Sweetened Foods and the Art of Uncertainty

It’s May 22, 1857, and a New York Times journalist is encouraging Americans to eat less sugar.

His concerns are financial – the price of sugar is about to spike – but at the end he shifts to public health:

“We are, beyond question, the greatest sugar-consumers in the world, and many of our diseases may be attributed to too free a use of sweet food.”

This could’ve been written yesterday.

By some metrics, America still leads the world in per capita intake of sugar and high-fructose corn syrup. Although consumption is down since the late 1990s (see below), an estimated 60% of us still eat too much added sugar. And, on a given day, more than a third of Americans consume non-nutritive sweeteners such as aspartame, sucralose, saccharine, or stevia.

Our “use” of sweet food is clearly too free. In the 20th century, scientists established that excess sugar intake contributes to diabetes, cardiovascular disease, obesity, and other health problems that tend to be more prevalent in the U.S. than in peer nations. Experts have also begun to question the safety of non-nutritive sweeteners. Just two weeks ago, a report from Tufts researchers suggested that these sweeteners promote cardiometabolic disease.

In this newsletter I’ll be discussing the new report and exploring whether non-nutritive sweeteners are undermining public health. (The TL;DR is that I see no clear evidence of harm at low levels of consumption.) At the end, I’ll touch on potential strategies for addressing America’s sweetness problem. Should we be changing sweeteners, or scaling back, or both? How should we deal with uncertainties about the health impacts of NNS – or anything else we consume?

A quick note on terminology before digging in.

Sweet words

Americans are said to consume too much added sugar, a term used in reference to table sugar as well high fructose corn syrup, dextrose, added fructose, concentrated fruit juice, and more.

The new report suggests that our health is also undermined by non-nutritive sweeteners (NNS), substances that sweeten foods and drinks but provide zero or very few calories.

NNS include artificial sweeteners such as aspartame, sucralose, saccharin, and acesulfame potassium, as well as naturally-derived substances like stevia and monk fruit extract.

(Side note: The new report concerns “artificial and other non-nutritive sweeteners”, which also include sugar alcohols such as xylitol and erythritol. I’ll use the term NNS to keep things simple.)

The new report

This meta-analysis and systematic review, led by Meng Wang and colleagues at the Tufts Food is Medicine Institute, was published in Current Atherosclerosis Reports on June 25th. Wang and colleagues examined much of the existing literature on the cardiometabolic impacts of non-nutritive sweetener (NNS) intake.

We know that when added sugars are replaced by NNS, people may benefit from the reduction in calories. A separate question is whether NNS in and of themselves are beneficial or cause harm. To address this question, Wang and colleagues only examined research that compared NNS to water or some other non-caloric substance.

Two kinds of studies were included: 21 short-term controlled experiments (RCTs), and more than three dozen observational studies, many of which were part of systematic reviews or meta-analyses.

Three key findings

1. NNS have no clear benefits.

Wang and colleagues found no consistent links between NNS intake and more favorable lipid levels, adiposity, or blood pressure.

2. NNS may impair glucose regulation.

Most cells in our bodies obtain energy from glucose, which comes from the carbs we eat. Insulin, a hormone secreted by the pancreas, helps move glucose from the bloodstream into our cells. Normally, relatively little insulin is needed for this process.

The RCTs that Wang and colleagues reviewed show that after consuming NNS, fasting insulin levels as well as average blood sugar levels increased. (The blood sugar data, obtained from HbA1c testing, represent averages over the previous 2 to 3 months.)

Together these findings suggest that NNS cause our bodies to regulate glucose less efficiently.

The effects are small; they don’t tell us that an occasional diet soda would be harmful. However, the implication is that over time, consumption of NNS could lead to metabolic problems that include prediabetes and increased risk of diabetes, heart disease, and stroke.

The data are far from conclusive though.

For one thing, none of the RCTs lasted longer than a few months. Peoples’ bodies might adjust if they continue to consume NNS over a period of years.

More importantly, we don’t know how much NNS study participants were consuming.

Wang and colleagues only analyzed RCTs where NNS intake fell within the FDA’s recommended limits. These limits, called Acceptable Daily Intakes (ADI), seem extremely high. For instance:

This puts the data in a different light.

Wang and colleagues reported that NNS have a slightly negative impact on glucose regulation. But their data mixes people who drink one or two cans of Diet Coke per day with people who drink more than a dozen. People who use two packets of Splenda per day are mixed with people who may use as many as 15 to 17.

Given such a wide range of intake, I have little confidence that low levels of NNS intake are metabolically harmful. I think this is still unknown.

Unfortunately, the RCTs don’t tell us much about dose-response relationships. In a typical study, the experimental group consumes a given amount of NNS and is compared to another group who consumes something else (e.g., a placebo). Even when NNS dose is high, harmful effects aren’t necessarily observed.

3. NNS are associated with increased risk of type 2 diabetes, cardiovascular disease, stroke, and premature death.

The RCTs I mentioned earlier were tightly controlled and followed participants for a few weeks or months after they consumed (or did not consume) a specific amount of NNS.

Wang and colleagues also reviewed observational studies that tracked people for roughly a decade or more. These studies linked self-reported NNS intake to undesirable cardiometabolic outcomes.

Many of these studies focused on beverages with artificial sweeteners. Broadly, each additional daily serving (roughly one 8.5 to 12-ounce can) of an artificially-sweetened beverage increased the risk of type 2 diabetes by 15% and the risk of cardiovascular disease and premature death by 5 to 10%.

These percentages are relative risk values. Given your particular age, health status, and other characteristics, if you currently have a 10% risk of developing type 2 diabetes over the next decade, then drinking one more Diet Coke per day would increase your risk to 11.5%, at least according to the observational data.

However, I don’t find this evidence any more trustworthy than what the RCTs showed. These studies don’t rule out reverse causality, a concern that Wang and colleagues themselves acknowledged:

“People who are overweight or already at high risk for diabetes often switch to diet drinks and other NNS-containing products. Consequently, higher NNS consumption may partly reflect pre-existing metabolic risk rather than cause it.”

One more thing. Wang and colleagues rely heavily on data from the ongoing Nutrinet-Santé study being carried out in France. As I’ve noted elsewhere, Nutrinet-Santé is quite rigorous for an observational study, but the findings are quirky. In some cases, moderate doses of food additives are more harmful than higher doses, but the researchers can’t explain why.

In sum, I see no clear evidence that NNS are harmful, at least at lower levels of consumption.

A perspective on the sweetness problem

What can Americans do about our sweetness problem – i.e., overconsumption of added sugars and (maybe) NNS?

There are healthier alternatives to added sweeteners. Thanks to Driscoll’s, for instance, we can eat fresh berries any month of the year. In theory, we would be healthier if the sweetest foods we ate were fresh fruit.

However, in the real world, most of us (I’m including myself here) are unlikely to give up the ice cream and diet sodas and other sweetened foods we routinely crave. Scaling back seems like a more realistic strategy. The question is what our goals should be.

Added sugars

The newest Dietary Guidelines for Americans (2025-2030) is more stringent than the previous version:

“While no amount of added sugars or non-nutritive sweeteners is recommended or considered part of a healthy or nutritious diet, one meal should contain no more than 10 grams of added sugars.”

To put that in context, 10 grams would be about 2.5 teaspoons of sugar, or roughly what you’d get from one Nature Valley Oats’n Honey granola bar, three-quarters of a cup of Chobani low-fat vanilla Greek yogurt, or one packet of Quaker Instant Oatmeal: Maple & Brown Sugar.

(I chose relatively healthy food options here. You can also get 10 grams of sugar from, say, just one tablespoon of Smucker’s Strawberry Jam.)

Guidance on added sugar intake is tough to negotiate, not just because the advice varies from one source to the next, but because it’s hard to measure how much added sugar you actually consume. (The WHO, for instance, would require an additional step of dividing free sugars by total caloric intake.) Not even a smart phone nutrient calculator is going to be accurate to the gram.

Fortunately, there’s not much evidence of harm for less than roughly 30 to 50 grams of added sugar per day. Most of us can have a little sugar without increasing health risks, assuming no health conditions that preclude it.

Non-nutrient sweeteners and the art of uncertainty

The Tufts researchers advise caution with respect to NNS; the Dietary Guidelines for Americans recommends that we avoid them altogether.

Caution makes sense to me, but only because of lingering uncertainties. Though I don’t believe the Tufts meta-analysis impugns moderate intake of NNS, I think it does at least justify gathering more data.

In the meantime, how should we live with uncertainty around NNS – or any other potentially harmful substance?

In The Art of Uncertainty (2024), the statistician David Spiegelhalter writes that

“We cannot avoid uncertainty. So we need to embrace it, be humbled by it, and even try to enjoy it.”

I don’t enjoy uncertainty around the healthiness of dietary choices – I’d rather know for sure whether NNS are safe or not – but I do appreciate its inevitability, as well as the call for humility that permeates Spiegelhalter’s book.

It’s not hard to find social media influencers who tell you that added sugars “are” bad, or that some NNS “is” carcinogenic. The data rarely supports these categorical statements.

Take aspartame, for instance, an NNS widely used in diet sodas, chewing gum, ice cream, breakfast cereals, etc. The WHO’s current position is that aspartame is “possibly” carcinogenic (IARC), but that up to 40 mg per kilogram of body weight per day is “acceptable” (JECFA).

This is slightly more restrictive than the FDA’s ADI of 50 mg/kg per day. Still, it’s a big number. According to the WHO guidelines, it would be acceptable for a 154-pound person to drink up to roughly 14 Diet Cokes per day.

That seems a bit much. I can’t prove it, but I suspect that the negative health impacts of NNS intake are more likely among folks who drink more than a dozen diet sodas per day, as opposed to, say, one or two. In any case, I wouldn’t conclude that aspartame – or any other NNS – is categorically bad for you. I’m comfortable with some degree of uncertainty here. The art of uncertainty includes acceptance of uncertain conclusions.

Final thoughts

A statistician might describe certainty and uncertainty as two ends of the same spectrum: More of one means, in effect, less of the other. What I look for in health data is just enough certainty that the uncertainty seems tolerable. Spiegelhalter quotes a line from John Stuart Mill on this:

“There is no such thing as absolute certainty, but there is assurance sufficient for the purposes of human life.”

I feel assured – though not absolutely certain – that moderate intake of sweetened foods doesn’t increase health risks for most people, including me.

I do suspect that my added sugar intake is too high. This is a gut feeling (no pun intended), as I don’t count grams, my bloodwork is normal, and I can eat a second helping of ice cream without feeling ill. Still, I’ve been trying to cut back, without stressing about the fact that I may never know the exact amount I should be aiming for.

Thanks for reading!