Analytical approach:
The analysis was based on a decision-analytic model with one- and five-year time horizons. The model consisted of three modules: prophylaxis period, post-prophylaxis period and long-term complications. The first two stages were analysed using a decision tree model. The third stage was represented by a Markov model. The authors stated that the perspective was that of the health care payer.
Effectiveness data:
Clinical inputs were taken from a selection of known relevant studies. Most data for the short-term model were derived from three of the four RECORD studies (REgulation of Coagulation in ORthopaedic surgery to prevent Deep vein thrombosis and pulmonary embolism), which were randomised controlled double-blind phase III trials that compared rivaroxaban with enoxaparin. RECORD1 compared 35 days of rivaroxaban to 35 days of enoxaparin in total hip replacement patients, RECORD2 compared 35 days of rivaroxaban to 10 to 14 days of enoxaparin in total hip replacement patients and RECORD3 compared 10 to 14 days of rivaroxaban to 10 to 14 days of enoxaparin in total knee replacement patients. RECORD1 and RECORD2 were pooled as they enrolled a comparable patient population. Rates of deep vein thrombosis (DVT) and pulmonary embolism were key inputs of the model. Additional data were taken from other studies using an indirect comparison methodology when required.
Monetary benefit and utility valuations:
Venous thromboembolism-related utility valuations were taken from published sources.
Measure of benefit:
The primary summary benefit measure was the number of symptomatic venous thromboembolism events avoided. Quality-adjusted life-years (QALYs) were used in an alternative analysis.
Cost data:
The economic analysis included the costs of drugs, administration, monitoring, diagnosis and treatment of venous thromboembolism as well as treatment of post-thrombotic syndrome and recurrent venous thromboembolism. Key unit costs were reported as well as data on quantities of resources used. Costs were derived mainly by using Medicare reimbursement rates supplemented with data from USA studies. Resource quantities were based on large USA databases for patients undergoing orthopaedic surgery. Costs were in US dollars ($). The price year was 2010. A 3% annual discount rate was used in the long-term analysis.
Analysis of uncertainty:
Alternative scenarios were considered in deterministic sensitivity analyses using confidence intervals derived from RECORD trials for specific events and clinical assumptions. A probabilistic sensitivity analysis investigated the issue of uncertainty using a multivariate approach and conventional probability distributions for groups of inputs based on published sources or authors’ assumptions. An alternative analysis was conducted using data from a fourth RECORD trial.