Cost-effectiveness of venous thromboembolism prophylaxis in total hip and knee replacement surgery: the evolving application of health economic modelling over 20 years

Abstract

OBJECTIVES: In the last two decades, there has been considerable evolution of methods for cost-effectiveness modelling. Some of the first models were developed in the area of venous thromboembolism (VTE) prophylaxis. Hence, this area can serve as an important example to illustrate evolving standards. Our objectives are to document evolving methodology by describing VTE models, assess their critical strengths and weaknesses, and inform future advances for models in this therapeutic area. RESEARCH DESIGN AND METHODS: A systematic review of economic models of primary VTE prevention following hip and knee replacement surgery was undertaken. Electronic searches of PubMed, EMBASE, the Cochrane library, and grey literature were conducted (1985-2006). Reference lists of included articles and reviews were examined for relevant studies. RESULTS: Twenty-nine cost-effectiveness models were identified. Nineteen other cost-effectiveness analyses were excluded because they were not model-based; 16 were simple cost calculations and three were analyses of resource use data collected alongside clinical trials. The majority of models (24) were constructed as decision trees, frequently utilising previously published model structures, with some adaptation for new comparators, and/or addition of relevant events omitted by earlier models (e.g., bleeding due to prophylactic treatment). Later models have included Markov processes to model potential long-term consequences of VTE (recurrent VTE and post-thrombotic syndrome) over longer time horizons. Systematic identification of clinical evidence and more sophisticated analysis methods (e.g., Bayesian mixed-treatment comparisons and probabilistic sensitivity analyses) have recently been introduced. CONCLUSIONS: Model structures have evolved substantially in this highly studied therapeutic area, with improvements made to the model structure, the comprehensiveness of clinical evidence included, and the underlying calculation methodology