A blood pressure diet for rheumatoid arthritis? After 12 weeks on DASH, people in an Istanbul trial said they felt much better. Their swollen joints didn't change

DASH stands for Dietary Approaches to Stop Hypertension. It came out of a US trial published in 1997: plenty of fruit, vegetables and low-fat dairy, whole grains, fish, poultry and nuts, and less saturated fat, red meat and sugar. It was designed to bring blood pressure down, and it did. Since then it has been tried on almost everything else. The latest is rheumatoid arthritis, the autoimmune disease that attacks the lining of the joints.

Researchers in Istanbul recruited 82 adults aged 18 to 64 with seropositive rheumatoid arthritis (their blood carries the antibodies typical of the disease) from two city hospitals. Everyone had active disease, had had it for at least a year, and had been on the same medicines for months. Forty were given a personalised DASH plan for 12 weeks; 42 were told to carry on eating as usual. The results came out in Clinical Nutrition on 1 October 2026.

On the trial's main measure, the DAS-28 disease activity score, the DASH group improved by 0.92 points and the usual-diet group by 0.17, a gap of 0.75 points. Pain on a 0 to 10 scale fell by about two points more in the DASH group, and so did people's rating of their general health. Daily function improved, and the quality-of-life questionnaire moved a long way: on its 'role-physical' scale, which asks whether health gets in the way of work and daily life, the difference was 47 points out of 100. People who stuck most closely to the diet saw the biggest falls in disease activity.

Now the parts that didn't move. The number of tender joints and the number of swollen joints, counted by a clinician, showed no real difference between the groups. Neither did CRP, the blood marker of inflammation. The other blood marker, ESR, fell by about 4 mm/h more in the DASH group.

That matters because of how DAS-28 is built. It combines four things: tender joints, swollen joints, ESR and the patient's own rating of their general health. If the joint counts didn't change, most of the gap has to come from the other two. By our arithmetic, a 2.1-point fall in the general-health rating on a 0 to 10 scale alone accounts for about 0.3 of the 0.75-point difference. The diet may well have made people feel better. It is less clear that it calmed the disease.

Feeling better counts for a lot in a painful chronic illness. But this trial couldn't separate the diet from the attention. Nobody was blinded: the DASH group knew they were on a special diet and had a dietitian planning their meals, while the control group was simply sent back to their usual food. Pain, energy and general health are exactly the scores that move when people expect them to.

The bigger problem is who was left at the end. Of the 82 people randomised, only 49 completed the 12 weeks, so 40% dropped out. The authors analysed everyone as randomised and flag the attrition themselves. But when that many people leave an unblinded diet trial, the ones who stay tend to be the ones it suits.

The registration record adds a few more questions. The trial was registered on ClinicalTrials.gov in September 2025, after recruitment had started in May 2024; the record says so openly, explaining that this was a doctoral project and the team didn't know registration had to come first. The registry lists 49 participants, the number who finished, not the 82 randomised. And it names DAS-28 with both CRP and ESR as the main outcome, while the paper makes the ESR version primary and reports the CRP version as secondary. Both improved in the paper (by 0.75 and 0.62 points more than usual diet), so the switch doesn't change the result, but it is the kind of change a registry is meant to catch.

None of this means DASH is bad for people with rheumatoid arthritis. It is a sensible way to eat, and people with rheumatoid arthritis run a higher risk of heart disease. What this trial can't show is that it changes the disease itself. That needs a larger trial where the people counting joints don't know who is on the diet, and where the control group gets equal attention.