News · Nutrition & Diet
Feeding adults with type 2 diabetes a DASH-style diet lowered three separate blood sugar markers
The DASH diet was built for blood pressure. A version tailored for diabetes was tested by cooking every meal for 101 people, which removes the usual problem of not knowing what anyone actually ate.
- All three blood sugar measures fell on the tailored diet.
- Every meal was prepared and provided, so intake was known rather than reported.
- The individual reductions were small, but they moved together.
- 101 adults, average age 67, and 87% were Black adults.
- A randomized crossover: every person ate all four diets, 5 weeks each.
Almost everything known about diet and disease rests on asking people what they ate, and people are unreliable narrators of their own meals.
They forget. They round down. They describe the week they meant to have. Every large nutrition study carries that error somewhere in it, which is why the field produces so many findings that fail to replicate.
A controlled feeding trial removes the problem by removing the choice. Researchers cook and provide every meal, and what a participant ate is a matter of record.
What was tested
Writing in Diabetes Care, researchers took the DASH diet, which was designed decades ago to lower blood pressure, and used a version tailored for diabetes.
Adults with type 2 diabetes were fed four diets in a random order: the tailored diet and a comparison diet meant to represent typical American eating, each at higher and lower sodium. Every participant ate every diet, so each person served as their own comparison.
Each period lasted five weeks.
What moved
Three separate measures of blood sugar, all in the same direction.
Fructosamine fell, which reflects average glucose over the previous two to three weeks. Fasting glucose fell. And the longer-term measure, which tracks roughly three months, fell as well.
None of these reductions is large on its own. The longer-term one in particular is a small fraction of what a diabetes medication would produce, and nobody should read it as a substitute.
Why three small changes beat one
A single marker moving slightly is the kind of result that turns out to be noise about as often as it turns out to be real.
Three markers covering different time windows, all shifting the same way in the same people, is harder to explain away. They are measuring the same underlying thing through different instruments, and agreement between instruments is what separates a signal from a wobble.
The five-week window is also why the shorter-term measure matters here. A three-month marker cannot fully register a five-week change, so the trial was partly measuring with the wrong ruler and found the effect anyway.
Who was in it
101 people, average age 67, and 87% of them Black adults.
That last figure is unusual and worth stating plainly. Type 2 diabetes falls disproportionately on Black Americans, and feeding trials have historically recruited populations that look nothing like the people most affected. This one did the opposite.
The trade is the ordinary one. A result grounded in the group that carries the burden is more useful for them and less certain for everyone else.
The limits
Five weeks is short. Diabetes is measured in decades, and what matters in the end is eyesight, kidneys, nerves and hearts rather than a laboratory value.
Feeding trials also cannot tell you whether anyone would keep it up. Handing someone every meal for five weeks establishes what the food does to the body; it says nothing about whether a person can shop, cook and sustain the same pattern for a year.
And 101 people is a small trial, which is what controlled feeding costs.
What it adds
Type 2 diabetes is a disease in which your blood glucose, or blood sugar, levels are too high, and the advice given for it has long included eating patterns supported by weaker evidence than this.
What this trial supplies is a diet tested under conditions where the eating is not in doubt, in the population most affected, with three markers agreeing. That is a firmer footing than most dietary advice in diabetes currently stands on.
People also ask
What did the trial find?
Among 101 participants (mean age 67 years, 65% female, 87% Black adults), compared with the comparison diet, the DASH4D diet significantly reduced end-of-period fructosamine (adjusted difference -5.6 micromol/L, P = 0.002), fasting glucose (adjusted difference -4.5 mg/dL, P = 0.02) and HbA1c (adjusted difference -0.09 percentage points, P = 0.04).
What is a controlled feeding trial?
One where the researchers prepare and provide every meal rather than advising people what to eat. It is expensive and hard to sustain, and it removes the largest source of error in nutrition research, which is that people misremember and misreport what they ate.
What is DASH?
Dietary Approaches to Stop Hypertension, an eating pattern developed to lower blood pressure. It emphasizes vegetables, fruit, whole grains and low-fat dairy while limiting saturated fat and sodium. This trial used a version adjusted for diabetes.
Are these reductions large?
No. Each one on its own is small, and the HbA1c change in particular is a fraction of what a medication would deliver. What makes them notable is that three different measures of the same thing all moved the same way over five weeks.
What is fructosamine?
A blood measure reflecting average glucose over roughly the previous two to three weeks, which is a shorter window than HbA1c covers. In a trial with five-week diet periods, it registers a change that HbA1c would only partly capture.
Why does the study population matter?
Because 87% of participants were Black adults, a group carrying a disproportionate burden of type 2 diabetes and usually underrepresented in feeding trials. It makes the result more relevant to that population and less certain for others.
Should someone with diabetes switch to this diet?
Diet changes in diabetes interact with medication doses, so this is a conversation with the clinician managing the condition rather than a change to make alone. This is general information rather than medical advice.