| Summary: | Links between health behaviours are potentially of great importance for cost-effectiveness estimates in economic evaluations of behaviour change strategies. This thesis investigates bias in economic evaluations due to the omission of links between health behaviours. A framework to evaluate the health costs and consequences of behaviour change strategies while accounting for links to other health-related behaviours is proposed, and tested using a case study of the link between tobacco and alcohol use. There is strong evidence of correlation between alcohol use and tobacco use. Nevertheless, while many economic evaluations of interventions for alcohol and tobacco behaviour change have been conducted and used to inform resource allocation decisions, none have considered implications of links between the two behaviours. Assumptions about behaviour beyond typical trial endpoints in historical economic evaluations have in general been based on limited follow-up data. Analysis of the joint dynamics of tobacco and alcohol use employing large-scale longitudinal survey data is used in this thesis as an alternative to inform assumptions about long-run smoking behaviour and its link to alcohol use, in a de novo individual-level simulation to appraise the cost-effectiveness of smoking cessation interventions. Smoking behaviour is found to be persistent and dynamically linked to alcohol use, but also influenced by numerous other factors, including unobserved time-invariant person-specific characteristics which determine the propensity to smoke. The ability of smoking cessation interventions to permanently reduce the propensity to smoke is important for their long-run cost-effectiveness. In the absence of data on this effect, historical economic evaluations may have misinformed decision makers. There is a need for robust and tested assumptions about long-term behaviour and case by case consideration of the importance of inter-behavioural links in future economic evaluations of behaviour change strategies, and for further investigation into the dynamics and interrelation of health-related behaviours.
|
|---|
