In a randomized controlled trial of 62 preschool children with autism spectrum disorder, a 2‑month program that combined rehabilitation with a modified ketogenic diet produced larger decreases in Autism Behavior Checklist and Childhood Autism Rating Scale scores than rehabilitation alone.
Study and findings
Sixty‑two children with ASD were enrolled at a single centre in China and randomly allocated (1:1) to a treatment arm (modified ketogenic diet plus standard rehabilitation) or a control arm (standard rehabilitation with a regular diet). Both groups received identical hospital‑based rehabilitation on weekdays and home‑based sessions on weekends for eight weeks. The ketogenic protocol consisted of a ketogenic nutritional powder and a structured meal plan; lunch was provided at the hospital, while other meals were prepared by caregivers. Primary outcomes were change scores on the Autism Behavior Checklist (ABC) and the Childhood Autism Rating Scale (CARS). After two months, each group showed significant within‑group improvement (p < 0.05). Between‑group analysis revealed that the diet‑plus‑rehabilitation group had greater reductions: ABC change score U = 324.0, p = 0.027; CARS change score U = 273.5, p = 0.003.
Clinical interpretation
The trial suggests that, in young children with ASD, adding a modified ketogenic regimen to an intensive rehabilitation program may enhance short‑term behavioral improvement beyond rehabilitation alone. The magnitude of the between‑group differences reached statistical significance for both rating scales, indicating that the diet contributed an effect detectable over the 8‑week period. Because the intervention was delivered alongside a standardized rehabilitation protocol, the observed benefit cannot be attributed solely to the diet, but the data support the hypothesis that metabolic modulation can augment behavioral outcomes in this population.
Limitations and open questions
The study was limited to a single centre and a relatively small sample, which restricts generalizability. Blinding of participants and caregivers was not feasible, raising the possibility of expectation bias influencing caregiver‑rated scales. The follow‑up duration was only two months, so durability of the effect remains unknown. Moreover, the trial did not include a third arm receiving a non‑ketogenic low‑carbohydrate diet, preventing isolation of the specific role of ketosis versus other dietary changes. Future multicentre trials with longer observation periods, objective neurobehavioral endpoints, and metabolic monitoring (e.g., ketone levels) are needed to clarify the mechanisms and long‑term clinical relevance.
Source
Liu L, Zhao Q, Zhou J, Liu Z, Wu D, He K. Evaluating the therapeutic efficacy of a modified ketogenic diet in children with autism spectrum disorder: a randomized controlled trial. BMC Pediatr. 2026;26(1):893. doi: 10.1186/s12887-026-07708-3. PMID: 42763398.