What's new?

In patients with chronic coronary artery disease (CAD) requiring long‑term oral anticoagulation (OAC), the optimal long‑term antithrombotic regimen remains debated. This meta‑analysis of 5 randomized trials, including the recent European AQUATIC trial, shows that OAC monotherapy reduces cardiovascular mortality and major bleeding, as compared with OAC combined with a single antiplatelet agent, without increasing the incidence of nonfatal myocardial infarction or ischemic stroke over median follow‑up of 23 months. Bayesian and trial‑sequential analyses confirm that the evidence for cardiovascular mortality reduction is statistically robust. The reduction in cardiovascular mortality and major bleeding, without excess nonfatal thrombotic events, generally supports OAC alone as the preferred long‑term antithrombotic strategy for this patient population.

Introduction

Long‑term prescription of full‑dose oral anticoagulation (OAC) with an antiplatelet agent is not uncommon in clinical practice. Such dual antithrombotic therapy typically involves patients with atrial fibrillation (AF) and chronic coronary artery disease (CAD). Despite growing evidence suggesting that less is more,^1,2 the recent AQUATIC trial (Aspirin in Patients with Chronic Coronary Syndrome Receiving Oral Anticoagulation) reported that 68% of patients receiving long‑term OAC, with a coronary stent implantation more than 6 months before, were receiving additional single antiplatelet therapy (SAPT) at trial enrollment.³

Combining OAC (to prevent systemic thromboembolism) and an antiplatelet agent (to mitigate atherothrombosis) has a rational basis. However, this strategy predictably increases bleeding without necessarily yielding ischemic protection after the acute or early‑intervention period.^3-7 In the above settings, the effect of OAC alone vs OAC+SAPT on cardiovascular mortality remains uncertain. Individual trials^3-7 and meta‑analyses⁸ have so far been underpowered for cardiovascular death, given low event rates, wide CIs, and noninclusion of the most recent evidence.

We performed a meta‑analysis of all available randomized controlled trials (RCTs) comparing OAC monotherapy against OAC+SAPT (as of August 2025), prespecifying cardiovascular mortality as the primary end point. Secondary end points included nonfatal ischemic or bleeding events and all‑cause death. To assess robustness and address residual uncertainty, we applied both frequentist and Bayesian random‑effects models.

Methods

Established methods recommended by the PRISMA statement were used.^9,10 A systematic literature search of PubMed, Embase, and the Cochrane Central Register of Controlled Trials was conducted from inception through August 2025. The full electronic search string for PubMed, and the corresponding search combinations for Embase and Cochrane base, included the key words and MeSH/Emtree terms of “oral anticoagulants,” “coronary artery disease,” “atrial fibrillation,” “antiplatelet therapy,” and “randomized trial.” References of the included studies and reviews were searched for additional eligible trials. Two independent reviewers (EPN and FA) screened the titles, abstracts, and full‑text articles for eligibility. The studies were included when CAD patients with established indications for anticoagulation were randomly allocated to either OAC alone or OAC+SAPT. Observational studies or those comparing triple vs dual antithrombotic therapy were not included. No language restrictions were applied. The meta‑analysis is registered in the PROSPERO database (CRD420251167661).

Results

Primary outcome of cardiovascular death

Cardiovascular mortality data were available in all trials. This outcome occurred in 68 of 2488 patients in the OAC alone group vs 94 of 2476 patients in the OAC+SAPT group (2.7% vs 3.8%; I² = 22.8%; HR, 0.69; 95% CI, 0.5–0.96; P = 0.03; Figure 1A). The Bayesian meta‑analysis confirmed a significant cardiovascular mortality reduction with OAC vs OAC+SAPT (HR, 0.75; 95% CrI, 0.61–0.93; Figure 1B). The TSA showed that the sample size of 4964 patients exceeded the required information size (approximately 4384 patients), indicating sufficient accrued evidence to demonstrate a cardiovascular mortality reduction with OAC alone (Figure 1C). The Bayesian probability density plot for the absolute difference indicated a 98.2% posterior probability of OAC alone reducing cardiovascular mortality, as compared with OAC+SAPT (Supplementary material, Figure S2).

Secondary outcomes

All‑cause mortality data were available in all trials. This outcome occurred in 123 of 2488 patients in the OAC arm and in 166 of 2476 patients in the OAC+SAPT arm, a large numerical difference that, however, did not achieve significance (4.9% vs 6.7%; I² = 59.7%; HR, 0.79; 95% CI, 0.5–1.24; P = 0.3; Figure 1D).

Data on major bleeding—defined in all trials according to the International Society on Thrombosis and Hemostasis—were available in all cases. Major bleeding occurred in 83 of 2488 patients in the OAC arm and in 160 of 2476 patients in the OAC+SAPT arm (3.3% vs 6.4%; I² = 43.9%; HR, 0.51; 95% CI, 0.39–0.66; P <⁠0.001; Figure 2A).

Nonfatal MI data were available in all trials. This outcome occurred in 30 of 2488 patients in the OAC arm and in 27 of 2476 patients in the OAC+SAPT arm, with a small difference (1.2% vs 1.1%; I² = 7.9%; HR, 1.06; 95% CI, 0.62–1.81; P = 0.84; Figure 2B).

Nonfatal IS data were available in all trials. This outcome occurred in 49 of 2488 patients in the OAC arm vs 49 of 2476 patients in the OAC+SAPT arm (1.9% vs 1.9%; I² = 0%; HR, 1; 95% CI, 0.68–1.49; P = 0.98; Figure 2C).

Discussion

This meta‑analysis is the first systematic evaluation to show reduced cardiovascular mortality, enhanced safety in terms of major bleeding, and no excess in major nonfatal ischemic events with OAC vs OAC+SAPT in patients with chronic CAD and a concomitant condition requiring long‑term anticoagulation. It differs from a previous work⁸ that summarized 4 randomized trials, by incorporating the more recent AQUATIC trial,³ by focusing a priori on cardiovascular mortality as the primary end point with time‑to‑event analyses, and by complementing conventional random‑effects pooling with the Bayesian and trial‑sequential methods, thereby narrowing the CIs for cardiovascular death and major bleeding, while confirming neutral effects on nonfatal MI and IS.

The reduction in cardiovascular mortality with OAC monotherapy is compelling, yielding a 31% relative risk reduction and an exploratory ARD of 1.1 percentage points over a median 2‑year follow‑up, resulting in a number needed to treat (NNT) of 94 to prevent 1 cardiovascular death (Figure 3). The findings, derived from the frequentist analyses and corroborated by the complementary Bayesian modeling to account for residual uncertainty, support the conclusions from recent trials that were underpowered to detect cardiovascular mortality as an individual end point. The concordance of the results across analytic paradigms increases confidence that the observed effect is not model‑dependent but reflects a true underlying effect.¹³ The Bayesian meta‑analysis indicates a 98.2% posterior probability that OAC alone reduces cardiovascular mortality, as compared with OAC+SAPT, complementing conventional CIs and providing useful information for patient‑centered discussions.¹⁴ The TSA further strengthened the validity of the finding. The cumulative Z‑score crossing the O’Brien–Fleming efficacy boundary provided evidence that sufficient statistical power was achieved to detect the observed treatment effect. TSA confirmation is particularly important in meta‑analyses, as it addresses concerns about random error and risks of false‑positive findings due to repeated testing across multiple studies.¹⁵ In the present analysis, TSA also addressed concerns about premature trial discontinuation. With the information‑size requirement exceeded, the conclusion that OAC monotherapy lowers cardiovascular mortality is unlikely to be significantly altered by further trials, supporting its translation into clinical practice and regulatory guidance.

The mechanism underlying the cardiovascular survival benefit with OAC alone appears multifactorial but is likely primarily driven by the substantial reduction in major bleeding observed in our analysis. The ARD of 3.1 percentage points for major bleeding—corresponding to 1 major bleeding event prevented for every 32 patients treated over the 2‑year period (Figure 3)—represents a substantial and meaningful reduction in bleeding risk across the trials. Major bleeding, especially intracranial and gastrointestinal, has consistently been associated with markedly increased short- and long‑term mortality across cardiovascular patient populations.^16,17 Major bleeding may be directly fatal or—by triggering antithrombotic therapy discontinuation— may lead to fatal ischemic complications, supporting the concept that bleeding prevention translates directly into survival benefit.

The cardiovascular mortality benefit with OAC alone was achieved without any tradeoff in major nonfatal thrombotic events. The absence of significant differences in nonfatal MI (ARD, 0.1 percentage points) or nonfatal IS (ARD, 0 percentage points) rates between treatment strategies suggests that anticoagulant therapy alone provides adequate protection against both cardioembolic and atherothrombotic events in most patients (Figure 3). Indeed, historical evidence, accrued before the coronary intervention era, had demonstrated survival benefit of vitamin K antagonists over placebo,¹⁸ and comparable protection of OAC vs OAC+SAPT,² in patients with previous MI. Importantly, the present data extend this concept to include DOACs.

Although the point estimate for all‑cause death favored OAC alone vs OAC+SAPT, the difference was not significant. The lack of significance may be related, at least in part, to competing causes of noncardiovascular deaths in older populations (eg, cancer, infections, frailty‑related cause, which are not uncommon in older patients and have grown over recent years). Competing risks of death can dilute the impact on all‑cause death of treatments directed specifically at cardiovascular mechanisms, blunting the observed differences and inflating the required sample size to detect a meaningful difference in all‑cause death rates.¹⁹ In comparison with a prior meta‑analysis,²⁰ this one was prospectively centered on cardiovascular mortality as the primary end point and analyzed it as a time‑to‑event outcome using HRs rather than risk ratios (RRs). HRs leverage the full follow‑up, appropriately handling censoring and differing event times, whereas simple RRs on accrued counts can be biased when follow‑up durations vary or when early event timing is prognostically informative. This approach is consistent with established guidance for survival end points in meta‑analyses.²¹ The clinical relevance of modeling event time is underscored by the AFIRE⁴ and AQUATIC³ trials, which were stopped early for excess deaths in the antiplatelet arm during follow‑up—an example where early hazard separation matters and HRs are the appropriate effect measure. Of note, the inclusion of the AQUATIC trial,³ the largest European trial in this setting, importantly extended the previously entirely East‑Asian evidence base, and supported generalizability of our findings to European patients, despite differences in DOAC dosing patterns.

In the included RCTs, SAPT composition varied substantially across trials (Table 1): in AQUATIC trial,³ the dual‑therapy arm used aspirin 100 mg daily in all patients; in the AFIRE,⁴ EPIC‑CAD,⁵ and OAC‑ALONE⁶ studies, the OAC+SAPT arm received aspirin as the predominant agent, with clopidogrel (or other P2Y₁₂ inhibitor) in a minority, whereas in the PRAEDO AF trial,⁷ the antiplatelet agent was clopidogrel. Despite this heterogeneity, the reduction in cardiovascular death and nonfatal major bleeding with OAC monotherapy was directionally consistent across SAPT regimens, and remained evident when the analysis was restricted to aspirin‑only dual‑therapy in the AQUATIC trial.³ These findings indicate that the cardiovascular death– and bleeding‑related benefits of OAC monotherapy are not confined to a specific SAPT agent. From an absolute‑risk perspective, cardiovascular death occurred in 2.7% of the patients assigned to OAC alone and 3.8% of those receiving OAC+SAPT, while major bleeding occurred in 3.3% and 6.4% of the patients, respectively. By contrast, the absolute risks of MI and IS were virtually identical in both groups (ARDs close to 0), indicating that the net clinical benefit of OAC monotherapy is driven by fewer fatal and major bleeding events without a loss of ischemic protection. In our study, the reduction in cardiovascular mortality with OAC alone was replicated in the Bayesian meta‑analysis (quantifying posterior uncertainty) and corroborated by TSA, where crossing of efficacy boundaries indicated that the evidence was conclusive rather than a product of random error.