Critique 218: Long-term study of alcohol intake and the risk of Alzheimer Disease or other types of dementia ——— 21 August 2018

Sabia S, Fayosse A, Dumurgier J, Dugravot A, Akbaraly T, Britton A, Kivimäki M, Singh-Manoux A. Alcohol consumption and risk of dementia: 23 year follow-up of Whitehall II cohort study. BMJ 2018;362:k2927. http://dx.doi.org/10.1136/bmj.k2927

Authors’ Abstract

Objective To examine the association between alcohol consumption and risk of dementia.
Design Prospective cohort study.
Setting Civil service departments in London (Whitehall II study).Participants 9087 participants aged 35-55 years at study inception (1985/88).
Main outcome measures Incident dementia, identified through linkage to hospital, mental health services, and mortality registers until 2017. Measures of alcohol consumption were the mean from three assessments between 1985/88 and 1991/93 (midlife), categorised as abstinence, 1-14 units/week, and >14 units/week; 17 year trajectories of alcohol consumption based on five assessments of alcohol consumption between 1985/88 and 2002/04; CAGE questionnaire for alcohol dependence assessed in 1991/93; and hospital admission for alcohol related chronic diseases between 1991 and 2017.
Results 397 cases of dementia were recorded over a mean follow-up of 23 years. Abstinence in midlife was associated with a higher risk of dementia (hazard ratio 1.47, 95% confidence interval 1.15 to 1.89) compared with consumption of 1-14 units/week. Among those drinking >14 units/week, a 7 unit increase in alcohol consumption was associated with a 17% (95% confidence interval 4% to 32%) increase in risk of dementia. CAGE score >2 (hazard ratio 2.19, 1.29 to 3.71) and alcohol related hospital admission (4.28, 2.72 to 6.73) were also associated with an increased risk of dementia. Alcohol consumption trajectories from midlife to early old age showed long term abstinence (1.74, 1.31 to 2.30), decrease in consumption (1.55, 1.08 to 2.22), and long term consumption >14 units/week (1.40, 1.02 to 1.93) to be associated with a higher risk of dementia compared with long term consumption of 1-14 units/week. Analysis using multistate models suggested that the excess risk of dementia associated with abstinence in midlife was partly explained by cardiometabolic disease over the follow-up as the hazard ratio of dementia in abstainers without cardiometabolic disease was 1.33 (0.88 to 2.02) compared with 1.47 (1.15 to 1.89) in the entire population.
Conclusions The risk of dementia was increased in people who abstained from alcohol in midlife or consumed >14 units/week. In several countries, guidelines define thresholds for harmful alcohol consumption much higher than 14 units/week. The present findings encourage the downward revision of such guidelines to promote cognitive health at older ages.

Forum Comments

The vast majority of well-done prospective studies indicate that, in comparison with non-drinkers, moderate, non-binge-drinking older adults have a lower risk of cardiovascular disease and total mortality. Most studies also suggest that moderate drinkers tend to have a lower risk of developing dementia, including Alzheimer disease.

The present study provides important information on the association of alcohol and dementia by following a large cohort of British civil servants over a mean period of 23 years, with repeated assessments of alcohol consumption. The main results indicate that abstinence in middle life is associated with a significantly higher risk of dementia than the risk among moderate drinkers, while subjects reporting the intake of larger amounts of alcohol or evidence of an alcohol use disorder are at increased risk of dementia. However, the data presented do not allow for a firm determination of a possible cut-point for increased risk from alcohol intake.

Overview of paper: All Forum members considered that this was a well-done analysis, as important strengths included a database consisting of a large, well-described cohort of British civil subjects followed by many years; assessments of alcohol consumption throughout midlife, as well as evidence of alcohol dependence and hospital admission for alcohol related disease; and with repeated assessments of alcohol intake (on 8 occasions during follow up) having the ability to construct trajectories of alcohol consumption over 17 years. Further, the authors were able to classify current non-drinkers into long-term abstainers versus ex-drinkers; finding similar associations with dementia for all current non-drinkers (including “ex-drinkers” and “occasional drinkers”), they combined all into one category, “abstinence”. Regarding long-term patterns of alcohol intake over “middle age” (the pattern between assessments at a median age of about 45 years to that at about 61 years), the authors used their repeated assessments of consumption to construct trajectories of intake. The identified trajectories were long-term abstinence, decreased alcohol consumption, long term consumption of 1-14 units/week, increased consumption, and long term consumption of >14 units/week.

Finally, data from electronic health records of three public databases were used to ascertain the diagnosis of dementia, using well-constructed and valid algorithms. In addition, the investigators examined whether cardiometabolic disease modifies the association between alcohol consumption and dementia, using appropriate measurements to determine cardiovascular risk factors and disease during the follow-up period.

The authors report: “Among the 10 231 participants alive in 1991/93, 9087 had at least two measurements of alcohol consumption between 1985/88 and 1991/93 and complete data on covariates. Among these participants, a total of 397 cases of dementia were recorded over a mean follow-up of 23.2 (SD 4.4, range 0.08-25.6) years. Mean age at dementia diagnosis was 75.6 (SD 5.8) years.” There were five trajectories of alcohol consumption: long term abstinence (9% of subjects), decreased consumption (6%), long-term consumption of 1-14 units/week group (59%), increased consumption (11%), and long-term consumption >14 units/week (14%). With a referent group of participants in the long-term consumption of 1-14 units/week group (“moderate drinkers”), those with long-term abstinence showed a relative risk of dementia of 1.74, CI 1.31 to 2.30; those who decreased consumption also showed an increase in risk, 1.55, CI 1.08 to 2.22. For subjects with-long-term consumption >14 units/week, there was also an increased risk of dementia (1.40, CI 1.02 to 1.93). These associations remained after adjustment for behavioural and health related factors.

The data suggest that some of the adverse effects of abstinence and greater amounts of alcohol related to effects of alcohol on cardiovascular disease. The authors report that among those without cardiometabolic disease, the risk of dementia for the abstinence category was less, HR 1.33 (CI 0.88 to 2.02), and with consumption >14 units/week it was 1.28 (CI 0.85 to 1.92). The investigators state: “Results for dementia from the modified Fine and Gray model that accounts for competing risks of mortality were similar to those in the main analysis.”

Specific comments by Forum members: Reviewer McEvoy noted: “I think this is a very well done study that provides further support to the growing literature that moderate alcohol consumption is associated with a reduced risk of dementia. Particular strengths of the study are the use of multiple assessments of reported alcohol use to determine mid-life drinking habits, with five subsequent assessments to characterize trajectories of alcohol use from midlife to early older age. They also had repeated assessment of health behaviors, and were able to adjust for socioeconomic status based on occupational position. The results are compelling in showing the protective association of moderate drinking at midlife and at later ages with reduced risk of dementia, and that continued moderate drinking with aging is associated with reduced dementia risk, while long term abstinence is associated with the highest risk. The longitudinal data and use of repeated assessments to characterize midlife drinking rules out the ‘sick quitter’ hypothesis to explain increased risk in the non-drinking group.”

Forum member Stockley stated: “This is a well done study which builds on the previous papers published by Sabia et al in 2010 and 2014 on different facets of cognitive decline, dementia, and alcohol. Although there is variation in methodology between observational studies, analyses consistently suggest that, on balance, there is a J- or U-shaped relationship between alcohol consumption and the risk of cognitive decline or dysfunction and the development of dementias such as Alzheimer’s disease. Consumption of approximately >14 units/week being associated with an increase in risk of dementia has also been observed to increase the risk of certain CVDs which gives credence to observations supporting similar or related biological mechanisms (lipids, blood clotting, blood flow) reducing risk between cognitive decline and cardiovascular disease.”

Forum member Lanzmann-Petithory had a number of comments: “In addition to the effects of underreporting of intake, which has been well described by Klatsky et al (2006, 2014) and others to relate to risk of disease, the drinking pattern seems to me an important factor among the >14 units/week group. This group probably consists of a mix of steady drinkers of 2-3 units/day and binge drinkers of >14 units during week-ends; this could represent a confounding factor and explain the non-significant increased risk in those reporting >14 units/week in the present study.

“Further, since a big component of dementia relates to cerebral atherosclerosis and stroke, it is not surprising that cardiovascular risk factors play a role in dementia. Also, binge drinking is an independent risk factor for all strokes and ischemic stroke (Sundell et al, Renaud). The type of alcohol plays a role as shown in the Framingham Heart Study for ischaemic stroke and wine (Djoussé et al), and to some extent in the present study as the J curve seems to me more pronounced for wine in the supplement material, with the most significant p-value (<0.001) by far for wine consumption, being lower in the dementia group than in the non-dementia group. In the text, the authors do not hide this fact, stating that in beverage-specific analyses, ‘The study also found a reduced risk of dementia for moderate wine consumption and a linear increased risk of dementia in those consuming spirits.’

“Finally, as mentioned by Van Velden, genetics play an important role for dementia. In some studies, risk of dementia increases with increasing alcohol consumption only in those individuals carrying the apolipoprotein E4 allele (Anttila et al). For all these reasons, the present study does not represent at all for me a solid support to a decrease of quantities in the new alcohol guidelines in the UK, except perhaps for subjects with the e4 allele.”

At what level of drinking does alcohol consumption increase the risk of dementia? The authors state: “Alcohol consumption >14 units/week increased the risk of dementia in a linear fashion; an excess risk that was evident when alcohol consumption was assessed at ages 50, 60, and 70 years. Data using hospital admission for chronic disease caused by high alcohol consumption showed a four times higher risk of dementia, supporting findings on the neurotoxic effects of alcohol consumption >14 units/week.” Forum members suggest that these findings are in line with earlier research indicating that heavy alcohol consumption (especially abuse) increases not only mortality but the risk of dementia, but the present data do not allow a determination of a firm cut-point for an increase in risk.

The authors state in the text, “Regardless of type of alcohol consumed, the risk of dementia increased linearly, starting around 14 units/week.” However, the figures in the text show that the lower 95% confidence level for wine and spirits remain below 1.0 (consistent even with perhaps a decrease in risk) regardless of the level of alcohol, and only beer shows that the lower 95% CI goes above 1.0 (statistically significant probably because of larger numbers). Also, there are very few subjects with greater consumption of alcohol-containing beverages. And, as stated, they do find healthier effects for wine consumers.

Forum members Ellison and Zhang state: “We question the authors’ interpretation of the linear increase shown in their data as indicating that >14 units/week leads to increased dementia. With only linear analyses and no spline analyses above 14 drinks/week, it is difficult to estimate any specific cut-point when the risk of dementia increases. Based on the data presented, it could be that only the true alcohol abusers (who had hospital admissions for abuse), who were included in the risk of heavier drinkers, increased their risk; we cannot determine from what is here what a reasonable cut-point for adverse effect might be.”

Also, Forum member McEvoy noted: “I disagree with the authors’ conclusion that this paper suggests that upper limits for drinking recommendations in other countries should be reduced to no more than 14 units per week. Not only is there the concern of under-reporting, which may be exaggerated in an occupational cohort, the graphs in the publication show that hazard ratios do not begin to exceed 1.0 until about 20 units of intake, and do not become statistically significant (with lower confidence limit above 1) until at least 30 units. Further, I agree with other forum members about the problem with reporting alcohol intake as ‘units’. The paper does not seem to provide the translation of units into grams, or into typical drinks, and this makes it hard for the general reader to understand what amount of drinking may be beneficial and what may be harmful.” (This concern is discussed below.)

Forum member Keil notes that in the Lancet paper by Ruitenberg et al, the largest protection against all types of dementia in the Rotterdam Study was 1-3 drinks/day. In the Keil et al study from Germany, the increase in CVD and mortality was between the 20-39 g/day and the 40 g/day categories of alcohol intake, and in some analyses only for the more than 80 g/day groups.”

The authors also state: “Multistate models showed that part of the excess risk of dementia in abstainers was attributable to the greater risk of cardiometabolic disease in this group. Taken together, these results suggest that abstention and excessive alcohol consumption are associated with an increased risk of dementia, although the underlying mechanisms are likely to be different in the two groups. Overall, no evidence was found that alcohol consumption between 1 unit/week and 14 units/week increases the risk of dementia.” Forum members agree that while risk factors for dementia are not well known, cardiovascular risk factors have previously been found to play some role in the risk of cognitive impairment.

What are reasonable measures of alcohol intake? Reviewer Finkel noted: “On a personal note, I take this opportunity to express my continuing annoyance with the use, chiefly in Britain, of units (defined as 8 grams of alcohol) as a quantitation of alcohol consumption. We already have a perfectly good near universal system of measurement: grams of alcohol. Remember that the metric system and, for some, the so-called English system (called in modern times the imperial system, except in break-away colonies) works well. To add another layer makes no sense to me, just requires more mental gyrations inserted between observation and understanding. It is beyond silly to seek increased precision by inventing new measuring standards when the basic data are by their nature imprecise. Who orders a 1.5-unit glass of Pinot Noir? Who can with straight face report alcohol consumption as, say, ‘15 units per week’”?

Forum member Keil stated: “I couldn`t agree more with Finkel concerning the paper and the unit system. The main findings reflect those in the paper by Ruitenberg et al from the Rotterdam group (which, surprisingly, is not cited by the authors). Perhaps the statement in the Ruitenberg et al paper is too clear for the authors of the present paper: ‘These findings suggest that light to moderate alcohol consumption is associated with a reduced risk of dementia in individuals aged 55 years or older. The effect seems to be unchanged by the source of alcohol.’”

Keil continued: “A ‘unit’ of alcohol in the UK is 8 grams. This is a really small amount of alcohol, as a 0.1 liter glass of wine amounts to about 10 grams of alcohol and the often consumed 0.2 liter glass of wine to 20 grams of alcohol. The smallest amount of beer you can regularly consume in Germany is from a 0.25 liter glass of beer which amounts to 10 grams of alcohol. Much more often you find the following glass sizes: 0.3, 0.4, 0.5, 1.0 liter corresponding to 12, 16, 20 and 40 grams of alcohol, respectively. This shows to me that the 8 grams of alcohol unit in the UK is purely theoretical. Any practitioner should laugh about this value for a unit. Why scientists do not accept the metric system counting alcohol consumption in grams per day or week is an enigma to me. Anyhow, the English will soon break away from the EU with a harsh or hard BREXIT. They can then also still keep their old fashioned and unscientific measures.

“In a paper from Augsburg in 1997 we counted alcohol in grams. A 0.3 liter glass of beer means 12 grams of alcohol and a one liter mug (eine Maß!!) at the Oktoberfest means 40 grams of alcohol (Keil et al). However, we should keep in mind that the Whitehall project is an occupational cohort study; it is well known that occupational studies are prone to underreporting of alcohol consumption. I have seen this for example in the well known PROCAM study in the region of Münster, which has also recruited civil servants and collected ‘peculiar’ data on alcohol consumption.”

Reviewer Skovenborg wrote: “I agree with the comments about using units of alcohol. In addition there is the general problem of underreporting and the specific British problem of underreporting the small British unit of 8 grams, as drinks usually have an alcohol content of more than 8 grams.” Others noted that the finding that females have more dementia than males is confirmed by many other studies. It can be partially explained by lower levels of enzymes metabolizing alcohol among women, but also it is known that females tend to have on average a lower socio-economic status and a lower level of education, which may make them more vulnerable to adverse effects of alcohol consumption (as shown most recently by Colpani et al).

Implications of the present study results on drinking guidelines: Forum member Keil noted: “Projections on the increase of dementia should be interpreted with caution because recent studies clearly show that dementia is on the decline when you look at age-specific rates of dementia. With increasing educational status, especially in women, age-specific rates of dementia will probably decline. (More educated women also drink more alcohol). Is it true that ‘What is good for the heart is good for the brain?’ The Mediterranean diet is obviously good for the heart and general well being and it is usually combined with a glass of wine, preferably a 0.2 liter glass = approximately 20 grams of alcohol.”

Forum member Van Velden wrote: “All that I can add is the fact that the authors did not take into consideration genetic risk factors for CVD. People with Apo-E4 polymorphisms are more at risk for CVD, and alcohol may increase their risk for cardio-metabolic disease. Diet-health implications cannot be simplified and generalized for all people with different genetic backgrounds.”

In their Discussion, the authors state: “First, the risk of dementia was higher in those abstaining from alcohol in midlife. Alcohol consumption trajectories from midlife to early old age supported these findings – both long term abstainers and those reporting decreased alcohol consumption had an increased risk of dementia.” Forum members conclude that the implication from these findings strongly support previous findings of a lower risk of dementia for moderate drinkers. Further, these results suggest that truly moderate drinkers in middle age should not be advised to stop their alcohol consumption.

References from Forum critique

Anttila T, Helkala EL, Viitanen M, Kåreholt I, Fratiglioni L, Winblad B, et al. Alcohol drinking in middle age and subsequent risk of mild cognitive impairment and dementia in old age: a prospective population based study. BMJ 2004;329(7465):539. doi: 10.1136/bmj.38181.418958.BE
https://www.ncbi.nlm.nih.gov/pmc/articles/PMC516103/pdf/bmj32900539.pdf

Colpani V, Baena CP, Jaspers L, van Dijk GM, Farajzadegan Z, Dhana K, Tielemans MJ, Voortman T, Freak-Poli R, Veloso GGV, Chowdhury R, Kavousi M, Muka T. Franco OH. Lifestyle factors, cardiovascular disease and all-cause mortality in middle-aged and elderly women: a systematic review and metaanalysis. Eur J Epidemiol 2018; doi: 0.1007/s10654-018-0374-z. [Epub ahead of print]

Djoussé L, Ellison RC, Beiser A, Scaramucci A, D’Agostino RB, Wolf PA. Alcohol consumption and risk of ischemic stroke: The Framingham Study. Stroke 2002;33:907-912.
https://www.ncbi.nlm.nih.gov/pubmed/11935035

Keil U1, Chambless LE, Döring A, Filipiak B, Stieber J. The relation of alcohol intake to coronary heart disease and all-cause mortality in a beer-drinking population. Epidemiology 1997;8:150-156.

Klatsky AL, Gunderson E, D G, Kipp H, Udaltsova N, Friedman GD. Higher prevalence of systemic HTN among moderate alcohol drinkers: exploring the role of under-reporting. J Stud Alcohol 2006;67:421–428.

Klatsky AL, Udaltsova N, Li Y, Baer D, Tran HN, Friedman GD. Moderate alcohol intake and cancer: the role of underreporting. Cancer Causes Control 2014;25:693. https://doi.org/10.1007/s10552-014-0372-8.

Renaud SC. Diet and stroke. J Nutr Health Aging 2001;5:167-172. https://www.ncbi.nlm.nih.gov/pubmed/11458287

Ruitenberg A, van Swieten JC, Witteman JC, Mehta KM, van Duijn CM, Hofman A, Breteler MM. Alcohol consumption and risk of dementia: the Rotterdam Study. Lancet 2002;359(9303):281-286.

Sabia S, Elbaz A, Britton A, Bell S, Dugravot A, Shipley M, Kivimaki M, Singh-Manoux A. Alcohol consumption and cognitive decline in early old age. Neurology 2014;82:332-339.

Sabia S, Kivimaki M, Kumari M, Shipley MJ, Singh-Manoux A. Effect of Apolipoprotein E epsilon 4 on the association between health behaviors and cognitive function in late midlife. Mol Neurodegener 2010;5:23.

Sundell L, Salomaa V, Vartiainen E, Poikolainen K, Laatikainen T. Increased stroke risk is related to a binge-drinking habit. Stroke 2008;39:3179-3184. doi: 10.1161/STROKEAHA.108.520817. Epub 2008 Oct 2. https://www.ncbi.nlm.nih.gov/pubmed/18832741

Forum Summary

While there is general agreement that the moderate intake of alcohol is associated with a significantly lower risk of cardiovascular disease, there are less data on the relation of moderate alcohol consumption to dementia. However, the vast majority of well-done prospective studies indicate that, in comparison with non-drinkers, moderate, non-binge-drinking older adults have lower risk of Alzheimer Disease and other types of dementia. The present study provides important information on this association by monitoring for the development of dementia among subjects in a large cohort of British civil servants (the Whitehall Study) over more than two decades; there were repeated assessments of alcohol consumption.

The main results indicate that abstinence in middle life is associated with a significantly higher risk of dementia than the risk among moderate drinkers, while subjects reporting the intake of larger amounts of alcohol or evidence of an alcohol use disorder are at increased risk of dementia. Specifically, the study shows that among the 397 cases of dementia that were recorded over a mean follow up of 23 years, abstinence in midlife was associated with a 47% higher risk of dementia compared with consumption of 1-14 units/week (a British unit is the equivalent of 8 grams of alcohol). There was a 17% increase in risk of dementia for those reporting more than 14 units/week. With repeated assessments of alcohol the authors also calculated trajectories of alcohol consumption from midlife to early old age, with continued abstinence being associated with an increase in dementia of 74% and a decrease in consumption with an increase of 55% in comparison with subjects reporting continued moderate consumption. In several analyses, wine consumption was associated with more favorable effects than those of other beverages.

Forum members thought that this was a well-done analysis, but considered that the data presented do not allow for a firm determination of a cut-point for increased risk of dementia from alcohol intake. While the authors provide estimates of a linear increase in dementia risk for subjects reporting more than 14 units/week or reporting evidence of alcohol abuse, they do not give data that permits an estimate of the level of intake where the risk of dementia exceeds that of non-drinkers.

Overall, the results showing a decreased risk of dementia for moderate drinkers support the findings from most well-done prospective cohort studies.  As for implications for policy, the study further shows that, in terms of the risk of dementia as well as cardiovascular disease, middle-aged and older individuals who are consuming alcohol moderately and without binge drinking should not be advised to stop drinking.


Contributions to this critique by the International Scientific Forum on Alcohol Research were provided by the following members:

Giovanni de Gaetano, MD, PhD, Department of Epidemiology and Prevention, IRCCS Istituto Neurologico Mediterraneo NEUROMED, Pozzilli, Italy

R. Curtis Ellison, MD, Professor of Medicine, Section of Preventive Medicine & Epidemiology, Boston University School of Medicine, Boston, MA, USA

Ramon Estruch, MD, PhD, Hospital Clinic, IDIBAPS, Associate Professor of Medicine, University of Barcelona, Spain

Harvey Finkel, MD, Hematology/Oncology, Retired (Formerly, Clinical Professor of Medicine, Boston University Medical Center, Boston, MA, USA)

Ulrich Keil, MD, PhD, Professor Emeritus, Institute of Epidemiology & Social Medicine, University of Muenster, Germany

Dominique Lanzmann-Petithory, MD, PhD, Nutrition Geriatrics, Hôpital Emile Roux, APHP Paris, Limeil-Brévannes, France

Linda McEvoy, PhD, Department of Radiology, University of California at San Diego (UCSD), La Jolla, CA, USA

Erik Skovenborg, MD, specialized in family medicine, member of the Scandinavian Medical Alcohol Board, Aarhus, Denmark

Creina Stockley, PhD, MSc Clinical Pharmacology, MBA; Health and Regulatory Information Manager, Australian Wine Research Institute, Glen Osmond, South Australia, Australia

Yuqing Zhang, MD, DSc, Clinical Epidemiology, Boston University School of Medicine; Massachusetts General Hospital, Harvard Medical School, Boston, MA, USA.