Table 1.

Summary of findings

Study author(s) and date	Study design, (methods used for ASD diagnosis)	Dose of medications	Participants (N, age, (mean ± s.d.), gender, IQ (mean ± s.d.), intervention, FU)	Outcome measures used	Findings, Jadad score	Comments
Divalproex sodium
Hollander et al (2006)24	Parallel-design, double-blind, placebo-controlled, DSM-IV, ADI-R, ADOS	Divalproex sodium: mean dose at end-point was 822.92 ± 326.21 mg/day (range 500–1500/day). Mean trough serum level of divalproex at end-point was 58.23 ± 21.63 μg/ml	Divalproex: 9 PBO: 4 Age: 9.5 (5–17) years (one adult, age 40 years) Gender: not specified IQ: Mean: 60 (range 30–104) FU: 8 weeks	C-YBOCS	Statistically significant improvement in the treatment group compared with PBO in C-YBOCS score at FU (P = 0.037). Effect size (d = 1.61) Jadad = 4	Very small sample size and short study duration. Also, P-value was 0.05 when data on one adult participant were excluded
Hollander et al (2010)25	Parallel-design double-blind placebo-controlled, DSM-IV-TR, ADI-R, ADOS-G. CGI-S score of at least 4 to justify exposure to the medication.	Divalproex sodium <40 kg: 125 mg/day to 250 mg twice daily >40 kg: 250 mg/day to 500 mg twice daily	Divalproex: 16 PBO: 11 Age: 9.66 years (intervention), 8.97 years (control) (5–17 years) Gender: intervention group, 13 boys, three girls; PBO, ten boys, one girl IQ: 52.92 (Range 30–89) FU: 12 weeks	CGI-I, ABC, OAS-M, CYBOCS, VABS, YMRS	Overall, 62.5% in the divalproex group compared with 9% in the PBO group were responders (CGI-irritability OR: 16.7, Fisher's exact P = 0.008). A marginally statistically significant improvement was also noted on the ABC-irritability subscale (P = 0.048). There was a trend for responders to have higher valproate blood levels compared with non-responders. Participants with an abnormal EEG may have responded better to divalproex than those without any abnormality, but no definitive conclusion could be drawn because of the very small number involved. No statistically significant inter-group differences in either the VABS communication and socialisation domains, or the YMRS scores. Controlling for IQ did not change these findings. No significant inter-group difference on the aggression score according to OAS-M or repetitiveness score as per C-YBOCS. Jadad = 4	Small sample size and short study duration. Also, contradictory findings depending on the outcome measures used; whereas the primary outcome measure showed a significant difference, four secondary outcome measures failed to do so.
Martsenkovsky (2014)27	Non inferiority parallel-design double-blind RCT, DSM-IV, ADI-R, ADOS	Divalproex sodium dispensed in the form of sprinkles and titrated to tolerance or serum valproate level between 50 and 100 pg/ml v. risperidone (0.01–0.07 mg/kg/day).	Divalproex: 43 Risperidone: 43 Age: 4.87 ± 0.41 years Gender: Dvalporex 27 boys, 16 girls; Risperidone 31 boys, 12 girls IQ: Not specified FU: 16 weeks	CGI-I, OAS-M, ABC	Risperidone was significantly better than divalproex sodium in improving aggression, impulsivity, hyperactivity/non-compliance and stereotypy; CGI-I-irritability (P = 0.002), OAS-M (P = 0.005), ABC-I (teacher rating) (P = 0.04). Jadad = 0	No information on the drop-out rate or inclusion/exclusion criteria
Lamotrigine
Belsito et al (2001)23	Parallel-design double-blind placebo-controlled, ADI-R	Lamotrigine twice daily dose (0.5–5 mg/kg/day)	Lamotrigine: 14 PBO: 14 Age: 5.8 ± 1.75 years; (3–11 years) Gender: 27 boys, one girl IQ: Not specified FU: 18 weeks	AUBC, ABC-C, VABS, PL-ADOS, ADOS, CARS	No significant inter-group difference in scores on the following outcome measures: AUBC, ABC-C, VABS communication or daily living domains, PL-ADOS and CARS. No improvement in stereotypies, lethargy, irritability, hyperactivity, emotional reciprocity, sharing pleasures, language and communication, socialisation, and daily living skills. Marginally significant improvement in VABS social domain (P = 0.045). Scores improved by 3.9% for lamotrigine and 0.2% for placebo. Jadad = 5	Unnecessary use of too many outcome measures and a large placebo effect
Levetiracetam
Wang et al (2017)33	Single-blind parallel-design non-placebo-controlled, DSM-5	Levetiracetam 60 mg/kg/day	Levetiracetam + psychoeducation: 32 Psychoeducation alone: 35 Age: 61.6 ± 13.8 months (combined group), 63.1 ± 12.7 months (control group) Gender: 57 boys, ten girls IQ: Not specified FU: 6 months	PEP-3, CVP, EL, RL, CCS, CARS, ABC/AUBC	Both groups showed significant improvements in their behavioural and cognitive functions at FU according to PEP-3, CARS and ABC scores, but all these outcomes were significantly better in the combined than the control group at FU (P < 0.05). Subclinical EEG abnormalities improved significantly in the combined (75%) compared with the control group (14.3%) (P < 0.05). Jadad = 4	Single-blind and not placebo-controlled
Wasserman et al (2006)26	Parallel-design double-blind placebo-controlled, DSM-IV, ADI-R, ADOS	Levetiracetam: 125 mg/day to 250 mg/day to 20–30 mg/kg/day	Levetiracetam: 10 PBO: 10 Age: 7.62 years (intervention group), 9.82 years (control group) Gender: 17 boys, three girls IQ: 75.75 ± 33.4 (range 29–107) FU: 10 weeks	CGI-AD, ABC, Parent C-YBOCS, CPRS	No significant inter-group difference according to any of the outcome measures, except for teacher ratings on ABC-I (P: 0.003). The ABC-I score in the placebo group was significantly reduced at FU (61.59 to 58.94) compared with the intervention group (56.41 to 59.96). Jadad = 4	Small sample size and short study duration
Topiramate
Rezaei et al (2010)31	Parallel-design double-blind placebo-controlled add-on, DSM IV-TR, ADI-R, and ABC-I score ≥12.	Topiramate (100 mg/day to 200 mg/day) + risperidone (0.5–2 mg/day for children up to 40 kg and 3 mg/day for children over 40 kg.	Risperidone + topiramate: 20 Risperidone + PBO: 20 Age: 8.17 years (intervention), 7.85 years (control) Gender: 27 boys, 13 girls IQ: Not specified FU: 8 weeks	ABC-C subscales, ESRS	Of the five subscales of ABC-C, there were significant improvements in the treatment group compared with the control group in three, namely irritability, hyperactivity/ noncompliance and stereotypy (all P < 0.05) but not in lethargy/social withdrawal and inappropriate speech, which showed non-significant inter-group difference. Jadad = 5	Small sample size and short study duration. Also, contradictory findings according to the different ABC-C subdomain scores
Valproate
Hellings et al (2005)32	Parallel-design double-blind placebo-controlled, DSM-IV, ADI-R, ADOS	Valproate liquid (250 mg/day to 20 mg/kg/day); mean trough serum level was 77.8 μg/ml at trial end-point.	Valproate: 16 (13 completed) PBO: 14 (12 completed) Age: 11.2 ± 4.2 years (6–20 years) Gender: 20 males, 10 females IQ: 54 (20–137) (24 with IDD, two with borderline IQ, and one each with average and above average IQ, two missing IQ data) FU: 8 weeks	ABC-I, OAS, CGI-I, CGI-S	No significant inter-group difference according to any of the outcome measure scores. The same result was found for children with IDD. Jadad = 4	Small sample size, short study duration, heterogeneous sample and large placebo effect

ABC-C, Aberrant Behaviour Checklist-Community version; ADI-R, Autism Diagnostic Interview-Revised; ADOS, Autism Diagnostic Observation Schedule; ASD, Autistic Spectrum Disorder; AUBC, Autism Behaviour Checklist; CARS, Childhood Autism Rating Scale; CCS, communication composite score; CGI, Clinical Global Impressions Scale; CGI-AD, CGI Scale for Autistic Disorder; CGI-I, CGI Scale - Improvement; CGI-S, CGI-Severity, CPRS, Conners’ Parents Rating Scale; CVP, cognitive verbal/preverbal; C-YBOCS, Children's Yale Brown Obsessive Compulsive Scale; DSM-IV-TR, DSM-IV Text Revision; EL, expressive language; ESRS, Extrapyramidal Symptom Rating Scale; FU, follow-up; IDD; intellectual developmental disabilities; IQ, Intelligence Quotient; OAS-M, Overt Aggression Scale-Modified; PBO, placebo; PL-ADOS, Pre-linguistic Autism Diagnostic Observation Schedule; PEP-3, Psychoeducational Profile third edition; RL, receptive language; VABS, Vineland Adaptive Behavior Scales; YBOCS, Yale-Brown Obsessive Compulsive Scale; YMRS, Young Mania Rating scale.