Current issues of ACP Journal Club are published in Annals of Internal Medicine


Review: Antiplatelet agents reduce risks for death, stroke, myocardial infarction, deep venous thrombosis, and arterial occlusion

ACP J Club. 1994 July-Aug;121:4. doi:10.7326/ACPJC-1994-121-1-004

Source Citations

Antiplatelet Trialists' Collaboration. Collaborative overview of randomised trials of antiplatelet therapy—I: Prevention of death, myocardial infarction, and stroke by prolonged antiplatelet therapy in various categories of patients. BMJ. 1994 Jan 8;308:83-108.

Antiplatelet Trialists' Collaboration. Collaborative overview of randomised trials of antiplatelet therapy—II: Maintenance of vascular graft or arterial patency by antiplatelet therapy. BMJ. 1994 Jan 15;308:159-68.

Antiplatelet Trialists' Collaboration. Collaborative overview of randomised trials of antiplatelet therapy—III: Reduction in venous thrombosis and pulmonary embolism by antiplatelet prophylaxis among surgical and medical patients. BMJ. 1994 Jan 22;308:235-46.



To determine the effect of long-term (> 1 month) antiplatelet therapy on total and cause-specific mortality, myocardial infarction (MI), and stroke for all patients and patients at high and low risk for occlusive vascular disease; to determine the efficacy of antiplatelet therapy in reducing vascular occlusion in similar groups of high- and low-risk patients; and to determine the efficacy of antiplatelet therapy as prophylaxis against deep venous thrombosis (DVT) or pulmonary embolism (PE) in surgical and high-risk medical patients.

Data sources

Randomized controlled trials were identified using MEDLINE and Current Contents databases; manual searches of selected journals; lists of conference abstracts and meeting reports; bibliographies of relevant studies and review articles; collaboration with the trial register of the International Committee on Thrombosis and Haemostasis; and correspondence with colleagues, manufacturers of antiplatelet drugs, and collaborating trialists.

Study selection

Trials with unbiased methods of randomization were included if they were available for review before March 1990. All trials comparing prophylactic antiplatelet therapy with no antiplatelet therapy or comparing 2 or more antiplatelet agents were included. Agents studied had a primary mode of action on the vascular system of either platelet aggregation inhibition, adhesion inhibition, or both: cyclooxygenase inhibitors (aspirin, flurbiprofen, ibuprofen, indobufen, naproxen, sulphinpyrazone, triflusal), phosphodiesterase inhibitors (dipyridamole, E5510, RA233), platelet calcium ion channel inhibitors (suloctidil), phosopholipase inhibitors (hydroxychloroquine), thromboxane synthase inhibitors, receptor blockers, or both (dazoxiben, piracetam, picotamide, ridogrel, sulotroban, daltroban, GR32191), and agents with direct effects on platelet membranes (ticlopidine). Agents with potent vasodilating or anticoagulant activity were excluded.

To examine the effect on occlusion, trials were included if they studied coronary artery bypass grafting, percutaneous transluminal coronary angioplasty, medical prevention of occlusion in symptomatic peripheral vascular disease, transluminal angioplasty of the legs, noncoronary arterial grafting procedures, or arteriovenous fistulas or shunts for hemodialysis access.

Studies of DVT and PE in surgical and high-risk medical patients were included if DVT was systematically monitored by venography, radiolabeled fibrinogen uptake, or both.

Data extraction

Numbers of patients studied, antiplatelet agent and dose used, timing and duration of therapy, and side effects including hemorrhage were extracted for all studies. Individual patient data were sought for certain categories of patients. For the studies of MI and stroke, the following were extracted: MI, nonfatal stroke, vascular deaths, nonvascular deaths, total deaths, and vascular events (nonfatal MI + nonfatal stroke + vascular deaths). Patients were classified as high risk (previous MI, current MI, previous stroke or transient ischemic attack, or other high risk) or low risk for occlusive vascular disease.

For the occlusion studies, additional data were extracted on vascular occlusion rates. For DVT and PE prophylaxis studies, the following were extracted: numbers and follow-up method of all and proximal DVTs, fatal and nonfatal PE, non-PE deaths, reoperations, wound hematomas, and major hemorrhage.

Main results

145 trials were included for evaluation of vascular events (70 000 high-risk and 30 000 low-risk patients). Antiplatelet therapy, compared with control, showed reductions in most outcomes: In patients with a previous MI, antiplatelet therapy led to a lower rate of MI, stroke, or vascular death after 1 month of treatment (P < 0.001) (Table). For all patients at high risk, antiplatelet therapy led to lower rates of MI, stroke, or vascular death (P < 0.001), nonfatal MI (P < 0.001), vascular death or death from unknown causes (P < 0.001), and all-cause mortality (P < 0.001) (Table). The most widely tested regimen was aspirin, 75 to 325 mg. No evidence was provided that high doses were more effective than medium doses. Low-risk recipients of primary prevention did have a reduced rate of nonfatal MI but did not have reduced rates of vascular events, vascular mortality, or all-cause mortality.

39 trials measuring vascular occlusion were included (8000 patients in trials comparing antiplatelet therapy with placebo or no therapy and 4000 patients in trials comparing ≥ 2 antiplatelet agents). Vascular occlusion rates were reduced with antiplatelet therapy compared with placebo: For coronary arteries (P < 0.001), peripheral arteries (P < 0.001), hemodialysis access (P < 0.001), and all 3 types of occlusion combined (P < 0.001) (Table). Occlusion rates did not vary across antiplatelet agents. Data on hemorrhage were incomplete.

Of 29 trials in 6691 general surgery patients, 11 trials studied traumatic orthopedic surgery (964 patients), 18 trials studied elective orthopedic surgery (1154 patients), and 11 trials studied high-risk medical patients (814 patients). All patients allocated to antiplatelet therapy had a reduced rate of DVT when compared with controls (P < 0.001) and PE (P < 0.001) (Table). For patients who had surgery, the reductions were similar for PE ((P < 0.001) and for deaths from PE (P < 0.001) (Table). For patients who had general surgery or orthopedic surgery, antiplatelet therapy showed similar reductions in PE (P < 0.001 for both) (Table). Antiplatelet therapy reduced the risk for DVT in high-risk medical patients (P = 0.02) (Table). The treatment groups did not differ for fatal hemorrhage, but patients treated with antiplatelet therapy compared with patients allocated to other therapies had more major hemorrhages (P = 0.04) and other complications (reoperations, wound hematomas, or infection caused by bleeding) (P = 0.003) (Table).


Antiplatelet agents given to a wide range of patients at risk for occlusive vascular disease reduce all-cause mortality, vascular mortality, vascular events (including nonfatal myocardial, nonfatal stroke, and vascular death), and nonfatal myocardial infarction. Benefit is observed among patients with unstable angina, suspected acute myocardial infarction, past history of myocardial infarction, nonfatal stroke, transient ischemic attack, peripheral vascular disease, and those having vascular procedures. Low-risk patients receiving antiplatelet agents have reduced nonfatal myocardial infarction only. Vascular occlusion rates are reduced with antiplatelet therapy when compared with no prophylaxis. Prophylactic antiplatelet therapy decreases the risk for deep venous thrombosis in high-risk medical patients whether or not they have surgery.

Sources of funding: Medical Research Council (UK); Chest, Heart, and Stroke Association (UK and Scotland); Edinburgh University; University of Oxford; British Heart Foundation; Imperial Cancer Research Fund; Boehringer Ingelheim (UK); Ciba Giegy (USA); EISAI (Europe); European Aspirin Foundation; Glaxo (UK); Glenbrook Laboratories Division of Sterling Drugs (USA); ICI (UK); Lilly (UK); Reckitt and Colman (UK); and Sanofi (Europe).

For article reprint: APT Statistical Secretariat, ICRF/BHF/MRC Clinical Trial Service Unit, Nuffield Department of Clinical Medicine, Radcliffe Infirmary, Oxford OX2 6HE, England, UK.

Table. Antiplatelet therapy vs control for vascular events at up to 5 years of follow-up*

Outcomes Antiplatelet therapy Adjusted control RRR (95% CI) NNT (CI)
MI, stroke, or vascular death (all patients) 13.5% 17.1% 21% (16 to 26) 28 (22 to 38)
MI, stroke or vascular deaths (high-risk patients) 11.4% 14.7% 22% (19 to 25) 31 (27 to 37)
Nonfatal MI (high-risk patients) 2.8% 4.2% 32% (26 to 37) 76 (63 to 96)
Vascular death or death from unknown causes (high-risk patients) 6.9% 8.3% 16% (12 to 20) 76 (59 to 106)
All-cause mortality (high-risk patients) 8.0% 9.5% 15% (11 to 19) 71 (55 to 100)

*Abbreviations defined in Glossary; RRR, NNT, and CI calculated from data in article.

Table. Antiplatelet therapy vs control for vascular occlusion at up to 5 years of follow-up†

Outcomes Antiplatelet therapy Adjusted control RRR (95% CI) NNT (CI)
Coronary artery occlusion 18.8% 27.3% 32% (25 to 38) 11 (9 to 15)
Peripheral artery occlusion 15.7% 24.9% 37% (28 to 45) 11 (8 to 15)
Hemodialysis access 16.8% 38.7% 57% (40 to 70) 5 (3 to 7)
All 3 types of occlusion 17.7% 27.0% 35% (29 to 39) 11 (9 to 15)

†Abbreviations defined in Glossary; RRR, NNT, and CI calculated from data in article.

Table. Antiplatelet therapy vs control for deep venous thrombosis and pulmonary embolism at up to 5 years of follow-up†

Outcomes Antiplatelet therapy Adjusted control RRR (95% CI) NNT (CI)
DVT (All patients) 24.8% 33.6% 26% (19 to 32) 11 (9 to 16)
DVT (High-risk medical patients) 14.9% 22.9% 35% (6.6 to 55) 13 ( 7 to 77)
PE ( All patients) 1.0% 2.7% 63% (49 to 74) 58 (44 to 83)
PE (Surgery patients) 1.0% 2.7% 64% (49 to 74) 58 (43 to 85)
PE (General surgery patients) 0.5% 1.7% 73% (52 to 84) 82 (57 to 132)
PE (Orthopedic surgery patients) 2.7% 6.1% 56% (32 to 71) 29 (19 to 60)
PE Mortality 0.2% 0.9% 72% (46 to 85) 159 (103 to 302)
Major hemorrhage 0.7% 0.4% 87% (1.1 to 246) 291 (141 to 15172)
Other complications 7.8% 5.6% 39% (12 to 74) 45 (27 to 131)

†Abbreviations defined in Glossary; RRR, RRI, NNT, NNH, and CI calculated from data in article.


As an active clinician, few things are clearer—my patients having an acute MI or those who survive an MI, stroke, or transient ischemic attack must receive antiplatelet agents to lower their risk for vascular events and death. Because antiplatelet agents for various high-risk patients cause beneficial outcomes more frequently than adverse outcomes, clinicians now must prescribe aspirin, an effective, inexpensive, and low-risk therapy, unless there are specific contraindications. Aspirin works at virtually any medium dose (75 to 325 mg/d), with efficacy independent of the sex, blood pressure, or glucose intolerance of the patient. The benefit extends to patients with unstable angina and to those who have received angioplasty. Testing of treatments that may work better than current antiplatelet drugs will not justify placebo controls for patients who tolerate aspirin.

As a clinical researcher, I recognize the labor-intensive tasks and investigators' thoroughness in collecting data from each randomized trial. The pursuit of individual patient data by the Antiplatelet Trialists' Collaboration made its overview more difficult to carry out than some included studies! This "blue-collar" meta-analysis of individual patients differs considerably from traditional "white-collar" meta-analyses of the literature that focus on summary measures alone. Describing their study as a "meta-analysis" inaccurately characterizes these investgators' greater efforts. A new and better word, for example "mega-analysis," seems more appropriate for the huge effort required when combining patient data across methodologically sound trials. The investigators' payoffs from this mega-analysis are recommendations extending aspirin use 1) over various risk factors and 2) for several outcomes. The risk factor information, for example, might not have been available from the summary data of a traditional meta-analysis.

The Collaboration's high-quality methods were mostly conservative in interpreting evidence, favoring finding no antiplatelet effect. The appropriateness of combining data for different antiplatelet agents with different mechanisms of action intended to produce the same outcome, however, may disturb clinicians. The authors contend that overall estimates of risk reduction apply to some subgroups of patients better than do actual data for these subgroups because of the unreliability of data for the smaller number of patients in subgroups (1). For most high-risk disorders, clinicians should accept as valid the estimates of benefit from overall findings. A more difficult decision about appropriateness occurs when examining high-risk categories with few patients studied, especially patients having coronary bypass surgery and those with stable angina, atrial fibrillation, or peripheral vascular disease. I find the conclusions about the efficacy of antiplatelet agents for these categories less certain because the 95% CIs all overlap "no effect" (i.e., the effect was not statistically significant). Unencumbered by rigid adherence to statistics, I find the recommendations particularly unconvincing for peripheral vascular disease and post-valve surgery because less than half the patients in these studies were enrolled in trials of aspirin. By comparison, aspirin was tested in trials involving 90% of patients after coronary bypass surgery, 76% of patients with stable angina, and 100% of patients with atrial fibrillation.

The investigators studied 3 primary prevention investigations. Although aspirin is statistically beneficial in preventing nonfatal MI, the clinical benefit is small. The low frequency of vascular events means that many patients need to be treated to prevent 1 bad outcome. I do not recommend primary prevention for low-risk patients based on these results.

Clinicians unimpressed that patients require aspirin to decrease the risk for death, MI, and stroke after arterial procedures must, nonetheless, acknowledge another fact that the mega-analysis makes clear—antiplatelet agents lower the risk for total arterial occlusions after arterial procedures, including arterial grafts, vessels after angioplasty, and in surgically created fistulas or shunts. Again, unless contraindications exist, I will now prescribe aspirin as an effective, inexpensive, and low-risk regimen for patients who have had these procedures and are at high risk for total vascular occlusion. Aspirin works best when started within 24 hours of the procedure, but the duration of treatment is still unknown because most trials lasted less than 1 year.

Conventional thinking would lead clinicians to believe the third review was unnecessary—antiplatelet agents are not routinely used to prevent DVT and PE in patients who are hospitalized. The investigators counter this by noting that the many small studies lacked statistical power to find meaningful effects. The analysis of surgical and high-risk medical patients, however, found the efficacy of aspirin similar to the effects observed in the first 2 reviews. A clinically important benefit from antiplatelet agents exists when they are used for preventing DVT among patients who have had general surgery, traumatic orthopedic surgery, or elective orthopedic surgery or who are high-risk general medical patients. For elective orthopedic surgery, treating only 7 patients prevents 1 DVT. Adverse bleeding events are almost always nonfatal and seldom lead to transfusions for less serious hemorrhage. A surgeon, however, would have 1 complication among 45 patients treated that is potentially attributable to antiplatelet agents.

Determining the most appropriate regimens for reducing thromboses, while minimizing adverse outcomes, requires large randomized trials with placebo controls. It is possible that dipyridamole enhances the effect of aspirin. Further, some drugs may be better than aspirin. For example, and surprising to me, hydroxychloroquine used alone appeared to be at least as effective as aspirin for preventing DVT and PE. The results of this mega-analysis should encourage clinical trialists to compare the efficacy of antiplatelet agents with anticoagulants among hospitalized high-risk medical patients and certain categories of patients having surgery.

David L. Simel, MD
Duke University Medical CenterDurham, North Carolina, USA


1. Efron B, Morris C. Stein's paradox in statistics. Sci Am. 1977;236:119-27.