Sotalol is a commonly prescribed antiarrhythmic drug (AAD) used for maintaining sinus rhythm in patients with atrial fibrillation (AF). Although randomized studies have found that sotalol can significantly delay time to AF recurrence, its association with mortality is less clear, particularly among those with coronary artery disease. We examined outcomes of 2,838 patients with coronary artery disease and AF. Using Cox proportional hazards modeling, landmark analysis, and time-dependent covariates for drug therapy, we compared cumulative survival among patients treated with sotalol (n = 226), amiodarone (n = 856), or no AAD (n = 1,756). Median follow-up was 4.2 years (interquartile range [IQR] 2.0–7.4). The median age was 68 years (IQR 60–75). Compared with those treated with amiodarone or no AAD, patients treated with sotalol were less likely to be black (6% vs 13% vs 13%) and have a previous myocardial infarction (35% vs 51% vs 48%) or a left ventricular ejection fraction <40% (13% vs 26% vs 21%). In follow-up, persistence of sotalol was limited; 97% of patients treated with sotalol were treated for <25% of the follow-up period. In adjusted analysis accounting for time on therapy, sotalol use was associated with an increased risk of all-cause death compared with no drug (hazard ratio 1.53, 95% confidence interval 1.19 to 1.96, p = 0.0009), but a decreased risk of death compared with amiodarone (hazard ratio 0.72, 95% confidence interval 0.55 to 0.91, p = 0.0141). In conclusion, sotalol therapy was more frequently used in patients with fewer co-morbidities, often discontinued early in follow-up, and was associated with increased mortality compared with no AAD but decreased mortality relative to amiodarone.
Atrial fibrillation (AF) is a common and difficult arrhythmia to treat. Projections suggest that the prevalence of AF will double by the year 2050. Sotalol is a Vaughan-Williams class III antiarrhythmic agent commonly prescribed for the maintenance of sinus rhythm in patients with AF. To date, randomized trials of sotalol have not been adequately powered to specifically address effects on mortality, yet meta-analyses indicate a trend toward increased arrhythmic death and all-cause mortality with sotalol therapy. Additionally, little is known about whether sotalol is associated with increased mortality when used in routine clinical practice outside of randomized controlled trials. This is particularly true in patients at greater risk for sudden cardiac death on antiarrhythmic therapy such as those with coronary artery disease (CAD). The objective of this study was to determine if sotalol therapy is associated with an increased risk of all-cause mortality relative to other therapeutic choices for the maintenance of sinus rhythm.
Methods
The Duke Databank for Cardiovascular Disease has captured the clinical course of >100,000 patients who have received cardiovascular care in the Duke University Health System. A longitudinal record is established for each patient with significant CAD, and follow-up data are collected at 6 months and then yearly thereafter (98% follow-up rate). Data collection includes demographics, clinical characteristics, medications, and major clinical events (including death, myocardial infarction, sustained ventricular arrhythmias, and sudden cardiac death). Mortality is confirmed yearly using the National Death Index.
In this study, patients were included if they had evidence of CAD (≥1 coronary arteries with stenosis ≥50%) and a diagnosis of AF from July 1, 1993 (the year sotalol became available for clinical use), to December 31, 2007. Patients with atrial flutter only (and no evidence of AF) were excluded. For the purpose of this analysis, patients with an implantable cardioverter-defibrillator, primary valvular heart disease or mitral valve stenosis, congenital heart disease, patients receiving membrane-active antiarrhythmic medications other than sotalol or amiodarone, and those receiving both amiodarone and sotalol were excluded. Based on the type of antiarrhythmic medication used at baseline, patients were stratified into 3 exposure groups (no antiarrhythmic drug [AAD], sotalol, or amiodarone). Medications used during the last 2 days of an inpatient stay were used as a surrogate for discharge medications. Although the primary comparison of interest was sotalol versus no antiarrhythmic therapy, an amiodarone cohort was included to help provide a referent group with active antiarrhythmic treatment.
The primary end point of interest was all-cause mortality at the last follow-up. Time 0 (baseline) for outcome ascertainment was the first documented episode of AF after July 1, 1993. The null hypothesis for this analysis was that there was no difference in all-cause survival in patients with AF and CAD, regardless of treatment with sotalol or amiodarone (H o ).
Clinical characteristics were compared among the 3 treatment groups (sotalol, amiodarone, or no antiarrhythmic therapy) using summary statistics (median and twenty-fifth and seventy-fifth quartiles) for continuous variables and percentage for categorical variables. The Kaplan-Meier method was used to evaluate mortality. Accordingly, patients who did not develop the primary end point were censored at the time of loss to follow-up or the end of the study period. The log-rank test was used to compare event-free survival between the treatment groups.
Multiple imputation (SAS Proc MI) was used to impute missing values for variables that were <15% missing. All multivariate modeling was completed on the analysis data set containing the imputed values. Baseline characteristics were compared using the analysis data set with nonimputed values. Variables with missing >15% were dropped from the analysis.
Logistic regression was used to identify covariates associated with treatment, including sotalol versus no AAD and amiodarone versus no AAD. The results of these analyses were incorporated in inverse probability-weighted analyses of all-cause mortality. Adjusted cumulative incidence rates were determined using inverse probability-weighted estimators to limit the influence of selection bias in this nonrandomized retrospective comparison. Weighted Cox proportional hazards regression models and adjusted cumulative incidence curves were generated using the inverse probability weights. Covariates included the following: female gender, race, age, carotid bruits, history of congestive heart failure, New York Heart Association class, Charlson index, body mass index, heart rate, baseline ejection fraction, systolic blood pressure, diastolic blood pressure, hypertension, diabetes, hyperlipidemia, smoking history, third heard sound, number of diseased vessels, history of stroke and/or transient ischemic attack, history of peripheral vascular disease, history of myocardial infarction, glomerular filtration rate, and β blockers (other than sotalol). To further explore the association between AAD therapy and outcomes, we used landmark analysis at 60 days and 1 year.
To account for temporal changes in drug exposure (e.g., discontinuation and initiation after baseline), proportional hazards regression models using AAD usage as time-dependent covariates were also used to model all-cause mortality. The candidate list of covariates was the same as that described in the adjusted analyses using inverse-weighted estimators. There were 3 exposure groups: sotalol, amiodarone, or no antiarrhythmic (none). There were 2 time-dependent covariates: one for sotalol and one for amiodarone. Variables for medication use at baseline, 60 and 180 days, and 1, 2, 3, and 4 years were created.
All tests were 2 tailed, and statistical significance was declared at α <0.05. All analyses were performed using SAS software, version 8.2 (SAS Institute, Cary, North Carolina). The study protocol was reviewed and approved by the Duke University Medical Center Institutional Review Board.
Results
The study cohort included 2,838 patients with AF and obstructive CAD. Median follow-up was 4.2 years (IQR 2.0–7.4). The study cohort formation is shown in Figure 1 . Within this cohort, 8% of patients (n = 226) were treated with sotalol, 30.2% (n = 856) were treated with amiodarone, and 61.9% (n = 1,756) were not on an AAD at baseline. The baseline characteristics of the cohort are listed in Table 1 .
| Characteristic | Sotalol (n = 226) | Amiodarone (n = 856) | No Antiarrhythmic (n = 1756) | p-Value |
|---|---|---|---|---|
| Median follow-up (years) | 7.5 (5.0–10.0) | 4.0 (2.1–6.2) | 4.1 (1.6–8.1) | <0.001 |
| Age, median (IQR) (years) | 68 (61–74) | 67 (59–74) | 69 (61–75) | 0.003 |
| Female | 31% | 28% | 33.3% | 0.024 |
| Race | 0.007 | |||
| Black/African American | 6.2% | 13.4% | 12.8% | |
| Native American | 0.9% | 3.3% | 2.7% | |
| Other | 2.2% | 0.8% | 0.8% | |
| White | 90.7% | 82.5% | 83.8% | |
| CHADS 2 , median (IQR) | 2 (1–2) | 2 (1–2) | 2 (1–3) | 0.049 |
| Hypertension | 70.4% | 69.6% | 71.6% | 0.557 |
| Diabetes | 28.3% | 31.3% | 31.9% | 0.550 |
| Prior coronary revascularization | 39.8% | 32% | 36% | 0.040 |
| Prior myocardial infarction | 35% | 51.2% | 48.3% | <0.001 |
| Cerebrovascular disease | 13.7% | 11.8% | 15.5% | 0.036 |
| Peripheral vascular disease | 18.6% | 11.6% | 15.3% | 0.007 |
| Smoker | 61.9% | 52.6% | 56.8% | 0.02 |
| Chronic obstructive pulmonary disease | 11.9% | 6.2% | 9.5% | 0.004 |
| Diagnosis of hyperlipidemia | 57.1% | 56% | 58.7% | 0.415 |
| Charlson comorbidity (≥2) | 24.8% | 21% | 23.5% | 0.292 |
| Heart failure severity | 0.081 | |||
| None | 74.1% | 76.3% | 77.0% | |
| NYHA class I | 2.3% | 3.2% | 4.5% | |
| NYHA class II | 10% | 6.4% | 6.1% | |
| NYHA class III | 9.1% | 7.7% | 8.1% | |
| NYHA class IV | 4.5% | 6.4% | 4.4% | |
| No. of coronary vessels narrowed: | <0.001 | |||
| 1 | 36.7% | 27.3% | 40.5% | |
| 2 | 21.7% | 24.3% | 25.2% | |
| 3 | 41.6% | 48.4% | 34.4% | |
| Systolic BP, median (IQR) (mm Hg) | 147 (130–169) | 141 (124–163) | 143 (127–163) | 0.02 |
| Diastolic BP, median (IQR) (mm Hg) | 80 (70–88) | 78 (69–87) | 78 (68–87) | 0.512 |
| Body mass index, median (IQR) (kg/m 2 ) | 27 (25–30) | 28 (25–32) | 27 (24–31) | 0.007 |
| Carotid bruits | 11.5% | 8.3% | 14.7% | <0.001 |
| S3 gallop | 1.8% | 4.1% | 3.7% | 0.253 |
| LVEF, median (IQR) (%) | 55 (48–62) | 53 (40–62) | 55 (43–64) | <0.001 |
| <40% | 13% | 26.1% | 21% | <0.001 |
| Estimated GFR, median (IQR) (mL/min/1.73 m 2 ) | 73 (63–88) | 70 (55–86) | 69 (55–82) | 0.002 |
| Baseline medications | ||||
| Aspirin | 93.4% | 93.6% | 86.8% | <0.001 |
| ACE-I/ARB | 55.8% | 75.7% | 52.7% | <0.001 |
| Beta-blocker | 86.7% | 89.6% | 72.3% | <0.001 |
| Warfarin | 18.1% | 18.2% | 9.2% | <0.001 |
The median age was 68 years (IQR 60–75). Sotalol was used less frequently in African-Americans. The frequency of left ventricular dysfunction, defined as a left ventricular ejection fraction <40%, was lesser in those treated with sotalol (13%) versus amiodarone (26%) or no antiarrhythmic (21%, p <0.001). Despite these differences in left ventricular dysfunction, the frequency of symptomatic heart failure was relatively similar across the 3 treatment groups. Previous myocardial infarction was less common in those treated with sotalol, whereas chronic obstructive pulmonary disease and peripheral vascular disease were more frequent in those treated with sotalol. Overall renal function, as measured by estimated glomerular filtration rate, was best or most preserved in the sotalol group. The median noncardiac Charlson index of co-morbid illness was 1.0 (IQR 0–1) in all 3 treatment groups.
Figure 2 shows the utilization of antiarrhythmic therapy in follow-up. Long-term exposure was limited. Of those treated with sotalol, 97% were taking the drug for ≤25% of the follow-up period and only 1% was taking it for most of the follow-up period. A similar relation was observed with amiodarone. Finally, among those who were not taking an AAD at baseline, only 44% remained off AAD therapy for most of follow-up.
Mortality rates in follow-up according to antiarrhythmic use (sotalol, amiodarone, and no AAD) are listed in Table 2 . In unadjusted and adjusted settings, mortality rates were lower in patients treated with sotalol compared with amiodarone or no AAD. After adjustment for baseline characteristics only, the 1-year mortality rate was 10% in those treated with sotalol, 20% in those treated with amiodarone, and 14% in those treated with no AAD.
| Assessment Time-Point | Unadjusted Kaplan-Meier Rates | Adjusted Kaplan-Meier Rates | ||||
|---|---|---|---|---|---|---|
| Sotalol | Amiodarone | No AAD | Sotalol | Amiodarone | No AAD | |
| 60 days | 4.0% | 13.4% | 7.4% | 3.2% | 11.0% | 5.8% |
| 1 year | 12.0% | 23.4% | 16.8% | 10.0% | 20.0% | 13.8% |
| 3 years | 21.5% | 32.5% | 30.3% | 18.8% | 29.2% | 26.9% |
| 5 years | 35.1% | 39.0% | 40.9% | 32.5% | 36.5% | 38.8% |
To further explore the association between AAD therapy and outcomes, we used landmark analysis at 60 days and 1 year ( Table 3 ). After adjustment and weighting, sotalol was associated with improved survival from 0 to 60 days compared with amiodarone (hazard ratio [HR] 0.14, 95% confidence interval [CI] 0.06 to 0.32) but not at later time points (≥60 days or ≥1 year). Similarly, compared with no AAD therapy, sotalol was not associated with improved survival beyond 60 days. Figure 3 shows the cumulative survival after 1 year in patients treated with sotalol versus no AAD (p = 0.64). A sensitivity analysis using the same landmark methods, restricted to those with heart failure and/or a left ventricular ejection fraction ≤40%, demonstrated similar results ( Table 4 ).
| Sotalol vs Amiodarone | Sotalol vs No Antiarrhythmic | |||||
|---|---|---|---|---|---|---|
| 0–60 Days | ≥60 Days | ≥1 yr | 0–60 Days | ≥60 Days | ≥1 yr | |
| N = 625 Events = 125 | N = 489 Events = 246 | N = 341 Events = 127 | N = 2727 Events = 179 | N = 2524 Events = 1074 | N = 2201 Events = 879 | |
| Unadjusted | 0.17 (0.09–0.30) | 0.67 (0.51–0.89) | 0.65 (0.45–0.93) | 0.66 (0.48–0.91) | 0.67 (0.58–0.78) | 0.98 (0.78–1.23) |
| Adjusted | 0.12 (0.06–0.23) | 0.82 (0.59–1.14) | 0.76 (0.49–1.18) | 0.77 (0.55–1.08) | 0.76 (0.66–0.89) | 1.00 (0.80–1.26) |
| Adjusted & weighted | 0.14 (0.06–0.32) | 0.96 (0.69–1.34) | 0.92 (0.50–1.69) | 0.88 (0.59–1.31) | 0.77 (0.65–0.92) | 0.97 (0.76–1.24) |
Stay updated, free articles. Join our Telegram channel
Full access? Get Clinical Tree
