The contribution of health behaviour to socioeconomic inequalities in alcohol harm: analysis of the UK biobank, a large cohort study with linked health outcomes

This is the first study to use the UK Biobank database to: 1) test whether participants of a low socioeconomic position (SEP) are less likely to drink, but more likely to suffer alcohol-related harm, and 2) test the contribution of behavioural factors. The database contains health-related information from 500,000 UK residents that were recruited aged 40–69 between 2006 and 2010. Our analysis focuses on participants resident in England (86% of the total sample). We obtained baseline demographics, survey data regarding alcohol consumption and other behaviours, and linked death and hospital-admission records. The primary outcome was time from study entry to experiencing an alcohol-attributable event (hospital admission or death). The relationship between alcohol-attributable harm and five measures of SEP (area-level deprivation, housing tenure, employment status, household income and qualifications) was investigated using time-to-event analysis. Average weekly alcohol consumption, other drinking behaviours (drinking history and beverage preference), and lifestyle factors (BMI and smoking status) were added incrementally as covariates in nested regression models to investigate whether they could explain the relationship between harm and SEP. 432,722 participants (197,449 men and 235,273 women) were included in the analysis with 3,496,431 person-years of follow-up. Those of a low SEP were most likely to be never/former drinkers or high-risk drinkers. However, alcohol consumption could not explain experiences of alcohol-attributable harm between SEP groups (Hazard Ratio (HR) 1.48; 95% Confidence Interval 1.45–1.51, after adjusting for alcohol consumption). Drinking history, drinking mostly spirits, an unhealthy Body Mass Index and smoking all increased the risk of alcohol-attributable harm. However, these factors only partially explain SEP differences in alcohol harm as the HR for the most deprived vs the least deprived was still 1.28 after adjustment. This suggests that improving wider health behaviour of the most deprived could reduce alcohol-related inequalities. However, a substantial proportion of the variance in alcohol harm remains unexplained.