The Trial That Got CBD Approved: 39% Fewer Seizures in Children With Dravet Syndrome
Trial of Cannabidiol for Drug-Resistant Seizures in the Dravet Syndrome
Bottom Line
The definitive randomized trial proved CBD reduces seizures in Dravet syndrome by 39% vs 13% with placebo — leading directly to FDA approval of Epidiolex.
Why It Matters
This NEJM trial provided the gold-standard evidence the FDA needed to approve Epidiolex — the first cannabis-derived drug ever approved in the United States. It completed the evidence chain from Charlotte Figi's case report through open-label trials to definitive proof.
The Backstory
The New England Journal of Medicine doesn't publish studies about cannabis. Not because of policy — because the evidence has never been good enough. Since the journal's founding in 1812, not a single cannabis-derived drug had produced the kind of randomized, controlled, blinded data that NEJM demands.
In May 2017, that changed. Orrin Devinsky's trial of cannabidiol for Dravet syndrome was the first cannabis-derived therapy to clear the highest bar in medicine — and the paper that forced the FDA to act.
The Trial
One hundred twenty children and young adults with Dravet syndrome — a severe, SCN1A-mutation-driven epilepsy characterized by prolonged, frequent, and life-threatening seizures — were randomized across 23 sites to receive either pharmaceutical-grade CBD oral solution (Epidiolex, 20 mg/kg/day) or placebo for 14 weeks. All patients had drug-resistant seizures: they were still seizing despite being on a median of three other antiepileptic drugs.
The design was everything Charlotte's Web was not: multicenter, randomized, double-blind, placebo-controlled. No parent co-authors. No dispensary products. No hope-driven narratives. Just a molecule, a placebo, and a protocol.
The primary endpoint — change in convulsive seizure frequency — showed a clear signal. The adjusted median difference between groups was -22.8 percentage points (95% CI: -41.1 to -5.4; p=0.01). In a population where patients were already on three drugs and still seizing, this additional reduction was clinically significant.
The Numbers Behind the Numbers
The headline "39% reduction" can feel abstract. For the families in this trial, the numbers meant something concrete:
12.4 → 5.9
convulsive seizures per month — the median change in the CBD group. That's roughly 7 fewer seizures per month. For a child having a seizure every 2-3 days, that could mean entire weeks without a convulsion.
The placebo group went from 14.9 to 14.1 — essentially no change. Dravet syndrome doesn't get better on its own. Without effective treatment, these children continue to seize.
Devinsky et al. (2017), N Engl J Med
Perhaps more striking: 5% of CBD patients — 3 out of 61 — became completely seizure-free during the trial. Zero placebo patients did. In Dravet syndrome, where seizure freedom is considered nearly impossible with available drugs, this was remarkable.
And 43% of the CBD group achieved at least a 50% reduction in seizures (the standard threshold for "responder" in epilepsy trials), compared to 27% on placebo. This means that for nearly half the patients who received CBD, their seizure burden was cut in half or more — on top of whatever their existing medications were already doing.
The Side Effects Were Real
This was not a free lunch. CBD at 20 mg/kg/day produced meaningful side effects:
The liver function findings were particularly important: CBD appears to interact with valproic acid (a common antiepileptic), causing elevated liver enzymes. This drug-drug interaction became a black box warning on the Epidiolex label and requires regular liver monitoring. It's a reminder that "natural" doesn't mean "harmless" — CBD is a pharmacologically active molecule that interacts with other drugs through CYP450 enzyme pathways.
From Trial to FDA Approval
The FDA Advisory Committee voted 13-0 to recommend approval — unanimous, which is unusual for any drug, let alone one derived from cannabis. The committee members were explicit: the evidence was strong, the unmet need was severe, and the risk-benefit ratio was favorable despite the side effects.
The DEA's subsequent rescheduling created one of the strangest legal situations in American drug policy. Epidiolex (purified CBD) was placed in Schedule V — the least restrictive category, alongside cough syrup with codeine. But CBD extracted from the cannabis plant remained Schedule I — the most restrictive category, alongside heroin. The same molecule, different legal status, depending on its source.
What This Study Doesn't Tell You
The trial was designed to answer one question: does CBD reduce convulsive seizures in Dravet syndrome? It answered that clearly. But several important questions remain:
Is whole-plant better than purified? Maa and Figi's case report argued it might be. This trial used purified CBD. Head-to-head comparisons of whole-plant extracts versus Epidiolex have not been conducted.
Does it work for other epilepsies? Epidiolex has since been approved for Lennox-Gastaut syndrome and tuberous sclerosis complex, but these are all severe, rare epilepsies. Whether CBD helps with more common forms of epilepsy remains unclear.
What about long-term use? The trial was 14 weeks. Children with Dravet syndrome need lifelong treatment. Laux's 2019 extension study provided some long-term data, but decades of follow-up data don't yet exist.
Is the dose right? Devinsky's 2018 dose-ranging study found that 10 mg/kg/day was as effective as 20 mg/kg/day with fewer side effects — suggesting the original trial may have used a higher dose than necessary.
Key Takeaways
Trial of Cannabidiol for Drug-Resistant Seizures in the Dravet Syndrome
Devinsky O, Cross JH, Laux L, Marsh E, Miller I, Nabbout R, Scheffer IE, Thiele EA, Wright S () · New England Journal of Medicine
Frequently Asked Questions
Cite this study
Devinsky, Orrin; Cross, J Helen; Laux, Linda; Marsh, Eric; Miller, Ian; Nabbout, Rima; Scheffer, Ingrid E; Thiele, Elizabeth A; Wright, Stephen. (2017). Trial of Cannabidiol for Drug-Resistant Seizures in the Dravet Syndrome. New England Journal of Medicine, 376(21), 2011-2020. https://doi.org/10.1056/NEJMoa1611618