Does Thromboprophylaxis Reduce Mortality in Medical Patients? A Systematic Review and Meta-Analysis of Randomized Cntrolled Tials.

Background: Despite good evidence that anticoagulants are effective in preventing venous thromboembolism in medical patients at risk for this disease, only one-third of such patients are receiving thromboprophylaxis. Underutilization of thromboprophylaxis in medical patients may be due to lack of evidence that thromboprophylaxis reduces mortality, concerns about anticoagulant-related bleeding, and questions about the clinical significance of surrogate (venographic) outcomes to assess efficacy in these trials. We performed a meta-analysis of randomized, placebo-controlled trials of anticoagulant thromboprophylaxis in medical patients to assess effects on mortality and bleeding.

Methods: The MEDLINE, EMBASE and Cochrane Central Register of Controlled Trials databases were searched until June 2005. Studies were included that were randomized, placebo-controlled trials investigated a prophylactic dose of unfractionated heparin, low-molecular-weight heparin, or fondaparinux in medical patients. Meta-analysis was done to obtain pooled estimates of the effects of anticoagulant thromboprophylaxis on mortality and clinically important (major) bleeding. The effect of treatment on venous thromboembolism was not pooled because of across-study difference in methods used to diagnose this, and use of asymptomatic (venographic) outcomes in some studies.

Results: There were 8 studies of 20,631 patients in the assessment of mortality and 4 studies of 5,428 patients in the assessment of major bleeding. Death occurred in 536 of 10,321 (5.2%) patients who received thromboprophylaxis, and in 608 of 10,510 (5.8%) who received placebo. Thromboprophylaxis was associated with 10% decreased risk for all-cause mortality, although this effect was not quite statistically significant (odds ratio [OR] = 0.90; 95% confidence interval [CI]: 0.81, 1.01). Major bleeding occurred in 22 of 2,726 (0.8%) patients who received thromboprophylaxis, and in 11 of 2,702 (0.4%) patients who received placebo. Thromboprophylaxis was associated with 2-fold increased risk for major bleeding, although this effect was not quite statistically significant (OR = 2.04; 95% CI: 0.98, 4.23).

Conclusion: In medical patients who are at increased risk for venous thromboembolism, anticoagulant thromboprophylaxis appears to confer a small reduction in mortality; this benefit should be balanced against an increased risk for major bleeding.