Summary of findings 4. Antidepressants compared to placebo for irritability, aggression, and self‐injury in autism spectrum disorder (ASD).

Antidepressants compared to placebo for irritability, aggression, and self‐injury in autism spectrum disorder (ASD)
Patient or population: participants (any age) with a clinical diagnosis of ASD who have displayed one or more unwanted or challenging behaviours at baseline assessment Setting: hospital inpatient or outpatient centres, education or disability settings, mental health settings, or clinics and research centres associated with universities Intervention: antidepressants Comparison: placebo
Outcomes		Anticipated absolute effects* (95% CI)		Relative effect (95% CI)	№ of participants (studies)	Certainty of the evidence (GRADE)	Comments
		Risk with placebo	Risk with antidepressants
Irritability Follow‐up: short term (up to six months) Measured via Aberrant Behaviour Checklist (Irritability subscale) (ABC‐I) (Aman 1985), Score range (0‐45) Lower scores indicate lower severity		The mean score in the placebo group ranged from 10.2 to 13.8	SMD 0.06 lower (95% CI 0.30 lower to 0.18 higher)	‐	267 (3 studies)	⊕⊕⊝⊝ Lowa	There was no evidence of a difference, but results are uncertain.
Aggression		No data were reported for this outcome in this comparison
Self‐injurious behaviour ‐ no data could be used for this outcome because of skewness (see Table 5)
Adverse effects Follow‐up: short term (up to six months)	Neurological	There was evidence of a higher rate of decreased attention in the intervention group		RR 4.16 (95% CI 1.07 higher to 16.11 higher); citalopram 1 study; clomipramine 1 study; fluoxetine 5 studies; fluvoxamine 1 study; sertraline 1 study; tianeptine 1 study	815 (10 studies)	⊕⊕⊝⊝ Lowb
		The was little to no evidence of a difference between groups for any of the other neurological adverse effects including activation syndrome (P = 0.64), agitation (P = 0.96), aggression or hostility (P = 0.83), anger or irritability (P = 0.35), autonomic disturbance (P = 0.83), CNS disturbance (P = 0.50), diaphoresis (sweating) (P = 0.49), drowsiness (P = 0.50), headache (P = 0.23), hyperactivity (P = 0.36), insomnia (P = 0.29), sedation (P = 0.16), sleep disturbance (P = 0.76), mood lability (P = 0.43), restlessness (P = 0.13), twitching (P = 0.17), tremor (P = 0.22), or vertigo (P = 0.65)
	Psychological	The was evidence of a higher rate of AEs in the intervention group for impulsive behaviour		RR 2.92 (95% CI 1.11 higher to 7.68 higher); citalopram 1 study	243 (4 studies)	⊕⊝⊝⊝ Very lowc
		The was evidence of a higher rate of AEs in the intervention group for stereotypy		RR 8.33 (95% CI 1.07 higher to 64.95 higher); citalopram 1 study
		The was little to no evidence of a difference between groups for anorexia (P = 0.42), verbal aggression (P = 0.36), suicidal ideation (P = 0.65), bad dreams (P = 0.28), unstable mood (P = 0.66), anxiety (P = 0.16) and depression (P = 0.79)
	Metabolic	There was evidence of a higher rate of decreased energy in the antidepressant group		RR 1.94 (95% CI 1.13 higher to 3.33 higher); citalopram 1 study	512 (7 studies)	⊕⊝⊝⊝ Very lowd
		The was little to no evidence of a difference between groups for appetite disturbance (P = 0.40), decreased appetite (P = 0.39), increased appetite (P = 0.85), and weight gain (P = 0.80)
	Musculoskeletal	The was little to no evidence of a difference between groups for motor disturbance (P = 0.30) or neck pain (P = 0.65)			202 (2 studies)	⊕⊝⊝⊝ Very lowe
*The risk in the intervention group (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI). AE: adverse effect; ASD: autism spectrum disorder; CI: confidence interval; CNS: central nervous system; RR: risk ratio; SMD: standardised mean difference
GRADE Working Group grades of evidence High certainty: we are very confident that the true effect lies close to that of the estimate of the effect. Moderate certainty: we are moderately confident in the effect estimate: the true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially different. Low certainty: our confidence in the effect estimate is limited: the true effect may be substantially different from the estimate of the effect. Very low certainty: we have very little confidence in the effect estimate: the true effect is likely to be substantially different from the estimate of effect.

^aDowngraded 1 level for imprecision (95% confidence intervals includes both benefit and harm), and 1 level for imprecision (small sample size of n = 267). 
^bDowngraded 1 level for study limitations (high risk of bias across multiple domains) and 1 level for imprecision (95% confidence intervals include both benefit and harm).
^cDowngraded 1 level for study limitations (high risk of bias across multiple domains), 1 level for inconsistency (direction of effect varied across studies) and 1 level for imprecision (small sample size of n = 279). 
^dDowngraded 1 level for study limitations (high risk of bias across multiple domains), 1 level for inconsistency (direction of effect varied across studies and 1 level for imprecision (95% confidence intervals includes both benefit and harm).
^eDowngraded 1 level for study limitations (high risk of bias across multiple domains), 1 level for imprecision (small sample size of n=202), and 1 level for imprecision (95% confidence intervals includes both benefit and harm).