Practice guideline: Treatment for insomnia and disrupted sleep behavior in children and adolescents with autism spectrum disorder

The guideline panel reviewed all available studies (through December 2017) testing interventions for sleep disturbances in children and adolescents with autism spectrum disorder (ASD). They examined two broad categories:

**Nonpharmacologic interventions:** Behavioral strategies (sleep hygiene education, parent training in bedtime routines, faded bedtime, positive reinforcement), weighted blankets, specialized mattress technology, and other environmental modifications.

**Pharmacologic/nutraceutical interventions:** Melatonin (immediate-release and sustained-release formulations), melatonin analogs (ramelteon), antihistamines, clonidine, trazodone, and other sleep medications.

The comparator conditions varied across studies: placebo-controlled trials, waitlist controls, treatment-as-usual, and pre-post designs. Outcome measures focused on sleep onset latency (time to fall asleep), total sleep time, night wakings, sleep efficiency, and parent-reported sleep quality.

The panel did not test a single intervention themselves — this is a systematic review that synthesizes evidence across multiple studies to produce clinical practice recommendations.

The systematic review included studies of children and adolescents with autism spectrum disorder (ASD) aged 0–21 years who had sleep disturbances. The exact total sample size across all included studies is not reported in a single number, but the review covered:

**Behavioral intervention studies:** Multiple randomized controlled trials (RCTs) and non-randomized studies, typically with sample sizes ranging from 20 to 80 children per study. Most studies recruited from autism clinics, developmental pediatrics centers, or community referrals.

**Melatonin studies:** Several RCTs and open-label trials, with sample sizes ranging from 16 to 125 children per study. The largest and most influential studies included children aged 2–18 years with ASD and chronic sleep onset insomnia.

**Weighted blanket studies:** One small RCT (n=67 children with ASD, aged 5–16 years) and a few smaller studies.

**Other pharmacologic studies:** Very limited — most had fewer than 30 participants.

The population was predominantly children with ASD diagnosed by DSM-IV or DSM-5 criteria, with co-occurring sleep disturbances defined by parent report or sleep diaries. Many studies excluded children with severe intellectual disability, epilepsy, or other significant medical comorbidities. Most studies were conducted in outpatient clinic settings in the United States, Canada, Europe, and Australia.

The guideline panel evaluated studies using a variety of sleep measurement tools:

**Parent-completed sleep diaries/logs:** The most common outcome measure. Parents recorded bedtime, sleep onset time, night wakings, and wake time daily for 1–4 weeks. This is subjective but practical for home-based studies.

**Actigraphy:** Wrist-worn devices that measure movement to estimate sleep-wake patterns. Used in some melatonin and behavioral studies. Provides objective data on sleep onset latency, total sleep time, and night wakings. Less accurate for detecting wakefulness during the night compared to polysomnography.

**Polysomnography (PSG):** Overnight sleep study in a lab. Used in a few small melatonin studies. Gold standard for objective sleep measurement but expensive, artificial environment, and difficult for children with ASD.

**Standardized questionnaires:** The Children's Sleep Habits Questionnaire (CSHQ, 33–52 items, parent-report, assesses bedtime resistance, sleep onset delay, sleep duration, night wakings, and daytime sleepiness). The Sleep Disturbance Scale for Children (SDSC, 26 items). The Family Inventory of Sleep Habits (FISH, 12 items, measures sleep hygiene practices).

**Clinical global impression scales:** Clinician-rated improvement in sleep problems (e.g., CGI-I, 1–7 scale, lower = much improved).

The panel rated the quality of evidence using the American Academy of Neurology's four-tier system (Class I = highest quality RCT, Class IV = case series or expert opinion). Most behavioral intervention studies were Class II or III. Melatonin studies included some Class I RCTs.

**Study design:** This is a systematic review and clinical practice guideline developed by the American Academy of Neurology (AAN). The panel followed the AAN's 2011 guideline development process, which includes:

1. **Systematic literature search:** PubMed, EMBASE, and Cochrane databases were searched for studies published through December 2017. Search terms covered autism, sleep disturbance, and various interventions.

2. **Study selection:** Two reviewers independently screened titles/abstracts and full texts. Disagreements were resolved by consensus. Inclusion criteria: original research (RCTs, non-randomized controlled trials, cohort studies, case-control studies, case series with ≥10 participants) in English, studying children/adolescents with ASD and sleep disturbance.

3. **Data extraction and quality rating:** Two reviewers extracted data on study design, sample characteristics, intervention details, outcomes, and adverse events. Each study was rated for quality using AAN's classification scheme (Class I–IV).

4. **Evidence synthesis:** The panel summarized findings across studies, noting consistency, effect sizes, and study limitations. They did not perform a formal meta-analysis (pooling effect sizes statistically) because of heterogeneity in interventions, outcome measures, and study designs.

5. **Recommendation development:** Recommendations were graded as Level A (established as effective/ineffective), Level B (probably effective/ineffective), Level C (possibly effective/ineffective), or Level U (insufficient evidence). The panel used a modified Delphi process to reach consensus.

**What this design can and cannot prove:**

**Can prove:** The systematic review can identify which interventions have the strongest evidence base across multiple studies. By synthesizing all available data, it provides a more reliable estimate of treatment effects than any single study. The structured quality rating helps distinguish well-conducted from poorly-conducted studies.

**Cannot prove:** A systematic review cannot establish causality for any single intervention — it depends on the quality of the underlying studies. The review cannot determine optimal dosing, duration, or patient subgroups that respond best, because individual studies varied too much. The review cannot detect rare adverse events that might only appear in very large populations. The review cannot account for publication bias (studies with null or negative results are less likely to be published) except through statistical tests that were not performed here.

**Major methodological weaknesses of this review:**

**No formal meta-analysis:** The panel did not pool effect sizes across studies, so we cannot get a single precise estimate of how much melatonin or behavioral interventions improve sleep. This limits the ability to compare interventions quantitatively.

**Search ended December 2017:** The review is now several years old. New studies on melatonin, behavioral interventions, and other treatments (e.g., cannabidiol, digital sleep interventions) have been published since.

**Heterogeneity of outcome measures:** Studies used different definitions of "sleep disturbance" and different measurement tools, making it difficult to compare results across studies.

**Limited evidence for pharmacologic options beyond melatonin:** The review found insufficient evidence for antihistamines, clonidine, trazodone, and other commonly prescribed sleep medications in children with ASD. This does not mean they are ineffective — it means no adequate studies exist.

**Industry funding:** Some melatonin studies were funded by melatonin manufacturers, which could introduce bias. The panel noted this but did not exclude those studies.

**Behavioral interventions (nonpharmacologic):**

**Level B recommendation:** Clinicians should counsel parents on sleep hygiene and behavioral strategies as first-line treatment. This is based on multiple Class II and III studies showing improvements in sleep onset latency, night wakings, and total sleep time.

**Specific behavioral strategies with evidence:** Consistent bedtime routine, faded bedtime (gradually moving bedtime earlier), positive reinforcement for staying in bed, and parent education on sleep hygiene. One Class II RCT (n=36) found that a 4-session parent training program reduced sleep onset latency by an average of 22 minutes (p<0.01) compared to waitlist control.

**Weighted blankets:** One Class II RCT (n=67 children with ASD, aged 5–16 years) found no significant difference in sleep onset latency, total sleep time, or night wakings between weighted blankets and regular blankets. The weighted blanket group had a mean sleep onset latency of 38 minutes vs. 36 minutes for regular blankets (p=0.72). No serious adverse events were reported.

**Specialized mattress technology:** No studies met inclusion criteria. Level U (insufficient evidence).

**Melatonin:**

**Level B recommendation:** Clinicians should offer melatonin if behavioral strategies have not been helpful, starting with a low dose (0.5–1 mg) and titrating up as needed. Clinicians should recommend pharmaceutical-grade melatonin if available.

**Effect on sleep onset latency:** Multiple Class I and II RCTs consistently found that melatonin reduces sleep onset latency by 30–60 minutes compared to placebo. The largest Class I RCT (n=125 children with ASD, aged 2–18 years, sustained-release melatonin 2–10 mg) found that melatonin reduced sleep onset latency from a mean of 68 minutes at baseline to 36 minutes at 12 weeks, compared to 68 to 62 minutes in the placebo group (p<0.001). This is a reduction of approximately 32 minutes.

**Effect on total sleep time:** The same Class I RCT found that melatonin increased total sleep time by approximately 30 minutes (from 7.5 hours to 8.0 hours) compared to placebo (p<0.001).

**Effect on night wakings:** Mixed results. Some studies found a small reduction in number of night wakings (1–2 fewer per night), while others found no significant difference.

**Dosing:** Effective doses ranged from 0.5 mg to 10 mg, with most children responding to 2–5 mg. Higher doses did not consistently produce better results. Immediate-release formulations were more effective for sleep onset; sustained-release formulations were more effective for maintaining sleep through the night.

**Adverse effects:** Generally mild and transient. Most common: headache (5–10%), dizziness (3–5%), morning drowsiness (10–15%), and gastrointestinal upset (3–5%). No serious adverse events were reported in any study. However, the panel noted that long-term safety data (>6–12 months) are lacking.

**Other pharmacologic interventions:**

**Antihistamines (diphenhydramine, hydroxyzine):** No adequate studies in children with ASD. Level U (insufficient evidence).

**Clonidine:** One small Class III study (n=20) suggested possible benefit for sleep onset, but no RCTs. Level U.

**Trazodone:** No studies in children with ASD. Level U.

**Ramelteon (melatonin receptor agonist):** One small Class III study (n=12) showed no significant benefit. Level U.

**Coexisting conditions and medications:**

**Level B recommendation:** Clinicians should assess for and address medications and coexisting conditions that could contribute to sleep disturbance. Common contributors include: stimulant medications for ADHD (can delay sleep onset), SSRIs (can cause insomnia or restless sleep), anxiety, depression, gastrointestinal issues (reflux, constipation), pain, and sensory sensitivities.

**Behavioral interventions:** A well-structured parent training program (4–8 sessions) can reduce the time it takes a child with ASD to fall asleep by roughly 20–30 minutes on average. This is a meaningful improvement for families — a child who previously took 60–90 minutes to fall asleep might now fall asleep in 30–60 minutes. The effect is roughly equivalent to what you might expect from a low-dose melatonin supplement, but without any medication side effects.

**Melatonin:** The most robust finding is that melatonin reduces sleep onset latency by about 30–60 minutes compared to placebo. To put this in perspective: if your child typically takes 70 minutes to fall asleep, melatonin might bring that down to 30–40 minutes. This is a large effect — comparable to the difference between having no sleep aid and having a prescription sleep medication in adults. The effect on total sleep time is more modest: about 30 minutes of additional sleep per night. This means a child who sleeps 7.5 hours might sleep 8 hours with melatonin.

**Weighted blankets:** The evidence suggests no meaningful effect. The difference between weighted and regular blankets was only 2 minutes in sleep onset latency — a trivial amount that could easily be due to chance. This is roughly the same as the effect of a placebo.

**Clinical significance:** For families struggling with a child who takes 1–2 hours to fall asleep, a 30–60 minute reduction is often life-changing. It can reduce parental stress, improve family functioning, and reduce the child's daytime behavioral problems. However, the panel emphasizes that these effects are averages — some children respond dramatically, others minimally.

**What the authors acknowledge:**

The evidence base is limited by small sample sizes, short follow-up durations (most studies <6 months), and heterogeneity in outcome measures.

Long-term safety data for melatonin in children are lacking — no studies followed children for more than 12 months.

The review could not assess the comparative effectiveness of different behavioral strategies or different melatonin formulations because of study heterogeneity.

Most studies excluded children with severe intellectual disability, epilepsy, or significant medical comorbidities, limiting generalizability.

The review did not include studies published after December 2017.

**What a critical reader would note:**

**Publication bias:** Studies with positive results are more likely to be published. The review did not formally test for publication bias (e.g., funnel plot analysis). The true effect of melatonin might be smaller than reported.

**Placebo response:** In the largest melatonin RCT, the placebo group also showed some improvement (sleep onset latency decreased from 68 to 62 minutes). This suggests a non-specific effect of participating in a sleep study (e.g., increased attention to sleep hygiene, expectation of benefit).

**Parent-report bias:** Most outcomes were based on parent-completed sleep diaries, which can be influenced by parental expectations and desire to see improvement. Objective measures (actigraphy) were used in fewer studies.

**Dosing variability:** Melatonin doses ranged from 0.5 mg to 10 mg across studies, and the optimal dose for individual children is unknown. Some children may respond to very low doses (0.5 mg), while others need higher doses. The panel's recommendation to "start low" is sensible but not based on strong dose-response data.

**Industry funding:** Some melatonin studies were funded by manufacturers. The panel noted this but did not exclude those studies. Industry-funded studies are more likely to report positive results.

**Lack of head-to-head comparisons:** No studies directly compared melatonin to behavioral interventions or to other pharmacologic options. We cannot say which is "best" for a given child.

**Generalizability to adults with ASD:** This review is specific to children and adolescents. Adults with ASD may respond differently to these interventions.

For someone running their own n=1 experiment (whether for a child with ASD or for yourself if you have ASD):

### What to test (specific intervention and dose)

**First-line: Behavioral sleep hygiene protocol**

**What:** Implement a consistent bedtime routine (same time every night, 30–60 minutes of calming activities before bed — bath, reading, quiet play). Remove screens 1 hour before bed. Keep bedroom dark, cool, and quiet. Use a visual schedule or social story if needed.

**Dose:** Do this every night for at least 4 weeks before evaluating.

**Why first:** No side effects, no cost, and the evidence supports it as effective for many children.

**Second-line: Melatonin (if behavioral strategies insufficient)**

**What:** Pharmaceutical-grade melatonin (immediate-release for sleep onset problems; sustained-release if the issue is staying asleep).

**Starting dose:** 0.5–1 mg, taken 30–60 minutes before desired bedtime.

**Titration:** If no effect after 1 week, increase by 0.5–1 mg every 1–2