Clinical Research

Why a Fall Counted in a Cannabidiol Epilepsy Trial

·

A caregiver and clinician reviewing a seizure-event diary.

A person can have a drop seizure without actually hitting the floor. A caregiver might catch them, or the seizure might occur while they are already sitting down. That seemingly small distinction mattered in a major trial of cannabidiol for Lennox–Gastaut syndrome. If only falls to the ground had counted, the same seizure could have disappeared from the record simply because someone was close enough to help.

Lennox–Gastaut syndrome is a severe epilepsy with multiple seizure types. Among the most immediately dangerous are attacks that make a person suddenly lose postural control. A rapid stiffening of muscles can do it, and so can a sudden loss of muscle tone. What links those events for a family is the possibility of an unprotected fall. What linked them for investigators was a definition that could be applied across households and clinics.

What went into the diary

In the randomized trial led by Orrin Devinsky, a drop seizure involved the body, trunk or head and caused, or could have caused, a fall, injury, slumping or a blow to the head. An ordinary stumble had no place in the count. Conversely, a caregiver could record a qualifying seizure that ended with the patient safely supported in a chair. The researchers were counting a defined clinical event with the potential for injury. Confirmed injuries would have required their own tally.

Before treatment assignment, families recorded seizures over a four-week baseline. The trial enrolled patients who had experienced at least two drop seizures per week during that period. The researchers then assigned 225 participants to one of two doses of purified cannabidiol oral solution or to matching placebo for fourteen weeks. Participants continued their existing antiseizure treatment. The trial registration preserves the distinction between the main outcome and the other outcomes the team planned to examine.

How to read a percentage reduction

Researchers compared each patient’s drop seizure frequency during treatment with that patient’s baseline. The median reduction was about 42 percent in the higher-dose group, 37 percent in the lower-dose group and 17 percent with placebo. Both cannabidiol groups differed significantly from placebo on the primary outcome. These percentages describe the middle of the distribution of individual changes. They are easy to mistake for the percentage of patients who became seizure-free, a different and much more demanding result.

Consider two imaginary diary entries. A fall from forty qualifying seizures to twenty in a fixed period is a 50 percent reduction. So is a fall from eight to four. The percentages capture proportional change, while the families have experienced very different numbers of events. The arithmetic also explains why the placebo group matters. Seizure frequency can vary across weeks, and comparing a person’s treatment record only with their own baseline offers no way to see how much change might have occurred in the same period under placebo.

What disappears into a total

Now suppose someone’s drop seizures decline while a different, more frequent seizure type remains unchanged. A grand total will give the more frequent type greater weight. That total may be worth reporting, but it can obscure a change the family would notice each time the patient stands up. Counting only drop seizures has its own cost. It leaves questions about other seizures, injuries, alertness and daily activities to other measurements. Sleepiness, decreased appetite and diarrhea were among the adverse events reported in this trial, information no seizure diary total can convey on its own.

Even a closely related study chose a different count. In the cannabidiol trial for Dravet syndrome, the primary outcome concerned convulsive seizures. That choice reflected a different epilepsy and a different clinical question. For the Lennox–Gastaut families, a qualifying entry might end with a child caught before a fall. Keeping that entry in the record was one reason the trial could measure the event it set out to study.