Analytical approach:
A stochastic smoking progression model was used to extrapolate the trial results to estimate the benefits of the prevention programme over the students' lifetime. The authors reported that a societal perspective was adopted.
Effectiveness data:
The effectiveness data were primarily obtained from a randomised controlled trial (RCT) on the SFC programme. This RCT included 131 classes, either in the SFC or control group, and an overall sample of 2,142 students. Smoking status (four-week smoking prevalence) was assessed at one month and one year of follow-up. Further details were reported elsewhere (Wiborg, et al. 2002, see ‘Other Publications of Related Interest’ below for bibliographic details). The smoking progression model combined data from two studies and official sources. The primary outcome was the number of participants who were prevented from becoming established smokers.
Monetary benefit and utility valuations:
The benefits were reported in monetary terms, for a prevented smoker, and these were derived from two studies. These benefits accounted for the direct health care costs and productivity losses, due to missed work, based on the human capital approach.
Measure of benefit:
The monetary value of preventing a person from becoming an established smoker was the measure of benefit. This was discounted at an annual rate of 5%.
Cost data:
The programme costs, smoking-related health care costs, and productivity costs were included. Programme costs were based on actual data, obtained from the agency that implemented the SFC, and included the costs of personnel, travel, materials, and overheads. At class level, the direct costs such as materials and communication were included as well as the productivity costs of classroom time. All costs were reported in Euros (EUR) for the price year 2001 to 2002.
Analysis of uncertainty:
One-way sensitivity analysis was performed on specific parameters and a number of modelling assumptions. All the specific parameters and the ranges over which they were tested were reported. Probabilistic sensitivity analysis was also conducted using Monte Carlo simulation and all the parameters that were investigated in the one-way sensitivity analysis, which were assigned triangular or normal distributions.