A cannabis extract improved some behavior measures in autistic children, but overall results were mixed | Research

The takeaway

In a 150-participant randomized trial, a whole-plant cannabis extract (20:1 CBD:THC) improved clinician-rated disruptive behavior and social responsiveness in autism, but did not improve the primary outcome measure of behavioral problems.

Parents of children with autism, developmental pediatricians, and researchers studying cannabinoid interventions for neurodevelopmental conditions.

49% improved disruptive behavior vs. 21% on placebo (p=0.005)

The Backstory

Thousands of families weren't waiting for the science. Long before Adi Aran's clinical trial enrolled its first participant, parents of children with severe autism were already using cannabis products — CBD oils, whole-plant extracts, dispensary tinctures — based on desperate hope and parent-community word of mouth. Online forums were filled with testimonials: reduced meltdowns, better sleep, spontaneous social engagement, the first eye contact in years.

The stories were compelling and impossible to evaluate. Were these genuine pharmacological effects, placebo responses from hopeful parents, or the natural fluctuation of symptoms misattributed to a new treatment? Without a controlled trial, nobody could say.

In 2021, Aran — a pediatric neurologist at Shaare Zedek Medical Center in Jerusalem — published the answer. Or rather, a complicated, honest, inconvenient fraction of an answer. The first randomized, placebo-controlled trial of cannabinoids for autism showed that the treatment was safe and well-tolerated. It showed real improvement in disruptive behavior and social responsiveness. And it failed to hit its primary endpoint. The evidence was, in the researchers' own words, "mixed and insufficient."

In medicine, this kind of result is often the most important kind. It tells you that something is happening — but not enough to declare victory. It tells you to keep looking.

The Trial Design

Aran's study was ambitious and carefully structured. 150 participants aged 5-21 with autism spectrum disorder were randomized to one of three groups:

150

children and young adults with ASD randomized in a double-blind, placebo-controlled trial — the first of its kind for cannabinoids in autism. Three groups received either whole-plant cannabis extract (CBD:THC 20:1), purified CBD+THC (same 20:1 ratio), or placebo for 12 weeks.

The inclusion of both whole-plant and purified formulations was a deliberate design choice, testing the entourage effect hypothesis: does the full spectrum of cannabis compounds work better than isolated cannabinoids? The 20:1 CBD:THC ratio ensured minimal psychoactive effects while maintaining the potential therapeutic contribution of THC.

Aran et al. (2021), Molecular Autism 12:6

After 12 weeks of treatment (the efficacy assessment period), participants underwent a 4-week washout and then crossed over for another 12 weeks to further assess tolerability. This crossover design meant every participant eventually tried both active treatment and placebo, maximizing the safety data available.

The study measured four outcomes:

HSQ-ASD (Home Situations Questionnaire — ASD version): Total behavioral problem score (primary outcome)
CGI-I (Clinical Global Impression — Improvement): Clinician-rated disruptive behavior improvement (co-primary outcome)
SRS-2 (Social Responsiveness Scale): Social communication deficits (secondary outcome)
APSI (Autism Parenting Stress Index): Parent stress (secondary outcome)

The Results: Promise and Disappointment

Efficacy Results

Mixed Outcomes: What Improved and What Didn't

Positive Signals

CGI-I: 49% much/very much improved vs. 21% placebo (p = 0.005)
SRS-2: 14.9-point improvement vs. 3.6 points placebo (p = 0.009)
Both measures showed clinically meaningful improvements in social behavior and disruptive behavior
Whole-plant extract outperformed purified cannabinoids on both measures

Significant

Negative Results

HSQ-ASD (primary outcome): no significant difference between groups
APSI (parent stress): no significant difference between groups
The trial technically 'failed' on its primary endpoint
Wide confidence intervals suggest high individual variability in response

Not significant

Aran et al. (2021), Molecular Autism

The most striking result was on the CGI-I for disruptive behavior: 49% of participants on whole-plant extract showed clinically meaningful improvement, compared to 21% on placebo. This is a statistically significant and clinically relevant difference — nearly half the treated children showed substantial behavioral improvement versus about one-fifth on placebo.

The SRS-2 social responsiveness data was similarly encouraging: a 14.9-point improvement versus 3.6 points on placebo. For families of children with autism, improvements in social communication are among the most meaningful outcomes.

But the primary outcome — the HSQ-ASD total behavioral problem score — showed no significant difference. In clinical trial methodology, when the primary endpoint fails, the positive secondary outcomes must be interpreted cautiously. The result is suggestive, not definitive.

Whole-Plant vs. Purified: The Entourage Effect

One of the study's most interesting findings was that the whole-plant extract consistently outperformed the purified cannabinoid formulation. The whole-plant extract — containing the full spectrum of cannabis compounds including minor cannabinoids and terpenes — showed stronger effects on both the CGI-I and SRS-2 than the purified CBD+THC at the same ratio.

This is consistent with the entourage effect hypothesis — the idea that cannabis compounds work better together than in isolation. If confirmed in larger trials, it would have significant implications for how cannabinoid medications are developed: the pharmaceutical preference for single-molecule drugs might not apply to cannabis therapeutics.

Safety Profile

Safety

Well-Tolerated, With Predictable Side Effects

Somnolence

28% on whole-plant extract, 23% on purified cannabinoids, 8% on placebo. Drowsiness is the most common CBD side effect and is consistent with all CBD clinical trials including Epidiolex.

Decreased appetite

25% on whole-plant extract, 21% on purified cannabinoids, 15% on placebo. Notable because THC typically increases appetite — the CBD dominance in the 20:1 ratio may explain this.

Serious adverse events

No treatment-related serious adverse events were reported. The cannabinoid treatments were well-tolerated across 3 months of administration.

Age range

Participants ranged from 5-21 years. The safety data in children as young as 5 is significant, as most cannabinoid trials exclude younger children.

Aran et al. (2021), Molecular Autism

Why the Endocannabinoid System Matters in Autism

The biological rationale for cannabinoid therapy in autism is stronger than most people realize. The endocannabinoid system plays documented roles in several processes disrupted in ASD:

Biological Mechanism

The ECS-Autism Connection

▼

Aran et al. (2021); Wei et al. (2015), Neuron; Zamberletti et al. (2017)

The Researcher

Adi Aran is a pediatric neurologist and head of the Neuropediatric Unit at Shaare Zedek Medical Center in Jerusalem. His journey into cannabis and autism research began with the families in his clinic. Parents of his patients were already using cannabis products, and he recognized that the absence of clinical evidence wasn't stopping families from experimenting — it was just ensuring they did so blindly.

Israel's position as a leader in both cannabis research (the field was essentially founded there by Raphael Mechoulam) and medical cannabis policy made it one of the few places where this trial could be conducted. Aran has since launched follow-up trials with larger sample sizes and longer durations, and his work has helped establish autism as a legitimate target for cannabinoid research rather than fringe speculation.

What This Means for Families

The results are genuinely complicated, and families deserve honesty about what they mean:

Something is happening. The improvements in disruptive behavior (49% vs. 21%) and social responsiveness (14.9 vs. 3.6 points) are real and statistically significant. This is not placebo.
It's not a miracle cure. The primary outcome was negative. Not all symptoms improved. Not all children responded. The effect sizes were modest.
Safety is encouraging. Three months of CBD-dominant treatment in children as young as 5 produced no serious adverse events. This matters enormously for families weighing risks.
The whole-plant extract worked better. For families choosing between CBD isolate and full-spectrum products, this data favors full-spectrum — though the evidence is preliminary.
More evidence is coming. Larger trials are underway. The question isn't whether to study cannabinoids for autism, but how to study them optimally.

Key Takeaways

Cannabinoid treatment for autism: a proof-of-concept randomized trial

Aran A, Harel M, Cassuto H, Polyansky L, Schnapp A, Wattad N, Shmueli D, Golan D, Castellanos FX (2021) · Molecular Autism

PubMed DOI

Related Research

Key studies in this area

()

Should I give my autistic child CBD?

This is a decision that should involve your child's medical team. The Aran trial provides encouraging but not definitive evidence. CBD was well-tolerated and showed improvements on some measures but not the primary outcome. If you do proceed, use a characterized product with known CBD content, start at a low dose, and work with a clinician who can monitor response and side effects.

Full-spectrum or CBD isolate?

In this trial, the whole-plant extract (full-spectrum) outperformed the purified cannabinoid formulation. This is consistent with the entourage effect hypothesis but is not yet definitive. If using CBD for autism, full-spectrum products have slightly more evidence behind them based on this trial.

Why didn't the primary outcome reach significance?

The HSQ-ASD measures a broad range of behavioral problems. The positive results were on more specific measures: clinician-rated disruptive behavior improvement and social responsiveness. It's possible that CBD helps specific autism symptoms rather than the entire behavioral profile — a common pattern in psychiatric medication trials.

Is there a follow-up trial?

Yes. Aran has launched larger trials based on these proof-of-concept results. Multiple other research groups worldwide are also conducting cannabinoid-autism trials. The field is moving rapidly.

What the researchers found

The primary outcome (HSQ-ASD Total Score) showed no significant difference between groups. However, 49% on whole-plant extract versus 21% on placebo showed much/very much improved disruptive behavior on CGI-I (co-primary, p=0.005). SRS Total Score improved by 14.9 vs. 3.6 points (secondary, p=0.009). Common side effects: somnolence (28%) and decreased appetite (25%). No serious treatment-related adverse events.

Why it matters

This is one of the first placebo-controlled trials of cannabinoids specifically for autism. The mixed results (positive on some measures, negative on the primary outcome) suggest cannabinoids may help certain autism symptoms but are not a broad-spectrum treatment.

The numbers in context

150 participants ages 5-21. Primary outcome (HSQ-ASD): no difference. CGI-I disruptive behavior: 49% vs. 21% improved (p=0.005). SRS: 14.9 vs. 3.6 point improvement (p=0.009). Somnolence 28%, decreased appetite 25% on extract. No serious AEs.

How the study worked

Placebo-controlled, double-blind trial of 150 participants ages 5-21 with ASD. Tested whole-plant extract (BOL-DP-O-01-W, CBD:THC 20:1) and purified cannabinoids (BOL-DP-O-01, same ratio). 12-week efficacy period, 4-week washout, then crossover for 12 more weeks.

What this study cannot tell us

Primary outcome was negative. Wide age range (5-21) and functional levels among participants. No pharmacokinetic data. The crossover design makes longer-term efficacy assessment difficult. Relatively high somnolence rate.

How to read the evidence

Well-designed RCT with 150 participants, but the primary outcome was negative and the positive findings were on secondary/co-primary measures.

When this study was published

2021 RCT (NCT02956226). First large placebo-controlled trial of cannabinoids specifically for autism.

The bigger picture

The autism community has shown enormous interest in cannabinoid treatments. This trial provides the first rigorous evidence that while cannabinoids appear safe and may help some symptoms, they do not universally improve behavioral problems in ASD.

Questions still open

Could specific ASD subgroups (by age, severity, or symptom profile) benefit more? Would different cannabinoid ratios or doses improve primary outcomes? Does the improvement in social responsiveness persist beyond the treatment period?

Common questions

Did cannabis help autism symptoms?

Results were mixed. Disruptive behavior improved significantly more on the cannabis extract (49% vs. 21%), and social responsiveness improved. But the primary measure of overall behavioral problems did not show a significant difference.

Was the treatment safe?

No serious treatment-related adverse events occurred. The most common side effects were somnolence (28%) and decreased appetite (25%), both mild to moderate.

Read the original research

Cannabinoid treatment for autism: a proof-of-concept randomized trial.

Molecular autism, 12(1), 6

View on PubMed ↗Publisher’s paper ↗

Citation

Aran, Adi; Harel, Moria; Cassuto, Hanoch; Polyansky, Lola; Schnapp, Aviad; Wattad, Nadia; Shmueli, Dorit; Golan, Daphna; Castellanos, F Xavier. (2021). Cannabinoid treatment for autism: a proof-of-concept randomized trial.. Molecular autism, 12(1), 6. https://doi.org/10.1186/s13229-021-00420-2