Like seemingly everything else in our food environment today, people view artificial sweeteners as either harmless or evil, sans middle ground. Like carbs, meat, or saturated fat, it can seem as though you have to pick a side when talking about nutrition, and artificial sweeteners (indeed, all artificial ingredients) are no exception.
Recently, renowned cardiologist and epidemiologist Dr. Harlan Krumholz wrote a piece entitled, "Why One Cardiologist has Drunk his Last Diet Soda." In it, he lays out the case that artificial sweeteners are not physiologically inert and that they are likely harmful due to consistent associations between artificial sweetener intake and risk for weight gain and adverse cardiometabolic outcomes.
So where is the science on this?
First, it makes sense to talk about what I suspect is the biggest fear most people have when it comes to artificial sweeteners (and artificial ingredients in general).
The greatest sweetener fear:
I’ll make an educated guess that the greatest fear associated with artificial sweeteners is cancer. Though cancer is outside the scope of this article (Dr. Krumholz doesn't mention cancer in his editorial), it seems worthwhile to briefly address.
Much of the fear surrounding artificial sweeteners and a link to cancer comes from news articles, rumors, or small animal studies that aren't relevant to humans. I definitely want to go into more detail on the misinformation associated with artificial sweeteners in the future but for now I’ve listed some coverage on the topic below. (And if you want additional reading material, the SciBabe is apparently writing a whole series on the subject, and she's a toxicology expert!)
As stated, cancer is not within the scope of this entry (but I will elaborate on the subject later). Let’s move on to other fears associated with artificial sweeteners!
Back to weight gain
As previously noted, Dr. Krumholz wasn't concerned with cancer; he was worried about the risks of weight gain and concurrent increased risk for heart disease. (Although, it's worth noting that overweight and obesity are associated with greater risks for some types of cancers.) His editorial presented two major points: the first noted an interesting meta-analysis and systematic review in the Canadian Medical Association Journal (CMAJ) about the use of non-nutritive sweeteners (NNS) in randomized controlled trials (RCTs) as well as in epidemiological studies. Next, he mentioned possible physiological effects or biological mechanisms these sweeteners could have that may contribute to weight gain in regular NNS users.
The evidence in all three of these areas - human RCTs, observational data, and physiological studies - is mixed, though it does seem that artificial sweetener intake is consistently associated with higher BMI and weight gain. But does this mean we should all drop our Diet Cokes and run? Or -gasp! - would people truly be better off drinking regular soda? Let's examine.
Some studies suggest that artificial sweeteners could alter taste perception, digestion, gut hormone secretion, and microbiota. They present a strong case that NNS’s likely have biological effects and then conclude "...amassing evidence suggests that NNSs are not biologically inert, and may influence feeding and metabolism through a variety of peripheral and central mechanisms."
These findings are concerning, but if you take the time to read the paper, you'll notice something interesting. Most of the results are in a small number of animals and are preliminary, and the findings are inconsistent. Throughout the Physiology and Behavior article, the authors are quick to point out inconsistent or non-replicated results. The other cited article, "Low-calorie sweetener use and energy balance: results from experimental studies in animals, and large-scale prospective studies in humans," also dives into inconsistent animal research. Some results seem to suggest that animals compensate for artificial sweeteners by eating more, while other studies seem to show weight gain in animals despite consistent energy intakes, some with weight gain dependent on sex, genetics, or obesity status prior to the experiment. None of the listed research in this paper, it's worth noting, was done in humans.
This isn't to say that there is no effect of artificial sweeteners on our brains or guts that encourages greater caloric intake or affects metabolism. But, if there is, it hasn't been consistently demonstrated.
Secondly, it's important to point out that lumping NNSs as a group when considering physiological effects (which two of these papers do) doesn't make a lot of sense considering that each sweetener is a separate molecule and is processed by the body differently. Even if all NNS effects on, say, taste receptors were the same, once the NNSs are swallowed effects could diverge significantly since some sweeteners are digested while others are excreted. Subsequent effects on microbiota would likely be different as well.
So, if artificial sweeteners DO cause weight gain, how do they do it? That part is still unknown. Once a consistent mechanism is determined, further study will be needed to replicate it and then demonstrate the effect in humans.
In the July CMAJ meta-analysis, authors found only seven RCTs that met their criteria for inclusion. Studies had to have lasted at least six months. The combined studies included 1003 total participants. Importantly, authors noted that most of the RCTs were at unclear or high risk of bias.
Not all of the RCTs compared the same responses to artificial sweeteners or employed similar methods. Two of the RCTs actually tested stevia in capsule form as a supplement for hypertension. (The results were mixed, with one finding a positive effect of the stevioside, and the other finding no significant effect.) The five remaining RCTs included in this CMAJ study looked at artificial sweeteners’ effects on weight loss. And though combined results of the RCTs found no benefit for weight loss, most were pretty short-term and each varied in methods.
One long-term study provided participants with either diet beverages or bottled water; enough to drink 24 oz per day at their convenience. (The study was sponsored by the beverage industry.) In this trial, the NNS group lost a greater percentage of body weight and regained less over the course of a year. Another industry-sponsored six-month trial was designed similarly and also found greater weight loss in the NNS arm. Interestingly, this study compared diet beverages against bottled water and seltzer and was funded not by a soda company, but by Nestlé Waters, which currently makes no sugar sweetened sodas, but instead owns brands of bottled and sparkling water as well as bottled iced teas. The funding source is something to consider when assessing the result of the study and since industry studies want to test whether or not their products have some benefit over a control; protocols often reflect this bias. Since soda companies frequently also bottle water, this study could be sort of a “win-win” from these companies’ standpoint.
The third industry study, funded by NutraSweet and lasting three years, looked at weight loss and maintenance in women with obesity. They found the aspartame group lost a statistically significantly greater percentage of body weight and regained less than the control group. This study provided aspartame-sweetened dairy products, drink packets, and sweetener for flavor to the NNS group. The non-NNS group was instead told to use up to 50 g (3 tsp or 45 kcals) or sugar or honey as a sweetener and to avoid artificial sweeteners. This group was provided with unsweetened flavored seltzers in lieu of sugar-sweetened or diet beverages.
Only one of the non-industry RCTs was a weight loss trial like those above. This study found that diet beverages were associated with poorer outcomes (i.e. women participating lost less weight). Participants were instructed to drink either 250 mL of water or diet beverage, but only after lunch. Further, the diet beverage group was instructed to drink their diet beverage after lunch only 5 days a week. While experimentally this makes sense, since it may have helped control for the time of day of intake and amount consumed between groups, in practice it probably doesn't represent how dieters consume diet soda in a real-world scenario. Interestingly, this study did find outcomes showing slightly less insulin resistance (calculated using HOMA-IR) and a trend toward better blood glucose control in the water group. Since the trial also involved weight loss, however, it would be hard to distinguish whether these effects were due to the relatively greater weight loss (about 1 kg) of the water group or the diet beverages group.
In the other non-industry study, 60 participants with overweight or obesity were split into four groups and each group was provided different beverages: sugar-sweetened soda, diet soda, a semi-skim milk, or still mineral water. All participants were asked to drink one liter of their respective drink per day. Why milk? Researchers knew that the milk had the same caloric content as the sugary beverage, but was associated with opposite health outcomes. They wanted to have two caloric and two non-caloric beverage groups to compare. No weight loss or lifestyle changes were prescribed to groups; the study looked at only the impact of the beverages. At six months, researchers found significant increases in visceral fat stores in the regular soda group. And weight gain? No differences in body weight were seen between the four groups during the six-month intervention.
We have to consider that none of these trials are blinded and many are at a high risk of bias, so while all of these studies used distinct and interesting methods, they are ultimately limited in what they can tell us. By combining the findings across the RCTs, the authors of the CMAJ paper found that there were no statistically significant effects for non-nutritive sweeteners for BMI or weight loss based on the RCTs. That being said, there were not many studies that fit the criteria.
Observational studies are where the bulk of Krumholz' data comes from for urging others to exercise caution when it comes to artificial sweeteners. The CMAJ article included and analyzed 30 qualifying cohort studies that assessed impact of NNS on various weight and heart health outcomes.
Authors reported that, "in contrast to RCTs, cohort studies with 4 to 9 years of follow-up [a total of 7,917 participants from the studies combined] showed that higher intake of non-nutritive sweeteners was associated with increasing weight circumference... higher incidence of abdominal obesity... and higher incidence of overweight for highest v. lowest intake quantiles."
Risk for incidence of diabetes was also higher in pooled data from 5 qualifying cohorts, though this data was possibly affected by publication bias. When authors imputed data from missing studies, the relationship became nonsignificant.
The authors conclude that "observational data suggest that routine consumption of non-nutritive sweeteners may be associated with a long-term increase in BMI and elevated risk of cardiometabolic disease, however these associations have not been confirmed in experimental studies and may be influenced by publication bias."
So What Should We Think?
There are several other limitations of these observational studies that should prevent someone from leaping to causal conclusions re: artificial sweeteners. Firstly, the observational studies typically measured only diet or artificially sweetened beverage intake, not total NNS intake. Though most people’s NNS intake comes from artificially sweetened beverages, many foods can contain artificial sweeteners to lower sugar intake. These include low-sugar candies, desserts, or energy bars. Because they only examine diet beverage intake, they’re also only looking at the artificial sweeteners typically used in beverages and not seeing impact of lower-calorie sugar alcohols, including mannitol, xylitol, and erythritol which are often used to flavor low-calorie candies and desserts.
The presumption most people make when they say that "NNSs are associated with weight gain" is that diet soda swillers are TRYING to maintain or lose weight with the diet soda choice. Thus, when they gain weight over time this is seen as artificial sweeteners backfiring on the user. But are most artificially sweetened beverage drinkers REALLY trying to control their weight with the drinks? Or are they just counting them as "free calories" and drinking diet sodas with whatever they'd like to eat (such as, say, fast food).
I conducted an unofficial Twitter poll with an extremely unrepresentative AND biased audience (people who follow me or people who follow nutrition people who retweeted this tweet) that suggests reasons for switching to diet beverages are split.
A number of replies noted Tweeters who drank diet soda out of necessity - they had diabetes or pre-diabetes and couldn't tolerate the real product. If many people make these kinds of switches after metabolic issues develop rather than as a preventive measure, this could possibly confound observational data finding worse outcomes for diet beverage drinkers (possible reverse causation).
Another consideration: what the studies largely don't measure is when people eat foods with artificial sweeteners (think low-sugar or no-sugar options like sugar-free Jello, etc.). Could these low-calorie foods be a better marker of people trying to control their weight than diet sodas? Possibly. And, since Halo Top is the new fastest selling pint of ice-cream in the United States, hopefully future food surveys will be able to capture this form of NNS intake.
So does the existing evidence justify going cold turkey on artificial sweeteners? Ultimately, that's an individual decision. While personally, I see no convincing physiological evidence showing any detriment attributed to artificial sweeteners, avoiding NNSs out of an abundance of caution is not unreasonable. But keep in mind that evidence largely suggests that sugar-sweetened beverages are worse than artificially sweetened ones for energy balance and weight maintenance, so swapping is not called for. Like most foods and food ingredients, the outcome probably is influenced by how the NNSs are used.
For another good look at this topic check out Colby Vorland's 2013 post, "Artificial Opinions."