Ankylosing spondylitis and diet

What the research found.

• The largest systematic review of nutrition across rheumatoid arthritis, axial spondyloarthritis, and psoriatic arthritis found that pooling its trials wasn't feasible. Heterogeneity between the interventions and the outcomes was the reason. So it reports a direction of effect and gives no pooled effect size at all.

  Van den Bruel and colleagues, Frontiers in Medicine, 2025

• The named findings in that review were in the other two diagnoses rather than in this one, where vegan, anti-inflammatory, and Mediterranean diets improved disease activity, inflammatory markers, and quality of life in rheumatoid arthritis. Synbiotics reduced interleukin-17 in psoriatic arthritis, which is a third diagnosis again. Nothing at all was named for axial spondyloarthritis.

  Van den Bruel and colleagues, Frontiers in Medicine, 2025

• A healthy overall way of eating was associated with lower incident rheumatoid arthritis in a meta-analysis, at an odds ratio of 0.54. That became 0.84 once only the three lowest-risk-of-bias studies were counted, with Mediterranean eating specifically at 0.88. The weaker studies are what produced the 0.54.

  Joerns and colleagues, Seminars in Arthritis and Rheumatism, 2025

• In 495 patients with ankylosing spondylitis, poor disease control was strongly associated with insomnia severity, at an odds ratio of 7.29. Corticosteroid use was associated with sleep disturbance at 3.98. Neither of those is a dietary finding, and both are in the cautions on this page for that reason.

  Tymms and colleagues, Clinical Rheumatology, 2021

What the largest review couldn't do

The biggest systematic review of nutrition across rheumatoid arthritis, axial spondyloarthritis, and psoriatic arthritis reached a conclusion about the state of its own literature rather than about diet. Pooling the trials wasn't feasible, because the interventions and the outcomes varied too widely to combine. So it reports a direction of effect and gives no pooled effect size at all.

That's the whole quantitative answer available for this condition. No diet has been tested in ankylosing spondylitis specifically, and the review that gathered what exists across the three inflammatory arthritis diagnoses couldn't combine those studies into a number. A review that can't pool is telling you about the research rather than about the food.

Its named findings were in the other two diseases rather than in this one. Vegan, anti-inflammatory, and Mediterranean diets improved disease activity, inflammatory markers, and quality of life in rheumatoid arthritis, and synbiotics reduced interleukin-17 in psoriatic arthritis. Nothing at all was reported for axial spondyloarthritis.

The low-starch diet, and an honest position

The most-discussed dietary claim in this condition is the low-starch or starch-free diet, and readers come looking for it. This site holds no source on it at all, which means it can neither support nor refute the claim. That's worth stating plainly.

Dismissing an untested claim is as unsupported as endorsing one, and a page that left out the thing readers came for would mislead by structure. What can be said is that the largest review of nutrition across these diagnoses produced no pooled estimate for any dietary approach. So nothing in this whole area rests on firm evidence.

If you've been given a reference for it, that's worth bringing to whoever manages your care. What this site won't do is manufacture a verdict in either direction out of nothing. An absence of evidence is a different thing from evidence of absence, and the difference is the whole of this section.

The numbers this page borrows

Almost every dietary figure with a confidence interval in this neighborhood comes from rheumatoid arthritis. A meta-analysis put a healthy overall way of eating at an odds ratio of 0.54 for developing rheumatoid arthritis, falling to 0.84 once only the three lowest-risk-of-bias studies were counted, with Mediterranean eating specifically at 0.88. The fall from 0.54 to 0.84 is what happens when the weaker studies drop out.

That figure describes who develops rheumatoid arthritis rather than what happens to somebody who already has ankylosing spondylitis. It's two removes from the question a reader on this page is asking. It's also the best-numbered thing available anywhere near this question, which is why it's here with the borrowing labeled every time.

Supplements borrow their numbers in that direction too. Marine oil reduced pain across 22 rheumatoid arthritis trials with a standardized mean difference of 0.21, whose upper confidence limit of 0.004 barely clears no effect, and function was unchanged. Not one of those was measured in this condition.

Where the effort is better spent

Ankylosing spondylitis is one of the few conditions on this site where exercise moves the measure clinicians use rather than only the way people feel. A network analysis found aquatic stretching reducing the BASDAI, a patient-reported disease activity score scored 0 to 10, by 1.42, land aerobic exercise by 0.94, and land stretching by 0.49. A separate meta-analysis found the BASFI, a similarly scored measure of everyday function, improving by 0.49 points.

Set against a dietary literature that couldn't be pooled at all, that's a clear difference in the strength of the evidence. If you've got limited energy and you have to choose where to put it, the exercise evidence in this condition is a great deal better than the dietary evidence. That's a judgment about where effort pays rather than a claim that diet does nothing.

That isn't an argument for ignoring diet. Cardiovascular risk is raised across inflammatory arthritis, which is a solid reason to eat well that doesn't depend on any trial in this disease. Eating well costs nothing you weren't spending anyway, so it isn't a trade against anything.

What is reasonable to do

Eat in a way that's good for your heart, because that reasoning holds up whatever the dietary trials eventually turn out to show. Mediterranean eating has the strongest general-population evidence of any approach discussed here. That's enough of a reason to try it on its own, without a disease-specific trial behind it.

Run any exclusion as an experiment rather than as a rule you intend to keep forever. A defined period, something specific tracked, and a planned reintroduction are what turn a change into information you can use. The alternative is living with a restriction for years without ever finding out whether it does anything at all, which happens more often than it should.

Tell your team about supplements, because interactions are real and they're rarely printed on the label. If you've taken corticosteroids at any point, ask about bone health as well. Corticosteroid use was associated with sleep disturbance in this population at an odds ratio of 3.98, and it brings bone risk separately from that.

Common misconceptions.

Cautions specific to this condition.

• Nothing in this condition supports cutting a food out permanently. Where people have learned something, it's been by dropping one food for a set time and then putting it back.
• Watch bone health if you've taken corticosteroids, since corticosteroid use was associated with sleep disturbance here at an odds ratio of 3.98 and brings bone risk of its own.
• Tell your team about supplements, because several interact with medication and the interactions are rarely on the label.
• Restriction is the route by which disordered eating gets worse. If that's part of your history, it's worth saying so to your team before anything changes.
• Ask about cardiovascular risk alongside dietary questions, because it's raised across inflammatory arthritis and it's the firmest reason to care about diet here.
• Put effort into exercise before diet if you have to choose, because exercise moves the disease activity score in this condition and no diet has been shown to.

Questions patients ask.