The Autoimmune Protocol (AIP) in Hashimoto's Thyroiditis: What Two Studies Actually Showed

Caveats

• Design. Both studies are single-arm: measurements before and after, with no one to compare against. No randomisation, no blinding, no group that carried on eating normally. No randomised controlled trial of the autoimmune protocol exists — not in thyroiditis, not in inflammatory bowel disease, not in rheumatoid arthritis, not in any condition; there are no meta-analyses of the protocol either. The entire evidence base consists of four interventional studies totalling 68 participants: two in thyroiditis, one in inflammatory bowel disease with 15 patients, and one in rheumatoid arthritis with 9 (Pardali EC, Metabol Open 2025).
• Well-being. Participants had volunteered for a programme they expected to work, and the authors list possible selection bias as a limitation. Separating the contribution of food, coaching, group support and physician attention is impossible in this design even in principle; the paper's own title reflects that: it studied the protocol as part of a multi-disciplinary intervention, not the diet by itself. So this design cannot attribute the improvement specifically to removing foods. A conflict of interest is disclosed in the publication itself: a co-author is a co-founder of the company whose coaching programme was the intervention under study.
• The thyroid in the first study. It did not move on a single measure: TSH 2.02 → 1.98 µIU/mL (p = 0.942), free T4 1.3 → 1.4 ng/dL (p = 0.418), free T3 2.4 → 2.4 pg/mL (p = 0.882), TPO antibodies 225 → 219 IU/mL (p = 0.455), thyroglobulin antibodies 110 → 124 IU/mL (p = 0.176). Thyroglobulin antibodies even rose on average, although the spread in that group was twice the mean itself, so the direction of the shift cannot be interpreted.
• Inflammation. The hs-CRP effect size was small, and the design does not permit «the protocol reduced inflammation» — only «it fell during the intervention». The authors also mention shifts in the white cell differential in their conclusions, but those hold up less well: the white cell count did not change significantly in the main analysis, and the rise in lymphocyte share did not survive correction for multiple comparisons, as the authors themselves state.
• The second study: antibodies went up. Over 12 weeks TSH and free T3 and T4 fell while staying within the reference range, thyroid volume decreased, thyroglobulin antibodies decreased slightly — and thyroid peroxidase antibodies rose significantly (Ihnatowicz P, Ann Agric Environ Med 2023). Participants' body weight fell, which the authors attribute to a caloric deficit: some of the shifts may follow from weight loss rather than from the composition of the diet. Across both studies: in one, antibodies did not change; in the other, they rose; a study showing them fall does not exist.
• What these studies do not contain. Follow-up longer than 12 weeks; hard outcomes — levothyroxine dose, progression to overt hypothyroidism, ultrasound change as a primary endpoint were not measured; there are almost no men: the first study was women-only, the second had one man among 28; and there are no data on what happens after the programme ends.
• The goal. The protocol promises to halt the autoimmune process with food. If the goal is lowering antibodies, coming off levothyroxine or halting the disease, neither study supports it, and the price is months of restriction, a risk of deficiencies and a celiac diagnosis deferred.
• Guidelines. The autoimmune protocol appears in no American or European Thyroid Association guideline — neither as a recommended option nor as one considered and rejected; in the most recent European Thyroid Association guideline on levothyroxine the phrases «autoimmune protocol», «elimination diet» and «Paleo» do not occur once, and diet is discussed there solely as a factor in drug absorption. This does not mean «proven not to work»: the evidence base was not even sufficient for the protocol to enter the discussion. A systematic review of the Paleolithic diet in autoimmune thyroid disease found eight studies in total, six of which were case reports (Hollywood JB, J Am Nutr Assoc 2023).
• The reintroduction phase. In practice it is the phase most often left unfinished, but there are no numbers on it: no data exist on what share of people outside a study protocol reach reintroduction, and there are no long-term data on the maintenance phase either.
• Microbiota. Both studies measured change against the participants’ own habitual diet. The earlier one is preliminary by the authors’ own description: ten volunteers, a single group, one month.
• Money. It is staple gluten-free products specifically that cost more: in Greek retail their median price per 100 g was clearly higher than for conventional ones. The remaining categories — snacks, sweets, sauces — were in fact slightly cheaper than conventional ones, while across the basket as a whole no significant difference was found (Bathrellou E, Nutr Bull 2025). The autoimmune protocol is broader than a gluten-free diet, and there are no cost data for it at all.
• Selenium. Heterogeneity between studies was very large, and the meta-analysis authors themselves phrase the conclusion cautiously; there was no effect on free T4 and T3, thyroglobulin antibodies or thyroid volume. A separate meta-analysis found no clinically meaningful benefit from selenium — not in well-being, not in TSH among untreated patients, not on ultrasound (Winther KH, Endocrine 2017). A Cochrane review of four early trials in 463 participants states the central point: not one of them tested whether the levothyroxine dose or quality of life changes (van Zuuren EJ, Cochrane Database Syst Rev 2013).
• Vitamin D and myo-inositol. Whether vitamin D affects thyroid function itself is answered differently by two meta-analyses. In the network meta-analysis the confidence intervals for vitamin D, myo-inositol and their combination crossed zero. There are no decent data on myo-inositol: no placebo-controlled trial has been done in thyroiditis.
• Iodine. The link between excess iodine and thyroiditis comes from observational data; no study has deliberately restricted iodine in people who already have thyroiditis.

This material is for information only and does not replace a consultation with a doctor.