POAF as a common phenomenon following CABG surgery is believed to increase stroke incidence [22,23,24]. According to European Society of Cardiology guidelines, long-term oral anticoagulation therapy is a reasonable treatment strategy for preventing stroke after POAF following cardiac surgery (Class of Recommendation IIb, Level of Evidence B) [8]. Additionally, the American Heart Association/American College of Cardiology guidelines recommend anticoagulation for POAF patients (Class of Recommendation IIa, Level of Evidence B) [9]. However, the risk of bleeding events has so far stunted this application from becoming the standard approach and the overall benefit or harm of oral anticoagulation following POAF remains uncertain [2, 25].

Our study examined the risk of bleeding in POAF patients after CABG surgery, following Warfarin or Rivaroxaban administration. Our main finding, the occurrence of bleeding events, was not statistically different between the 2 groups, and overall, it was only reported in 6% of the sample. This low prevalence has also been reported by similar studies examining the effects of anticoagulants on POAF patients [26], although other studies have found this ratio to be even lower [27, 28]. This discrepancy can be attributed to our lower sample size. The insignificant difference between bleeding events for Warfarin and Rivaroxaban (P value = 0.614) in our study, although in a smaller sample than the studies by Nauffal et al., or Patel et al., is nevertheless in agreement with their findings that bleeding events rates are similar between Warfarin and DOACs [27, 29].

The variability in bleeding rates among DOACs has been studied, revealing Rivaroxaban to exhibit the lowest safety profile [30]. In this study, we chose Rivaroxaban among other DOACs (e.g. Apixaban) as the intervention due to several reasons. Firstly, Rivaroxaban is administered as a single dose, making it simpler for patients compared to Apixaban, which requires twice-daily dosing. This may improve patient compliance with medication regimens.

Secondly, considering the challenges of drug availability and high costs in Iran, providing Rivaroxaban was more feasible within our research budget compared to Apixaban. This decision was made to ensure that the study intervention could be implemented effectively and efficiently within the resources available.

Overall, the selection of Rivaroxaban over other DOACs in this study was based on considerations of dosing convenience for patients and cost-effectiveness within the study context.

Our study detected comparable bleeding rates between Rivaroxaban and Warfarin. Nevertheless, additional research is imperative to corroborate these observations.

In this study, none of the patients experienced thromboembolic events, and this finding is congruent with a study by Butt et al., who found anticoagulant therapy reduces the risk of thromboembolic events in POAF patients [31]. Other studies on the effects of anticoagulants on POAF have also found similar results [26,27,28]. While we could not compare the protective effects of Rivaroxaban and Warfarin on thromboembolic events with each other due to the fortunate lack of events in either treatment arm, similar studies have found that Rivaroxaban is either more effective at reducing the incidence of CVAs [32] or similarly effective at CVA prevention [5].

While our study found no difference in the bleeding rates of Rivaroxaban and Warfarin, a study comparing these two in non-valvular atrial fibrillation patients found that patients receiving Rivaroxaban incur less bleeding [5]. Although our study does not fully corroborate this finding, the use of Rivaroxaban over Warfarin should still be favored, since Warfarin’s therapeutic effects may be altered by the use of over-the-counter medications, vitamin K deficiency, a narrow therapeutic index, and a constant need for routine INR monitoring. Rivaroxaban on the other hand is administered orally, does not have as many drug interactions, and isn’t dependent on food intake. It’s also important to note that among the few bleeding events we observed, they were all being administered at least one form of antiplatelet therapy along with either of the anticoagulants.

On a final note, the apparent LA enlargement in those who suffered atrial fibrillation in comparison to those who experienced normal sinus rhythms is as stated by existing literature, further highlighting the correlation between LA diameter and atrial fibrillation [33, 34]. We plan to follow up on this association in our future prospective studies to see the extent of this association and its predictive ability regarding atrial fibrillation.

One of the strengths of our study is that, unlike most prior research, our study was a prospective and randomized design. Prospective trials in this are lacking, due to the physicians’ reluctance regarding the difficult maintenance of Warfarin; however, because of the importance of this comparison of novel anticoagulants, we decided to perform a randomized trial to improve the existing literature foundation in decision-making, rather than choosing the medications based on convenience. Also, we observed no protocol violations, and our intent-to-treat analysis is the same as the per-treatment analysis.

One limitation of our study is that CABG surgery generally has the lowest risk of postoperative bleeding in comparison to other cardiac surgeries [35], therefore this work lacks generalizability to all cardiac surgeries, and further studies are warranted to compare anticoagulant-related bleeding events in other cardiac surgeries. Also, since our study lacks a placebo control group and most of the observed bleeding events were minor, we cannot state whether or not bleeding events are more common in those who take Warfarin or Rivaroxaban, in comparison to those who do not take any anticoagulants, yet it would be reasonable to assume a higher incidence among those taking any anticoagulants since a much larger study has found it to be almost 3 times more likely [27], although the risk of bleeding remains relatively low.

The lack of blinding introduces a potential source of bias in the study, as the awareness of treatment assignment could influence subjective outcomes, such as patient-reported symptoms or healthcare provider assessments. To mitigate the impact of unblinding on the study results, efforts were made to standardize outcome assessments and data collection procedures across treatment groups. Additionally, statistical analyses were conducted by researchers blinded to treatment allocation to minimize the influence of bias on data interpretation and conclusions.

While blinding is a fundamental methodological principle in clinical research to reduce bias and enhance the validity of study findings, the unique nature of the interventions in this study necessitated a pragmatic approach that prioritized patient safety and optimal clinical care.

The small sample size can be considered a limitation due to the reduced statistical power and potential difficulty in generalizing the findings to a larger population.

We performed echocardiography on all patients before surgery. However, due to some patients requiring emergency surgery and limited time, LA volume index measurement which is a more accurate method was not conducted on all patients. The parameter that was assessable in all patients in this regard was LA diameter in the parasternal long-axis view. This issue can be considered as a limitation.