Study Design May Bias the Estimation of the Health Effects of Low-dose Alcohol Consumption

Whether or not low-dose* alcohol consumption confers health benefits has been a recurring question. This systematic review and meta-analysis assessed data from 107 longitudinal studies to investigate whether certain study characteristics may bias estimates of all-cause mortality risk associated with low-dose alcohol consumption by comparing people with low-dose drinking* with those who are not drinking due to health reasons. The authors hypothesized that studies following people from a younger age, and characterizing abstinence as the total absence of drinking over the lifetime, would not find health benefits for low-dose drinking.

  • Studies with younger cohorts (i.e., mean cohort age of ≤55, followed up beyond 55 years), and studies separating people with former drinking from lifetime abstainers did not find health benefits associated with low-dose drinking (risk ratio [RR], 0.98).
  • Conversely, studies not meeting these criteria estimated lower risk among people with low-dose drinking compared with abstainers (RR, 0.84).

* Defined as consuming “between one drink per week (>1.30 g ethanol/day) and two drinks per day (<25 g ethanol/day).”

Comments: These results indicate that studying older cohorts and using comparison groups that include people with former drinking leads to biased estimates of the impact of low-dose alcohol consumption, making it appear to be potentially beneficial. These findings have important implications for appropriately assessing the impact of alcohol on population health, and providing credible drinking guidelines to the public.

Nicolas Bertholet, MD, MSc

Reference: Stockwell T, Zhao J, Clay J, et al. Why do only some cohort studies find health benefits from low-volume alcohol use? A systematic review and meta-analysis of study characteristics that may bias mortality risk estimates. J Stud Alcohol Drugs. 2024;85(4):441–452.

Post Your Comment

Comments are moderated and will not appear immediately.
Email address is for verification only; it will not be displayed.