LAUSR

Background Disease incidence and premature deaths tend to be influenced by multiple health risky behaviours, including smoking, excessive alcohol consumption and unhealthy diet. Risky behaviours tend not to be independent and may have a multiplicative effect on disease incidence and healthcare cost. Thus, understanding the interrelationship between health behaviours and their effect on health outcomes is crucial in designing behavioural intervention programmes. Objective To examine the interrelationship between health risky behaviours and associated disease outcomes amongst Scottish adults. Methods We use hospitalisation episode data from the Scottish Morbidity Records, (SMR), that have been administratively linked to Scottish Health Surveys (SHeS) respondents with target disease defined by relevant ICD9 and 10 codes. We apply a recursive multivariate probit model to jointly estimate the health risky behaviours and disease incidence to adequately control for unobserved heterogeneity. The model is estimated separately by gender. Results Modelling health risk behaviours and disease incidence equations independently rather than jointly may be misleading. We find a clear socioeconomic gradient predicting health risky behaviours and the results differ by gender. Specifically, smoking appears to be a key driver of other health risky behaviours. Current smokers are more likely to be drinking above the recommended limit, physically inactive, and eating inadequate diet. Conclusions Interventions targeting current smokers to quit could spillover to other behaviours by reducing excessive drinking, improve physical activity and adequate diet. Thus, improvements in one behaviour may increase the likelihood of adopting other healthier lifestyle behaviours.