49*
Age at index‡	54.3±15.3	55.9±16.4	0.590	0.036
Women	53 (75.7%)	37 (75.5%)	0.980	0.005
Race			0.212	0.338
White	51 (72.9%)	39 (79.6%)
Black	4 (5.7%)	5 (10.2%)
Other clinical characteristics/unknown	15 (21.4%)	5 (10.2%)
Body mass index‡, kg/m2; n=117	29.1±7.2	29.2±9.5	0.949	0.012
WHO functional class, n=100			0.541	0.298
I	2 (3.5%)	2 (4.8%)
II	9 (15.5%)	4 (9.5%)
III	41 (70.7%)	28 (66.7%)
IV	6 (10.3%)	8 (19.1%)
Etiology			0.838	0.174
Idiopathic	16 (22.9%)	10 (20.4%)
Heritable	1 (1.4%)	2 (4.1%)
Anorexigen	5 (7.1%)	3 (6.1%)
APAH	48 (68.6%)	34 (69.4%)
Medications at baseline
Prostacyclin	6 (8.6%)	6 (12.2%)	0.548	0.121
Phosphodiesterase‐5 inhibitor	21 (30.0%)	19 (38.8%)	0.319	0.186
Endothelin receptor antagonist	8 (11.4%)	6 (12.2%)	0.892	0.025
Riociguat	0 (0.0%)	1 (2.0%)	0.412	0.204
Warfarin	13 (18.6%)	10 (20.4%)	0.803	0.046
Calcium channel blocker	12 (17.1%)	9 (18.4%)	0.863	0.032
Supplemental oxygen	11 (15.7%)	10 (20.4%)	0.509	0.122
Echocardiography
LV ejection fraction‡, n=106	61.5±5.8	62.5±5.4	0.387	0.173
RV enlargement, n=109	46 (71.9%)	41 (91.1%)	0.014	0.511
RV enlargement severity, n = 87			0.996	0.018
Mild	7 (15.2%)	6 (14.6%)
Moderate	21 (45.7%)	19 (46.3%)
Severe	18 (39.1%)	16 (39.0%)
RV dysfunction (n=108)	40 (61.5%)	33 (76.7%)	0.098	0.334
RV dysfunction severity (n=72)			0.339	0.357
Mild	10 (25.6%)	4 (12.1%)
Moderate	19 (48.7%)	20 (60.6%)
Severe	10 (25.6%)	9 (27.3%)
TAPSE‡, n=73	1.7±0.5	1.7±0.4	0.957	0.013
S′‡, n=49	10.4±2.2	10.7±2.5	0.632	0.138
RV fractional area change‡, n=64	33.4±11.3	33.3±10.2	0.950	0.016
Right heart catheterization
Mean right atrial pressure, mm Hg‡	9.2±5.1	9.7±4.5	0.566	0.108
Mean PA pressure, mm Hg‡	49.2±14.5	50.1±11.1	0.690	0.076
Pulmonary capillary wedge pressure, mm Hg‡	10.8±4.2	11.4±6.9	0.526	0.114
Mixed venous oxygen saturation, %‡ (n=104)	65.4±8.3	62.0±8.3	0.039	0.415
Cardiac index, L/min per m2 ‡	2.4±0.83	2.2±0.83	0.273	0.227
Cardiac output, L/min‡	4.4±1.6	4.2±1.8	0.424	0.148
Pulmonary vascular resistance, Wood units‡	10.1±5.8	10.8±5.2	0.487	0.131
Other clinical characteristics
Six‐minute walk test distance, m†, n=68	356 (291–457)	323 (249–406)	0.220	0.306
NT‐proBNP, pg/mL†, n=106	531 (218–1842)	2539 (789–6043)	0.001	0.586
GFR (mL/min)†, n=114	75 (56–90)	71 (54–90)	0.596	0.050
Atrial fibrillation	6 (8.6%)	2 (4.1%)	0.468	0.185

APAH indicates associated pulmonary arterial hypertension; GFR, glomerular filtration rate; LV, left ventricular; NT‐proBNP, N‐terminal pro–brain natriuretic peptide; PA, pulmonary artery; RV, right ventricular; SMD, standardized mean difference; TAPSE, tricuspid annular plane systolic excursion; and WHO, World Health Organization.

^*n in the column header indicates the entire sample. The n for each covariate is listed in the row label. Covariates without n specified in row labels had no missing data.

^† Reported as median (interquartile range).

^‡ Reported as mean±SD.

Primary Outcome—All‐Cause Mortality or HF Hospitalization

The median follow‐up time was 2.1 (interquartile range, 0.6–5.0) years for the matched cohort. Over the course of follow‐up, 60.5% (72/119) of patients met the primary end point, of which 18 were deceased and 54 were hospitalized for HF. Forty‐three percent (31/72) of these patients were from the digoxin cohort while 56.9% (41/72) were from the nondigoxin group. On Cox proportional hazards regression analyses, digoxin therapy was associated with higher risk of combined all‐cause mortality or HF hospitalization after accounting for age, sex, index date, and propensity score (hazard ratio [HR], 1.82 [95% CI, 1.11–2.99]; Table 2). This estimate is slightly attenuated from the results obtained from sensitivity analyses of the unmatched cohort (HR, 1.95 [95% CI, 1.24–3.07]; Table S3) and the less stringently matched cohort (HR, 1.97 [95% CI, 1.23–3.12]; Table S4). Kaplan–Meier survival curves of the matched cohort for the primary end point are shown in Figure 2A.

Table 2.

Association of Digoxin Users Versus Nonusers With Primary and Secondary Outcomes in Matched Cohort

Outcome	Events		HR (95% CI)
	Digoxin	Nondigoxin
Primary: all‐cause mortality+HF hospitalization	31	41	1.82 (1.11–2.99)
Secondary (1): all‐cause mortality	23	26	1.92 (1.06–3.49)
Secondary (2): all‐cause mortality+lung transplant	25	27	2.00 (1.12–3.58)
Secondary (3): HF hospitalization	24	30	1.89 (1.07–3.35)

Models adjusted for age, sex, index date, and propensity score. HF indicates heart failure; and HR, hazard ratio.

Figure 2. Kaplan–Meier survival estimates stratified by digoxin use status.

A, Primary end point: combined all‐cause mortality and HF hospitalization. B, Secondary end point: all‐cause mortality. C, All‐cause mortality and lung transplantation (transplant‐free survival). D, HF hospitalization. HF indicates heart failure.

Secondary Outcomes—All‐Cause Mortality, Transplant‐Free Survival, HF Hospitalization

Similar to the findings observed in the primary end point, digoxin therapy was associated with higher all‐cause mortality (HR, 1.92 [95% CI, 1.06–3.49]), higher rates of HF hospitalization (HR, 1.89 [95% CI, 1.07–3.35]), and worse transplant‐free survival (HR, 2.00 [95% CI, 1.12–3.58]) in the matched cohort after adjusting for age, sex, index date, and propensity score (Table 2). These estimates are slightly attenuated from the results obtained in sensitivity analyses with unmatched and less stringently matched cohorts (Tables S3 and S4). Kaplan–Meier survival curves of the matched cohort for the secondary end points are shown in Figure 2B through 2D.

Subgroup Analysis—Cardiac Index <2.5 or mRAP >8 mm Hg

In a subgroup analysis of the matched cohort with RV dysfunction and right HF, defined by cardiac index <2.5 L/min per m² or mRAP >8 mm Hg, there were a total of 96 patients with 41 (42.7%) on digoxin and 55 (57.3%) never on digoxin. Twenty‐six patients in the digoxin group (6 deaths and 20 HF hospitalizations) and 30 patients in the nondigoxin group (10 deaths and 20 HF hospitalizations) met the primary end point. On Cox proportional hazards regression analyses, digoxin use was associated with a heightened risk of combined end point of all‐cause mortality or HF hospitalization (HR, 1.85 [95% CI, 1.07–3.21]), greater all‐cause mortality (HR, 1.95 [95% CI, 0.99–3.86]), higher HF hospitalization (HR, 2.02 [95% CI, 1.07–3.83]), and worse transplant‐free survival (HR, 2.04 [95% CI, 1.06–3.92]) when adjusted for age, sex, index date, and propensity score (Table 3).

Table 3.

Association of Digoxin Users Versus Nonusers With Primary and Secondary Outcomes in Subgroup With Cardiac Index <2.5 L/min per m² or Mean Right Atrial Pressure >8 mm Hg

Outcome	Events		HR (95% CI)
	Digoxin	Nondigoxin
Primary: all‐cause mortality+HF hospitalization	26	30	1.85 (1.07–3.21)
Secondary (1): all‐cause mortality	19	18	1.95 (0.99–3.86)
Secondary (2): all‐cause mortality+lung transplant	21	19	2.04 (1.06–3.92)
Secondary (3): HF hospitalization	20	20	2.02 (1.07–3.83)

Models adjusted for age, sex, index date, and propensity score. HF indicates heart failure; and HR, hazard ratio.

DISCUSSION

In this single‐center, contemporary study of consecutive patients with PAH treated at a tertiary referral center, the following findings were observed: (1) chronic digoxin use is more prevalent in those with more severe PAH and right HF, as defined by hemodynamics on right heart catheterization, and (2) chronic digoxin therapy did not improve any of the examined outcomes (ie, combined all‐cause mortality or HF hospitalization, all‐cause mortality alone, HF hospitalization alone, and transplant‐free survival) in an analysis which attempted to control for confounding through propensity‐score matching for demographics, etiology of PAH, comorbidities, medications, severity of PAH, and right heart dysfunction. In fact, digoxin use is associated with poorer outcomes. While interesting, the observation of an association between digoxin use and adverse outcomes should be interpreted cautiously given the potential for residual confounding effects of the more severe RV dysfunction at baseline in patients prescribed digoxin in this nonrandomized cohort.

In the DIG (Digitalis Investigational Group) trial, chronic digoxin use decreases the risk of HF hospitalization in patients with reduced LV ejection fraction and sinus rhythm. ¹⁴ , ¹⁵ Unlike the findings in this trial, there are no clinical data supporting the chronic use of digoxin in patients with PAH with or without concomitant right HF. The use of digoxin in PAH remains mainly at the discretion of medical providers and is largely based on limited preclinical and acute clinical hemodynamic studies. In 1954, Davis et al showed in a canine model of isolated RV failure induced by pulmonary artery banding that digoxin increased RV systolic pressure and CO, whilst decreasing right atrial pressure and total peripheral resistance. ²¹ Subsequently, in 1998, Rich et al demonstrated in 17 patients with PAH and right HF that intravenous digoxin acutely elevates CO and reduces circulating norepinephrine. ²²

In this analysis, chronic digoxin therapy was not associated with a reduction in all‐cause mortality or HF hospitalization in patients with PAH with RV failure. Paradoxically, digoxin use was associated with higher rates of mortality and HF hospitalization. There are several possible explanations for our observation. First, despite our attempts to control for confounding by propensity‐score matching, there is likely residual confounding whereby chronic digoxin therapy was more frequently prescribed in patients with more advanced PAH and right HF, a group destined for adverse outcomes. ²³ Thus, chronic digoxin use may just be a marker of severe PAH and right HF. However, this study raises the possibility that digoxin worsens mortality and morbidity in patients with PAH. Perhaps the use of digoxin heightens the risk of ventricular and atrial arrhythmias, which are known complications of digoxin. ²⁴ While the DIG trial shows there is a reduction in death because of HF with chronic digoxin use, there was a reciprocal heightened risk of mortality attributable to arrhythmias. ¹⁵ In addition, myocardial ischemia increases the risk of fatal ventricular arrhythmias with digoxin therapy, ²⁵ which may be problematic for patients with PAH as they are at risk for RV ischemia because of loss of right coronary artery perfusion during systole and capillary rarefaction. ²⁶ , ²⁷ , ²⁸ Hence, chronic digoxin use may increase the risk of mortality attributable to arrhythmias in patients with PAH.

To determine whether chronic digoxin use is beneficial only in patients with PAH with right HF, we performed a subgroup analysis including only those with either a cardiac index <2.5 L/min per m² or mRAP >8 mm Hg (Table 3). Even in this subgroup analysis, chronic digoxin use is associated with increased risk of all‐cause mortality and HF hospitalization, which supports the concern that chronic digoxin therapy may worsen morbidity and mortality in patients with PAH. The lack of benefit of digoxin therapy in PAH, as seen in our study, is not entirely surprising as most beneficial associations of digoxin use are observed in patients with primarily LV HF, regardless of the presence of RV failure. ¹⁵ , ²⁹ , ³⁰ In addition, the effect size of digoxin treatment even in patients with left HF is small, requiring a large sample size as well as a long follow‐up time to demonstrate beneficial effects. ¹⁵ Furthermore, chronic digoxin does not improve RV function, exercise capacity, or WHO functional class in patients with RV dysfunction secondary to pulmonary hypertension associated with chronic lung disease (cor pulmonale). ²⁹ , ³¹ , ³² In a placebo‐controlled, randomized, double‐blind, crossover study of 12 patients with cor pulmonale, digoxin did not improve exercise capacity or RV ejection fraction at rest or during exercise. ³² Similarly, in a double‐blind, randomized, placebo‐controlled study of 15 patients with cor pulmonale, digoxin did not augment RV ejection fraction when compared with placebo unless the LV ejection fraction was also concomitantly reduced. ²⁹ Finally, in a placebo‐controlled, double‐blind, crossover, randomized trial of 34 patients with cor pulmonale, digoxin had no effects on HF symptoms and WHO functional class except in those with atrial fibrillation. ³¹

This study has several limitations that warrant discussion. Most importantly, this was an observational study and as we did not randomize to digoxin, causal inference is limited. Despite our best effort to appropriately address differences in baseline covariates by using the propensity‐score matching, it is possible that there are unmeasured residual confounders. As noted by Aguirre Dávila et al, prescription bias plays an important role in any observational studies investigating potential increased mortality attributable to digoxin use, and thus it may be interpreted as an indicator of severity of HF rather than an effect of treatment. ³³ As an additional limitation, some patients were excluded because of missing data on some hemodynamic variables and etiology or failure to find a reasonable control for matching. This may have impacted the generalizability of our findings. However, there were no statistically significant differences in baseline clinical characteristics between those with and without missing data (Table S5), and sensitivity analyses show that effect estimates from a more general sample are similar (Tables S3 and S4). Also, our study does not assess the effects of digoxin on other end points such as exercise capacity, serum NT‐proBNP, RV function by imaging, or hemodynamics at follow‐up visits. Lastly, events may have been missed among patients who received care outside our hospital system.

In conclusion, in this observational study, chronic digoxin therapy in patients with PAH did not reduce all‐cause mortality, HF hospitalizations, and lung transplant. Contrary to our initial hypothesis, it was associated with worse outcomes. Future prospective studies should assess the safety and efficacy of chronic digoxin use in PAH.

Sources of Funding

This study is supported by the Vikki Auzzene Research Award. Dr. Prins is funded by US National Institutes of Health (NIH) K08 HL140100 and R01s HL158795 and HL162927, and an Innovation Award from the American Lung Association.  Dr. Archer is supported, in part, by the NIH grants, NIH R01HL113003 and NIH R01HL071115, Canada Foundation for Innovation 229252 and 33012; Tier 1 Canada Research Chair in Mitochondrial Dynamics and Translational Medicine 950‐229252; and the William J. Henderson Foundation. Dr. Lutsey is supported by NIH R01 R01HL131579 and NIH K24 HL159246. Dr. Thenappan is funded by the MN Futures award, Cardiovascular Medical Education and Research funds, and the University of Minnesota Faculty Development Research Grant.

Disclosures

K.W. Prins served as a consultant to Edwards and receives grant funding from Bayer. S.L. Archer is supported by research grants from the Canadian Institutes of Health Research. T. Thenappan served as a consultant to United Therapeutics, J&J, Merck, Aria CV, and Altvant Science. The remaining authors have no disclosures to report.

Supporting information

Tables S1–S5

Figure S1