A flawed analysis finds low-carb benefits disappear after six months


A new meta-analysis of randomized trials published in Acta Diabetologica suggests low-carb diets offer no long-term benefits for blood sugar control or weight loss. This is at odds with many other recent studies suggesting the opposite.

So, which claim is more likely to be accurate?

As we have discussed previously, before drawing any conclusions, we need to know how the authors defined low carb.

For the new meta-analysis, they defined low carb as between 26% and 45% of calories and very low carb as less than 130 grams or less than 26% of calories. They defined the control group, or as they called it the “balanced carb” group, as 45 to 60% of calories.

Once again, those definitions are inconsistent with the definitions of low carb and very low carb used by clinicians and researchers who work with patients using these diets.

At Diet Doctor, we consider anything below 100 grams per day as low carb and anything below 20 grams of net carbs as very low carb. For comparison, on a 2,000 calorie diet that equates to 20% of energy for low carb and 4% of energy for very low carb. That’s a big difference from the definitions in the recent study!

To understand the data further, here’s the breakdown for the low-carb diet studies included in the analysis:

Number of “low-carb” diet studies and percentages of calories from carbohydrates
Calories from carbohydrates Number of studies
≥40% 14
30-39% 6
26-29% 0

You can see the data are heavily weighted toward studies with greater than 40% of calories from carbohydrate. As a result, we have to dismiss any findings from the “low-carb” diet studies as they didn’t actually study diets low in carbohydrate!

Even with the poor definitions, they concluded that low-carb diets benefit Hemoglobin A1c (HbA1c) at three and six months, but not at 12 months and slightly worsen HbA1c at 24 months. And they found no difference in BMI at 12 and 24 months.

It is perplexing that the HbA1c would increase compared to the control group at 24 months, but given the definitions, I am not sure we can draw any meaningful conclusions from any of the findings.

The studies the authors included did not clarify the quality of carbs eaten, and when 40% of calories are from carbs, that leaves plenty of room for high-sugar, highly processed carbs.

What about the very low-carb diets? They also had beneficial effects on HbA1c at 3 and 6 months but were neutral at 12 and 24 months. They also had a benefit for BMI at 3 and 6 months, but only two studies had follow-up beyond 6 months, limiting the statistical strength of the analysis.

The claims for the very low-carb group are likewise questionable.

For instance, the authors list the 2009 study from Davis et. al. as being very low carb, claiming participants ate 20 to 40 grams of carbs per day. That may be how the study started, but the study shows by six months the low-carb group was eating 138 grams of carbs per day. So, as the study progressed, participants were clearly not eating a very low-carb diet.

The same is true for Goldstein 2011, where at six months, low-carb subjects were averaging 93 grams of carbs per day. Another study, Dyson 2007, was a little better in that its participants averaged 55 grams of carbs per day, but that is still a far cry from a true very low-carb diet.

One potential conclusion from this meta-analysis is that for people with type 2 diabetes, the degree of carb reduction matters. It matters a lot. The majority of the beneficial science for treating and reversing type 2 diabetes with carbohydrate reduction seems to be in strict low-carb diets. It is unlikely any of the studies in this meta-analysis limited carbohydrate to the extent considered to be clinically effective for improving metabolic health.

It is disheartening to see another group of researchers claiming to study low-carb diets, but doing so while using misleading definitions. Articles like this increase confusion about the efficacy and safety of low-carb diets for treating type 2 diabetes.

We wish we didn’t have to keep writing posts pointing out these errors, but we’ll continue to do so as long as journals continue to publish studies like this one. We hope it helps bring clarity to what the science says — and doesn’t say — about diets that are truly low in carbohydrate.

Rest assured, if you have type 2 diabetes and are following a very low-carb diet with the support of your healthcare team, science supports that you will likely improve your blood sugar control with less medication, and perhaps even reverse your diabetes.

Thanks for reading,
Bret Scher, MD FACC

More posts

Observational study suggests low-carb diets are associated with higher coronary calcium scores

Ketones improve cognitive function in a small randomized controlled trial

Low-carb improves pancreatic function and type 2 diabetes

Advanced low-carb topics