The causal effect of cigarette smoking on healthcare costs

Knowledge of the impact of smoking on healthcare costs is important for establishing the external effects of smoking and for evaluating policies intended to modify this behavior. Conventional analysis of this association is difficult because of omitted variable bias, reverse causality, and measurement error. We approached these challenges using a Mendelian Randomization study design, in which genetic variants associated with smoking behaviors were used as instrumental variables. We undertook genome wide association studies to identify genetic variants associated with smoking initiation and a composite index of lifetime smoking on up to 300,045 individuals in the UK Biobank cohort. These variants were used in two-stage least square models and a variety of sensitivity analyses. All results were concordant in indicating a substantial impact of each smoking exposure on annual inpatient hospital costs Our results indicate a substantial impact of smoking on hospital costs. Genetic liability to initiate smoking - ever versus never having smoked - was estimated to increase mean per-patient annual hospital costs by {pound}477 (95% confidence interval (CI): {pound}187 to {pound}766). A one unit change in genetic liability a composite index reflecting the cumulative health impacts of smoking was estimated to increase these costs by {pound}204 (95% CI: {pound}105 to {pound}303). Models conditioning on the causal effect of risk tolerance were not robust to weak instruments for this exposure. Our findings have implications for the scale of external effects that smokers impose on others, and on the probable cost-effectiveness of smoking interventions.