A systematic review of Cochrane anticoagulation reviews.

Context: I coauthored a published review of anticoagulation for venous thromboembolism in the Cochrane Database of Systematic Reviews and published a review on the same topic in MedGenMed (now the Medscape Journal of Medicine). In contrast to the article in Medscape, the discussion and conclusions in the Cochrane review were altered appreciably during the review process. Consequently, I decided to critique all anticoagulation drug-related reviews and protocols in the Cochrane database with feedback letters concerning any issues of potential controversy.

Evidence Acquisition: Using key words in the search engine of the Cochrane Reviews, I located reviews and protocols involving anticoagulant drugs. I critiqued each anticoagulation review and protocol and sent a total of 57 feedback letters to Cochrane concerning each publication to elicit a response/rebuttal from the authors.

Evidence Synthesis: Cochrane anticoagulation review editors acknowledged receipt of all letters. As of 12 months after receipt of my last letter, the Cochrane authors have replied to 13 of the 57 and agreed with many of my points. Two protocols were withdrawn after my feedback letters were acknowledged. The 58 Cochrane anticoagulation drug reviews, including mine, contained 9 categories of methodological errors (207 total instances) and 4 types of biases (18 total instances). This review of those Cochrane reviews suggests that the effectiveness of anticoagulants for 30 medical indications is questionable.

Conclusions: The efficacy of anticoagulants for treatment and prophylaxis for 30 current medical indications should be reconsidered by the scientific community and medical regulatory agencies. At least 50,000 people per year worldwide have fatal bleeding due to anticoagulant treatment or prophylaxis for these indications.