Cost-Effectiveness of Screening for Colorectal Cancer : Results from a Randomized Trial

In many European countries screening for colorectal cancer is introduced without a proper evidence of cost-effectiveness. In this paper we present a cost-effectiveness analysis of screening for colorectal cancer. A comprehensive dataset is collected from a Norwegian randomized trial of screening with once only flexible sigmoidoscopy in combination with faecal occult blood test. The trial gives the opportunity to study three unique features: Comparison of costs and effects between all individuals invited to the screening and the control group (intention-to-treat), and estimation of transition probabilities and development of a Markov model partially using observed data where uncertainty is explicitly modelled. Three questions are examined: How do the transitions between the Markov states well, colorectal cancer and death depend on age, screening method and participation in screening? What is the impact on cost of screening of including avoided treatment costs and production gains? What is the cost of screening per life year gained? Data at the level of each individual include information about trial costs, use of health care services, labour market participation, travel expenses, incidences of cancer with stage description, time of death and socio-economic variables as income, education, age, and gender. The transitions from well to colorectal cancer and from well to death are estimated by means of a multiple logit model, while the transition from colorectal cancer to death is estimated by means of duration analysis. The robustness of the results is tested with probabilistic sensitivity analyses. Preliminary results show that the transitions from well to both colorectal cancer and death are increasing with age. The cost per life year gained from sigmoidoscopy in combination with faecal occult blood test is favourable compared with other preventive health measures, but the results are not robust