OUP user menu

Aspirin plus warfarin compared to aspirin alone after acute coronary syndromes: an updated and comprehensive meta-analysis of 25 307 patients

Felicita Andreotti , Luca Testa , Giuseppe G.L. Biondi-Zoccai , Filippo Crea
DOI: http://dx.doi.org/10.1093/eurheartj/ehi485 519-526 First published online: 5 September 2005


Aims In patients recovering from acute coronary syndromes (ACS) the role of oral anticoagulation (and its intensity) in addition to aspirin remains controversial. We conducted a specific meta-analysis of randomized trials comparing aspirin plus warfarin (A+W) with aspirin alone in such patients.

Methods and results MEDLINE and Cochrane databases yielded 14 (of 148 potentially relevant) articles enrolling 25 307 patients. Follow-up ranged from 3 months to 5 years. Irrespective of International normalized ratio (INR), A+W did not significantly affect the risk of major adverse events (MAE: all cause death, non-fatal myocardial infarction, and non-fatal thrombo-embolic stroke) when compared with aspirin alone [OR 0.96 (0.90–1.03), P=0.30], but increased the risk of major bleeds (MB): OR 1.77 (1.47–2.13), P<0.00001. However, in studies with INR of 2–3, A+W was associated with a significant reduction of MAE [OR 0.73 (0.63–0.84), P<0.0001, number needed to treat to avoid one MAE=33], albeit at an increased risk of MB [OR 2.32 (1.63–3.29), P<0.00001; number needed to harm by causing one MB=100]. In both analyses, intracranial bleeding was not significantly increased by A+W when compared with aspirin alone.

Conclusion For patients recovering from ACS, a combined strategy of A+W at INR values of 2–3 doubles the risk of MB, but is nonetheless superior to aspirin alone in preventing MAE. Whether this combined regimen is also superior to a ‘double’ anti-platelet strategy or to newer evolving treatments warrants further investigation.

  • Aspirin
  • Warfarin
  • Acute coronary syndromes
  • Meta-analysis
  • Outcome
  • Risk/benefit ratio


The role of aspirin in the secondary prevention of ischaemic cardiovascular events is universally accepted. Aspirin is known to reduce, by approximately 25%, the risk of any serious vascular accident, with greatest protection among patients with acute coronary syndromes (ACS).1 In contrast, in such patients, the potential added benefit of long-term oral anticoagulants on top of aspirin to prevent recurrent ischaemia is still poorly appreciated despite supportive data.2,3 Reasons that favour the addition of anticoagulants include the prevention of new thrombus formation within the coronary tree,4 avoidance of intracardiac mural thrombosis and systemic embolization,4 protection against deep venous thrombosis and pulmonary embolism,4 and, possibly, inhibition of thrombin's proinflammatory5 and vasoconstrictor6 effects.

Two recent randomized studies have shown an improved global outcome by the combination of aspirin plus warfarin (A+W) when compared with aspirin alone for patients recovering from ACS.7,8 However, this evidence has to be weighed against the presence of prior conflicting reports,9,10 the increased risk of major bleeds (MB),710 and the need to monitor anticoagulation intensity. In patients with stable coronary artery disease or ACS, long-term oral anticoagulant therapy targeted to International normalized ratio (INR) values between 2 and 3 has been found to optimize the risk/benefit ratio of combined therapy, compared with INR values <22,3 and to aspirin alone, with a reasonable cost in terms of excess bleeding.

One of the commonest current anti-thrombotic strategies (aspirin+clopidogrel) in patients recovering from ACS is based on a limited number of studies.11,12 An alternative approach (aspirin+an oral direct thrombin inhibitor)13 is under scrutiny for its hepatotoxicity. Thus, there is a need to accurately and comprehensively define the risk/benefit profile of A+W (at different intensities of anticoagulation) vs. aspirin alone for the secondary prevention of major adverse events (MAE). We therefore conducted a formal and updated meta-analysis of eligible randomized controlled studies. We pre-specified per-protocol first to extend our investigation to all eligible studies independently of the INR achieved in the A+W arm, and then to conduct an additional analysis restricted to those studies with a target or measured INR between 2 and 3.


Search strategy

A systematic search for eligible studies involving MEDLINE and The Cochrane Collaboration CENTRAL database was conducted by three trained investigators (F.A., L.T., and G.B.Z.) according to a modified Robinson and Dickersin strategy,14 using the terms ‘warfarin’, ‘coumarin’, ‘aspirin’, ‘myocardial infarction’, ‘unstable angina’, and ‘acute’ (updated March 2005). No language restriction was applied. In addition, pertinent reviews and quoted papers were sought by hand.

Selection criteria

Inclusion criteria for the retrieved studies were: (i) controlled comparison of A+W vs. aspirin alone in patients recovering from ACS; (ii) randomized treatment allocation; and (iii) intention-to-treat analysis. Exclusion criteria were: (i) an equivocal treatment allocation process; (ii) statistically significant imbalances in major baseline characteristics among study groups; and (iii) incomplete (<80%) follow-up.

Data abstraction and quality assessment

Data abstraction was performed by three independent reviewers (F.A., L.T., and G.B.Z.). Divergences were resolved by consensus. The outcomes of interest were: (i) the combined rate of MAE, defined as all cause death, non-fatal myocardial infarction, or non-fatal thrombo-embolic stroke; and (ii) the rate of MB, defined as intracranial haemorrhages, bleeds requiring transfusions, or a ≥2 g/dL drop in haemoglobin. Whenever possible, intracranial haemorrhages were distinguished from extracranial ones. According to The Cochrane Collaboration, a three-letter grading (A, B, and C) was used for quality assessment, to indicate a low, moderate, or high probability of bias, respectively.15

Statistical analysis

This was inspired by the Quality of Reporting of Meta-analyses (QUORUM)16 statement. As per-protocol, we calculated: (a) the odds ratio for single and composite MAE and MB; (b) the number needed to treat (NNT) to prevent a MAE and the number needed to harm (NNH) by causing a MB. Statistical analyses were performed initially for all studies, and then for those with target or measured INR values between 2 and 3, using the Review Manager freeware package (RevMan 4.2).17 Binary outcomes from individual studies were combined with both fixed and random effect models. Odds ratios with 95% confidence intervals (95% CI) were used as summary statistics. The NNT and NNH were calculated as the inverse of the fixed effect risk difference for MAE and MB, respectively. To assess the contribution of each study to the pooled estimates, a sensitivity analysis was performed by excluding trials one at a time, starting from those with lowest quality score.15 The likelihood of publication bias was assessed graphically by generating a funnel plot for the combined endpoint of MAE and mathematically by means of Egger's test.18 The ‘z’-test values reported in Figures 3 and 4 are two-tailed. Hypothesis testing results were considered significant at the 0.05 level. Cochrane Q heterogeneity tests were also computed, and considered significant at P<0.10.15 The degree of inconsistency among studies (I2) was estimated through scores of 25, 50, and 75%, representing low, moderate, or high inconsistency, respectively.19


Search results

From 148 potentially relevant citations, our search finally identified 14 studies that randomized a total of 25 307 patients with ACS to combined treatment with A+W or to aspirin alone710,2028 (Figure 1).

Figure 1 QUORUM flow diagram.

Study characteristics

The main characteristics of the 14 trials are listed in Table 1. Recruited patients were admitted to hospital for unstable angina or for ST-elevation or non-ST-elevation myocardial infarction. In most trials enrolment took place within 1 week, but in some cases occurred 6 or 8 weeks after the acute event. Follow-up varied from 3 months to 5 years, totalling 52 936 patient-years. Completeness of follow-up was greater than 97% in all but one study. The dose of aspirin ranged from 75 to 325 mg/day. The target or measured INR in the combined-treatment arm was between 2 and 3 in 10 reports and below 2 in four.

View this table:
Table 1

Study characteristics. Trials are listed by publication date

Several of the included trials showed a number of additional specific items. Three studies—the Anti-thrombotics in the Secondary Prevention of Events in Coronary Thrombosis-2 (ASPECT-2),7 the Warfarin, Aspirin, Reinfarction Study (WARIS-II),8 and the study by Huynh et al.24—also included a warfarin-alone treatment arm (involving a total of 1586 patients), which was not considered in our meta-analysis. Two studies—the Anti-thrombotics in the Prevention of Reocclusion in Coronary Thrombolysis (APRICOT-2)26 and the study by Williams et al.23—primarily focused on angiographic endpoints, but additionally reported MAE and MB, and were, therefore, included. The Organization to Assess Strategies for Ischaemic Syndromes (OASIS) pilot study20 was conducted in two phases: the first aimed at INR values <2, and the second at INR values between 2 and 3; for the specific purpose of our analysis, these two phases were considered as separate studies. Huynh et al.24 selected patients with prior coronary artery bypass surgery who were ineligible for further revascularization. Finally, the Coumadin Aspirin Reinfarction Study (CARS)9 and the LoWASA27 investigation used fixed, low doses of warfarin (1 or 3 mg/day in CARS and 1.25 mg/day in LoWASA). The CARS9 trial was stopped prematurely for lack of benefit of combination therapy when compared with aspirin alone. Thirteen of the 14 trials were of good quality, with a low or moderate risk of bias. A high or moderate risk was mostly due to unblinded adjudication of events and handling of dropouts.

Quantitative findings

The pooled estimates by the random effect model did not differ significantly from those obtained by the fixed effect model. The presented results are according to the latter.

In the overall analysis (irrespective of INR), combination therapy when compared with aspirin alone had no significant effect on the risk of MAE [OR 0.96 (0.90–1.03), P=0.30; Figure 2A], although marked heterogeneity was found among studies (P=0.001). The risk of MB was increased by A+W [OR 1.77 (1.47–2.13), P<0.00001; NNH=100; Figure 3A]. More specifically, the rate of extracranial bleeds with combination therapy was associated with an OR of 2.20 (95% CI 1.64–2.96, P<0.00001; Figure 3B), whereas that of intracranial bleeds was associated with an OR of 1.37 (95% CI 0.79–2.37, P=0.27, Figure 3C).

Figure 2 Individual and summary odds ratios for MAE (all cause death/non-fatal myocardial infarction/non-fatal thrombo-embolic stroke): (A) overall analysis of all studies, regardless of INR values and (B) analysis restricted to studies with INR values between 2 and 3.

Figure 3 Individual and summary odds ratios for MB in the overall analysis: (A) all bleeds; (B) extracranial; and (C) intracranial.

No significant heterogeneity was found in the analyses restricted to studies with target or measured INR values between 2 and 3. Combination therapy was associated with a reduced risk of MAE [OR 0.73 (0.63–0.84), P<0.0001; NNT=33; Figure 2B] and with an increased risk of MB [OR 2.32 (1.63–3.29), P<0.00001; NNH=100; Figure 4A). More specifically, with combination therapy, the rate of extracranial bleeds was associated with an OR of 2.37 (95% CI 1.37–4.10, P=0.002; Figure 4B), whereas that of intracranial bleeds was associated with an OR of 3.02 (95% CI, 0.61–15.02, P=0.18; Figure 4C).

Figure 4 Individual and summary odds ratios for MB in the restricted analysis: (A) all bleeds; (B) extracranial; and (C) intracranial.

Table 2 gives the absolute incidences and odds ratios for individual MAE. In the overall analysis, A+W when compared with aspirin alone did not reduce the risk of all cause death nor that of non-fatal myocardial infarction, but did significantly reduce the risk of non-fatal thrombo-embolic stroke (NNT=100). In the restricted analysis, combination therapy was associated with a 30% reduction in the risk of non-fatal myocardial infarction (NNT=50) and with a 57% reduction in the risk of non-fatal thrombo-embolic stroke (NNT=100), whereas all cause mortality was not significantly affected. On sensitivity analysis, the exclusion of any single trial did not substantively alter the overall results.

View this table:
Table 2

Absolute incidences and odds ratios of individual MAE with aspirin plus warfarin compared to aspirin alone

Assessment of possible biases

Funnel plot for all studies and for those with INR between 2 and 3 showed a degree of asymmetry (Figure 5), possibly consistent with publication bias; the latter, however, was excluded by means of Egger's test (P=0.141 for all studies and P=0.646 for studies with INR between 2 and 3). Our pre-defined decision to exclude studies with a follow-up <80% could have introduced a selection bias into the results; however, none of the possible relevant citations was excluded for this reason.

Figure 5 Funnel plot of all studies for MAE.


This meta-analysis involving 25 307 patients provides a comprehensive and updated assessment of the risk/benefit profile of A+W when compared with aspirin alone in patients recovering from an ACS. It differs substantially from previous reports2,3 in several ways: it specifically addresses patients recovering from ACS; it provides a quantitative assessment of both efficacy and safety outcomes indicating, through the NNT and NNH, the absolute magnitude of benefits and harms; it differentiates intracranial from extracranial bleeds; finally, it systematically assesses the effect of A+W vs. aspirin alone with respect to individual clinical endpoints.

Heterogeneity was found in the overall analysis, but not in the one restricted to studies with INR values of 2–3. This implies that anticoagulation intensities ranging from <2 to >3 contributed in determining large differences among individual ORs (Figure 2A), thus representing a major cause of statistical heterogeneity. At INR values between 2 and 3, the present report indicates that treating 1000 patients with combination therapy should avoid approximately 30 MAE (20 non-fatal myocardial infarctions and 10 non-fatal thrombo-embolic strokes), while causing 10 MB. More specifically, as shown in Figure 4C, the rate of intracranial haemorrhages increased—in absolute terms—from 0.05% with aspirin alone to 0.25% with combination therapy. On the other hand, the absolute rate of thrombo-embolic stroke decreased from 1.48% with aspirin alone to 0.6% with A+W (Table 2). These estimates again suggest that allocating 1000 patients to A+W would avoid approximately nine thrombo-embolic strokes while causing two intracranial haemorrhages, thus yielding a net benefit in favour of A+W.

The use of oral anticoagulants in patients with ischaemic heart disease is commonly restricted to individuals at high risk of thrombo-embolic complications, e.g. those with atrial fibrillation, valve prostheses, intracardiac thrombi, recurrent thrombo-embolism, or anti-phospholipid syndrome.2931 In these groups, the prescription of oral anticoagulants usually induces clinicians to refrain from adding aspirin, given the uncertainties surrounding the risks and benefits of combined treatment. A recent American Heart Association/American College of Cardiology Statement dedicated to warfarin therapy31 presented a number of anti-thrombotic strategies for patients with ACS, ranging from aspirin alone, to aspirin plus clopidogrel, or A+W. Yet, indications for the choice of one rather than another strategy were left to future recommendations.31

Currently, the combined use of aspirin and a thienopyridine (ticlopidine or clopidogrel) is recommended among patients with ACS for 9–12 months after the acute event. Such an approach is mainly based on the results of the Clopidogrel in Unstable angina to prevent Recurrent Events (CURE) study.11 The latter reported at 9 months a 20% relative reduction in the combined rate of cardiovascular death, non-fatal myocardial infarction, and non-fatal stroke with the ‘double’ anti-platelet regimen of aspirin plus clopidogrel when compared with aspirin alone. After this period of 9–12 months, the anti-thrombotic regimen of choice still remains aspirin alone. Recent data show that a novel oral anticoagulant, the direct thrombin inhibitor ximelagatran, when added to aspirin, may yield a 22% relative reduction in the rate of MAE at 6 months when compared with aspirin alone in patients with ACS.13 Taken together, these premises call for: (1) a head-to-head comparison of aspirin combined with properly-dosed oral anticoagulants (e.g. warfarin or a direct thrombin inhibitor) against a dual anti-platelet regimen (e.g. aspirin with a thienopyridine) for the first 9–12 months after an ACS; (2) a triple-armed comparison among aspirin alone, aspirin+an oral anticoagulant, and aspirin+a thienopyridine for the period beyond the first 9–12 months after an ACS.

In conclusion, the present study makes a compelling case for the combined use of aspirin and intermediate-intensity oral anticoagulation for the prevention of MAE in patient with ACS, particularly in those at high-risk of recurrent cardiac and cerebrovascular events who are willing to face the logistic hurdles and the predictable but uncommon bleeding risks inherent to such a treatment strategy.

Conflict of interest: no conflict of interest.


View Abstract