Does moderate alcohol consumption have long-term health benefits?: A causal inference perspective

Abstract

Heavy alcohol consumption is a risk factor for numerous health conditions, but at low-to-moderate levels, a substantial body of evidence suggests drinking might be associated with health benefits. Whether these are genuine protective effects remains unclear – in large part due to the difficulty of drawing causal inferences from observational research. To address this, this thesis applies novel statistical and design-based methods that promote causal inference to the analysis of alcohol–long-term health relationships.

Study 1 systematically reviews observational studies that employ enhanced causal inference methods in assessing alcohol’s relationship with any long-term health outcome. Limited evidence is found for protective effects for certain health outcomes, with a dearth of studies employing the approaches of interest. Clear gaps where these methods must be applied are identified. Study 2 assesses consistency of associational findings – a necessary precursor to causal inference – in the alcohol–inflammation relationship. The association of moderate drinking with reduced inflammatory biomarker levels seems robust to data processing and analysis decisions. Study 3 tests whether the association between moderate drinking and lower risk for depression is causal, using a marginal structural model. A small, protective effect for occasional and moderate alcohol consumption is found. Study 4 applies Mendelian Randomisation analyses to the alcohol–type 2 diabetes relationship. It finds a small protective effect for low-level drinking against risk for incident diabetes, but results are not robust to all sensitivity analyses. At most, any protective effect is limited to a small benefit in women only.

This thesis advances content knowledge relating to several complex, unanswered questions about alcohol–health relationships, as well as methodological applications, laying the foundations for causal inference approaches in this field.