Risk of bleeding with concomitant use of oral anticoagulants and aspirin: A systematic review and meta-analysis

Abstract

Purpose

Oral anticoagulants (OACs) and aspirin can trigger bleeding events when used alone or in combination. The purpose of this study was to compare the risk of any type of bleeding in individuals exposed to a combination of OAC and aspirin with the risk in those taking an OAC or aspirin alone.

Methods

MEDLINE and Web of Science were queried in January 2021 for eligible articles. Studies were included if they were either randomized controlled trials (RCTs) or observational studies and evaluated the number of any bleeding events in two groups, one with exposure to both OAC and aspirin and one with exposure to OAC alone or aspirin alone. Pooled odds ratios were calculated using a random-effects model.

Results

Forty-two studies were included. In an analysis of 15 RCTs and 19 observational studies evaluating OAC plus aspirin versus OAC alone, a significant difference in the risk of bleeding was observed in the combination groups, with an odds ratio [OR] of, 1.36 (95% CI, 1.15-1.59) for RCTs and an OR of 1.42 (95% CI-, 1.09-1.87) for observational studies. When OAC plus aspirin was compared to aspirin alone, a higher rate of bleeding was found in the combination group (OR, 2.36; 95%CI, 1.91-2.92) in the analysis of 15 RCTs, but no significant difference was found among 10 observational studies (OR, 1.93; 95% Cl, 0.99-3.75).

Conclusion

The risk of any type of bleeding was significantly increased among patients taking aspirin plus OAC compared to those taking OAC alone in both RCTs and observational studies. Evaluation of RCTs comparing OAC plus aspirin to aspirin alone suggests increased bleeding risk as well.

Warfarin, an oral anticoagulant (OAC), is prescribed for stroke prevention in atrial fibrillation and for treatment and secondary prevention of venous thromboembolism.¹ Direct oral anticoagulants (DOACs) were introduced into the market in 2010.² The first DOAC approved by the US Food and Drug Administration (FDA) was dabigatran in October 2010. Following the approval of dabigatran, other DOACs like rivaroxaban, apixaban, and edoxaban were subsequently approved by FDA. DOACs are advantageous compared to warfarin due to fewer monitoring requirements and simpler dosing.³ However, all OACs increase the risk of bleeding events when taken individually or in combination with other medications that also increase bleeding risk.⁴

Aspirin is widely prescribed to prevent primary cardiovascular events⁵ and colorectal cancer (an off-label use).⁶ Aspirin monotherapy, however, is also associated with a risk of bleeding.⁷ The 2022 US Preventive Services Taskforce (USPSTF) recommendations state that aspirin can be used for the primary prevention of cardiovascular disease (CVD) in adults 40 to 59 years of age who are at high risk for a cardiovascular event and who are not thought to have a high risk of bleeding; for adults 60 years of age or older, the task force recommends against prescribing aspirin.⁵ Although this guideline states that aspirin is advisable in some patients younger than 60, various studies have shown that even younger patients can experience bleeding events while taking aspirin monotherapy.^7,8

Polypharmacy is a significant issue for many patients. When medications have interactions, the likelihood that patients will experience an adverse drug event is higher.^9,10 Historically, OACs and aspirin were used concomitantly to manage various clotting and cardiovascular conditions. Although older guidelines^11-13 recommended the concomitant use of aspirin and OACs for certain indications, most current guidelines^14,15 recommend against long-term concomitant use of OAC and aspirin due to the additive risk of bleeding. Numerous studies evaluating the concomitant use of OAC and aspirin have been conducted. Evidence for DOACs is more limited, but a randomized controlled trial (RCT) comparing rivaroxaban plus aspirin to aspirin alone showed an increase in the risk of bleeding (hazard ratio [HR], 1.70; 95% CI, 1.40-2.05).¹⁶ A retrospective case-control study found that the risk of gastrointestinal bleeding was significantly greater in patients taking warfarin plus aspirin than in patients taking each drug alone, with an odds ratio (OR) of 6.5 (95% CI, 4.3-9.9).¹⁷ Despite the numerous studies showing the risk of bleeding, in addition to the guideline recommendations, long-term concomitant use of aspirin and OACs persists.^18,19 To our knowledge, there has been no systematic review and meta-analysis to evaluate the risk of any type of bleeding with concurrent OAC and aspirin use versus use of OAC or aspirin alone across all indications, although there are several such studies focusing on specific indications.^20-25

Therefore, the purpose of this study was to conduct a systematic review and meta-analysis to examine the risk of any bleeding in individuals exposed to a combination of OAC and aspirin compared to those taking OAC or aspirin alone in all patient populations.

Methods

This systematic review and meta-analysis were reported in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) statement (eTable 1).²⁶

Search strategy.

We conducted a search of abstracting databases (from database inception to January 2021) in MEDLINE and Web of Science to identify relevant articles. The search strategy included terms related to OACs, salicylates, and bleeding. Search phrases used included “aspirin, bismuth subsalicylate, salicylate, methyl salicylate, trolamine salicylate, phenyl salicylate, potassium salicylate, sodium aminosalicylate, or sodium salicylate AND warfarin, apixaban, dabigatran, edoxaban, or rivaroxaban AND gastrointestinal (GI) bleeding, gastrointestinal hemorrhage, GI hemorrhage, intracranial bleeding, hemorrhage, bleeding, interaction, drug interaction, or drug-drug interaction” (eTable 2). No restriction on publication date was applied. No language restriction was used. Searches were conducted by clinical pharmacists with the assistance of a librarian and reviewed by a third researcher with experience in conducting meta-analyses.

Study selection.

After studies were identified for potential inclusion, titles and abstracts were screened, followed by full text review of relevant articles. Two researchers (P.G. and J.P.) screened potential studies, with the final decision made in consultation with a third researcher (D.C.M.) in case of disagreement. Studies were included if they evaluated the number of any bleeding events in two groups, one with exposure to both an OAC and aspirin and one with exposure to an OAC alone or aspirin alone. Both RCTs and observational studies were eligible to be included. Studies were excluded if they did not report the number of bleeding events in each group or if patients in the aspirin arm were treated with dual antiplatelet therapy. Articles were also excluded if they were news or review articles, case series, case reports, meta-analyses, double-reporting studies, or studies conducted on animals.

Data extraction and quality assessment.

A standardized data extraction form was used to gather data including citation, year of publication, study design (RCT or observational study) and follow-up duration, OAC and aspirin (or salicylate) dose, indication for OAC, mean age of participants, number of participants in each group and corresponding bleeding events, bleeding events by location and severity, target international normalized ratio (INR) for warfarin-treated patients, other endpoints studied, and author conclusions. To assess the risk of bias for RCTs, the Cochrane Risk of Bias Tool for Randomized Trials (RoB 2), which includes 5 domains, was used.²⁷ The Cochrane Risk of Bias In Non-randomized Studies of Interventions (ROBINS-I) tool, which includes 7 domains, was used to assess the risk of bias for observational studies.²⁸ Risk of bias in each domain was categorized as low, moderate, or serious based on the quality of each study.

Statistical analysis.

The primary outcome of interest was the OR for any bleeding. A meta-analysis was conducted to pool ORs across individual studies to estimate the overall risk of any bleeding. We calculated a pooled OR using random effects based on the DerSimonian and Laird methods.²⁹ Study heterogeneity was assessed using I² statistics and Cochran’s Q. Meta-regressions were performed to explore the source of heterogeneity. Specifically, if I² decreased by more than 50%, a given covariate was deemed a potential source of heterogeneity. Small-study effects were assessed using Egger’s test and a funnel plot. In addition, we performed a series of subgroup analyses: (1) “serious” bleeding as defined by each article, (2) gastrointestinal (GI) bleeding, and (3) indication for OAC. Analyses were performed using Microsoft Excel (Microsoft Corportation, Redmond, WA) and Stata version 17.0 (StataCorp LLC, College Station, TX), and a 2-sided statistical test with a P value of ≤0.05 was considered statistically significant.

Results

Study characteristics.

A total of 3,050 studies were identified from MEDLINE and Web of Science databases. A total of 2,541 studies were excluded based on title and abstract screening, with another 467 studies excluded after full-text review, resulting in inclusion of 42 studies in the meta-analysis (Figure 1). Eighteen of the included studies^30-47 were observational cohort studies, while 24 studies^16,48-70 were RCTs. Characteristics of the studies are summarized in Tables 1 through 4. One study was international,¹⁶ while the other studies were conducted in countries including Australia, China, Iran, Japan, Turkey, and the United States. Included studies were published between the years of 1982 and 2021. The durations of the RCTs and observational studies ranged from 18 to 60 months and 16.5 to 120 months, respectively. The OAC in most included studies (n = 38) was warfarin, with one study including all OACs,³⁶ one study examining direct-acting oral anticoagulants in general,⁴⁴ and 2 studies examining rivaroxaban.^16,70 The majority of studies (n = 32) reported the daily dose of aspirin, while 11 did not.^{32-34,36,38,40-45} Among the studies that reported the dose of aspirin, 22 studies^{16,31,35,37,39,46,49,50,52-54,58-61,63-67,69,70} reported a daily dose of 75 to 162 mg. For two studies,^52,60 the dose varied depending on whether the patient was also taking an anticoagulant, with a lower dose (80 or 81 mg/day) for patients on aspirin plus an anticoagulant versus patients taking aspirin alone (160 or 162 mg/day). Doses in the other 9 studies^{30,47,48,51,55-57,62,68} ranged from 200 mg to 600 mg. Among studies reporting the mean age of included patients (n = 36), the range was 35 to 75 years.^{16,30,31,33-37,39,41-43,45,47,49,50,52-70} The sample sizes ranged from 78 to 206,118 patients. The most common indications for OAC use were heart valve replacement (n = 12), atrial fibrillation (n = 11),^{34,36-40,44,45,51,55,68} myocardial infarction (n = 9),^{31,50,52,58,60,61,63-65} and coronary artery disease (n = 3)^16,70; two studies^32,41 did not report OAC indication. Ten studies^{16,31,38-40,48,49,54,61,62} reported GI bleeding, while 13 studies^{16,31,38,39,41,48,51,55,60-62,67,69} reported intracranial hemorrhage. Twenty-one studies reported the severity of bleeding events.^{16,31,36,43,49-55,57-65,70} The location and severity of bleeding events varied from study to study, with 4 not defining the bleeding events included at all^35,41,67,69; other studies had a clear method of describing the bleeding event, such as using the International Society on Thrombosis and Haemostasis criteria for major bleeding^16,36-38 or setting a threshold of needing a certain number of transfusions due to the bleeding event.^{33,43,45,46,48,57} Minor bleeding episodes were not life-threatening.

Bleeding events.

OAC plus aspirin vs OAC alone. In the meta-analysis of 15 RCTs^{16,48,49,51,53-55,57,59,61,63,66-69} evaluating OAC plus aspirin versus OAC alone, a significant difference in bleeding risk was found (OR, 1.36; 95% CI, 1.15-1.59) (Figure 2). Of note, the study of Chesebro et al⁴⁸ reported statistically significant findings and the highest OR, possibly due to the high daily dose of aspirin (500 mg/day), while the study of Pengo et al.⁶⁶ reported a wide OR confidence interval crossing 1 (nonsignificant) but was a small study (n = 198) and the INR target differed between the 2 study groups (2.5 for the warfarin plus aspirin group vs 3.7 for the warfarin monotherapy group) (Table 1, Figure 2). Additionally, the number needed to harm (NNH) was calculated for the 15 RCTs. Approximately 36 patients would have to receive a combination of OAC and aspirin instead of OAC alone for 1 bleeding event to occur.

Similarly, in the meta-analysis of 19 observational studies^30-46 comparing OAC plus aspirin with OAC alone, a significant bleeding risk increase was also observed (OR, 1.42; 95% CI, 1.09-1.87) (Figure 3). Zhang et al⁴⁶ reported a very high OR for bleeding (16.5; 95% CI, 5.9-46.18), likely due to the small sample size in the combination arm, while Gulati et al⁴¹ observed a protective effect of combined therapy (OR, 0.44; 95% CI, 0.38-0.50), possibly because the dose of aspirin varied among patients in the study (Table 2, Figure 3).

OAC plus aspirin vs aspirin alone.

A more than 2-fold increase in the odds of bleeding events was found in an analysis of 15 RCTs^{16,49,50,52,54-56,58,60-65,70} for the combination of an OAC and aspirin as compared to aspirin alone (OR, 2.36; 95% CI, 1.91-2.92) (Figure 4), but a nonsignificant difference was found for the 10 observational studies^{30,32,33,35,37,40-42,46,47} (OR, 1.93; 95% CI, 0.99-3.75) (Figure 5). In the RCTs for which NNH was calculated, approximately 16 patients would have to receive a combination of OAC and aspirin instead of aspirin alone for one bleeding event to occur. Among the RCTs, an outlier with a high OR was the study of Zibaeenezhad et al⁶⁵ (OR, 5.71; 95% CI, 1.82-17.91), likely because bleeding events of any severity were measured, whereas some studies focused only on major bleeding events (Table 3, Figure 4). Among the observational studies, the study of Gulati et al⁴¹ observed a protective effect of combined therapy for all 3 evaluated drugs (warfarin, rivaroxaban, and dabigatran), possibly because the both OAC and aspirin doses varied across studies and the only bleeding event reported was intracranial bleeding. The study of Zhang et al⁴⁶ was again an extreme outlier (OR, 185.50; 95% CI, 46.22-744.48) among the observational studies, possibly due to the low sample size of the group on aspirin alone (n = 95) versus the warfarin plus aspirin group (n = 395) (Table 4, Figure 5).

Subgroup analyses.

Serious and GI bleeding with OAC plus aspirin vs OAC alone. We examined serious and GI bleeding based on data reported in RCT publications. In 9 RCTs^{16,49,53-55,57,59,61,63} reporting serious bleeding, no significant difference between the 2 groups was found (OR, 0.98; 95% CI, 0.83-1.15) (eFigure 1). In a subgroup analysis of 5 RCTs^{16,48,49,54,61} reporting GI bleeding, the difference between groups was also not statistically significant (OR, 1.31; 95% CI, 0.93-1.85) (eFigure 2).

Indication for OAC:

OAC plus aspirin vs OAC alone. Additionally, we examined indications for OAC in a subgroup analysis. In 7 RCTs^{48,53,57,59,66,67,69} in which valve replacement was the indication for OAC, no significant difference in bleeding risk between the 2 groups was found (OR, 1.38; 95% CI, 0.86-2.21) (eFigure 3). In 5 RCTs^{16,49,54,61,63} where the indication for OAC was ischemic heart disease, myocardial infarction, or coronary heart disease, there was no significant difference in bleeding (OR, 1.42; 95% CI, 1.17-1.73) (eFigure 4).

Serious and GI bleeding:

OAC plus aspirin vs aspirin alone. In the analysis of 14 RCTs that reported serious bleeding,^{16,49,50,52,54,55,58,60-65,70} comparison of OAC plus aspirin to aspirin alone found no significant difference in bleeding risk (OR, 0.94; 95% CI, 0.70-1.25) (eFigure 5). In 5 RCTs^{16,49,54,61,62} reporting GI bleeding, no significant risk difference was found with use of OAC plus aspirin versus aspirin alone (OR, 0.89; 95% CI, 0.59-1.35) (eFigure 6).

Few observational studies reported the severity of bleeding or the indication for OAC use; therefore, it was not possible to conduct subgroup analyses on bleeding severity or OAC indication for these studies.

Heterogeneity.

There was evidence of heterogeneity across the analyses we conducted. In the analysis of 15 RCTs^{16,48,49,51,53-55,57,59,61,63,66-69} for OAC plus concomitant aspirin versus OAC alone and 15 RCTs^{16,49,50,52,54-56,58,60-65,70} for OAC plus concomitant aspirin versus aspirin alone, I² was 65.3% (Figure 2) and 85.8% (Figure 4), respectively, indicating moderate to high heterogeneity. We further explored reasons for heterogeneity by performing a meta-regression for aspirin dose for those studies reporting this information. However, aspirin dose was not a potential source of heterogeneity (P = 0.908 for OAC plus aspirin vs OAC, P = 0.249 for OAC plus aspirin vs aspirin). Thus, other potential sources of heterogeneity were likely present. Our analysis of 19 observational studies^30-46 comparing OAC plus aspirin with OAC alone had an I² of 95.6% (Figure 3), indicating a high degree of heterogeneity. When OAC plus aspirin was compared to aspirin alone as the referent group, I² was 96.9% (Figure 5), again indicating a high amount of heterogeneity.^{30,32,33,35,37,40-42,46,47} However, we could not perform meta-regression for aspirin dose, as most studies (n = 11) did not report this information. The reasons for high heterogeneity in observational studies may be methodological differences in the conduct of the study, the extended timeframe for study inclusion (ranging from 1983 to 2019), the lack of data on aspirin dose, the varying definitions of bleeding, and the limited quality of data in observational studies.

Visual inspection of funnel plots of 15 RCTs^{16,48,49,51,53-55,57,59,61,63,66-69} for OAC plus aspirin versus OAC alone (eFigure 7) showed a modest degree of asymmetry with respect to effect size and sample size. However, further exploration of small-study effects using Egger’s test found no evidence of significant small-study effect (P = 0.44). When funnel plots of 15 RCTs^{16,49,50,52,54-56,58,60-65,70} for OAC plus aspirin compared to aspirin alone (eFigure 8) were inspected, appearance of asymmetry was present, but Egger’s test found no evidence of a significant small-study effect (P = 0.43). Funnel plots of 19 observational studies^30–46 comparing OAC plus aspirin with OAC alone (eFigure 9) and 10 observational studies^{30,32,33,35,37,40-42,46,47} comparing OAC plus aspirin with aspirin alone (eFigure 10) also had asymmetry. Again, Egger’s test found no evidence of a significant small-study effect for either study subset. Thus, no statistically significant evidence of publication bias was observed in either RCTs or observational studies.

Risk of bias assessment.

For the RCTs, most studies had a similar risk of bias profile for the item randomization process (eFigure 11). Most of these RCTs were not double-blind (ie, both the patients and outcome assessors were not blinded). Two studies, those of Casais et al⁵⁹ and Zibaeenezhad et al,⁶⁵ had a high risk of bias for item measurement of the outcome, as the studies did not report a detailed description of how outcomes were measured.

Most observational studies had an overall low risk of bias (eFigure 12). For studies published before 2012, there was a moderate risk of bias in the domain of confounding, while 2 studies, those of Holden et al³³ and Zhang et al,⁴⁶ had serious risk of bias.

Discussion

This study found that the combination of OAC plus aspirin conferred a higher risk of bleeding events than either OAC or aspirin monotherapy. These findings were consistent across RCTs and observational studies, except when observational studies compared aspirin plus OAC to aspirin alone. When OAC plus aspirin was compared to OAC monotherapy, a significantly higher risk of bleeding was found in the combination group in both RCTs (OR, 1.36) and observational studies (OR, 1.42). Similarly, when OAC plus aspirin was compared to aspirin monotherapy, a significantly higher risk of bleeding was observed in the combination group in RCTs (OR, 2.36). In light of these findings, providers and patients receiving OACs should be cognizant of this risk while using aspirin concomitantly. The findings from this analysis provide evidence that can guide healthcare practitioners in their decision-making and help them counsel patients effectively. However, it is important to note that individual patient circumstances may vary. In certain cases, the potential benefits of concomitant OAC and aspirin use may outweigh the risks. Therefore, it is essential to consider each patient’s specific clinical situation, including their underlying medical conditions and the indications for OAC and aspirin therapy.

Avoiding the use of aspirin is challenging, and even when taken alone aspirin can increase the risk of bleeding relative to the risk in patients not taking aspirin.^7,71 Aspirin has been a widely promoted therapy and widely used for many indications, including analgesia, fever, prevention and treatment of atherosclerotic CVD, prevention of colorectal cancer, and prevention of venous thromboembolism.⁷²

Several guidelines address the risks posed by this combination and propose management strategies if the interacting medications must be prescribed concomitantly. The 2020 American College of Cardiology (ACC) expert consensus pathway for anticoagulant and antiplatelet therapy recommends against the concomitant use of aspirin and OAC in patients taking an anticoagulant at baseline who undergo percutaneous coronary intervention (PCI) or who are diagnosed with atherosclerotic CVD, recommending instead the use of a single P2Y₁₂ inhibitor with an OAC rather than “triple therapy” (dual antiplatelet therapy plus an OAC) in most patients due to the increased risk of bleeding.¹⁵ If triple therapy is deemed necessary, it should be limited to less than 30 days’ duration immediately following PCI, when patients are at highest risk for stent thrombosis. In addition, this guideline recommends the daily use of a proton pump inhibitor and avoidance of other nonsteroidal anti-inflammatory drugs in patients taking multiple antithrombotic agents. The ACC/American Heart Association guideline on atrial fibrillation recommends against the use of aspirin to treat atrial fibrillation and recommends that triple therapy should be avoided in patients undergoing PCI, regardless of the indication for aspirin.¹⁴ Similarly, an American College of Chest Physicians (ACCP) guideline from 2012 recommends the use of concomitant aspirin and OAC only when the benefit is known or highly likely to be greater than the harm from bleeding, with the provided example of an appropriate use scenario being recent acute coronary syndrome.⁷³ In a 2016 update to the ACCP guidelines, a recommendation to avoid the use of aspirin and OAC concomitantly for the indication of deep vein thrombosis prophylaxis was added.⁷⁴

Another guideline discusses the issue of whether aspirin is indicated for certain subgroups of patients. The USPSTF recommends against the use of aspirin for the primary prevention of colorectal cancer and CVD in patients 40 to 59 years of age unless the patient has a 10% or greater 10-year CVD risk, is not at increased risk for bleeding, has a life expectancy of greater than 10 years, and is willing to take aspirin for at least 10 years. There was insufficient evidence to provide a recommendation for patients younger than 60 years of age.⁵

This study has some limitations that should be considered when interpreting the findings. First is a lack of evidence for DOACs and aspirin use; the majority of included studies (n = 38) focused on warfarin as the anticoagulant. DOACs are relatively newer drugs and have not been evaluated in the context of concurrent aspirin use. Furthermore, several studies involving warfarin were conducted decades ago. In addition, slightly over 40% of the studies were observational studies, a weaker study design than the RCT. There was high heterogeneity observed across the studies, specifically in observational studies. However, this quantitative information of the risk of bleeding will be valuable for clinicians to counsel patients, especially concerning use of nonprescription aspirin. Another limitation is that the definitions of bleeding events varied across studies, with some studies focusing on only major events while others counted bleeding events of any severity level. Similarly, the dose of aspirin varied greatly across studies, from as low as 75 mg per day to as high as 600 mg per day. These limitations can explain in part the moderate to high level of heterogeneity observed across included studies.

Conclusion

Taking an OAC plus aspirin increases the odds of bleeding compared to monotherapy with either agent and should be avoided unless the benefits clearly outweigh the risks, such as after an acute coronary syndrome event. The results for increased risk of bleeding were consistent across RCTs, and the findings were similar for observational studies with an OAC alone comparator group. The evaluation of observational studies comparing use of OAC plus aspirin versus aspirin alone did not indicate a significant difference in bleeding risk.

Data availability

The data underlying this article are available in the article and its online supplementary material.

Disclosures

This project was supported by grant R01HS025984 from the Agency for Healthcare Research and Quality (AHRQ). The content is solely the responsibility of the authors and does not necessarily represent the official views of AHRQ. The authors have declared no potential conflicts of interest.

Additional information

The authors submitted the following information on author contributions: concept and design, Drs. Malone, Panic, and Ghule; identification of relevant studies, Drs. Ghule and Panic; statistical analysis. Dr. Ghule; interpretation of results, Drs. Malone, Ghule, and Panic; drafting of manuscript, Drs. Ghule, Panic, and Malone.

References