Hajizadeh‐Zaker 2018.

Study characteristics
Methods	10‐week trial of L‐carnosine versus placebo
Participants	Inclusion criteria: children 3‐12 years clinical diagnosis of autism based on the DSM‐5 criteria outpatients at a speciality clinic ABC‐I score of at least 12 Exclusion criteria: co‐occurring schizophrenia or psychotic disorder severe intellectual disability history of seizures presence of hepatic disease history of alcohol or drug abuse tardive dyskinesia any significant medical condition Location/setting: speciality clinic for autism in the children’s outpatient clinic (Iran) Sample size: 50 Number of withdrawals/dropouts: risperidone + L‐carnosine (n = 4) 4 discontinued treatment (withdrew consent), risperidone + placebo (n = 4) 4 discontinued treatment (withdrew consent) Gender: 35 male, 7 female Mean age: L‐carnosine + risperidone 8.24 (2.22), placebo + risperidone 7.90 (1.89) Mean IQ: details not provided Baseline ABC‐I or other BoC: ABC‐I > 22.0 at baseline Concurrent medications: details not provided History of previous medications: details not provided
Interventions	Intervention (L‐carnosine + risperidone): L‐carnosine was administered in tablet form, 400 mg twice daily, + tablet form of risperidone, 1‐3.5 mg/day, for 10 weeks Comparator (placebo + risperidone): risperidone was administered in tablet form, 1‐3.5 mg/day, plus placebo in tablet form for 10 weeks.
Outcomes	Primary outcomes: irritability, measured using the ABC‐I subscale (Aman 1985) AEs Secondary outcomes: tolerability
Notes	Study start date: December 2015 Study end date: November 2016 Funding: "This study was supported by a grant from Tehran University of Medical Sciences to Prof. S.A. (Grant No. 29571). This study was supported by a grant from Tehran University of Medical Sciences (Grant No. 29571)". Conflicts of interest: "No competing financial interests exist". Other ‐ trial registry: IRCT201512081556N83
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	A computerised random number generator (allocation ratio 1:1) was applied to generate randomisation codes
Allocation concealment (selection bias)	Low risk	Sealed opaque envelopes were used to keep the assignments to mask the allocation throughout the study.
Blinding of participants and personnel (performance bias) All outcomes	Low risk	"The children, parents, the physician who referred the children, the physician who arranged the medications, the rater, and the statistician were all blinded to the allocated treatment."
Blinding of outcome assessment (detection bias) All outcomes	Low risk	Quote: "The children, parents, the physician who referred the children, the physician who arranged the medications, the rater, and the statistician were all blinded to the allocated treatment."
Incomplete outcome data (attrition bias) All outcomes	Unclear risk	16% dropout in both groups.
Selective reporting (reporting bias)	Low risk	Outcome measures reported per protocol
Other bias	High risk	The contact author is also on the ethics committee at the university funding the study. The contact author is a peer‐reviewer for one of the journals in which some of their studies are published.