The analysis was based on a probabilistic decision-analytic model. The time horizon was the length of hospital stay, which was an average of seven days. The authors stated that the analysis was carried out from an institutional perspective.
The clinical data were from a selection of relevant studies. The probabilities of deep vein thrombosis and major bleeding while on LMWH or unfractionated heparin were the key inputs. These data were from a published meta-analysis of nine clinical trials. The remaining data were from published economic evaluations, meta-analyses, and diagnostic studies.
Monetary benefit and utility valuations:
Measure of benefit:
The summary benefit measures were the number of uncomplicated deep vein thromboses and a combined measure of untoward events, which included pulmonary embolism, major bleeding, and death.
The economic analysis included the direct medical costs incurred during initial hospitalisation, which were drugs, diagnostic tests, and the treatment of major bleeding or VTE (deep vein thrombosis or pulmonary embolism). The drug costs used their official prices at the Vancouver General Hospital in-patient pharmacy. The cost of major bleeding was from published pharmacoeconomic analyses conducted in Canada. The remaining costs were based on official rates in the province of British Columbia. All costs were in Canadian dollars (CAD) and the price year was 2009.
Analysis of uncertainty:
One-way sensitivity analyses were carried on the acquisition costs of LMWH and unfractionated heparin, the cost of managing a pulmonary embolism and major bleed, the rate of deep vein thrombosis, the probability of progression to pulmonary embolism without treatment, and the length of stay for complicated or uncomplicated deep vein thrombosis. The ranges of values were from published sources or were assigned by the authors. An alternative scenario included the nursing and pharmacy labour costs. Subgroup analyses were carried out, using published data, for patients older than 75 years, and for those with a history of deep vein thrombosis, chronic respiratory disease, pneumonia, heart failure, or malignancy. A probabilistic sensitivity analysis was carried out, using a Monte Carlo simulation and predetermined probability distributions for each type of model input. Cost-effectiveness acceptability curves were plotted.