The Role of Sick-Quitter Bias in the Association Between Alcohol Use and Anxiety and Depression Symptoms: A Causal Analysis in a U.S. National Cohort

Background: Observational studies frequently report a J- or U- shaped association between alcohol use and anxiety and depression, with moderate drinkers appearing to have lower risk than abstainers. However, this pattern may be driven in part by confounding due to health-related selection into abstinence or low-risk drinking (sick-quitter bias), whereby individuals reduce or stop drinking in response to declining health that is independently associated with worse mental health outcomes. We aimed to evaluate whether this association reflects a causal effect or residual confounding driven by sick-quitter bias. Methods: We analyzed data from 4,673 participants in the CHASING COVID Cohort, a U.S. longitudinal study with repeated measures from September 2021 through December 2023. Alcohol use was assessed at five timepoints using the AUDIT-C and categorized as abstinent, low-risk, moderate-risk, or high/severe-risk. Using a target trial emulation framework, we estimated the effects of sustained alcohol use strategies on anxiety and depression symptom severity (GAD-7 and PHQ-8) at follow-up. Artificial censoring and stabilized inverse probability weighting were applied to account for time-varying confounding and loss to follow-up. To assess the role of sick-quitter bias, we compared the abstinent versus moderate-risk contrast across strata of health status at time zero. Results: In the full sample, a J-shaped pattern was observed, with moderate-risk drinkers having the lowest predicted symptom scores. The abstinent versus moderate-risk symptom score contrast was 1.38 for depression (PHQ-8: 95% CI: 0.65, 2.14), and 0.95 for anxiety (GAD-7: 95% CI 0.24, 1.65). Among participants without underlying conditions or poor self-rated health, this pattern attenuated, with contrasts near zero for both outcomes. In contrast, the J-shaped pattern was amplified among participants with poorer baseline health, with abstinent versus moderate-risk contrasts of 3.31 for depression (95% CI: 2.14, 4.32) and 2.83 for anxiety (95% CI: 1.61, 3.86). High-risk drinking was consistently associated with higher symptom scores across all analyses. Conclusions: The J-shaped pattern between alcohol use and anxiety and depression symptoms observed in the full cohort was attenuated among participants without poorer baseline health, consistent with sick-quitter bias. Findings underscore the importance of addressing confounding due to declining health in observational alcohol research and caution against interpreting moderate drinking as beneficial for symptoms of anxiety and depression.