Alcohol intake → Cognitive performance

Mechanism

Observational papers have long sold a J-curve: abstainers look worse than moderate drinkers, so a glass looks protective. Sick quitters, education, and who still drinks at 70 sit on that backdoor. Researchers use a genetic shortcut called Mendelian randomization. It is good at ruling out diet, income, and the decision to stop drinking. It can still be fooled when one gene affects two traits. Almeida, Hankey, Yeap, Golledge, and Flicker (2014) used ADH1B rs1229984 in 3542 Australian men aged 65 to 83. The A allele slows ethanol oxidation and, in this sample, cut regular drinking. If alcohol caused later impairment, A carriers should have been safer. They were not. Six years later, 502 men (14.2%) scored 23 or less on the Mini-Mental State Examination. AA versus GG odds ratio 1.35 (95% CI 0.29 to 6.27); GA versus GG 1.05 (0.71 to 1.55). Abstainers and irregular drinkers still looked worse than regular drinkers after adjustment (OR 1.23, 1.00 to 1.51). That is the J-curve. The instrument does not walk it in. Campbell, Fu, MacDonald, Zawistowski, Bakulski, and Ware (2024) ran two-sample Mendelian randomization of weekly alcohol (Liu 2019, 941,280 people, 59 SNPs, median F 37.8) on late-onset Alzheimer disease (Kunkle 2019: 21,982 cases, 41,944 controls). Inverse-variance weighted odds ratio 1.15 (0.78 to 1.72). Egger, weighted median, and MR-PRESSO were also null. In the 2012 Health and Retirement Study, doubling weekly drinks was not associated with dementia or cognitively impaired, non-dementia once instruments were used. The raw European-ancestry correlation still looked protective (dementia OR 0.62, 0.51 to 0.73). Cui, Si, Zhu, and Zhao (2022) used ALDH2 rs671 in 235 genotyped adults from rural China (mean age 49.7, 78% men, Mini-Cog). Ordinary least squares was null. Two-stage least squares was not: current drinking coefficient 0.271 (0.073 to 0.469) on the mild cognitive impairment indicator; first-stage F 34.88. That file is why confidence is contested.

Caveats

This page is adult alcohol intake and later cognition, mild cognitive impairment, or dementia. It is not prenatal alcohol and a child's school score. That exposure would need their own pages. Almeida is older men in Perth, one SNP, no baseline Mini-Mental State Examination, and a healthier returner sample. The A allele is uncommon. The authors say they would need about 196,000 men to call a 35% rise significant. PubMed typed the paper as a randomized trial. The design is Mendelian randomization. Cui genotyped 235 people. Mini-Cog is not a dementia workup. Drinking is the past 30 days. The model is linear on a yes/no outcome. Women drink little in that sample. Campbell's record here is a medRxiv preprint. The two-sample analysis is the stronger piece. African-ancestry instruments in the Health and Retirement Study were weak. European-ancestry 70-SNP scores were stronger (F 18.71). Mendelian randomization assumes no pleiotropy and a linear lifelong effect. It can miss a threshold in very heavy drinkers, and it does not test a night of intoxication. This is not medical advice and not a recommendation to drink, to abstain, or to treat a score.

Effect

odds_ratio = 1. Campbell two-sample AD IVW OR 1.15 (0.78-1.72). Almeida ADH1B AA/GG OR 1.35 (0.29-6.27). Cui 2SLS drinking-on-MCI b=0.271 (N=235).

Nodes

Named confounders

• Educational attainment
• Sleep duration and regularity
• Disposable income

Papers

• Alcohol consumption and cognitive impairment in older men: a mendelian randomization study.

  OP Almeida, GJ Hankey, BB Yeap, J Golledge, L Flicker · 2014 · Neurology

• Alcohol Consumption and Mild Cognitive Impairment: A Mendelian Randomization Study from Rural China.

  Y Cui, W Si, C Zhu, Q Zhao · 2022 · Nutrients

• Relationship between alcohol consumption and dementia with Mendelian randomization approaches among older adults in the United States.

  KA Campbell, M Fu, E MacDonald, M Zawistowski, KM Bakulski, EB Ware · 2024 · medRxiv : the preprint server for health sciences