Prevention of pulmonary embolism and deep vein thrombosis with low dose aspirin: Pulmonary Embolism Prevention (PEP) trial

Summary Background Previous trials of antiplatelet therapy for the prevention of venous thromboembolism have individually been inconclusive, but a meta-analysis of their results indicated reductions in the risks of deep-vein thrombosis and of pulmonary embolism in various high-risk groups. The aim of this large randomised placebo-controlled trial was to confirm or refute these apparent benefits. Methods During 1992–1998, 148 hospitals in Australia, New Zealand, South Africa, Sweden and the UK randomised 13 356 patients undergoing surgery for hip fracture, and 22 hospitals in New Zealand randomised a further 4088 patients undergoing elective arthroplasty. Study treatment was 160 mg daily aspirin or placebo, started preoperatively and continued for 35 days. Patients received any other thromboprophylaxis thought necessary. Follow-up was of mortality and of in-hospital morbidity up to day 35. Findings Among the patients with hip fracture, allocation to aspirin produced proportional reductions in pulmonary embolism of 43% (95% CI 18–60; p=0·002) and in symptomatic deep-vein thrombosis of 29% (3–48; p=0·03). Pulmonary embolism or deep-vein thrombosis was confirmed in 105 (1·6%) of 6679 patients assigned aspirin compared with 165 (2·5%) of 6677 assigned placebo, which represents an absolute reduction of 9 (SE 2) per 1000 and a proportional reduction of 36% (19–50; p=0·0003). Similar proportional effects were seen in all major subgroups, including patients receiving subcutaneous heparin. Aspirin prevented 4 (1) fatal pulmonary emboli per 1000 patients (18 aspirin-group vs 43 placebo-group deaths), representing a proportional reduction of 58% (27–76; p=0·002), with no apparent effect on deaths from any other vascular cause (hazard ratio 1·04 [95% CI 0·86–1·26]) or non-vascular cause (1·01 [0·84–1·23]). Deaths due to bleeding were few (13 aspirin vs 15 placebo), but there was an excess of 6 (3) postoperative transfused bleeding episodes per 1000 patients assigned aspirin (p=0·04). Among electivearthroplasty patients, rates of venous thromboembolism were lower, but the proportional effects of aspirin were compatible with those among patients with hip fracture.