Do diet drinks really increase risk of type 2 diabetes

A friend sent me a news article and I immediately called BS on it. So pivoted on this week's email. The headline alone was enough to raise red flags. A new Australian study published in Diabetes & Metabolism linked both sugar-sweetened (SSBs) and artificially sweetened beverages (ASBs) to a higher risk of developing type 2 diabetes. Over 36,000 people were followed for nearly 14 years. It sounds impressive — but as always, the devil’s in the details.

This type of study - where they look at the habits and health parameters of people across a long time period - is designed to test for associations that researchers are then able to test in a randomised controlled trial. It’s hard to run a randomised controlled trial in this instance, the type of research that is considered gold standard, to truly determine if diet drinks have a causal role in type 2 diabetes. The outcome of interest takes years if not decades to develop. So this can only show associations, and it can’t be ruled out that what we are really seeing is reverse causality - people who are trying to lose weight or manage their blood sugar are more likely to drink diet soda as part of this strategy (and not that the diet soda was the cause of the diabetes).

How do we know which is which? Well, we don’t. We can look at similar studies (such as this) and it shows a similar finding (which suffers from the same limitations as the present study), but not always (such as this large-scale study). We also want to think about the mechanism: if type 2 diabetes is largely a condition where blood sugar can’t be controlled by insulin, then surely any food or drink that helps people reduce blood sugar should be a good choice. The researchers point to altered gut microbiome (as seen in studies like this) as the potential cause for this, however (again) this isn’t set in stone, with this study finding no meaningful difference in gut microbiota when consuming high quantities of sucralose and aspartame (two most commonly used sweeteners, especially in diet drinks). In truth, studies looking at gut microbiome are somewhat limited as the gut microbiome is changing all of the time and even if it wasn’t so labile, few could definitively say what the clinical significance of a change in this microbiome actually is. The researchers of the present study also posited that the change in insulin response when consuming diet sodas could influence type 2 diabetes risk, though the study they referenced to suggest this doesn’t show a change in acute food intake or glucose response with the authors of that study suggesting diet soda could be helpful for glucose regulation, not harmful. Given the benefit of consuming diet soda for weight loss (which reduces risk of metabolic disease such as type 2 diabetes) this suggestion that diet sodas contributes to risk doesn’t make sense to me.

So what are we seeing in a study like this? Let’s look at what they did and what they found.

What the study actually did:

•  Participants were asked, once, in the early 1990s, how often they drank sugary or diet soft drinks.
•  They were then followed for ~14 years. No further dietary data was collected.
•  Diabetes was self-reported via mailed questionnaires — not clinically verified with labs or glucose tests.

They found that people who drank ≥1 artificially sweetened beverage per day had a 38% increased risk of developing type 2 diabetes — even after adjusting for body mass and waist size.
All that said:

This is observational data — not a randomised trial — and the people who drank the most diet drinks weren’t metabolically average. They were, in fact:

•  More likely to have a higher BMI
•  More likely to carry central fat
•  More likely to smoke
•  Less physically active
•  Had poorer overall diet quality
•  And, notably, still drank more sugar-sweetened beverages too (this one made me LOL)

Even though the researchers adjusted for these variables, statistical models can’t scrub out all the overlap. These aren't isolated behaviours — they're parts of broader metabolic patterns. I mentioned reversed causality above: people at higher risk of diabetes may drink more diet sodas in an attempt to manage weight or blood sugar.

Interestingly, the link between sugary drinks and diabetes did disappear once they adjusted for body weight and waist size — reinforcing the idea that excess adiposity, not the sugar alone, may be the more direct driver of diabetes risk.

But for artificially sweetened beverages, the association persisted, though weaker — which headlines spun into “proof” that diet drinks cause diabetes.

As Dr. John Ioannidis, professor at Stanford and one of the most respected critics of medical research, points out in his landmark paper Why Most Published Research Findings Are False, small risk increases — like the 38% reported in this study — are especially prone to bias, confounding, and measurement error. Even after adjusting for known variables, unmeasured or imprecisely measured factors can still distort results. In studies based on self-reported data and one-off food questionnaires, it’s almost impossible to rule out these kinds of distortions — making it risky to treat correlation as causation.

In plain English? Associations like this one are often driven by bias, behaviour, and background noise, not biology.

If you’re metabolically healthy, active, and eating a nutrient-dense diet, a splash of stevia or a diet tonic isn’t going to give you diabetes. This study doesn’t prove causation — and shouldn’t drive public policy or fear-based headlines