Surgeons consider the benefit-to-harm ratio when making decisions regarding the use of
anticoagulant venous thromboembolism (VTE) prophylaxis. We evaluated the benefit-to-harm ratio of the use of newer
anticoagulants as thromboprophylaxis in patients undergoing major orthopaedic surgery using the likelihood of being helped or harmed (LHH), and assessed the effects of variation in the definition of major
bleeding on the results. A systematic literature search was performed to identify phase II and phase III studies that compared regulatory authority-approved newer
anticoagulants to the
low-molecular-weight heparin enoxaparin in patients undergoing major orthopaedic surgery. Analysis of outcomes data estimated the clinical benefit (number-needed-to-treat [NNT] to prevent one symptomatic VTE) and clinical harm (number-needed-to-harm [NNH] or the NNT to cause one major
bleeding event) of
therapies. We estimated each trial's benefit-to-harm ratio from NNT and NNH values, and expressed this as LHH = (1/NNT)/(1/NNH) = NNH/NNT. Based on reporting of efficacy and safety outcomes, most studies favoured
enoxaparin over
fondaparinux, and
rivaroxaban over
enoxaparin. However, when using the LHH metric, most trials favoured
enoxaparin over both
fondaparinux and
rivaroxaban when they included surgical-site
bleeding that did not require reoperation in the definition of major
bleeding. The exclusion of
bleeding at surgical site which did not require reoperation shifted the benefit-to-harm ratio in favour of the newer agents. Variations in the definitions of major
bleeding may change the benefit-to-harm ratio and subsequently affect its interpretation. Clinical trials should attempt to improve the consistency of major
bleeding reporting.