Skip to content

Johnsen PH, Juul FE, Hoff DAL, et al. Faecal microbiota transplantation in irritable bowel syndrome (REFIT2): a randomised, double-blind, placebo-controlled, phase 3 trial. Lancet. 2026. PMID 42785338

One-paragraph summary

REFIT2 randomised adults with Rome IV irritable bowel syndrome and an IBS Severity Scoring System (IBS-SSS) of at least 175, aged 18–65, at five Norwegian hospitals, 2:1 to a once-only rectal enema of healthy-donor faeces or their own faeces. The primary endpoint was a reduction of 75 points or more in IBS-SSS at 90 days, in everyone who received the allocated treatment. Of 2,304 people assessed, 450 were enrolled (299 donor, 151 autologous). Two were excluded because the allocated treatments were switched, leaving 448 analysed: 293 (65%) women and 155 (35%) men, median age 36 years (IQR 29–44). Improvement of at least 75 points occurred in 119 (40%) of the donor group and 57 (38%) of the autologous group. The absolute difference was 1.9 percentage points (95% CI −8.1 to 12.0; p=0.76). The proportion with adverse events was similar; the abstract does not give counts. The trial was funded by KLINBEFORSK and is registered as NCT04691544. No volume, issue or page numbers were in the PubMed record on 2026-09-30.

Key findings

  • No clinical benefit of healthy-donor over autologous FMT on the prespecified ≥75-point IBS-SSS endpoint at 90 days.
  • The comparison isolates donor material from the procedure: both arms received a rectal enema.
  • The authors' interpretation is that microbiota modulation alone might be insufficient for symptom improvement in IBS.
  • An accompanying Lancet commentary is indexed (Liu J, Shin A. Healthy donor faecal microbiota transplantation in irritable bowel syndrome: interpreting negative findings. Lancet. 2026. PMID 42785337). Its abstract is empty, so no argument is taken from it.

Limitations

  • One enema, and a 90-day primary endpoint, in a relapsing condition.
  • The responder threshold is ≥75 IBS-SSS points, not the ≥50-point threshold used by El-Salhy 2020 (PMID 31852769), Lahtinen 2020 (PMID 32343000) and Yau 2023 (PMID 37667968). The abstract does not report a 50-point result.
  • Rome IV only, ages 18–65, five Norwegian hospitals.
  • Route and dose differ from El-Salhy's gastroscopic 30 g and 60 g single-donor trial. A null donor-versus-autologous enema does not explain that result, and it does not randomise recipients among several allogenic donors.
  • Two participants were excluded after their allocated treatments were switched. Adverse-event counts are not in the abstract.
  • On 2026-09-30 ClinicalTrials.gov still listed NCT04691544 as active, not recruiting, enrolment 450, last update 2024-02-26, with no results posted.

Why it matters

This is the powered phase 3 the field had been waiting on to separate donor effect from procedure effect. It is negative. Smaller negative trials (Lahtinen 2020, PMID 32343000; Yau 2023, PMID 37667968) can no longer be set aside as underpowered, and donor-versus-procedure is not a sufficient account of why El-Salhy 2020 (PMID 31852769) was strongly positive. The rates are not directly comparable across those trials, because the responder definition differs.

Cited by wiki pages

  • microbiome
  • clinical-trials-landscape