Analytical approach:
The analysis was based on a decision-analytic model with a hypothetical cohort of hospitalised patients requiring a central venous catheter. A one-year time horizon was considered. The authors stated that the analysis was carried out from the perspective of the hospital.
Effectiveness data:
The clinical inputs were from relevant published sources. The key input was the relative risk reduction in catheter-related bloodstream infections (CR-BSIs) with the sponge dressing. This was the efficacy of the preventive treatment and was from two randomised controlled trials. The rates of colonisation were from a published meta-analysis. Other incidence rates were from the Centers for Disease Control and Prevention, National Healthcare Safety Network. Some assumptions were made.
Monetary benefit and utility valuations:
Not considered.
Measure of benefit:
The number of CR-BSIs was the main benefit measure. Other clinical outcomes, such as the number of local site infections and the number of deaths from CR-BSIs, were reported.
Cost data:
The economic analysis included the hospital costs of the sponge dressing, the management of local site infections, and the management of CR-BSIs. These costs were from published studies and an estimate provided by the Centers for Disease Control and Prevention. The cost of a local site infection included the costs of blood, catheter tip, cultures, intravenous or oral antimicrobial, and the replacement of the central venous catheter. All costs were in US dollars ($).
Analysis of uncertainty:
One-way sensitivity analyses were carried out to assess how robust the model results were to variations in the following inputs: the proportion of impregnated versus standard catheters, the average duration of catheter implantation, the rate of CR-BSIs, the rate of local infections, the percentage decrease in risk of CR-BSIs with the sponge dressing, the average number of dressings per catheter, the cost of a CR-BSI, the cost of a local infection, and the price of the sponge dressing. Plausible ranges or confidence intervals were used, where available, otherwise a range of ±25% of the base case value was applied.