Analytical approach:
This economic evaluation was based on a decision model with two periods: the prophylaxis period, which was up to 35 days for hip replacement or 10 to 14 days for knee replacement, and the post-prophylaxis period, which continued up to 90 days after surgery. The authors stated that they took the perspective of the US payer.
Effectiveness data:
The clinical inputs were from phase III clinical trials of rivaroxaban that formed part of the REgulation of Coagulation in ORthopaedic surgery to prevent Deep vein thrombosis and pulmonary embolism (RECORD) programme, which included over 12,500 patients. RECORD1 had 4,541 patients and compared extended rivaroxaban therapy with extended enoxaparin therapy in hip replacement patients. RECORD2 had 2,509 patients and compared extended rivaroxaban therapy with short enoxaparin therapy in hip replacement patients. RECORD3 had 2,531 knee replacement patients who received short therapy with either drug. The primary endpoint was the rate of total venous thromboembolism, defined as a composite of any deep vein thrombosis (DVT), non-fatal pulmonary embolism, or death. The main analysis assumed that the two treatments were equally effective for those endpoints where the trials did not find statistically significant differences between them.
Monetary benefit and utility valuations:
Not considered.
Measure of benefit:
The number of symptomatic events was the summary benefit measure.
Cost data:
The economic analysis included the costs of prophylaxis and treatment for venous thromboembolism. The quantities of resources were mainly from published studies, including the RECORD trials and other economic analyses conducted in the USA. Medicare reimbursement rates were used to estimate the costs of diagnosing venous thromboembolism or pulmonary embolism. All costs were in US $ and the price year was 2010.
Analysis of uncertainty:
Various one-way sensitivity analyses were carried out on selected inputs. The results of RECORD1 and RECORD2 were pooled, and data from RECORD4 were included with those of RECORD3 for the knee replacement population. A probabilistic sensitivity analysis was carried out using 95% confidence intervals from the clinical trials or published ranges of values.