Abstract
Primary membranous nephropathy is associated with increased risk of venous thromboembolic events, which are inversely correlated with serum albumin levels. To evaluate the potential benefit of prophylactic anticoagulation (venous thromboembolic events prevented) relative to the risk (major bleeds), we constructed a Markov decision model. The venous thromboembolic event risk according to serum albumin was obtained from an inception cohort of 898 patients with primary membranous nephropathy. Risk estimates of hemorrhage were obtained from a systematic literature review. Benefit-to-risk ratios were predicted according to bleeding risk and serum albumin. This ratio increased with worsening hypoalbuminemia from 4.5:1 for an albumin under 3 g/dl to 13.1:1 for an albumin under 2 g/dl in patients at low bleeding risk. Patients at intermediate bleeding risk with an albumin under 2 g/dl have a moderately favorable benefit-to-risk ratio (under 5:1). Patients at high bleeding risk are unlikely to benefit from prophylactic anticoagulation regardless of albuminemia. Probabilistic sensitivity analysis, to account for uncertainty in risk estimates, confirmed these trends. From these data, we constructed a tool to estimate the likelihood of benefit based on an individual's bleeding risk profile, serum albumin level, and acceptable benefit-to-risk ratio (www.gntools.com). This tool provides an approach to the decision of prophylactic anticoagulation personalized to the individual's needs and adaptable to dynamic changes in health status and risk profile.