Substance Use
Revisiting the Protective Association of Moderate Alcohol Use: A Positive Control Analysis Anna Shchetinina* Anna Shchetinina Shchetinina Shchetinina Shchetinina Shchetinina Shchetinina Harvard T.H. Chan School of Public Health
Introduction
Whether moderate alcohol consumption confers protective health effects has been extensively debated. This study is motivated by the hypothesis that frequently observed inverse associations between moderate alcohol use and health outcomes primarily reflect bias that is not fully addressed by conventional adjustment strategies.
Methods
Using the Nurses’ Health Study II, we utilized a positive control design leveraging prenatal alcohol exposure (PAE) reported by mothers of nurse-participants. PAE is a well-established teratogen with no plausible protective health effects. We compared associations of adult light-to-moderate alcohol use and PAE with cancers and tumors, cardiometabolic, respiratory, and other chronic conditions later in life. E-values quantified the minimum strength of unmeasured confounding required to fully explain observed associations.
Results
Analytic sample included 21,106 nurse-participants, of whom 62% reported light or moderate alcohol use. Participants who engaged in drinking had slightly higher prevalences of cancers and tumors, while most other conditions were more prevalent among non-drinkers. In positive control analyses, PAE exhibited patterns of association that closely paralleled those observed for adult alcohol use across a wide range of health outcomes. Specifically, PAE was associated with modestly elevated odds of selected cancers, whereas most cardiometabolic, gastrointestinal, endocrine, and other chronic conditions showed inverse associations. E-values indicated that modest unmeasured confounding (generally<2) could explain the observed associations.
Conclusions
Inverse associations commonly interpreted as protective effects of moderate alcohol use were also observed for a known harmful exposure, providing strong evidence that these findings are likely driven by shared residual confounding rather than causal effects. These results underscore caution against interpreting moderate alcohol consumption as health-promoting.
