Artificially Sweetened Beverages and the Risk of Stroke and Dementia

I’m often asked which is worse for health, sugar- or artificially sweetened beverages. My most common response is to recommend skipping both and choose water. However, in the era of fructose phobia and the apparent association between high fructose consumption and the risk of metabolic syndrome and diabetes, many tend to prefer artificially sweetened beverages (1,2). However, recently published scientific evidence suggests that this may be a wrong approach and that artificially sweetened soft drinks may indeed cause more harm than those containing sugar.

Artificially Sweetened Beverages and the Risk of Stroke and Dementia
The main advantage of artificially sweetened beverages is typically considered to be the lack of calories compared with sugars such as sucrose or fructose.

Five “non-nutritive” artificial, sweeteners used in soft drink production have FDA approval; saccharin, acesulfame, aspartame, neotame, and sucralose. Stevia, a natural extract from the plant Stevia rebaudiana, was approved by the FDA in 2008. These synthetic substances are hundreds to thousands of times more potent than sucrose and evoke an intense sensation of sweetness in trace amounts (3).


Artificially Sweetened Beverages and the Risk of Stroke and Dementia

A recently published paper by Matthew P. Pase PhD and coworkers from Boston University School of Medicine suggests that high consumption of artificially sweetened soft drinks may be associated with increased risk of both stroke and dementia (4). In fact, it is the first study to suggest that artificially sweetened beverages, but not sugar-sweetened beverages, may be associated with risk of stroke and dementia, including Alzheimer’s disease.

Artificially Sweetened Beverages and the Risk of Stroke and Dementia
Cumulative consumption of artificially sweetened beverages and event-free survival of stroke (A) and dementia (B). Green, red, and blue lines denote intake of 0/wk, 6/wk, and 1/d, respectively. DOI:

The investigators studied 2.888 participants from the Framingham Heart Study Offspring Cohort between 1991 and 2001. The mean age was 62 years, 55% were women. Beverage intake was quantified using food-frequency questionnaires. During the ten year follow-up period, 82 cases of incident stroke were observed and 81 cases of incident dementia, 63 of which were consistent with Alzheimer’s disease.

After adjustments for age, sex, education, caloric intake, diet quality, physical activity, and smoking, higher recent and higher cumulative intake of artificially sweetened soft drinks were associated with an increased risk of ischemic stroke, all-cause dementia, and Alzheimer’s disease dementia.

Those who consumed artificially sweetened soft drinks daily were approximately three times more likely to develop stroke or dementia compared to those who consumed no artificially sweetened beverages. Sugar-sweetened beverages were not associated with stroke or dementia.

The Bottom Line

The Nurses Health Study and Health Professionals Follow-Up Study found that greater consumption of both sugar-sweetened and low-calorie sodas was associated with a significantly higher risk of stroke (5).

A 2011 statement from the American Heart Association and American Diabetes Association concluded that when used judiciously, non-nutritive sweeteners (including very low-calorie sweeteners, artificial sweeteners, and non-caloric sweeteners) might help with weight loss or control, and could also have beneficial metabolic effects (6).

So clearly, the discussion about artificially and sugar-sweetened beverages has typically focused on the energy issue. And of course, the soft drink producers highlight the lack of calories as the main benefit of artificially sweetened beverages.

According to Coca-Cola; Diet Coke is the most popular calorie-free soft drink in America. It’s the original sparkling beverage for those who want great flavor without the calories – a drink for those with great taste.

According to Pepsi; Pepsi Zero Sugar or Pepsi Max is the only soda with zero calories and maximum Pepsi taste.

Interestingly, the study by Pase and coworkers also found that diabetes mellitus, which is a known risk factor for dementia, was more prevalent in those who regularly consumed artificially sweetened soft drinks.

On the other hand, it should be pointed out that this was an observational study and therefore can not determine whether there is a causal relationship between artificially sweetened soft drink intake and diabetes and stroke or dementia.

However, the fact that those who consumed artificially sweetened soft drinks daily, but not those who consumed sugar-sweetened beverages, were approximately three times more likely to develop stroke or dementia is intriguing.

About 25 percent of children and more than 41 percent of adults in the United States reported consuming foods and beverages containing low-calorie sweeteners such as aspartame, sucralose, and saccharin in a recent nationwide nutritional survey (7). Those numbers represent a 200 percent increase in the consumption of artificially sweetened beverages for children and a 54 percent jump for adults from 1999 to 2012.

These numbers highlight the potential implications of the hair-raising results presented by Pase and coworkers.

Unfortunately, in my opinion, the study has some methodological flaws. The limited number of events is clearly a problem. Furthermore, no adjustments were made for multiple statistical testing. Hence it can not be rule out that some of the findings are due to chance alone.

Another study by the same group, published online in Alzheimer’s and Dementia on March 5, shows a link between consumption of both sugar-sweetened and artificially sweetened beverages and reduction in brain volume in a middle-aged cohort (8). In the cross-sectional study, the sugary drinks, which included both soda and fruit juice, were also associated with worse episodic memory.

Apparently, the jury’s still out on the issue which is worse, sugar-sweetened or artificially sweetened. So, why not skip both and stick with water.


13 thoughts on “Artificially Sweetened Beverages and the Risk of Stroke and Dementia”

  1. I certainly agree with the thrust of your post. But I’d emphasize even more than you did that hypotheses generated from observational data are almost always wrong. People with met syndrome are probably more likely to choose artificially sweetened drinks than are healthy people.

  2. Thanks Axel for this article. I think there is another elephant in the room which might have “caused” this outcome. The group that consumed least amount of artificial sweeteners had much less diabetes than the group that consumed the most, the difference is more than 3 fold (from 7% to 22%). Same goes for for those who drank least amount of sugary drinks they had more diabetes (15%) compared to the group who consumed most (8%). I think the rate of dm amongst these groups is a major confounder.

    • Perhaps related to your diabetes observation: The authors of the paper did not control for obesity. Maybe overweight people are more likely to drink artificially sweetened drinks because it seems like an easy way to keep their total calorie intake lower.

      • You may be right Mark. Those who consumed most artificially sweetened drinks had a mean BMI of 33 while those who consumed no such drinks had a mean BMI of 30.

      • I was also wondering why they did not control for BMI. However, they did have a set of models that controlled for waist-to-hip ratio, which they may have felt was a better measure, especially for examining abdominal obesity.

        I am somewhat puzzled by exactly what exposure they are interested in? For the “sugary beverages,” is it sugar or is it something about sugar in beverages? There are plenty of other sources of sugar in the American diet. Then we have the artificially-sweetened beverages. Okay, they do refer to a mouse study where they fed them artificial sweeteners and changed gut microbiota and induced glucose intolerance. But that’s quite preliminary.

        I was intrigued by the graph showing the survival curves and notice they exaggerate the drop in the curves by having the Y-axis start at 96% for stroke and 92% for dementia. So at 10 years the curves, although separated, drop (at the worst) to 97% for stroke and 96% for dementia. I suppose this doesn’t exactly translate, but when I went to the ACC/AHA 10-year risk calculator for stroke and plug in some values that were typical for the study participants, I got 9.4% (age 62, male, non-smoker, total chol = 200, HDL = 55, no diabetes, SBP = 127). I found another paper which suggests 10-year risk of stroke for the US is about 10%. So, this Framingham cohort seems pretty healthy?

      • You’re absolutely right Lynn. I hadn’t noticed that the graph only covers the interval between 0.96 and 1.0. Of course this visually exaggerates the difference between the groups.

        Another thing pointed out by a colleague of mine in the Facebook discussion about my post is that the investigators did not adjust for multiple statistical testing. Hence it can not be rule out that some of the findings are due to chance. It is strange that the peer review process doesn’t pick that up.

      • Reflecting on the small decline in the survival curves, it makes me wonder about the relative contribution of the artificially sweetened beverages compared to other factors such as smoking, hypertension and obesity. Of course, there are interconnections between these factors. However, I don’t intend to encourage consumption of artificially-sweetened or sugary beverages – I am a water drinker myself. (And also a non-clinician so not qualified to give health advice.)

        Another thing I just noticed in the article is that for many of the models they ran, the authors do not have what would be considered a sufficient number of events (strokes, dementia) for the number of predictor variables they used. That is, the “rule of thumb” is 10 events for every predictor variable in the model. I did not look at all models presented, but for dementia the number of events per predictor variable ranged from 5 to 9. For stroke, there were a couple of models that had 10 and 12 events, so okay, but others had 6 to 8. This would be called “overfitting,” and bring up the possibility that the hazard ratios found are exaggerated away from zero.

        Thank you for your posts, they are interesting and informative.

      • Thanks Lynn.
        I think the study has several statistical flaws. One is that no adjustment was made for multiple testing. Another is the limited number of events which you mention and I do agree with you on that one. The low number of events may not allow multiple testing in the way it’s done here. So I really do wonder how the study got published in such a respected paper.

    • A good point Kjartan. The authors of the paper address this issue shortly in their discussion; “Indeed, in our study, diabetes mellitus-a known risk factor for dementia-was more prevalent in those who regularly consumed artificially sweetened soft drinks.”

      So the question asked is whether artificially sweetened soft drink intake increased the risk of dementia through diabetes, or whether people with diabetes were simply more likely to consume dietary beverages. Obviously, an observational study can not provide an answer to that question.

  3. I am kind of curious – could some of those who got dementia and Alzheimer’s already had it, but at an early stage and not yet noticeable? I read that sugar craving is one of the symptoms. People with Dementia may just be the one who drink more Coca Cola…

  4. Does stevia fit in category of other artificial sweeteners? I don’t use it often, but several people I know do. Seeking more information on that.

    • Yes, it probably does. But I assume Stevia is different because it is a natural extract.

      So, it may be misleading to talk about artificial sweeteners ass a group because they may be very different from each other.


Leave a Comment

This site uses Akismet to reduce spam. Learn how your comment data is processed.