The GBD 2020 Alcohol Collaborators study1GBD 2020 Alcohol CollaboratorsPopulation-level risks of alcohol consumption by amount, geography, age, sex, and year: a systematic analysis for the Global Burden of Disease Study 2020.Lancet. 2022; 400: 185-235Summary Full Text Full Text PDF PubMed Scopus (18) Google Scholar suggests alcohol use is harmful in younger adults but protective in older adults.1GBD 2020 Alcohol CollaboratorsPopulation-level risks of alcohol consumption by amount, geography, age, sex, and year: a systematic analysis for the Global Burden of Disease Study 2020.Lancet. 2022; 400: 185-235Summary Full Text Full Text PDF PubMed Scopus (18) Google Scholar The number of drinks per day equivalent to non-drinking is shown to rise from approximately one and a half standard drinks for individuals aged 50 years to approximately four standard drinks for individuals 80 years and older.1GBD 2020 Alcohol CollaboratorsPopulation-level risks of alcohol consumption by amount, geography, age, sex, and year: a systematic analysis for the Global Burden of Disease Study 2020.Lancet. 2022; 400: 185-235Summary Full Text Full Text PDF PubMed Scopus (18) Google Scholar Associations of alcohol use with disease and death are from cohort studies and case-control studies.1GBD 2020 Alcohol CollaboratorsPopulation-level risks of alcohol consumption by amount, geography, age, sex, and year: a systematic analysis for the Global Burden of Disease Study 2020.Lancet. 2022; 400: 185-235Summary Full Text Full Text PDF PubMed Scopus (18) Google Scholar Confounding undoubtedly introduces biases into estimates of moderate drinking because moderate drinkers probably have many protective moderate habits.2Naimi TS Brown DW Brewer RD et al.Cardiovascular risk factors and confounders among nondrinking and moderate-drinking U.S. adults.Am J Prev Med. 2005; 28: 369-373Summary Full Text Full Text PDF PubMed Scopus (260) Google Scholar More insidiously, alcohol use is typically a lifelong habit in adulthood, which could start before recruitment into a study. Specifically, for those who survive to recruitment, exposure time is too short. For those who do not survive to recruitment, exposure time and alcohol-related deaths are missing. Missing non-survivors can attenuate or reverse estimates.3WikipediaSurvivorship bias.https://en.wikipedia.org/wiki/Survivorship_biasDate: 2022Date accessed: July 16, 2022Google Scholar The older the age at recruitment the more the study is biased by only representing survivors of alcohol use. In a hypothetical cohort of people recruited at age 20 years, the effect of alcohol to age 60 years (at 40 years follow-up) could be harmful, but the effect in survivors from the same cohort recruited at 50 years of age could appear protective because of missing data for people who had already died from alcohol-related causes (appendix). Considering that a trial4Beckett NS Peters R Fletcher AE et al.Treatment of hypertension in patients 80 years of age or older.N Engl J Med. 2008; 358: 1887-1898Crossref PubMed Scopus (2467) Google Scholar on the treatment of hypertension in older people (80 years and older) showed that treatment had benefits despite contradictory observations, observational studies on the effects of alcohol use on death or disease that recruit older participants are open to unrecoverable bias and cannot provide a reliable guide to alcohol policy. We declare no competing interests. Download .pdf (.13 MB) Help with pdf files Supplementary appendix