Devinsky's clinical program produced the evidence that turned CBD from an anecdote into an FDA-approved drug. The open-label trial (214 patients, 36.5% seizure reduction) paved the way for the NEJM RCT (120 Dravet patients, 38.9% vs 13.3% placebo, p=0.01) and Epidiolex approval in June 2018.
Anyone who wants to understand how CBD went from folk remedy to FDA-approved drug. Parents of children with treatment-resistant epilepsy. Anyone evaluating CBD product claims.
12.4 → 5.9 seizures/month (CBD) vs 14.9 → 14.1 (placebo), p = 0.01
The Backstory
Before this trial, parents were fleeing across state lines. Families with children suffering from Dravet syndrome — a rare, catastrophic epilepsy that kills 15-20% of those it afflicts — were relocating to Colorado, a state where they could legally access cannabis oil. Some were dosing their seizing toddlers with homemade extracts. Some were breaking federal law. All were desperate.
In 2013, CNN's Sanjay Gupta told the story of Charlotte Figi, a Colorado girl with Dravet syndrome whose seizures dropped from 300 per week to less than one per month after her parents began giving her a CBD-rich cannabis extract. The segment was watched by millions. Families flooded Colorado. States began passing "CBD-only" laws. Charlotte's Web — the cannabis strain developed for her — became a household name.
The science was trailing the desperation. There were no randomized clinical trials. No placebo controls. No FDA-approved CBD product. Just parent reports and a tidal wave of hope.
Orrin Devinsky, director of the NYU Comprehensive Epilepsy Center, set out to produce the evidence that the families needed and the regulators required. The result was the clinical program that turned CBD from an anecdote into a medicine.
The Disease
The Open-Label Trial (2016)
Devinsky's program began with an open-label expanded access trial — the largest clinical study of CBD for epilepsy at that time.
36.5%
median reduction in monthly motor seizure frequency over 12 weeks in 137 patients with treatment-resistant epilepsy (including Dravet and Lennox-Gastaut syndromes) who received add-on CBD at 2-5 mg/kg/day titrated up to 25-50 mg/kg/day. Baseline median seizure frequency was 30 per month; at 12 weeks it dropped to 15.8. This was the first large-scale clinical evidence that CBD reduces seizures.
These were patients who had failed multiple anti-epileptic drugs. A 36.5% reduction in a treatment-resistant population is clinically significant — many conventional drugs fail to achieve any meaningful reduction in these patients.
Devinsky et al. (2016), Lancet Neurol 15:270-278
The trial enrolled 214 patients aged 1-30 across 11 US epilepsy centers. It was open-label — everyone knew they were getting CBD — which meant the placebo effect couldn't be ruled out. But the results were encouraging enough to justify what mattered next: a proper randomized controlled trial.
The NEJM Trial (2017)
The study that changed everything was published in the New England Journal of Medicine on May 25, 2017 — the most prestigious medical journal in the world.
How They Did It
The Trial That Made CBD a Medicine
Design
Double-blind, placebo-controlled, randomized trial. The gold standard of clinical evidence. Neither patients nor doctors knew who received CBD and who received placebo.
This design eliminates the placebo effect, expectation bias, and regression to the mean that plagued all prior CBD-epilepsy evidence
Population
120 children and young adults (ages 2-18) with Dravet syndrome confirmed by clinical criteria. All had drug-resistant seizures despite stable anti-epileptic drug regimens. Randomized 1:1 to CBD or placebo.
Drug-resistant = failed ≥2 anti-epileptic drugs at adequate doses
Intervention
Oral CBD solution (Epidiolex) at 20 mg/kg/day or matched placebo, added to existing anti-epileptic drugs. Treatment period: 14 weeks.
Pharmaceutical-grade purified CBD — no THC, no other cannabinoids, no terpenes
Primary endpoint
Change in convulsive seizure frequency during the treatment period compared to a 4-week baseline.
Convulsive seizures are the most dangerous type in Dravet — the kind that can cause SUDEP
Assessment
Caregivers recorded seizures in daily diaries. Independent adjudication. Intention-to-treat analysis.
Rigorous methodology matching any Phase III pharmaceutical trial
Devinsky et al. (2017), NEJM 376:2011-2020; NCT02091375
The results:
CBD Group (n=61)
- Median convulsive seizures: 12.4/month at baseline → 5.9/month
- Median reduction: 38.9%
- 5% of patients became seizure-free
- 62% rated as improved on Caregiver Global Impression of Change
- Adverse events: somnolence, decreased appetite, diarrhea, elevated liver enzymes
Clinically significant seizure reduction (p = 0.01)
Placebo Group (n=59)
- Median convulsive seizures: 14.9/month at baseline → 14.1/month
- Median reduction: 13.3%
- 0% became seizure-free
- 34% rated as improved on Caregiver Global Impression of Change
- Adverse events: less frequent than CBD group
Minimal change — confirming these seizures don't resolve spontaneously
Devinsky et al. (2017), NEJM 376:2011-2020
The p-value was 0.01. In a patient population where nothing else had worked, CBD cut convulsive seizures by nearly 40%. The placebo group showed essentially no change — confirming that the effect was real, not wishful thinking.
From Trial to Approval
Research Timeline
CBD Becomes a Medicine
Charlotte Figi's story airs on CNN
Sanjay Gupta's documentary shows a child with Dravet syndrome improving dramatically on CBD oil — triggers a national movement
Devinsky launches expanded access program
The first systematic clinical study of CBD for epilepsy begins across 11 US centers
Open-label results published in Lancet Neurology
214 patients, 36.5% seizure reduction — enough evidence to justify RCTs
NEJM publishes the Dravet RCT
120 patients, double-blind, placebo-controlled: 38.9% seizure reduction vs 13.3% placebo (p=0.01)
Thiele et al. publish Lennox-Gastaut RCT
Second indication confirmed — CBD also works for LGS, another catastrophic childhood epilepsy
FDA approves Epidiolex
First cannabis-derived prescription drug approved by the FDA. Approved for Dravet syndrome and Lennox-Gastaut syndrome in patients ≥2 years old.
DEA reschedules Epidiolex to Schedule V
A cannabis-derived compound is given the lowest federal drug schedule — a legal milestone
Charlotte Figi dies at age 13
The girl whose story started the movement dies from complications likely related to COVID-19. The medicine she inspired is now helping thousands.
Multiple sources
On June 25, 2018, the FDA approved Epidiolex — purified CBD oral solution — for the treatment of seizures associated with Dravet syndrome and Lennox-Gastaut syndrome in patients two years of age and older. It was the first cannabis-derived drug ever approved by the FDA. Three months later, the DEA placed it in Schedule V — the least restrictive federal drug schedule.
What the Trial Proved — and What It Didn't
Myth vs. Reality
This trial proved that cannabis oil cures epilepsy.
This trial proved that pharmaceutical-grade purified CBD, at specific doses (20 mg/kg/day), significantly reduces convulsive seizure frequency in Dravet syndrome when added to existing anti-epileptic drugs. It is not a cure — most patients still had seizures, and only 5% achieved seizure freedom. It used purified CBD (Epidiolex), not cannabis oil — no THC, no other cannabinoids, no terpenes. The results cannot be directly extrapolated to dispensary CBD products, which vary in purity, dosage, and formulation.
The Evidence
Devinsky et al. (2017): 38.9% median reduction in convulsive seizures vs 13.3% placebo (p=0.01). 5% seizure-free in CBD group vs 0% placebo. Epidiolex is ≥98% pure CBD.
Devinsky et al. (2017), NEJM 376:2011-2020
Several important nuances:
It used pure CBD, not whole-plant cannabis. Epidiolex is pharmaceutical-grade cannabidiol — no THC, no terpenes, no minor cannabinoids. This is significant for the entourage effect debate: pure CBD works for epilepsy without any "entourage." Whatever makes CBD anticonvulsant, it doesn't require other cannabis compounds.
The doses are high. 20 mg/kg/day for a 30 kg child is 600 mg of CBD daily. Typical over-the-counter CBD products contain 10-50 mg per serving. The clinical doses that produced seizure reduction are an order of magnitude higher than what most consumers take.
Adverse events were real. Somnolence, decreased appetite, diarrhea, and elevated liver enzymes (especially in patients also taking valproate) occurred at clinically significant rates. CBD is not a supplement without side effects — at therapeutic doses, it's a drug with a drug's risk profile.
It doesn't generalize to all epilepsy. The trial studied Dravet syndrome specifically. Subsequent trials confirmed efficacy in Lennox-Gastaut syndrome. Extrapolating to all forms of epilepsy — or to other conditions entirely — is not supported by this data.
Devinsky: The Right Scientist at the Right Time
Orrin Devinsky was uniquely positioned to run this program. A Harvard Medical School graduate mentored by Norman Geschwind, he had spent decades building the NYU Comprehensive Epilepsy Center into one of the premier programs in the country. He founded Finding A Cure for Epilepsy and Seizures (FACES) and co-founded epilepsy.com.
Devinsky understood the desperation of epilepsy families. He also understood that desperation wasn't evidence. When families began moving to Colorado for CBD oil, he recognized that the movement would either be validated by rigorous science or collapse under the weight of uncontrolled anecdotes.
He chose to produce the science. GW Pharmaceuticals provided the pharmaceutical-grade CBD (Epidiolex) and funding. The Epilepsy Foundation and FACES provided infrastructure. Eleven epilepsy centers across the United States enrolled patients. The result was the clinical evidence that turned a folk remedy into an FDA-approved drug.
Related Research
From Plant Chemistry to Prescription Pad
This trial sits at the end of a sixty-year arc — from THC isolation to the first FDA-approved cannabis-derived drug.
Isolation, Structure, and Partial Synthesis of an Active Constituent of Hashish
Gaoni & Mechoulam (1964)
Where it started — THC isolation. CBD was identified even earlier (1940) but ignored for decades.
A tale of two cannabinoids: THC and CBD therapeutic rationale
Russo & Guy (2006)
The paper that argued CBD had independent therapeutic value beyond modifying THC — Epidiolex proved it.
Taming THC: potential cannabis synergy and phytocannabinoid-terpenoid entourage effects
Russo (2011)
The entourage effect review — but Epidiolex works as pure CBD, challenging the 'whole plant is always better' claim
Cannabinoid pharmacology: the first 66 years
Pertwee (2006)
The pharmacology review that documented CBD's anticonvulsant properties decades before clinical trials existed
The ECS as an emerging target of pharmacotherapy
Pacher, Batkai & Kunos (2006)
The comprehensive review of cannabinoid therapeutics — Epidiolex vindicated the approach
What is Epidiolex and how is it different from CBD oil?
Epidiolex is a pharmaceutical-grade, FDA-approved oral solution of purified cannabidiol (CBD). It contains ≥98% pure CBD with no THC, no other cannabinoids, and no terpenes. It undergoes the same manufacturing, quality control, and regulatory scrutiny as any prescription drug. Over-the-counter CBD oils vary enormously in purity, concentration, and formulation — studies have found that many contain less CBD than labeled, and some contain undisclosed THC. The clinical trial results for Epidiolex cannot be directly applied to unregulated CBD products.
Can CBD cure epilepsy?
No. CBD can significantly reduce seizure frequency in certain types of drug-resistant epilepsy (Dravet syndrome, Lennox-Gastaut syndrome). In the NEJM trial, the median reduction was 38.9%, and only 5% of patients became completely seizure-free. Most patients still had seizures — just fewer of them. CBD is an effective add-on therapy, not a cure. For more on CBD and epilepsy, see our detailed article on cannabis, epilepsy, and Epidiolex.
Does this mean dispensary CBD products work for seizures?
Not necessarily. The trial used pharmaceutical-grade CBD at high doses (20 mg/kg/day — that's 600 mg daily for a 30 kg child) with precise quality control. Dispensary and over-the-counter CBD products vary in purity, concentration, and formulation. Many contain far lower doses than what was used in the trial. Some contain THC or contaminants. If you or a family member has epilepsy, work with a neurologist — Epidiolex is available by prescription for qualifying diagnoses.
Does Epidiolex working without THC disprove the entourage effect?
It complicates it. Epidiolex is pure CBD — no THC, no terpenes, no minor cannabinoids — and it works. This proves that CBD can be therapeutically effective without any "entourage." However, it doesn't prove the entourage effect is wrong in all contexts. Some patients and some conditions might benefit from multi-compound preparations. What the Epidiolex data does show is that the strongest version of the entourage claim — that isolated cannabinoids never work as well as whole plant — is incorrect.
What the researchers found
This landmark open-label trial enrolled 214 patients (aged 1-30 years) with severe, treatment-resistant epilepsy across 11 US epilepsy centers. Patients received oral CBD at 2-5 mg/kg per day, titrated up to 25-50 mg/kg per day, added to their existing anti-epileptic drugs.
Among the 137 patients included in the efficacy analysis, the median monthly motor seizure frequency dropped from 30.0 at baseline to 15.8 over 12 weeks, representing a median reduction of 36.5%. Patients with Dravet syndrome (20%) and Lennox-Gastaut syndrome (19%) were included.
Adverse events occurred in 79% of patients (most commonly somnolence, decreased appetite, and diarrhea), but only 3% discontinued treatment due to adverse events. Serious adverse events occurred in 30% of patients. One death (sudden unexpected death in epilepsy) was considered unrelated to CBD.
Why it matters
This study was pivotal in building the evidence base that ultimately led to FDA approval of Epidiolex (pharmaceutical CBD) for certain epilepsy syndromes. It provided the first large-scale clinical data on CBD for epilepsy and demonstrated both efficacy and a manageable safety profile in a severely affected population.
The numbers in context
214 enrolled, 162 in safety analysis, 137 in efficacy analysis. Median baseline seizures: 30/month. Median at 12 weeks: 15.8/month. Median reduction: 36.5%. Adverse events: 79%. Serious adverse events: 30%. Discontinuation from AEs: 3%. Status epilepticus in 6%. One death (unrelated).
How the study worked
Open-label interventional trial across 11 US epilepsy centers as part of an expanded access program. Patients aged 1-30 with severe childhood-onset treatment-resistant epilepsy on stable anti-epileptic drug regimens received add-on CBD. The primary endpoint was median percentage change in motor seizure frequency at 12 weeks.
What this study cannot tell us
Open-label design without a placebo control means the placebo effect and regression to the mean cannot be excluded. Patients and clinicians knew they were receiving CBD, which could influence reporting. The patient population was heterogeneous. The 12-week timeframe is relatively short for a chronic condition.
How to read the evidence
The open-label trial (this entry) provides moderate evidence — large sample but no placebo control. The subsequent NEJM RCT (2017, covered in this pillar MDX) provides strong evidence — double-blind, placebo-controlled, in the world's most prestigious medical journal. Together, this program produced the highest level of clinical evidence for any cannabis-derived compound.
When this study was published
Open-label published 2016. NEJM RCT published 2017. FDA approval June 2018. Epidiolex is now established standard-of-care for Dravet and Lennox-Gastaut syndromes.
The bigger picture
This is the most consequential clinical trial program in cannabis research history. It produced the first cannabis-derived FDA-approved drug. It proved CBD works for epilepsy in the most rigorous trial design. It also showed pure CBD works without any entourage — complicating the 'whole plant is always better' claim. And it happened because families were desperate enough to move across state lines for an unproven treatment, forcing the science to catch up.
Questions still open
- Which other epilepsy types respond to CBD? Can lower doses be effective? Would whole-plant preparations work differently? Why does CBD work as an anticonvulsant — the mechanism is still not fully understood.
Common questions
Read the original research
Cannabidiol in patients with treatment-resistant epilepsy: an open-label interventional trial.
The Lancet. Neurology, 15(3), 270-8
Citation
Devinsky, Orrin; Marsh, Eric; Friedman, Daniel; Thiele, Elizabeth; Laux, Linda; Sullivan, Joseph; Miller, Ian; Flamini, Robert; Wilfong, Angus; Filloux, Francis; Wong, Matthew; Tilton, Nicole; Bruno, Patricia; Bluvstein, Judith; Hedlund, Julie; Kamens, Rebecca; Maclean, Jane; Nangia, Srishti; Singhal, Nilika Shah; Wilson, Carey A; Patel, Anup; Cilio, Maria Roberta. (2016). Cannabidiol in patients with treatment-resistant epilepsy: an open-label interventional trial.. The Lancet. Neurology, 15(3), 270-8. https://doi.org/10.1016/S1474-4422(15)00379-8
Explore the wider topic
- CBD Oil Quality Guide: How to Avoid Snake Oil
- Anxiety After Quitting Weed: When to Consider Medication
- Cannabis for Chemotherapy Nausea: What the Evidence Actually Shows
- Cannabis for Chronic Pain: What the Research Actually Supports
- Cannabis and Epilepsy: The Epidiolex Story and What It Means
- Does CBD Actually Work for Anxiety? What the Evidence Shows
- Does CBD Help with Weed Withdrawal? What Studies Show
- CBD vs THC: The Differences That Actually Matter
- The Proven Medical Benefits of Cannabis: What Research Supports
- Quitting Weed Before Surgery
- Weed and Medications: What Changes When You Quit
- Quitting Weed During Pregnancy: What You Need to Know
- Quitting Weed While Pregnant: What You Need to Know
- Cannabis and Older Adults: Risks Seniors Should Know
- Cannabis and Breastfeeding: THC in Breast Milk