The headlines recently have been unmistakable:
Drinking alcohol key to living past 90, study says
Drinking Tied To Long Life In New Study
Could these headlines be true? After all, wasn't there a recent headline that said drinking alcohol was the largest single modifiable risk factor for dementia (1)? Buried in some of those headlines are also secondary stories about political decisions that did not go well for the producers of some alcoholic beverages. France's Health Minister Agnès Buzyn - a physician stated recently that alcohol is alcohol. She went on to say that contrary to what French citizens are taught to believe about the health effects of wine it is no different than drinking beer or distilled spirits and it is bad for health. I think that we have been in the midst of a tremendous amount of hype about alcohol, the specific types of alcohol, secondary natural products, the purported metabolic effects and the effect of alcohol on longevity. The current headlines were the only ones I can recall where the positive effects of drinking alcohol was estimated to be on par with exercise.
I come at the problem from the perspective of an acute care and addiction psychiatrist. For 22 years, I worked at a tertiary care center that was also a Level 1 Trauma Center and it contained a burn unit. At one point our medical director surveyed our admissions and determined that at least 50% across the entire hospital were there because of drugs or alcohol. We saw every type of injury and chronic illness due to intoxicants and the patients with those insults often had markedly shorter life spans than expected. How could alcohol use extend life? Why was it seen as a common finding? Most importantly - why were all of these headlines surfacing right now?
Some of the articles named Claudia Kawas, MD and her work in the 90+ Study and Leisure World Cohort Study as the source for the headlines (2-4). The Leisure World Cohort Study (LWCS) followed a group of 8,371 women and 4,828 men from a media baseline age of 74 for a period of 28 years or until death. The group was located in a retirement community and were described as predominately white, middle class and well educated. They were sampled at intervals with questionnaires that asked about their dietary habit including beverage intake in terms of alcohol and caffeine containing beverages and other types, a number of activity levels, and total amount of exercise. A large number of papers resulted from this study and are still being written as the continuation study of the members that survived into their 90s. Dr. Kawas gave a presentation at a recent American Association for the Advancement of Science (AAAS) meeting on some of her findings and that appears to be what the headlines based on.
From the LWCS group, there were several notable findings. In terms of activity level (2), any activity of 1/2 hour per day or more reduced mortality risk 15-30%. A broad range of exercise of various levels of intensity and whether they were done inside or outside. Level of activity at age 40 was a predictor of activity in old age. Relative Risks (RR) for all cause mortality were calculated for the activities and their duration. as well as the time spent. After 3/4 of an hour per day the RR effect tapered off. Sedentary activities like watching television had no significant impact on the RR. The greatest observed risk occurred when activity levels were reduced due to injury or illness. They found no survival advantage for a high activity level (1+ hours per day) compared to a moderate level of 1/2 to 3/4 hours per day.
The same group looked at the issue of alcohol intake in the LWCS group. In their introduction they note that 4% of the annual mortality in the world is caused by alcohol. They review some of the previous literature and the purported J - or - U shaped mortality curves for alcohol consumption - meaning higher mortality rates for abstainers, lower mortality rates for moderate drinkers (1-2 standard drinks per day), and higher mortality rates for higher levels of drinking. The response choices on the survey were for 1, 2, 3, and 4 or more drinks per day. They also broke the sample down based on their responses drinking surveys in 1992 and 1998 to to stable non-drinkers, stable drinkers, starters, and quitters based on comparing their survey answers. Three quarter of the sample drank. Two drinks a day conferred a 14-16% in decreased mortality irrespective of the type of alcohol. At follow up there were more non drinkers than at baseline (36% versus 26%). The quitters and starters acquired the expected mortality risks in each group. They conclude that there was a small beneficial risk of alcohol on mortality of about 15% but qualify the result based on the study limitations.
The final dimension in this sample of the LWCS paper was a look at non-alcoholic beverages and caffeine content. They looked a coffee, decaffeinated coffee, black or green tea, cola drinks (sugar or artificially sweetened), other soft drinks and sweetener combinations, and the amount of chocolate eaten (daily versus a few times per month.) They found that there was a U-shaped mortality curve for caffeine consumption with peak protection at about the 100-399 mg/day. They also found that consuming as little as one can a week of artificially sweetened soft drinks had a small increased risk of death (11-24%). They looked at specifics and determined that 1-3 cups of regular coffee/day reduced mortality risk by 5-10% and drinking decaffeinated coffee or tea reduced risk by 5-9%. Drinking sugar sweetened cola - had an 8% lower risk of death. Infrequent chocolate users also had a reduced risk of death (3-9%).
Taken all together these three papers suggest that moderate levels of alcohol, caffeine, and activity are all consistent with longevity. In order to look at the alcohol findings in perspective, I searched the literature for a meta-analysis of all of the alcohol x longevity studies and came up with an outstanding paper by Stockwell, et al (5). In it the authors look at and extensive analysis of existing alcohol effect on mortality studies and initially duplicated a J-shaped mortality curve based on 87 studies they included in their analysis. They went back into that sample and corrected for abstainer biases such as including including former and occasional drinkers in the abstainer category. They model four types of abstainer bias in their in the paper. When those corrections are made or when only very high quality studies are used - the purported mortality advantage of moderate (1-2 standard drinks per day) - disappears completely. I could not find any data from the LWCS studies used in this meta-analysis. According to the author's selection criteria the LWCS data probably would have been eliminated because it was a cross sectional study.
That alcohol is not a heath food should not come as a surprise. Any cohort of drinkers in their 90s suggests to me that they are biologically selected to survive the alcohol and that is probably why they are drinking into their 90s and not because of it. Since the activity, caffeine, and diet soda effects noted in this study were collected using similar methodologies, that can be a cause for concern. The authors were careful to cite supporting data and discuss the limitations. Observational studies like the LWCS and 90+ Study add to the literature but it is necessary to keep these findings in perspective and consider the potential biases of the design.
At this time I have not found a similar meta-analysis for each of the other cases (activity level, caffeine consumption).
George Dawson, MD, DFAPA
All linked papers below are to free full text articles.
1: Schwarzinger M, Pollock BG, Hasan OSM, Dufouil C, Rehm J; QalyDays Study Group. Contribution of alcohol use disorders to the burden of dementia in France 2008-13: a nationwide retrospective cohort study. Lancet Public Health. 2018 Feb 20. pii: S2468-2667(18)30022-7. doi: 10.1016/S2468-2667(18)30022-7. [Epub ahead of print] PubMed PMID: 29475810.
5: Stockwell T, Zhao J, Panwar S, Roemer A, Naimi T, Chikritzhs T. Do "Moderate" Drinkers Have Reduced Mortality Risk? A Systematic Review and Meta-Analysis of Alcohol Consumption and All-Cause Mortality. J Stud Alcohol Drugs. 2016 Mar;77(2):185-98. Review. PubMed PMID: 26997174.