PIH4 Network Meta-Analysis of Pharmacological Interventions to Prevent Venous Thromboembolism Following Elective Knee and Hip Replacement Surgery: Why Extend the Network?

Abstract

In 2010, NICE published a clinical guideline on reducing the risk of venous thromboembolism (VTE) in patients admitted to hospital. At the time, five pharmological VTE prophylaxis were recommended following elective knee or hip replacement surgery - dabigatran etexilate, fondaparinux sodium, low molecular weight heparin (LMWH), rivaroxaban, and unfractionated heparin (UFH) for patients with renal failure - in 2011, apixaban was also recommended for the same indication in the UK. Although apixaban, dabigatran, and rivaroxaban were all assessed at different times by NICE, each technology appraisal used a distinct methodology to compare relevant interventions, directly and indirectly. OBJECTIVES: To evaluate the impact of study identification methods and network size on indirect and mixed treatment comparisons of pharmacological interventions for the prevention of VTE in patients undergoing elective orthopaedic surgery. METHODS: Data identification and synthesis methods used in the NICE TA157, TA170, and TA245 were reviewed and compared to best practice searching strategies and network meta-analysis methodology. Literature searches specifically designed to collect data for use in network meta-analysis were conducted and compared to original search results. Various network sizes were tested to synthesise available evidence using Winbugs. PRELIMINARY. RESULTS: Specifically designed search strategies to inform the evidence base for network meta-analysis led to increased numbers of studies identified and included for synthesis increasing the total number of indirect comparisons between existing VTE interventions. The impact of additional comparisons on relative treatment estimates was limited and precision did not increase. CONCLUSIONS: Using a specific search strategy designed to optimise the identification of studies for network meta-analysis allowed us to extend the network of relevant studies; however, extending the network increased uncertainty in the results and led to inconsistencies in the direct and indirect comparisons.