https://orcid.org/0000-0002-6819-6818
Abstract
The use of digitalis has been plagued by controversy since its initial use. We aimed to determine the relationship between digoxin use and outcomes in hospitalized patients with acute coronary syndromes (ACSs) complicated by heart failure (HF) accounting for sex difference and prior heart diseases.
Of the 25 187 patients presenting with acute HF (Killip class ≥2) in the International Survey of Acute Coronary Syndromes Archives (NCT04008173) registry, 4722 (18.7%) received digoxin on hospital admission. The main outcome measure was all-cause 30-day mortality. Estimates were evaluated by inverse probability of treatment weighting models. Women who received digoxin had a higher rate of death than women who did not receive it [33.8% vs. 29.2%; relative risk (RR) ratio: 1.24; 95% confidence interval (CI): 1.12–1.37]. Similar odds for mortality with digoxin were observed in men (28.5% vs. 24.9%; RR ratio: 1.20; 95% CI: 1.10–1.32). Comparable results were obtained in patients with no prior coronary heart disease (RR ratio: 1.26; 95% CI: 1.10–1.45 in women and RR ratio: 1.21; 95% CI: 1.06–1.39 in men) and those in sinus rhythm at admission (RR ratio: 1.34; 95% CI: 1.15–1.54 in women and RR ratio: 1.26; 95% CI: 1.10–1.45 in men).
Digoxin therapy is associated with an increased risk of early death among women and men with ACS complicated by HF. This finding highlights the need for re-examination of digoxin use in the clinical setting of ACS.
Inverse probability of treatment weighting: relative risk for association between digoxin use and 30-day mortality by sex and acute coronary syndrome types in patients presenting with acute heart failure on hospital admission. Abbreviations: ACS, acute coronary syndrome; CHD, coronary heart disease; HF, heart failure; NSTE-ACS, non-ST segment elevation acute coronary syndromes; STEMI, ST segment elevation myocardial infarction.
Introduction
The use of digoxin in patients with heart failure (HF) is controversial. The Digitalis Investigation Group (DIG) trial offered reassurance that digoxin1 can be safely used in patients with chronic HF, but concern remains about the effects of digoxin on mortality in patients with incident HF in the acute setting, especially in those with myocardial infarction.2–5
Controversy over the use of digoxin in myocardial infarction stems from prior observations. Digoxin does not improve left ventricular performance in myocardial infarction and the risk of digoxin-toxic ventricular tachycardias may be greater in the acute phase of HF.6 Digoxin increases myocardial oxygen consumption. Since oxygen delivery is impaired with myocardial infarction, digoxin may contribute to cause progressive infarct expansion.7 In a study of patients who had survived a myocardial infarction, digoxin therapy was one of the strongest predictors of fatal arrhythmic events.8 Yet, these studies are not contemporary and recent guidelines for myocardial infarction still support digoxin use in patients with severe HF and/or supraventricular arrhythmias.9–13
At least two other sources of uncertainty merit attention. First, a new post-hoc analysis of the DIG trial and registry1 suggests that many published data on digoxin use are affected by prescription bias, so they cannot yield any firm conclusions.14 Prescription bias is a form of selection bias. Patients who were on digoxin before enrolment in almost all studies had much higher rates of mortality compared with those not previously treated with digoxin. Digoxin is often prescribed in the sicker patients. In such patients, it is their being sicker, not digoxin, that may cause their worse outcomes. The second issue relates to the role of sex. Another post-hoc analysis of the DIG trial examined the effects of digoxin on mortality based on the sex of patients.15 Digoxin therapy was found to be associated with an increased risk of death from any cause among women, but not men. Yet, subsequent studies did not confirm such observation.16,17
In view of the uncertainty regarding the safety of digoxin and the possible sex-specific heterogeneity of its effects, we explored the association between digoxin use and 30-day mortality in a large international cohort of women and men presenting with acute coronary syndromes (ACSs) and HF on hospital admission. A requirement of the study was that the outcomes of people treated with digoxin could not be influenced by the other concomitant comorbidities and therapies for acute HF or revascularization strategies.
Methods
From October 2003 to January 2019, we analysed information from the International Survey of Acute Coronary Syndromes (ISACS) Archives (NCT04008173). The ISACS Archives provide access to de-identified research cohorts and clinical trials in ACS. Registries enrolled in the ISACS Archives use data definition for the measures/experiments that are harmonized to the standard variables of the ISACS-TC (NCT01218776).18,19 Registries had independent source documentation. All supporting data and analyses are available within the article and its supplemental material. This study complies with the Declaration of Helsinki. The local research ethics committee from each hospital approved the study. Because patient information was collected anonymously, institutional review boards waived the need for individual informed consent. All data were transferred to the Department of Electrical and Computer Engineering, University of California, Los Angeles, where final statistical analyses were done.
Patient population
The cohort study consisted of 88 774 ACS patients with information on Killip class at hospital presentation. Patients were admitted with a diagnosis of ACS and had at least one of the following: ECG changes consistent with ACS, increases in serum biochemical markers of cardiac necrosis, and/or documentation of coronary artery disease.20 Acute HF was the subgroup of interest. Patients were classified as receiving or not receiving digoxin on admission. We excluded patients with pre-existing non-ischaemic congestive cardiomyopathy (n = 962) to distinguish HF complication on admission for ACS from decompensated chronic HF. We also excluded patients without acute HF leaving a final study population of 25 187 patients (Supplementary material online, Figure S1).
Outcome measures and definitions
The primary outcome measure was all-cause mortality within 30 days of hospital admission. The 30-day window was selected to enrich the data over those acquired during the index hospitalization while mitigating survivor bias. The diagnosis of acute HF was based on clinical symptoms or signs and radiographic evidence of pulmonary congestion. The presence and severity of acute HF at the time of hospital presentation was formally categorized by use of the Killip classification.10 Acute HF was defined as Killip class ≥2. We also noted the use of percutaneous coronary intervention (PCI) and the type of medications given on hospital admission: aspirin, clopidogrel, unfractionated heparin, glycoprotein IIb/IIIa inhibitors, nitrates (nitroglycerin, nitroprusside), diuretics (furosemide, torasemide, bumetanide), inotropic agents (dopamine, dobutamine, milrinone), and digoxin. We also noted the use of angiotensin receptor blockers (ARBs) (candesartan, valsartan, losartan) and angiotensin-converting enzyme (ACE) inhibitors (captopril, enalapril, lisinopril, ramipril, trandolapril) during hospitalization.
Dose and blood levels of digoxin were not collected. Coronary artery bypass graft was performed as a need for urgent surgery after PCI. So, outcomes of coronary artery bypass graft interventions were included in the subgroup of patients with PCI. Smoking habits were self-reported (Supplementary material online, Methods).
Statistical analysis
Patient characteristics were examined according to sex and stratified according to the treatment group: digoxin recipients vs. digoxin non-recipients. Our analysis focused solely on the possible interaction between sex and digoxin therapy prescribed on hospital admission; no other subgroup analyses on medications were conducted. Baseline characteristics were reported as percentages for categorical variables and means with standard deviation for continuous variables. We had complete data on sex, age, index event, and outcomes. Some patients had missing data on other variables. We used k-nearest neighbour algorithms as the imputation method to treat missing data21,22 (Supplementary material online, Methods). The existence of an association between 30-day mortality and digoxin therapy was evaluated with the use of inverse probability of treatment weighting models23 in which patients who did or did not receive digoxin therapy would be balanced on key measured baseline characteristics. A notable advantage of using the propensity-based weighted approach is that as in a randomized controlled trial the process of cohort assembly is outcome blinded. In the model, use of digoxin was the dependent variable and 26 baseline characteristics listed in Table1 and Supplementary material online, Table S3 were used as covariates (Supplementary material online, Methods). We calculated digoxin vs. no-digoxin relative risk (RR) ratios with their 95% confidence interval (CI). Standardized differences after weighting were calculated to ensure balanced treatment groups with respect to baseline characteristics.24 Groups were considered balanced when the standardized difference was <10% (Supplementary material online, Methods). Comparisons of outcomes between groups were made by two-sided P-value. Subsidiary analyses were also conducted to estimate the sex-specific effect of digoxin therapy in two subsets of patients presenting to hospital with different baseline risk of 30-day mortality: ST segment elevation myocardial infarction (STEMI) and non-ST segment elevation acute coronary syndrome (NSTE-ACS). Further analyses were also performed in patients with STEMI undergoing reperfusion therapies and in patients without pre-existing diagnosis of coronary heart disease (CHD). Because supraventricular arrhythmias may worsen HF during hospitalization, we also assembled a sensitivity cohort that accounted for patients with sinus rhythm excluding those with supraventricular arrhythmias, specifically atrial fibrillation and flutter, multifocal atrial tachycardia, and re-entrant supraventricular tachycardia. To minimize concern about comparison of the treatment effect in subgroups, estimates were compared by test of interaction on the log scale. A P-value <0.05 for outcomes was taken to indicate that the difference between the effects in women and men was unlikely to have occurred simply by chance (Supplementary material online, Methods).25
Inverse probability of treatment weighting in patients with acute coronary syndrome complicated by acute heart failure: outcomes sorted by digoxin therapy on hospital admission and sex
| Women (N = 10 442) | Men (N = 14 745) | |||||
|---|---|---|---|---|---|---|
| Characteristics | Digoxin recipients (N = 2075) | Digoxin non-recipients (N = 8367) | Standardized difference | Digoxin recipients (N = 2647) | Digoxin non-recipients (N = 12 098) | Standardized difference |
| Age, years | 70.9 ± 9.9 | 70.8 ± 9.8 | 0.01 | 66.8 ± 10.8 | 66.1 ± 11.1 | 0.06 |
| Cardiovascular risk factors | ||||||
| Family history of CAD | 41.5 | 42.7 | −0.03 | 41.3 | 41.0 | 0.006 |
| Diabetes | 39.7 | 39.3 | 0.009 | 31.8 | 31.1 | 0.02 |
| Hypertension | 79.0 | 79.1 | −0.003 | 69.1 | 69.6 | −0.01 |
| Hypercholesterolaemia | 51.3 | 51.8 | −0.01 | 48.7 | 48.2 | 0.01 |
| Current smokers | 20.9 | 21.7 | −0.02 | 42.7 | 43.5 | −0.02 |
| Former smokers | 0.7 | 0.7 | 0.005 | 2.3 | 2.1 | 0.009 |
| Clinical history of CHD | ||||||
| Previous angina pectoris | 36.3 | 36.6 | −0.007 | 33.6 | 32.5 | 0.02 |
| Previous myocardial infarction | 21.4 | 21.3 | 0.001 | 28.2 | 27.1 | 0.03 |
| Previous PCI | 4.5 | 4.5 | −0.001 | 6.9 | 6.7 | 0.006 |
| Previous CABG | 3.1 | 2.8 | 0.02 | 5.7 | 5.6 | 0.003 |
| Clinical history of CVD | ||||||
| Peripheral artery disease | 8.9 | 8.9 | 0.002 | 10.2 | 9.4 | 0.03 |
| Previous stroke | 8.8 | 8.4 | 0.01 | 7.0 | 6.7 | 0.01 |
| Medications on admission | ||||||
| Nitrates | 72.2 | 72.8 | −0.01 | 72.8 | 72.7 | 0.001 |
| Diuretics | 71.5 | 70.9 | 0.01 | 70.7 | 68.1 | 0.06 |
| Inotropes | 23.0 | 22.2 | 0.02 | 21.8 | 20.7 | 0.03 |
| Aspirin | 91.7 | 91.7 | 0.002 | 92.1 | 92.3 | −0.006 |
| Clopidogrel | 61.3 | 61.6 | −0.005 | 65.7 | 65.9 | −0.004 |
| Unfractionated heparin | 49.1 | 49.4 | −0.006 | 52.7 | 51.1 | 0.03 |
| Glycoprotein IIb/IIIa inhibitors | 8.3 | 8.1 | 0.009 | 8.0 | 8.1 | −0.003 |
| Clinical presentation on admission | ||||||
| STEMI | 66.7 | 66.5 | 0.005 | 68.8 | 68.9 | −0.002 |
| ST segment shifts in anterior leads (at ECG) | 52.7 | 52.3 | 0.008 | 53.9 | 52.7 | 0.02 |
| Systolic BP, mmHg | 131.1 ± 30.4 | 131.4 ± 30.1 | −0.01 | 130.0 ± 29.7 | 130.2 ± 29.3 | −0.008 |
| Heart rate, b.p.m. | 91.7 ± 25.7 | 91.1 ± 25.9 | 0.02 | 89.6 ± 25.1 | 89.3 ± 24.4 | 0.01 |
| Serum creatinine at baseline (mg/dL) | 1.3 ± 0.9 | 1.3 ± 0.9 | −0.006 | 1.4 ± 1.0 | 1.4 ± 1.1 | 0.03 |
| Outcome | P-value | P-value | ||||
| 30-day mortality | 33.8 | 29.2 | 0.0001 | 28.5 | 24.9 | 0.0002 |
| Relative risk ratio (95% CI) | 1.24 (1.12–1.37) | 0.0001 | 1.20 (1.10–1.32) | 0.0001 | ||
| Women (N = 10 442) | Men (N = 14 745) | |||||
|---|---|---|---|---|---|---|
| Characteristics | Digoxin recipients (N = 2075) | Digoxin non-recipients (N = 8367) | Standardized difference | Digoxin recipients (N = 2647) | Digoxin non-recipients (N = 12 098) | Standardized difference |
| Age, years | 70.9 ± 9.9 | 70.8 ± 9.8 | 0.01 | 66.8 ± 10.8 | 66.1 ± 11.1 | 0.06 |
| Cardiovascular risk factors | ||||||
| Family history of CAD | 41.5 | 42.7 | −0.03 | 41.3 | 41.0 | 0.006 |
| Diabetes | 39.7 | 39.3 | 0.009 | 31.8 | 31.1 | 0.02 |
| Hypertension | 79.0 | 79.1 | −0.003 | 69.1 | 69.6 | −0.01 |
| Hypercholesterolaemia | 51.3 | 51.8 | −0.01 | 48.7 | 48.2 | 0.01 |
| Current smokers | 20.9 | 21.7 | −0.02 | 42.7 | 43.5 | −0.02 |
| Former smokers | 0.7 | 0.7 | 0.005 | 2.3 | 2.1 | 0.009 |
| Clinical history of CHD | ||||||
| Previous angina pectoris | 36.3 | 36.6 | −0.007 | 33.6 | 32.5 | 0.02 |
| Previous myocardial infarction | 21.4 | 21.3 | 0.001 | 28.2 | 27.1 | 0.03 |
| Previous PCI | 4.5 | 4.5 | −0.001 | 6.9 | 6.7 | 0.006 |
| Previous CABG | 3.1 | 2.8 | 0.02 | 5.7 | 5.6 | 0.003 |
| Clinical history of CVD | ||||||
| Peripheral artery disease | 8.9 | 8.9 | 0.002 | 10.2 | 9.4 | 0.03 |
| Previous stroke | 8.8 | 8.4 | 0.01 | 7.0 | 6.7 | 0.01 |
| Medications on admission | ||||||
| Nitrates | 72.2 | 72.8 | −0.01 | 72.8 | 72.7 | 0.001 |
| Diuretics | 71.5 | 70.9 | 0.01 | 70.7 | 68.1 | 0.06 |
| Inotropes | 23.0 | 22.2 | 0.02 | 21.8 | 20.7 | 0.03 |
| Aspirin | 91.7 | 91.7 | 0.002 | 92.1 | 92.3 | −0.006 |
| Clopidogrel | 61.3 | 61.6 | −0.005 | 65.7 | 65.9 | −0.004 |
| Unfractionated heparin | 49.1 | 49.4 | −0.006 | 52.7 | 51.1 | 0.03 |
| Glycoprotein IIb/IIIa inhibitors | 8.3 | 8.1 | 0.009 | 8.0 | 8.1 | −0.003 |
| Clinical presentation on admission | ||||||
| STEMI | 66.7 | 66.5 | 0.005 | 68.8 | 68.9 | −0.002 |
| ST segment shifts in anterior leads (at ECG) | 52.7 | 52.3 | 0.008 | 53.9 | 52.7 | 0.02 |
| Systolic BP, mmHg | 131.1 ± 30.4 | 131.4 ± 30.1 | −0.01 | 130.0 ± 29.7 | 130.2 ± 29.3 | −0.008 |
| Heart rate, b.p.m. | 91.7 ± 25.7 | 91.1 ± 25.9 | 0.02 | 89.6 ± 25.1 | 89.3 ± 24.4 | 0.01 |
| Serum creatinine at baseline (mg/dL) | 1.3 ± 0.9 | 1.3 ± 0.9 | −0.006 | 1.4 ± 1.0 | 1.4 ± 1.1 | 0.03 |
| Outcome | P-value | P-value | ||||
| 30-day mortality | 33.8 | 29.2 | 0.0001 | 28.5 | 24.9 | 0.0002 |
| Relative risk ratio (95% CI) | 1.24 (1.12–1.37) | 0.0001 | 1.20 (1.10–1.32) | 0.0001 | ||
Values are percentages or means ± standard deviation unless stated otherwise. Abbreviations: BP, blood pressure; CABG, coronary artery bypass graft; CAD, coronary artery disease; CHD, coronary heart disease; CI, confidence interval; CVD, cardiovascular disease; PCI, percutaneous coronary intervention; STEMI, ST segment elevation myocardial infarction.
Inverse probability of treatment weighting in patients with acute coronary syndrome complicated by acute heart failure: outcomes sorted by digoxin therapy on hospital admission and sex
| Women (N = 10 442) | Men (N = 14 745) | |||||
|---|---|---|---|---|---|---|
| Characteristics | Digoxin recipients (N = 2075) | Digoxin non-recipients (N = 8367) | Standardized difference | Digoxin recipients (N = 2647) | Digoxin non-recipients (N = 12 098) | Standardized difference |
| Age, years | 70.9 ± 9.9 | 70.8 ± 9.8 | 0.01 | 66.8 ± 10.8 | 66.1 ± 11.1 | 0.06 |
| Cardiovascular risk factors | ||||||
| Family history of CAD | 41.5 | 42.7 | −0.03 | 41.3 | 41.0 | 0.006 |
| Diabetes | 39.7 | 39.3 | 0.009 | 31.8 | 31.1 | 0.02 |
| Hypertension | 79.0 | 79.1 | −0.003 | 69.1 | 69.6 | −0.01 |
| Hypercholesterolaemia | 51.3 | 51.8 | −0.01 | 48.7 | 48.2 | 0.01 |
| Current smokers | 20.9 | 21.7 | −0.02 | 42.7 | 43.5 | −0.02 |
| Former smokers | 0.7 | 0.7 | 0.005 | 2.3 | 2.1 | 0.009 |
| Clinical history of CHD | ||||||
| Previous angina pectoris | 36.3 | 36.6 | −0.007 | 33.6 | 32.5 | 0.02 |
| Previous myocardial infarction | 21.4 | 21.3 | 0.001 | 28.2 | 27.1 | 0.03 |
| Previous PCI | 4.5 | 4.5 | −0.001 | 6.9 | 6.7 | 0.006 |
| Previous CABG | 3.1 | 2.8 | 0.02 | 5.7 | 5.6 | 0.003 |
| Clinical history of CVD | ||||||
| Peripheral artery disease | 8.9 | 8.9 | 0.002 | 10.2 | 9.4 | 0.03 |
| Previous stroke | 8.8 | 8.4 | 0.01 | 7.0 | 6.7 | 0.01 |
| Medications on admission | ||||||
| Nitrates | 72.2 | 72.8 | −0.01 | 72.8 | 72.7 | 0.001 |
| Diuretics | 71.5 | 70.9 | 0.01 | 70.7 | 68.1 | 0.06 |
| Inotropes | 23.0 | 22.2 | 0.02 | 21.8 | 20.7 | 0.03 |
| Aspirin | 91.7 | 91.7 | 0.002 | 92.1 | 92.3 | −0.006 |
| Clopidogrel | 61.3 | 61.6 | −0.005 | 65.7 | 65.9 | −0.004 |
| Unfractionated heparin | 49.1 | 49.4 | −0.006 | 52.7 | 51.1 | 0.03 |
| Glycoprotein IIb/IIIa inhibitors | 8.3 | 8.1 | 0.009 | 8.0 | 8.1 | −0.003 |
| Clinical presentation on admission | ||||||
| STEMI | 66.7 | 66.5 | 0.005 | 68.8 | 68.9 | −0.002 |
| ST segment shifts in anterior leads (at ECG) | 52.7 | 52.3 | 0.008 | 53.9 | 52.7 | 0.02 |
| Systolic BP, mmHg | 131.1 ± 30.4 | 131.4 ± 30.1 | −0.01 | 130.0 ± 29.7 | 130.2 ± 29.3 | −0.008 |
| Heart rate, b.p.m. | 91.7 ± 25.7 | 91.1 ± 25.9 | 0.02 | 89.6 ± 25.1 | 89.3 ± 24.4 | 0.01 |
| Serum creatinine at baseline (mg/dL) | 1.3 ± 0.9 | 1.3 ± 0.9 | −0.006 | 1.4 ± 1.0 | 1.4 ± 1.1 | 0.03 |
| Outcome | P-value | P-value | ||||
| 30-day mortality | 33.8 | 29.2 | 0.0001 | 28.5 | 24.9 | 0.0002 |
| Relative risk ratio (95% CI) | 1.24 (1.12–1.37) | 0.0001 | 1.20 (1.10–1.32) | 0.0001 | ||
| Women (N = 10 442) | Men (N = 14 745) | |||||
|---|---|---|---|---|---|---|
| Characteristics | Digoxin recipients (N = 2075) | Digoxin non-recipients (N = 8367) | Standardized difference | Digoxin recipients (N = 2647) | Digoxin non-recipients (N = 12 098) | Standardized difference |
| Age, years | 70.9 ± 9.9 | 70.8 ± 9.8 | 0.01 | 66.8 ± 10.8 | 66.1 ± 11.1 | 0.06 |
| Cardiovascular risk factors | ||||||
| Family history of CAD | 41.5 | 42.7 | −0.03 | 41.3 | 41.0 | 0.006 |
| Diabetes | 39.7 | 39.3 | 0.009 | 31.8 | 31.1 | 0.02 |
| Hypertension | 79.0 | 79.1 | −0.003 | 69.1 | 69.6 | −0.01 |
| Hypercholesterolaemia | 51.3 | 51.8 | −0.01 | 48.7 | 48.2 | 0.01 |
| Current smokers | 20.9 | 21.7 | −0.02 | 42.7 | 43.5 | −0.02 |
| Former smokers | 0.7 | 0.7 | 0.005 | 2.3 | 2.1 | 0.009 |
| Clinical history of CHD | ||||||
| Previous angina pectoris | 36.3 | 36.6 | −0.007 | 33.6 | 32.5 | 0.02 |
| Previous myocardial infarction | 21.4 | 21.3 | 0.001 | 28.2 | 27.1 | 0.03 |
| Previous PCI | 4.5 | 4.5 | −0.001 | 6.9 | 6.7 | 0.006 |
| Previous CABG | 3.1 | 2.8 | 0.02 | 5.7 | 5.6 | 0.003 |
| Clinical history of CVD | ||||||
| Peripheral artery disease | 8.9 | 8.9 | 0.002 | 10.2 | 9.4 | 0.03 |
| Previous stroke | 8.8 | 8.4 | 0.01 | 7.0 | 6.7 | 0.01 |
| Medications on admission | ||||||
| Nitrates | 72.2 | 72.8 | −0.01 | 72.8 | 72.7 | 0.001 |
| Diuretics | 71.5 | 70.9 | 0.01 | 70.7 | 68.1 | 0.06 |
| Inotropes | 23.0 | 22.2 | 0.02 | 21.8 | 20.7 | 0.03 |
| Aspirin | 91.7 | 91.7 | 0.002 | 92.1 | 92.3 | −0.006 |
| Clopidogrel | 61.3 | 61.6 | −0.005 | 65.7 | 65.9 | −0.004 |
| Unfractionated heparin | 49.1 | 49.4 | −0.006 | 52.7 | 51.1 | 0.03 |
| Glycoprotein IIb/IIIa inhibitors | 8.3 | 8.1 | 0.009 | 8.0 | 8.1 | −0.003 |
| Clinical presentation on admission | ||||||
| STEMI | 66.7 | 66.5 | 0.005 | 68.8 | 68.9 | −0.002 |
| ST segment shifts in anterior leads (at ECG) | 52.7 | 52.3 | 0.008 | 53.9 | 52.7 | 0.02 |
| Systolic BP, mmHg | 131.1 ± 30.4 | 131.4 ± 30.1 | −0.01 | 130.0 ± 29.7 | 130.2 ± 29.3 | −0.008 |
| Heart rate, b.p.m. | 91.7 ± 25.7 | 91.1 ± 25.9 | 0.02 | 89.6 ± 25.1 | 89.3 ± 24.4 | 0.01 |
| Serum creatinine at baseline (mg/dL) | 1.3 ± 0.9 | 1.3 ± 0.9 | −0.006 | 1.4 ± 1.0 | 1.4 ± 1.1 | 0.03 |
| Outcome | P-value | P-value | ||||
| 30-day mortality | 33.8 | 29.2 | 0.0001 | 28.5 | 24.9 | 0.0002 |
| Relative risk ratio (95% CI) | 1.24 (1.12–1.37) | 0.0001 | 1.20 (1.10–1.32) | 0.0001 | ||
Values are percentages or means ± standard deviation unless stated otherwise. Abbreviations: BP, blood pressure; CABG, coronary artery bypass graft; CAD, coronary artery disease; CHD, coronary heart disease; CI, confidence interval; CVD, cardiovascular disease; PCI, percutaneous coronary intervention; STEMI, ST segment elevation myocardial infarction.
Results
Baseline characteristics
The study cohort consisted of 10 442 women and 14 745 men. Patients with Killip class 4 constituted only 14.9% of the overall population of women with HF and accounted for ∼12.8% of its male counterpart with HF (Supplementary material online, Figure S2). Women as a whole were older than men. A greater proportion of women had a history of diabetes, hypertension, and hypercholesterolaemia than men, whereas fewer women presented as current or former smokers. Many of the sex differences were not consistent between digoxin-exposed and unexposed groups, as reflected by the low standardized difference (<10%) for all variables. The only notable difference was older age. There were high rates of diuretic use among women and men undergoing digoxin therapy (Supplementary material online, Table S1). Higher Killip class assessed on admission was not a predictor of greater dioxin use among women and men (Supplementary material online, Figure S3).
Balancing covariates
We balanced the distribution of covariates using a parametric strategy by inverse probability of treatment weighting (Table1). After weighting, there were no significant standardized differences in baseline characteristics and medications between the 2075 women who received and the 8367 women who did not receive digoxin therapy and the 2647 men who received and the 12 098 men who did not receive digoxin therapy.
Clinical outcomes in the overall population
Survival differed significantly among digoxin recipients and non-recipients (Table1). Women who received digoxin had a higher rate of death than women who did not receive it (33.8% vs. 29.2%; RR ratio: 1.24; 95% CI: 1.12–1.37). Similar odds for mortality with digoxin were observed in men (28.5% in recipients vs. 24.9% in non-recipients; RR ratio: 1.20; 95% CI: 1.10–1.32). There was no significant interaction between sex and digoxin therapy with respect to death (P-value for sex interaction = 0.3180) (Supplementary material online, Table S2). The RR ratios for mortality associated with digoxin did not change when controlling for the use of ARB/ACE inhibitors during hospitalization (Supplementary material online, Tables S3 and S4).
Clinical outcomes and type of acute coronary syndrome
We tested the type of ACS as potential effect modifier (Figure1). In STEMI, digoxin treatment was significantly associated with an increased rate of death in the overall population with women and men having similar mortality risks (absolute differences: 5% and 4.6%, respectively; RR ratio: 1.25; 95% CI: 1.10–1.41 and RR ratio: 1.26; 95% CI: 1.13–1.41, respectively) (Supplementary material online, Table S5). The rate of death was also higher among women with NSTE-ACS receiving digoxin compared with those in the control group (absolute difference: 3.7%; RR ratio: 1.22; 95% CI: 1.01–1.48) (Supplementary material online, Table S6). In contrast, the rate of death was not significantly different among men with NSTE-ACS who received digoxin and those who did not receive digoxin (absolute difference: 2.2%; RR ratio: 1.12; 95% CI: 0.95–1.33). There was no interaction between the outcomes of women and men (P-values = 0.4624 for STEMI and 0.2548 for NSTE-ACS) (Supplementary material online, Tables S7 and S8).
Inverse probability of treatment weighting: relative risks for association between digoxin use and 30-day mortality by sex and acute coronary syndrome types in patients presenting with acute heart failure on hospital admission. Abbreviations: ACS, acute coronary syndrome; HF, heart failure; NSTE-ACS, non-ST segment elevation acute coronary syndrome; RRR, relative risk ratio; STEMI, ST segment elevation myocardial infarction.
Clinical outcomes in patients with reperfusion therapy
Because differences in reperfusion rates for STEMI may result in disparities in outcomes, we restricted the analysis to those patients presenting with acute HF who underwent such therapy. The 30-day mortality rate consistently decreased in both sexes, but remained higher in the digoxin-treated group vs. the untreated group (28.7% vs. 24.5%; RR ratio: 1.24; 95% CI: 1.00–1.53 in women and 22.5% vs. 19.2%; RR ratio: 1.22; 95% CI: 1.02–1.46 in men) (Supplementary material online, Table S9). There was no evidence of effect modification by sex (interaction test; P-value = 0.4545) (Supplementary material online, Table S10).
Patients without prior history of coronary heart disease
A total of 13 424 patients (53.3%) did not have prior history of CHD. Of these, 2323 patients (17.3%) were treated with digoxin on admission. Overall, subgroup findings were similar to the point estimates for the full matched cohorts. In patients with ACS, digoxin was associated with a significant increase in the rate of death in both women (absolute difference: 5.2%; RR ratio: 1.26; 95% CI: 1.10–1.45) and men (absolute difference: 3.8%; RR ratio: 1.21; 95% CI: 1.06–1.39) (Supplementary material online, Table S11) After subgrouping for type of ACS (Figure 2), digoxin therapy was still associated with a significantly increased risk of death among STEMI in both women (absolute difference: 5.4%; RR ratio: 1.27; 95% CI: 1.08–1.49) and men (absolute difference: 3.9%; RR: 1.22; 95% CI: 1.05–1.42). On the other hand, in NSTE-ACS digoxin therapy was associated with a small, non-significant reduction in the risk of death among women (absolute difference: 2.8%; RR ratio: 1.15; 95% CI: 0.86–1.55) and men (absolute difference: 4.6%; RR ratio: 1.26; 95% CI: 0.96–1.67) (Supplementary material online, Tables S12 and S13). There was no significant heterogeneity in the association between digoxin use and sex across subgroups (P-values = 0.3408 in ACS, 0.3609 in STEMI, and 0.3290 in NSTE-ACS) (Supplementary material online, Tables S14–S16).
Inverse probability of treatment weighting: relative risks for association between digoxin use and 30-day mortality by sex and acute coronary syndrome types in patients presenting with acute heart failure on hospital admission and no prior history of coronary heart disease. Abbreviations: ACS, acute coronary syndrome; CHD, coronary heart disease; HF, heart failure; NSTE-ACS, non-ST segment elevation acute coronary syndrome; RRR, relative risk ratio; STEMI, ST segment elevation myocardial infarction.
Patients without supraventricular arrhythmias on admission and during hospital stay
Among the 14 670 HF patients with ACS but without supraventricular arrhythmias (Supplementary material online, Table S17 and Figure 3), digoxin use was significantly associated with higher risk of 30-day mortality (absolute difference: 6.1%; RR ratio: 1.34; 95% CI: 1.15–1.54 in women and absolute difference: 4.4%; RR ratio: 1.26; 95% CI: 1.10–1.45 in men; P = 0.2741 for the interaction) (Supplementary material online, Table S18). Breakdown of patients in the two categories of ACS showed qualitatively similar results in STEMI (absolute difference: 7.2%; RR ratio: 1.38; 95% CI: 1.17–1.64 in women and absolute difference: 5.3%; RR ratio: 1.33; 95% CI: 1.13–1.57 in men; P = 0.3794 for the interaction) (Supplementary material online, Tables S19 and S20), but not in NSTE-ACS (absolute difference: 4%; RR ratio: 1.26; 95% CI: 0.94–1.68 in women and absolute difference: 2.8%; RR ratio: 1.16; 95% CI: 0.91–1.48 in men; P = 0.3343 for the interaction) (Supplementary material online, Tables S21 and S22).
Inverse probability of treatment weighting: relative risks for association between digoxin use and 30-day mortality by sex and acute coronary syndrome types in patients presenting with acute heart failure on hospital admission and without supraventricular arrhythmias. Abbreviations: ACS, acute coronary syndrome; NSTE-ACS, non-ST segment elevation acute coronary syndrome; RRR, relative risk ratio; STEMI, ST segment elevation myocardial infarction.
Discussion
Although retrospective analysis has limitations, the current study supports concerns about the safety of digoxin in patients with HF on hospital presentation after an ACS. To keep confounding to a minimum, we conducted a comparative cohort study of digoxin recipients vs. non-recipients using a parametric balancing strategy with 26 baseline clinical and therapeutic variables. Reperfusion therapy was also taken into account. Digoxin was firmly associated with an increased risk of 30-day mortality among women and men presenting to hospital with HF and a diagnosis of ACS. The potential harm of digoxin was magnified by the severity of clinical presentation. The absolute difference in the rate of death after STEMI was 5.0% for women and 4.6% for men, whereas estimates after NSTE-ACS decreased to 3.7% and 2.2%, respectively. Although digoxin was associated with an increased risk of death among women compared with men, we did not note any statistically significant interaction between outcomes and sex.
A contemporary randomized trial to examine the safety of digoxin on HF after an ACS seems unlikely. Prior randomized work focused almost exclusively on patients with chronic HF conditions. The DIG trial was performed in 1997 and assigned 6800 patients with chronic systolic HF and sinus rhythm to digoxin or placebo.1 The trial showed neutral effects on mortality. Some of the controversy regarding this trial relates to the question of whether it should be subject to ‘narrow interpretation’ or ‘broad interpretation’. A very narrow interpretation would be that only patients similar to those enrolled in the trial would have neutral effects on mortality. The trial, instead, has received the broader interpretation that digoxin is safe in a variety of acute and chronic conditions. So, digoxin therapy is still recommended in guidelines. In general, it should only be used for rate control in the presence of HF with reduced ejection fraction, but in clinical practice it is still widely used for acute HF itself.
When randomized controlled trials are absent, a comprehensive health care data source like a shared data repository may provide an alternative for complementary analyses. The current study evaluated patients with ACS complicated by acute HF using a network that provides a shared infrastructure and standardized data collection across multiple registries on ACS. We balanced differences between groups using propensity weighting, which helps reducing biases inherent in retrospective designs. We assessed variation in the effects of digoxin therapy across sexes and severity of ACS clinical presentations comparing the RRs within the separate strata of women vs. men and STEMI vs. NSTE-ACS. Comparisons were measured by formal tests of treatment interaction. On this background, the size of the association between digoxin therapy and 30-day mortality and the precision of its estimation in STEMI and NSTE-ACS lead us to believe that the associations we recognized are clinically significant and not spurious statistical associations.
To our knowledge, there are no published large data concerning associations between ACS and digoxin harmful effects in acute HF with which to compare our results. Several studies performed in the 1990s evaluated outcomes among patients with chronic HF surviving a myocardial infarction and found increased long-term mortality with digoxin.26–28 Although patients in these older studies were not on current standard therapies, there is no reason to believe that their observations and conclusions are not applicable to contemporary patients. Our study supports early findings and provides new evidence of an association between 30-day mortality after an ACS complicated by acute HF and digoxin therapy. It also suggests a higher increase of risk among those ACS patients with more severe clinical presentation, specifically among patients with STEMI.
Although we cannot identify the mechanism of effect modification between type of ACS and digoxin therapy, our data indicate that some hypotheses can be declined. The effect modification is not attributable to older age or comorbidities29 as we weighted subjects with STEMI vs. NSTE-ACS by the inverse probability of treatment received and created a sample in which treatment was independent of measured baseline covariates. The effect modification does not reflect a proxy for a more common occurrence of atrial fibrillation and worse outcomes in STEMI, since in a subgroup analysis we excluded patients with supraventricular arrhythmias, and subgroup findings were similar to the point estimates for the full matched cohorts.
A possible mechanism for the increased risk of death among STEMI on digoxin therapy may involve an enhanced myocardial sensitivity to the toxic effect of digoxin during severe and persisting ischaemia as that produced by complete coronary occlusion. Myocardial sensitivity to cardiac glycosides has been well documented in experimentally induced STEMI. There was a 43% decrease in the dose of cardiac glycosides needed to produce ventricular arrhythmias in dogs or pigs after acute ligation compared with the pre-ischaemic dose.30,31 Furthermore, digoxin may enhance systemic vascular resistance due to its direct arteriolar vasoconstrictive effect, which, in turn, may decrease coronary blood flow and exacerbate myocardial injury.32 NSTE-ACS patients present smaller infarct size and higher collateral blood supply than STEMI patients.33 These factors may potentially weaken the association between digoxin and mortality causing relatively better outcomes in NSTE-ACS compared with STEMI.
Prior work on chronic HF has suggested that digoxin may increase mortality by ∼20% in women but not in men.15 Our subgroup analyses showed that there were sex-based differences in survival resulting from digoxin therapy only in patients with NSTE-ACS. Digoxin was associated with a significantly higher risk of death among women, but it had no significant effect among men. We wondered about whether the RRs from the subgroups of women and men really differed from each other. This information is commonly guided by a statistical test for interaction.25 Based on this calculation, there was insufficient evidence to support a different treatment effect in women and men. However, failure to find significant interactions does not demonstrate definitively that the treatment effect seen overall applies equally to women and men. Lack of significance can arise because of differences in sample sizes or differences in standard errors or a combination of the two. The existence of interaction between sex and digoxin-related outcomes thus remains in doubt. Further studies would be needed to confirm this finding (or not).
Digoxin is particularly prone to prescription bias as clinicians have been trained to use it in the sickest patients.34 Thus, our results in the overall population of ACS patients complicated by acute HF might be interpreted with caution. To circumvent this limitation, we assessed the safety of digoxin in two more homogeneous groups of patients in whom prescription bias is unlikely: those without prior history of CHD and those without supraventricular arrhythmias at hospital presentation. Yet, our subgroup analyses showed that not only the association between digoxin use and death remained significant, but also the absolute differences in the rates of death were quantitatively similar to those estimated for the whole population. These findings reinforce concerns regarding the use of digoxin for treatment of acute HF after an ACS. The present study also fits in the common notion of harm from digoxin therapy as evaluated by contemporary post-hoc analyses and meta-analyses in other disease entities. In the ARISTOTLE (Apixaban for Reduction in Stroke and Other Thromboembolic Events in Atrial Fibrillation) trial,35 patients with atrial fibrillation who are taking digoxin had an increased risk of death, whether or not they had HF, compared with patients not taking this drug. The hazard inherently present in patients undergoing digoxin therapy has been further demonstrated by a large meta-analysis including 235 047 patients with atrial fibrillation and 91 379 patients with chronic HF.36 The findings of these studies parallel our results and call into question the use of digoxin in a variety of acute and chronic clinical conditions.
Study limitations
The current study has some limitations. Being observational in nature, it is therefore subject to potential for confounding and bias by intent to treat. Inverse probability of treatment weighting helps to minimize such confounding. Prior digoxin use is associated with worse outcomes, but secondary analyses excluding patients who potentially may have been previously treated with digoxin showed similar results. Several other limitations should be noted. We acknowledge the inability to standardize determination of Killip class as it reflects clinical practice. On the other hand, the relevance of Killip classification among patients with ACSs is well established by guidelines.10 Information on ejection fraction on admission cannot be systematically available as most patients died soon after admission. Exclusion of patients due to missing measurements can introduce systematic bias. On the other hand, previous studies have shown that male and female patients with reduced and preserved ejection fraction have similar rates of in-hospital mortality,37 and increasing evidence suggests that assessment for the presence and severity of HF by physical examination provides significant independent prognostic information, much more than that provided by echocardiography.38 Dose and drug serum digoxin concentration were not collected. However, it is unlikely that variations of serum digoxin levels may differentially affect STEMI over NSTE-ACS patients and, thus, they are unlikely to have modified the associations that we found. Clinical endpoints were ascertained through case report forms by the investigators, which may be affected by the complexity of endpoint definitions. Finally, we evaluated the potential interaction of digoxin with ARB/ACE inhibitors during hospitalization. However, information on timing of and reasons for ARB/ACE inhibitors’ initiation was not systematically available in the database. So, analyses on ARB/ACE inhibitors should be interpreted with caution due to the possible persistence of immortal time bias. The vast majority of physicians delayed the administration of aldosterone blockade until patients were haemodynamically stable, and most patients received aldosterone blockade soon after discharge. As such, we did not note the use aldosterone blockade as our outcomes were assessed at 30 days from admission.
Conclusions
Digoxin has not been well studied in acute HF syndromes. Our findings highlight the strong association between digoxin use, HF after an ACS, and worse 30-day clinical outcomes in patients with and without supraventricular arrhythmias. The interaction between ACS and digoxin therapy on early mortality could be confirmed by a randomized controlled trial of digoxin therapy. However, new prospective randomized trials of digoxin are unlikely. Yet, such a randomized study may not be considered ethical, since it would be designed to confirm risks and not benefits. In the absence of definitive evidence from trials, our data may represent a strong case for a re-examination of the use of digoxin therapy for patients with HF in the clinical setting of ACS.
Funding
None.
Conflict of interest: L.B. reports personal fees from Bayer, International Aspirin Foundation, UK, Sanofi, Lilly, BMS/Pfizer, FICYE, and PACE; grants from AstraZeneca; and other from GlyCardial, outside the submitted work. L.B. has a patent APOj-Gly licensed, IV_STATIN and DJ1-F pending. C.P.G. reports personal fees from AstraZeneca, Amgen, Bayer, and Daiichi Sankyo, and grants from Abbott and BMS outside the submitted work. All other authors reported no conflicts of interest.
Acknowledgement
Graphical abstract created using Biorender.com
