Meta-analysis of controlled trials of anticoagulant therapies in patients with sepsis

Although coagulation abnormalities may partly underlie the physiologic derangements of the sepsis syndrome, anticoagulant therapies have produced mixed results on survival in clinical studies. We hypothesized that a meta-analysis of clinical trials of anticoagulants in sepsis may provide insight as to the therapeutic utility of targeting the clotting cascade in this syndrome. We searched electronic databases and reviewed bibliographies of pertinent articles to identify controlled clinical studies in which anticoagulants had been administered as adjunctive therapy to patients with sepsis. After establishing statistical homogeneity, odds ratios (OR; with 95% confidence intervals [Cl]) for effect of these agents on mortality and bleeding complications were determined using Mantel-Haenszel methodology. Potential for publication bias was assessed by the use of a statistical test of funnel plot asymmetry. Weighted linear regression was performed to examine the effect of control group mortality rate on treatment efficacy. We identified 11 studies that satisfied our inclusion criteria. Collectively, these studies enrolled 4690 patients (range of 29-2314) and examined three agents: antithrombin 111 (2659 patients), tissue factor pathway inhibitor (210 patients), and activated protein C (1821 patients). After establishing statistical homogeneity (P > 0.10, chi-square), we found that the OR (with 95% Cl) for effect on mortality for these agents, relative to control treatment, was 0.8692 (0.7519-1.006). Weighted linear regression analysis was consistent with a control group mortality dependent effect for these agents (P = 0.02). Only five of the studies reported bleeding complications. Pooling the results of these five studies (4376 patients) resulted in an OR (with 95% Cl) of 1.70 (1.40-2.07) relative to control treatment for bleeding risk. Anticoagulants as adjuvant therapy do not appear to improve outcome in sepsis and are associated with a significant risk of bleeding complications. To the extent that their treatment effect is dependent upon disease severity, the safety and efficacy of these agents may be enhanced by refinement in techniques of clinical stratification.