Diet Soda Is Fine

2019 has been an abundant year for observational studies of diet soda. Three new papers have re-analyzed data from four gigantic, long-term, prospective observational trials of nutrition and health. Three of the four trial cohorts ended up showing a correlation between consuming more than two diet sodas per day and an increase in the risk of all-cause mortality. The increase in deaths was attributed to cardiovascular disease (CVD), with no change in cancer or neurodegenerative disease.

These papers show a few strange inconsistencies with large differences in outcomes between groups that don’t make sense. Besides these inconsistencies, there are two big questions about these findings. First, could it be reverse causal? In other words, could it be that being unhealthy causes people to consume diet soda rather than the other way around. Second, what’s the mechanism? How would drinking diet soda harm you?

The first study, published in February, looked at the effect of diet soda in the Women’s Health Initiative Observational Study, which included 82,000 American women who were between 50 and 79 years old when they enrolled in the study and were followed for an average of 12 years. Overall, women who drank over two diet sodas per day were at about a 30% higher risk of stroke or heart disease and a 16% higher risk of all-cause mortality. Women who drank 1 diet soda per day were not at any higher risk for health problems.

Strangely, the results included gigantic differences by race. White women, who made up 85% of the study, didn’t have a higher risk of any health problems at any level of diet soda consumption. The entire result is attributable to a gigantic effect for the remaining 15% of non-white women. There’s no reason non-white women should be harmed by diet soda more than white women, so this should make us suspicious that the health problems are actually caused by something else that is correlated with diet soda consumption in non-white people.

The researchers did check for reverse causality. One way of testing for it – also used for measuring the optimal BMI – is by excluding people who develop health problems in the early years of the study. In this case, they excluded anyone who developed diabetes or CVD within 3 years of the beginning of the study. The hope is that anyone who had started drinking diet soda because they were on the verge of developing health problems gets excluded this way. Making that change didn’t alter the outcomes above: 1 diet soda per day wasn’t linked with any problems, but 2 or more diet sodas per day was.

The second study, from April, covered both the Health Professional’s Follow-up Study, which included 38,000 American men, and the Nurses’ Health Study, which included 81,000 American women. Combining these two studies, they find that people consuming over 2 diet sodas per day were at 13% higher risk of CVD mortality and 4% higher risk of all-cause mortality, smaller effect sizes than in the previous study, but roughly consistent. In comparison, they found that the same level of consumption of regular soda was correlated with a 31% increase in CVD mortality and a 21% increase in total mortality. Just like the previous study found large differences by race, this study found large differences by sex: diet soda was correlated with harm in women, but not in men. Another result that doesn’t make sense: consuming 1-4 diet sodas per month was actually correlated with lower mortality than consuming none.

The most recent study, from last week, analyzed data from the EPIC study, which included 450,000 people, both men and women, from 10 European countries. Like the previous study, they find that 1-4 diet sodas per month is correlated with a benefit, but more than two glasses per day is correlated with a 17% increase in all-cause mortality. This paper came the provocative finding that the same level of consumption of regular soda actually causes a much smaller 8% increase in all-cause mortality.

These studies might seem like they’re getting redundant. Vasanti Malik, the author of the study April study on doctors and nurses, told the New York Times, “I don’t think it adds much to what we already know.” John Ioannidis from Stanford, who studies the way we study health and nutrition, argues that we rely far too heavily on repeated observational trials that don’t tell us anything new. Instead, he argues scientists and funding agencies should focus on a smaller number of high quality randomized trials. I completely agree with him, but I’m also a little more sanguine about this European study. With such a large sample they can conduct a stronger test of reverse causality. First, they excluded everyone who died in the first eight years of the study; the results were unchanged. Second, they conducted separate analyses for people grouped by BMI instead of including BMI as a control variable; again, the results were unchanged. These results contradict the claim that people drink diet soda because they are unhealthy or overweight rather than the other way around.

Still, this study leaves us with some strange contradictions. It is implausible that a small amount of diet soda is beneficial or that a lot of diet soda is worse for your health than a lot of sugary soda. A soda tax advocate told the NYT, “Gosh, at this point, you probably want to go with water, tea or unsweetened coffee and not take a chance on beverages we don’t know much about.” But his premise is wrong. The FDA says, “Aspartame [the sweetener in Diet Coke] is one of the most exhaustively studied substances in the human food supply.” Some people argue that aspartame causes cancer. They’re wrong. The American Cancer Society refers to these arguments as “rumors” before quoting the FDA and its European equivalent as saying that aspartame is safe. The ACS blames the rumors on one set of poorly-conducted mouse studies. All four of these new studies show no connection between diet soda and cancer.

Alternatively, people argue that diet soda harms your gut bacteria. The studies on how artificial sweeteners affect the gut microbiome are mostly in petri dishes and mice, often at doses that are “physiologically irrelevant.” The most famous human study got a lot of press with bold claims, but it didn’t include a proper control group or even show that diet soda harm the gut microbiome. All it showed was that people who drank diet soda had changes over time in their gut microbiome. Turns out, though, exactly the same thing happens in people who don’t drink diet soda. This new analysis of the EPIC study evaluated the correlation between diet soda consumption and digestive health problems, but found none.

Taken at face value, these results mean that diet soda harms black women but not white women and harms women but not men. The last paper, which found the largest harm from diet soda, concludes, “Possible biological mechanisms that may explain the positive associations between artificially sweetened soft drinks and mortality outcomes are unclear.” In other words, they have no idea why this would happen. Really, it doesn’t make sense. Vasanti Malik, from the April study, argues that the effect is probably explained by other unhealthy behaviors people justify by drinking diet soda. In contrast, sugary beverages are highly caloric and cause an immediate spike in blood sugar. The evidence is strongly in favor of artificial sweeteners over sugar. As for artificial sweeteners over water, it probably doesn’t make a big difference for healthy adults (children and people with kidney problems might be a different story), but you could limit your consumption to below two glasses per day to be safe.


Observational Trial


Diet soda…

European Prospective Investigation into Cancer and Nutrition (EPIC)

450,000 Europeans, 70% female, average age 51, 16-year follow-up

correlated with CVD and all-cause mortality

but not with cancer, neurodegenerative disease, or digestive diseases.

Women’s Health Initiative Observational Study

94,000 American women aged 50 to 79, 11 year follow-up

correlated with CVD and all-cause mortality in non-white women

but not in white women

Health Professional’s Follow-up Study

38,000 American men

not correlated with CVD or all-cause mortality.

Nurses’ Health Study

81,000 American women

correlated with CVD and all-cause mortality, but not cancer.


Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s