Decision-analysis modelling of effectiveness and cost-effectiveness of pharmacological thromboprophylaxis for surgical inpatients, using variable risk assessment models or other strategies

Background Surgical inpatients are at risk of venous thromboembolism (VTE) which can be life-threatening or result in chronic complications. Thromboprophylaxis reduces VTE risk but incurs costs and may increase bleeding risk. Risk assessment models (RAMs) are currently used to target thromboprophylaxis at high-risk patients. Objective To determine the balance of cost, risk, and benefit for different thromboprophylaxis strategies in adult surgical inpatients, excluding major orthopaedic surgery, critical care and pregnant women. Methods Decision analytic modelling to estimate the following outcomes for alternative thromboprophylaxis strategies: thromboprophylaxis usage; VTE incidence and treatment; major bleeding; chronic thromboembolic complications; and overall survival. Strategies compared were: no thromboprophylaxis; thromboprophylaxis for all; and thromboprophylaxis given according to RAMs (Caprini and Pannucci). Thromboprophylaxis is assumed to be given for the duration of hospitalisation. The model evaluates life-time costs and quality-adjusted life-years (QALYs) within England’s health and social care services. Results Thromboprophylaxis for all surgical inpatients had a 70% probability of being the most cost-effective strategy (at a £20,000 per QALY threshold). RAM-based prophylaxis would be the most cost-effective strategy if a RAM with higher sensitivity (99.9%) were available for surgical inpatients. QALY gains were mainly due to reduced post-thrombotic complications. The optimal strategy was sensitive to several other factors including: risk of VTE, bleeding and post thrombotic syndrome; duration of prophylaxis and patient age. Conclusions Thromboprophylaxis for all eligible surgical inpatients appeared to be the most cost-effective strategy. Default recommendations for pharmacological thromboprophylaxis, with the potential to ‘opt-out’, may be superior to a complex risk-based ‘opt-in’ approach