New Study: Diet Soda meets Straw Man

A new study in the journal Obesity, comparing diet soda to water for weight loss, has resulted in extensive and worldwide media coverage -- some of it, to my surprise, directly involving me. My involvement derives from my published opinions about diet soda and artificial sweeteners, which thus far remain unchanged after reading the study in question.
This post was published on the now-closed HuffPost Contributor platform. Contributors control their own work and posted freely to our site. If you need to flag this entry as abusive, send us an email.

A new study in the journal Obesity, comparing diet soda to water for weight loss, has resulted in extensive and worldwide media coverage -- some of it, to my surprise, directly involving me. My involvement derives from my published opinions about diet soda and artificial sweeteners, which thus far remain unchanged after reading the study in question. Here are the particulars.

The study reported weight loss results over 12 weeks in a group of approximately 300 adults, all overweight or obese at baseline. The study participants all received general weight loss guidance in once weekly, one-hour group meetings. The usual items -- diet, physical activity, and self-monitoring -- were all addressed.

The study was designed to test the hypothesis that when those other factors pertaining to weight loss were equal between groups, as one would expect in a randomized trial, then weight loss would also be equal between groups when one was assigned to drink beverages with non-nutritive sweeteners (NNS) daily (i.e., what most of us would refer to as "diet sodas") and the other group assigned to water. This design is called an equivalence study, since it attempts to show lack of difference between treatments.

The investigators must have been gratified to overshoot the equivalence target. Not only was weight loss as good in the diet soda group -- it was better. The group assigned to sweetened beverages lost nearly 2 kg more (over 4 pounds) on average than the group assigned to water, and reported less hunger into the bargain.

If at this point you are dashing out for a diet soda -- let me quickly append: not so fast.

One of the stipulations in the study was that at the start, all participants already be routine diet soda drinkers; the protocol required several diet sodas per week at a minimum. The group assigned to diet sodas could drink all the water they wanted in addition; the group assigned to water could not, of course, drink any diet sodas -- nor add any non-nutritive (or "artificial") sweeteners to their foods or drinks on purpose. They could consume foods with such sweeteners in them, or chew gum.

So, at this point, we might summarize as follows: from a group that routinely drank diet sodas, half were allowed to keep doing that, and the other half had to stop. Both received guidance to facilitate weight loss.

Personally, I don't like non-nutritive sweeteners for the most part, because I think food should actually be food. I don't think these compounds are especially toxic, but I do think they are chemical compounds, and clearly some people don't tolerate them. Chemicals in our food certainly have at least some potential to cause harm, some of it rather hard to measure.

But mostly, I don't like these sweeteners because they are intensely sweet. That figures into the prevailing concerns about sugar intake, food addiction, and what we call in the vernacular a "sweet tooth." One of the properties of appetite that fosters questions about food addiction is that for certain flavors, sweet salient among them, the more we get, the more we need to feel satisfied. This is really what a "sweet tooth" means, and it is quite apt. The preference for sweet is, indeed, for sweet -- not necessarily sugar. Artificial sweeteners, which range from 600 to 1300 times as sweet as sugar, also feed -- and grow a sweet tooth.

What are the basic implications? Those who routinely drink diet sodas are getting a lot of sweet stuff in their diets, and likely prefer it that way. Take away that sweetness, and they will likely need to get it elsewhere, or feel very frustrated.

What information would we need to know whether or not that occurred in the new study? We would need details of dietary intake for the two treatment groups. What information is totally absent from the article? Exactly that. The investigators provide no information at all about dietary intake.

Of note, the study was funded by the American Beverage Association, and two of the investigators are paid consultants for Coca-Cola. This may explain why the study was designed to enroll only routine diet soda drinkers in the first place. Why, for instance, wasn't it designed to enroll only routine water drinkers, then assigned to drink diet soda -- as opposed to routine diet soda drinkers, assigned to drink only water? That does suggest potential bias to me.

The authors do address this issue in their discussion. In their own words: "It should be noted that because eligible subjects were already NNS drinkers assignment to the NNS treatment did not require as great a behavior change as the Water group who had to abstain from NNS beverages for the trial. We chose this design rather than admitting all comers in order to ensure that subjects assigned to NNS would adhere to the treatment giving us the ability to see if NNS adversely affected weight loss." I am not sure I buy this argument entirely.

I hasten to add that while funding source should always be disclosed, we often make too much of it. Yes, of course, every corporate funder of research is biased, hoping to show good results for the product in question. But what the public and media may overlook is that NIH is biased, too. NIH isn't accountable to shareholders, but it is accountable to Congress, and taxpayers, and it, too, benefits when study results are positive and exciting. If NIH funded studies with disappointing results too often, the NIH budget itself might prove disappointing.

Perhaps even more germane is that we researchers are always biased, meaning we are hoping for a particular outcome. Why would anyone waste time conducting research -- which is hard work -- unless they cared about the outcome? We always do, no matter who is paying the bills. So funder, and investigator, are inevitably biased.

What matters then is that the right people generate and interpret the study results, and that the methods reliably defend against the bias present at the start. In fact, as the author of textbooks on research methods, I teach my own laboratory staff that bias when contemplating a new study is both healthy and unavoidable, but biased methods are totally unacceptable. We must impose methods to defend us from our biases.

The current study could only do that to a very limited degree. Yes, it was randomized -- and that's important. But randomization is a defense against a problem called confounding (which I won't get into here), not against bias. So-called "blinding" is a defense against bias, and this study was not blinded. Both the study participants and investigators knew who was drinking just water, and who was drinking diet soda. That is an important chink in the armor defending against bias. But the bigger problem, as noted above, is that the study enrolled a population of habitual diet soda drinkers in the first place.

In the absence of any details about dietary intake and dietary change, I am left to speculate about why the diet soda group lost more weight. My inference is that all of these people were accustomed to a fairly high intake of "sweet," and those assigned to water and not allowed diet soda missed it. They therefore indulged their sweet teeth some other way, adding other sweet foods -- and presumably calories -- to their diets. They thus lost less weight under the controlled conditions of an intervention trial than a group allowed to keep assuaging their sweet tooth in the familiar manner.

This is exactly the effect I have suggested diet soda consumption is likely to have, based both on relevant literature, and my experience with patients over nearly 25 years. I believe artificial sweeteners cultivate a sweet tooth, and cause people to want -- or need -- a sweeter diet overall. Consequently, I think there are far better ways to reduce sugar and calorie intake that are not dependent on the uncertain effects of these chemical compounds.

It is possible, if not highly plausible, that diet sodas actually work better for weight loss than water, as the authors suggest. But I don't think so. What I think this study shows is that it's hard for people accustomed to intensely sweet items in their diet to give them up. At best, I would say that until or unless this study is repeated with a group of overweight, habitual water drinkers -- we can't draw any reliable conclusions.

But of course, it may be hard to find a group of 300 overweight or obese people who drink mostly water and no diet sodas. The habitual water drinkers may be less prone to obesity in the first place.

The bottom line here is that while we can't say for sure that diet soda drinking doesn't help with weight loss, we certainly do not have proof that it does. The study design to make that case was something of a straw man.


Dr. David L. Katz has authored three editions of a nutrition textbook for health care professionals. He is editor-in-chief of the peer-reviewed journal, Childhood Obesity, and President of the American College of Lifestyle Medicine. He was commissioned by Annual Review in Public Health to write the review article, Can We Say What Diet is Best for Health? He is the author of Disease Proof, and most recently, of the epic novel, reVision.