We retrospectively analysed 12-year follow-up data on 250 patients who were included in the randomized control trial, RASCABG [12]. Inclusion criteria of the original study were elective CABG, age older than 18 years, willingness to be randomized, and provision of informed consent. Exclusion criteria were other types of heart surgery, previous heart surgery, resting heart rate of less than 40 beats/min, atrial-ventricular blockage of any degree, known AF or atrial flutter, hepatic dysfunction (alanine aminotransferase level more than twice the upper normal limit), hyperthyroidism, pregnancy, breast-feeding, or known adverse reactions to amiodarone [12]. No adverse events were found in the primary study. In the current study, we furthermore excluded patients with protocol violation.

Baseline data were extracted from the original dataset and the Western Denmark Heart Registry [13]. Data on follow-up were obtained from the Danish Register of Causes of Death, the Western Denmark Heart Registry, and the Danish National Patient Registry [13] [14] [15]. Follow-up data included heart failure, late AF, readmission, stroke, long-term side effects and, long-term all-cause mortality and cardiovascular mortality. Unless specified otherwise, heart failure, late AF, readmission and stroke, will be referred to as HARS (Heart failure, late Atrial fibrillation, Readmission and Stroke). Late AF, stroke, heart failure, long-term side effects, long-term all-cause mortality and cardiovascular mortality were evaluated as binary variables defined by diagnostic codes, classified by the ICD-10 classification. Diagnostic codes were clustered into categories e.g. “cancer”, and subcategories e.g. “lung cancer”, and converted into variables. Diagnostic codes that only occurred once and could not be categorised with others, were excluded. For the variable of late AF, we included diagnostic codes for atrial flutter, AF and unspecified atrial flutter or fibrillation. Late AF was defined as atrial fibrillation during the follow-up period, occurring after the 30. postoperative day. Long-term mortality was divided into all-cause mortality, defined as death of any cause during the follow-up period, and cardiovascular mortality defined as death due to cardiovascular diseases. Readmission was evaluated as the number of patients readmitted during the follow-up period. Time to readmission and AF, were defined as the time interval from the surgery date, to the first readmission and AF during the follow-up period, respectively.

Data access and processing were approved by Danish Data Protection Agency (2017-41-5275) and the Danish Patient Safety Authority (3-3013-2374/1).

Numerical variables were tested for normality of distribution using the Shapiro-Wilk test, and reported as median and quartiles (interquartile range (IQR) from 25th to 75th quartile) or mean and standard deviation (SD). For categorical variables, the number and percentage of subjects in each category was reported. Baseline differences and differences in stroke, heart failure, readmission, late AF, long-term all-cause mortality and long-term side effects among groups were tested using Students t-test or Mann-Whitney U test, for numerical variables, and Pearson’s χ 2 test of independence or Fischer’s exact test for categorical variables.

We analysed prophylactic groups (intervention vs. placebo) and subgroups (POAF vs. sinus rhythm) separately.

Cox multivariable proportional hazard regression analysis, adjusted for potential confounders, was used to investigate the association between 1) amiodarone and readmission, late AF, long-term all-cause mortality and cardiovascular mortality 2) POAF and readmission, late AF, long-term all-cause mortality and cardiovascular mortality. Results were reported as hazard ratios (HR) with 95% confidence intervals (CI).

Kaplan-Meier survival curves were used to estimate long-term all-cause mortality for prophylactic groups and subgroups respectively, during the follow-up period. Survival time was determined from the date of surgery to date of event or 12 years, having survived the follow-up. Differences in mortality rates between prophylactic groups and subgroups were compared using Log-rank test.

The time-to-event data included time to first AF and time to the first readmission from the date of surgery. Cumulative incidence function (CIF) was used in a Fine and Gray Proportional Hazard Regression Model to investigate the time to event data, in the presence of competing events [16].

All p-values are two-tailed, and a p-value < 0.05 was considered as statistically significant. All the data were analysed using Stata Statistical Software 15.0 (StataCorp LLC, College Station, TX).

In the baseline characteristics, triple vessel disease was overrepresented in the amiodarone group, and the number of grafts used per case, as the use of two grafts was more frequent in the placebo group in Table 1. Patients who developed

POAF = Postoperative atrial fibrillation. IQR = Interquartile range. IHD = Ischemic heart disease. CNS = Central nerve system. COPD = Chronic obstructive pulmonary disease. NYHA = New York Heart Association. CCS = Canadian Cardiovascular Society grading of angina pectoris. MI = Myocardial infarction. PCI = Percutaneous coronary intervention. EF = Ejection fraction (%). BMI = Body Mass Index. LAD = Left anterior descending. CX = Circumflex artery. RCA = Right coronary artery. XCL time = Aortic Cross-clamp time. CPB time = Cardiopulmonary bypass time. LIMA = Left internal mammary artery. RIMA = Right internal mammary artery. ICU = Intensive Care Unit. **Three or fewer cases. * Significant p-value < 0.05. N/A = Not applicable.

POAF were older, received placebo, more likely to have higher EuroSCORE, and left main stenosis compared to patients who stayed in sinus rhythm. Further, patients with POAF had a longer stay in referring department.

Potential long-term side effects of amiodarone were evaluated based on registered diagnostic codes as described in the method section. When comparing the subcategories, we found no statistically significant difference between prophylactic groups. Furthermore, no statistically significant difference between prophylactic groups was found, when comparing the uncategorized diagnostic codes.

There were no statistically significant differences between prophylactic groups regarding HARS in Table 2. When comparing subgroups, though, patients who developed POAF, were more likely to develop late AF. None of the other variables of HARS were statistically significant different between subgroups. Among the 195 patients included in the follow-up analysis, 36% died during the 12-year follow-up. Of those, 39% died of cardiovascular diseases and 58% of others causes with no significant differences between the prophylactic groups. Cause of death was not registered in two patients (3%). Patients who developed POAF, had a higher long-term all-cause mortality compared with those who stayed in sinus rhythm in Table 2.

In both the uni- and multivariable Cox regression analysis, amiodarone was not statistically significant associated with an increased risk of readmission, late AF, long-term all-cause mortality or cardiovascular mortality in Table 3. In the univariable Cox regression analysis POAF was associated with a higher risk of late AF [95% CI 2.28 1.15; 4.56], long-term cardiovascular mortality HR 2.34 [95% CI 1.10; 5.00] and all-cause mortality HR 1.74 [95% CI 1.05; 2.88], although not significant in the adjusted analysis in Table 4. POAF was not associated with a higher risk of readmission in Table 4.

POAF = Postoperative atrial fibrillation. AF = Atrial fibrillation. *Significant p-value < 0.05.

HR = Hazard ratio. CI = Confidence interval. POAF = Postoperative atrial fibrillation. AF = Atrial fibrillation. ^aAdjusted for age, POAF and Vessels affected 3; ^bAdjusted for age and POAF; ^c Adjusted for POAF.

Despite non-violation of proportional hazard assumption, the survival curves for the prophylactic groups crossed more than once in Figure 2(a).

HR = Hazard ratio. CI = Confidence interval. POAF = Postoperative atrial fibrillation. AF = Atrial fibrillation. ^aAdjusted for age, EuroSCORE, Left main stenosis, and Length of stay in referring department; ^bAdjusted for age, EuroSCORE; ^cAdjusted for age.

Since the overall survival of our cohort was greater than 50% at the last time point, we were not able to calculate the median survival time. When comparing the survival curves of the prophylactic groups, we found no statistically significant difference in the overall long-term survival in Figure 2(a). However, we did find a decreased overall long-term survival, in patients who developed POAF in Figure 2(b).

We investigated the cumulative incidence function (CIF) of readmission and late AF with death as a possible competing risk within prophylactic groups (a) and subgroups (b), respectively in Figure 3 and Figure 4. No statistically significant differences in the risk of readmission were found. POAF was associated with an increased risk of late AF HR 2.28 [95% CI 1.15; 4.55], when accounting for death as a competing risk. Amiodarone was not associated with late AF.