Overview
- Authors Kevin Mitchell, Dorothy Bishop and colleagues review prior observational work, animal models and clinical trials and conclude the proposed causal link is unsupported.
- Highly cited comparative studies often included only 7–43 participants per group despite statistical guidance that thousands are needed, according to biostatistician Darren Dahly.
- Reported microbiome differences frequently vanished after controlling for diet or when comparing autistic children with their neurotypical siblings, pointing to likely reverse causation.
- Mouse findings based on "autistic-like" behaviors and human interventions such as fecal transplants or probiotics were faulted for small samples, poor controls and weak statistics.
- The authors call the hypothesis a dead end and urge either stopping such research or conducting rigorously powered studies, with resources prioritized for genetic investigations.