Joint effects of alcohol use, smoking and body mass index as an explanation for the alcohol harm paradox: causal mediation analysis of eight cohort studies

Abstract Background and aims Lower socio‐economic status (SES) is associated with higher alcohol‐related harm despite lower levels of alcohol use. Differential vulnerability due to joint effects of behavioural risk factors is one potential explanation for this ‘alcohol harm paradox’. We analysed to what extent socio‐economic inequalities in alcohol‐mortality are mediated by alcohol, smoking and body mass index (BMI), and their joint effects with each other and with SES. Design Cohort study of eight health examination surveys (1978–2007) linked to mortality data. Setting Finland. Participants A total of 53 632 Finnish residents aged 25+ years. Measurements The primary outcome was alcohol‐attributable mortality. We used income as an indicator of SES. We assessed the joint effects between income and mediators (alcohol use, smoking and BMI) and between the mediators, adjusting for socio‐demographic indicators. We used causal mediation analysis to calculate the total, direct, indirect and mediated interactive effects using Aalen's additive hazards models. Findings During 1 085 839 person‐years of follow‐up, we identified 865 alcohol‐attributable deaths. We found joint effects for income and alcohol use and income and smoking, resulting in 46.8 and 11.4 extra deaths due to the interaction per 10 000 person‐years. No interactions were observed for income and BMI or between alcohol and other mediators. The lowest compared with the highest income quintile was associated with 5.5 additional alcohol deaths per 10 000 person‐years (95% confidence interval = 3.7, 7.3) after adjusting for confounders. The proportion mediated by alcohol use was negative (−69.3%), consistent with the alcohol harm paradox. The proportion mediated by smoking and BMI and their additive interactions with income explained 18.1% of the total effect of income on alcohol‐attributable mortality. Conclusions People of lower socio‐economic status appear to be more vulnerable to the effects of alcohol use and smoking on alcohol‐attributable mortality. Behavioural risk factors and their joint effects with income may explain part of the alcohol harm paradox.