Neal S. Parikh, MD
@NealSParikhMD
In this entry, I discuss a recent publication by Matthew Pase and colleagues regarding the risks of stroke and dementia associated with the consumption of sugar-sweetened and artificially sweetened beverages.
Citing conflicting data, the authors sought to examine the association of sugar- or artificially sweetened soft drink intake with incident stroke and dementia in the Framingham Heart study.
From among 3,539 participants who attended a Framingham Heart Study examination between 1998 and 2001, the authors included 2,690 participants for stroke follow-up and 1,395 participants for dementia follow-up. Participants with prevalent stroke, dementia, cognitive impairment, and other neurological disease, in addition to younger participants, were excluded.
Sweetened beverage intake was measured with a validated food-frequency questionnaire. The questionnaire included sugar-sweetened soft drinks, fruit juice, and artificially sweetened soft drinks. Recent intake and cumulative intake (averaging at least two separate Framingham examinations) were analyzed separately. Participants were followed for up to 10 years for incident stroke and dementia.
In a stepwise fashion, the authors constructed Cox proportional hazards regression models. In Model 1, the covariates were age, sex, education, and total caloric intake. In Model 2, additional covariates reflected lifestyle factors such as diet quality, physical activity, and smoking status. In Model 3, they additionally included hypertension, cardiac disease, atrial fibrillation, cholesterol, diabetes, apolipoprotein E genotype, and waist-to-hip ratio.
A key observation of baseline characteristics was that greater artificially sweetened soft drink consumption was associated with greater baseline cardiovascular disease and diabetes, whereas the inverse relationship was observed for total sugary beverages.
The regression models revealed that total sugary beverage and sugar-sweetened soft drink consumption was not associated with stroke, whereas high recent and cumulative artificially sweetened soft drink intake were associated with incidence stroke. Participants who had at least 1 artificially sweetened soft drink per day faced a nearly two-fold risk of incident stroke (HR, 1.97; 95% CI, 1.10-3.55) compared to participants with no intake, after adjustment for all covariates in Model 3. The association with dementia appeared weaker, with loss of statistical significance with adjustment for all covariates (Model 3). Importantly, prevalent diabetes—or diabetes already present at the beginning of observation—partially mediated the association between artificially sweetened soft drink consumption and stroke.
Because participants who frequently consumed artificially sweetened beverages had higher rates of diabetes at baseline, and considering the mediation analysis results, the authors rightfully state that reverse causality may explain their results. Simply, one explanation for these findings is that participants with diabetes and other pre-existing and dominant stroke/dementia risk factors drank more artificially sweetened beverages, perhaps in response to physician counsel.
Readers may wonder why these results deserve our attention despite the possibility of reverse causality. There are at least two good reasons. First, bench research suggests that artificial sweeteners cause metabolic disorders, so the findings of this paper have biological plausibility. Second, even a specter of an association between artificial sweeteners and stroke/dementia deserves notice. Artificially sweetened soft drinks have no nutritional value yet are consumed widely as an ostensibly healthy alternative to sugary beverages; the public health implications of a true association between artificial sweeteners and stroke/dementia cannot be understated.
However, by no means do these findings suggest that drinking sugar-sweetened beverages instead of artificially sweetened sodas is beneficial. It may be prudent to advocate consumption of unsweetened and minimally sweetened beverages instead.
Were any kind of sweeteners (other than sugar) considered as a whole?
They did and reported: “Neither intake of total sugary beverages nor intake of sugar-sweetened soft drink was associated with the risks of stroke.”
“Artificially sweetened soft drinks have no nutritional value yet are consumed widely as an ostensibly healthy alternative to sugary beverages; the public health implications of a true association between artificial sweeteners and stroke/dementia cannot be understated.”
If artificial sweeteners help to keep weight down by avoiding extra calories, then would they not be valuable for health despite having no nutritional value?
That would be a possible benefit, but only if it is determined that that translates into a reduction of important cardiovascular outcomes.
Reverse causality and the possibility that factors such as obesity and genetics related to diabetes are predisposing are huge challenges for this research to overcome. The mediation of diabetes and stroke favors other explanations than artificial sweetened beverages. Diabetics for example eat many other foods with artificial sweeteners. A stronger research model would have considered all artificial sweetener use not just in beverages. It would also have parsed different types of artificial sweeteners. The dementia finding is the weakest and the most suspect conclusion, especially in an aging population.
Fully agree – many limitations of retrospective food and nutrition research. Such studies must be interpreted in the context of all related studies, as no single study can provide sufficient evidence for inference.
Were there differences between different artificial sweeteners? Do saccharin, acesulfame, aspartame, neotame, and sucralose have similar or different effects?
I do not believe they report on these sweeteners separately. I agree with your insinuation that it may be inappropriate to lump all artificial sweeteners together, but this current study did not address that.