“…Data were summarized using frequency (percent) for categorical variables and median (interquartile range [IQR]) for continuous variables. While the number of days with seizures and side effects in the last 28 days were collected at a more granular level (1 day, 2–3 days, 4–14 days, 15–27 days, all 28 days, uncertain), aggregated categories of no days, some days ( 1 , 2 , 3 , 4 , 5 , 6 , 7 , 8 , 9 , 10 , 11 , 12 , 13 , 14 , 15 , 16 , 17 , 18 , 19 , 20 , 21 , 22 , 23 , 24 , 25 , 26 , 27 ), all 28 days, or uncertain were used for modeling because of small numbers in several categories. A composite variable was created using information about vagus nerve stimulation (VNS) and epilepsy surgery data, grouping participants into having had no surgery for epilepsy, VNS, or intracranial epilepsy surgeries (‘non‐VNS surgery’).…”