Dementia Risk With Low Amounts of Alcohol
Observational studies of the health effects of alcohol have many methodological challenges. Investigators attempted to address one by examining midlife drinking and dementia risk over 23 years. Of 10,308 adults studied prospectively, 1221 were excluded due to death, cardiometabolic disease, or missing data on alcohol (all of which could be related to dementia); 397 cases of dementia were detected in medical records codes.
- In analyses adjusted for socioeconomic status, health behaviors, and comorbidity, the hazard ratio for dementia was 1.45 for drinking <0.7 drinks in a week compared with 0.7–10 in a week, and 1.18 for every 5-drink increase over 10 in a week (1 drink=14g of alcohol, standard in the US).
Comments: In analyses adjusted for socioeconomic status, health behaviors, and comorbidity, the hazard ratio for dementia was 1.45 for drinking 8 drinks in a week was associated with dementia. These associations may not be causal—other factors associated with alcohol use and dementia may be responsible for the risk. This study translated self-reported measures into weekly averages and did not measure heavy episodes (common among people who drink). Furthermore, most people categorized as “abstinent” drank as recently as one week prior. These decisions raise questions about the ability to interpret and attribute differences in dementia to abstinence or drinking low amounts. Clearly one cannot rely on such observational studies to dose a carcinogen with a narrow therapeutic index to achieve or avoid any health outcome. And if the inflection point for increasing dementia risk is 8 drinks in a week, then most guidelines are too liberal.
Richard Saitz, MD, MPH
Reference: Sabia S, Fayosse A, Dumurgier J, et al. Alcohol consumption and risk of dementia: 23-year follow-up of Whitehall II cohort study. BMJ. 2018;362:k2927.
2 comments
Is it implicit that you adjusted for the English definition of a drink (standard alcohol unit)? Is this correct that US standard unit is 12-14 grams ethanol and UK standard is 8-10g? So your 10 drinks per week threshold (US) is roughly 14 drinks per week (UK). Accurate?
In our summary all of the drinks are in standard US 14g size–we calculated them based on the units in the paper. You could back-calculate to UK drinks from our US drinks (10 US drinks would be 10*14g=140g and 140g is 17.5 UK units) but even better would be to just use what is published in the original paper if you want to speak to the British public. HOWEVER, we noticed that there is an error in the summary–Thank you!
The corrected version is:
In analyses adjusted for socioeconomic status, health behaviors, and comorbidity, the hazard ratio for dementia was 1.45 for drinking 8 drinks in a week was associated with dementia. These associations may not be causal—other factors associated with alcohol use and dementia may be responsible for the risk. This study translated self-reported measures into weekly averages and did not measure heavy episodes (common among people who drink). Furthermore, most people categorized as “abstinent” drank as recently as one week prior. These decisions raise questions about the ability to interpret and attribute differences in dementia to abstinence or drinking low amounts. Clearly one cannot rely on such observational studies to dose a carcinogen with a narrow therapeutic index to achieve or avoid any health outcome. And if the inflection point for increasing dementia risk is 8 drinks in a week then most guidelines are too liberal.