Introduction
Attention-deficit/hyperactivity disorder (ADHD) is among the most common comorbidities in children with epilepsy. Approximately 11% [
1] of children in the general population are diagnosed with ADHD, compared with up to 50% of children with epilepsy [
2,
3]. ADHD is associated with cognitive impairment and behavioral difficulties that can limit children’s psychosocial and academic development [
4,
5]. Effective treatment of ADHD is essential for optimizing academic, social, and developmental outcomes [
6,
7].
Although withholding treatment is associated with poorer outcomes, concerns that stimulants may lower the seizure threshold and increase seizure frequency often complicate ADHD treatment. In a retrospective study of patients aged 6 to 25 years, Man et al. [
8] identified a seizure risk associated with stimulants, particularly immediately after methylphenidate initiation. This finding has contributed to clinician hesitancy and undertreatment of ADHD in children with epilepsy [
2,
8]. Clinical data have not consistently supported these concerns: several studies have shown that stimulants, including methylphenidate, may be used safely in people with epilepsy without consistent increases in seizure activity [
9-
11]. However, large pediatric studies remain limited [
11]. Our retrospective study of 107 pediatric patients, comprising 131 treatment observations, examined whether ADHD medications increase seizure frequency or burden in children with epilepsy, with the aim of clarifying their safety and informing treatment strategies.
Materials and Methods
This study was approved by the Institutional Review Board of Mayo Clinic, approval number 08-008137, with written informed consent waived due to retrospective nature of study. We used the Rochester Epidemiology Project (REP) database to retrospectively review seizure frequency in children with epilepsy and ADHD before and after exposure to ADHD medications. The REP is a collaborative database containing over 1.4 million patient records and has supported population-based studies since its establishment in 1966 [
12]. From a cohort of pediatric patients diagnosed with epilepsy between 2005 and 2020, we identified those subsequently diagnosed with comorbid ADHD, defined by the presence of a diagnostic code for ADHD or attention deficit disorder. The resulting study population comprised 107 patients, with each patient’s pretreatment status serving as their own control. All included patients had documented neurology follow-up within 6–12 months after ADHD treatment initiation. Patients were excluded if epilepsy or ADHD was diagnosed after age 18 years or if their seizures occurred only in the neonatal period. We also excluded patients who received ADHD medications for other indications, such as clonidine for weaning from sedation, mood stabilization, or treatment of sleep disorders. Patients whose clinical seizure descriptions or documentation did not meet the 2014 International League Against Epilepsy definition of epilepsy were excluded [
13].
We used the Children’s Health Council’s comprehensive list of available ADHD medications (
Table 1) [
14] to classify medications as stimulants or nonstimulants. Patients receiving only one class were assigned to the corresponding cohort; those receiving both classes of medicines simultaneously were assigned to the combination cohort. For patients who switched between stimulant and nonstimulant treatment without overlap of therapies, seizure burden was evaluated after each medication trial and compared to seizure burden prior to initiation of any ADHD treatment. Patients contributed to both cohorts only if the exposures occurred during distinct, nonoverlapping periods. To clarify the timing and minimize potential carryover effects, seizure burden during each treatment period was compared only with the patient’s baseline before any ADHD medication was initiated. Each observation therefore represented an independent within-patient comparison before and after treatment with one medication class, rather than a comparison between classes during the same interval.
During the retrospective chart review, we searched for ‘Epilepsy,’ ‘Seizure,’ and ‘Neurology’ to identify epilepsy histories and seizure burden. Seizure burden was defined as seizure incidence per unit of time, using intervals of 0–6, 6–12, 12–24, or more than 24 months before the clinic visit; the same measure was applied before and after treatment. Posttreatment seizure burden was recorded from any visit or communication immediately following ADHD medication initiation, within 6–12 months, and the start date was obtained from outpatient pediatric neurology or primary care records. This approach was used to detect clinically significant changes in patients who had infrequent seizures before ADHD treatment (every 6–12, 12–24, or >24 months). To capture patients with more frequent seizures, we also recorded the numbers of antiseizure medications that had failed and were being used at the last visit as proxies for seizure severity.
Seizure-frequency categories were assigned scores of 1, 2, 3, and 4 for intervals of >24, 12–24, 6–12, and 0–6 months, respectively. Paired differences between pretreatment and posttreatment seizure-frequency scores were calculated. These differences were treated as a continuous variable and compared between groups using the Kruskal–Wallis test in SAS version 9.4 (SAS Inc., Cary, NC, USA).
Results
1. Demographic data
We identified 107 patients with both epilepsy and ADHD who had received ADHD medication trials. This represented an ADHD incidence of 16% in the pediatric epilepsy population (107 of 685 patients). Initially, 72 patients were assigned to the stimulant cohort, eight to the nonstimulant cohort, and 27 to the combination cohort. Because 12 patients had received separate stimulant and nonstimulant trials, the total number examined was 131, with final stimulant and nonstimulant cohorts of 84 and 20, respectively. These 12 patients' seizure frequency during treatment was compared to their seizure burden prior to treatment of any kind. Overall, 68.7% of the children were male and 88.5% were non-Hispanic White, reflecting the higher incidence of ADHD in boys and the demographic composition of Olmsted County [
15] and the upper Midwest, where the study was conducted. Medication regimen did not differ significantly by sex (
P=0.50) or race (
P=0.56). Developmental delay was present in most children in the nonstimulant cohort (n=15, 75%), a significant difference (
P=0.02) from the stimulant and combination cohorts, in which most children did not have developmental delay. Intelligence quotient scores were inconsistently documented, with data available for only 68 patients; no significant between-group difference was observed. Epilepsy syndrome classification was also examined, but most patients had not been classified, and no significant between-group difference was found.
2. Stimulant treatment
In the stimulant cohort (n=84), the median ages at seizure onset and epilepsy diagnosis were 67 and 72 months, respectively (
Table 2). The median ages at ADHD diagnosis and treatment initiation were both 96 months (
P=0.01 and
P=0.02, respectively) (
Table 2). Most patients had focal/multifocal seizures (n=49, 58%). Generalized epilepsy (n=41, 49%) was more common than focal epilepsy with or without intact awareness (n=7, 35%) (
Table 3). At the last visit, 74 patients (88%) were receiving 0–1 antiseizure medications, and 19 (23%) had experienced failure of more than two antiseizure medications. Before stimulant treatment, 17 patients (20%) had experienced seizures within the preceding 0–6 months, compared with 14 (17%) after treatment. Two patients (2%) had experienced seizures within the preceding 6–12 months both before and after treatment. Seizures within the preceding 12–24 months were reported in 12 patients (14%) before treatment and 11 (13%) after treatment. Before treatment, 53 patients (63%) had last experienced a seizure more than 24 months earlier; after treatment, 57 (68%) had been seizure-free for more than 24 months.
3. Nonstimulant treatment
In the nonstimulant cohort (n=20), the median ages at seizure onset and epilepsy diagnosis were both 36 months (
Table 2). The median ages at ADHD diagnosis and treatment initiation were both 72 months (
P=0.01 and
P=0.02, respectively) (
Table 2). Most patients had focal/multifocal seizures (n=13, 65%). Generalized epilepsy (n=12, 60%) was more common than focal epilepsy with or without intact awareness (n=30, 36%) (
Table 3). At the last visit, 15 patients (75%) were receiving 0–1 antiseizure medications, and five (25%) had experienced failure of more than two antiseizure medications. Before treatment, nine patients (45%) had experienced seizures within the preceding 0–6 months, compared with five (25%) after treatment. Seizures within the preceding 6–12 months were reported in one patient (5%) before treatment and three (15%) after treatment. Seizures within the preceding 12–24 months were reported in two patients (10%) before treatment and five (25%) after treatment. Before treatment, eight patients (40%) had last experienced a seizure more than 24 months earlier; after treatment, seven (35%) had been seizure-free for more than 24 months.
4. Combination treatment
In the combination cohort (n=27), the median ages at seizure onset and epilepsy diagnosis were both 48 months (
Table 2). The median ages at ADHD diagnosis and treatment initiation were both 72 months (
P=0.01 and
P=0.02, respectively) (
Table 2). Approximately half the patients had focal/multifocal seizures (n=13, 48%), and approximately half had generalized seizures (n=14, 51%). Generalized epilepsy was present in 18 patients (67%), compared with combined focal and generalized epilepsy in six (22%) (
Table 3). At the last visit, 24 patients (89%) were receiving 0–1 antiseizure medications, and three (11%) had experienced failure of more than two antiseizure medications. Before treatment, 10 patients (37%) had experienced seizures within the preceding 0–6 months, compared with six (22%) after treatment. Two patients (7%) had experienced seizures within the preceding 6–12 months both before and after treatment. Seizures within the preceding 12–24 months were reported in three patients (11%) before treatment and one (4%) after treatment. Before treatment, 12 patients (44%) had last experienced a seizure more than 24 months earlier; after treatment, 18 (67%) had been seizure-free for more than 24 months.
5. All cohorts
Seizure burden did not differ significantly among the three cohorts before ADHD medication initiation (
P=0.18) (
Fig. 1). After treatment began, patients who received stimulant or combination therapy were more likely to have been seizure-free for more than 24 months (
P=0.05). No cohort showed a significant increase in seizure burden after ADHD medication initiation (
Fig. 1). We also calculated paired differences between pretreatment and posttreatment seizure-frequency scores and compared these differences as a continuous variable using the Kruskal–Wallis test (
P=0.15). The median difference was zero in each cohort, with ranges of (−1 to 2), (−3 to 3), and (0 to 3) in the nonstimulant, stimulant, and combination cohorts, respectively. These differences were also classified as worsened, unchanged, or improved seizure frequency and compared using the chi-square test. Overall, seizure burden improved in 16% (21 of 131), remained stable in 81.7% (107 of 131), and worsened in 2.3% (3 of 131) of patients. The distribution did not differ significantly among the three medication cohorts (
P=0.35).
Discussion
Three large retrospective population-based cohort studies previously examined the risks and outcomes of ADHD pharmacotherapy in children with epilepsy. In 2018, Liu et al. [
16] used billing records from 26 United States collected between 1999 and 2010 to compare seizure-related hospitalizations over 12 months in 18,166 children with epilepsy who received any ADHD medication and 54,197 who did not. They found no significant increase or difference in hospitalization incidence for epilepsy-related concerns among patients receiving ADHD therapy. However, they did not specifically compare hospitalization rates between stimulant and nonstimulant treatment groups. Using hospitalization as a proxy for seizure frequency may also have missed clinically significant changes in patients who were not admitted. In another large retrospective population-based study, Wiggs et al. [
17] tracked monthly seizure incidence in 801,831 children and adults with ADHD. Participants could have epilepsy or no epilepsy, and seizure incidence was not compared among stimulant, nonstimulant, and combination therapies. Patients were less likely to experience seizures while receiving ADHD treatment. Finally, Brikell et al. [
9] followed 995 children with epilepsy who received ADHD medications, using unplanned hospitalizations or outpatient neurology appointments as proxies for seizure frequency. Seizure frequency did not differ during the 24 months before and after ADHD medication initiation. As in the other studies, however, seizure frequency was not compared among stimulant, nonstimulant, and combination treatment groups.
Although other studies have examined breakthrough seizures during ADHD treatment, our study uniquely focused on children with both epilepsy and ADHD. Our measure of seizure burden, defined as seizure incidence per unit of time, is more sensitive to clinical changes in seizure control than hospitalization for breakthrough seizures, because many children with seizures are not hospitalized. We analyzed changes in seizure burden with methylphenidate and other stimulants, including amphetamine, and compared these changes with those in patients receiving nonstimulant medications or combined stimulant and nonstimulant therapy. These features, particularly the inclusion of nonstimulant medications, support the applicability of our findings across clinical settings that treat children with epilepsy and ADHD.
Although several features support this study’s clinical applicability, it has important limitations. The cohort comprised 131 patients, with smaller nonstimulant and combination subgroups (n=20 and n=27, respectively), compared with the thousands of participants in large multicenter studies [
9,
16,
17]. The sample size limited more complex multivariable modeling and may reduce the statistical power of subgroup analyses.
The retrospective design restricted the available data to information documented in the electronic medical record (EMR). Documentation may not fully represent the patient’s clinical course, so seizure frequency, treatment response, and medication effects may not have been captured accurately. To address this limitation, we used multiple sources within the EMR, including clinic notes, in-basket communications, and outside records, to validate key data elements.
Treatment was selected according to clinician judgment rather than randomized assignment, creating potential confounding by indication. The choice of a stimulant or nonstimulant medication may have been influenced by epilepsy severity, perceived seizure risk, comorbid behavioral conditions, and family or clinician preferences.
Twelve patients also contributed data to both the stimulant and nonstimulant cohorts at different time points. Although these exposures occurred in distinct, nonoverlapping periods and were analyzed as separate pretreatment–posttreatment intervals, observations may not have been fully independent in between-group comparisons.
The definition of seizure burden was another limitation: it was measured as the interval since the last seizure. Although this measure does not capture every dimension of seizure burden, it was selected to permit a detailed assessment of the early period after medication initiation and to detect clinically significant seizure provocation that might otherwise be obscured.
The predominantly male and White study population reflected the demographics of Olmsted County [
15], which may limit generalizability to more diverse populations. This distribution was broadly consistent with United States epidemiological trends in ADHD diagnosis: boys are diagnosed more often than girls (13.3% vs. 5.6%), and reported diagnosis rates are higher among individuals of European descent than among those of African descent (11.5% vs. 8.9%) [
5].
This cohort study contributes to understanding the effects of ADHD medications in children with epilepsy by showing no indication of worsening preexisting seizure burden and providing further evidence for the safety of stimulants in pediatric patients. Its clinical relevance also extends to nonstimulant and combination ADHD therapies, which were not associated with increased seizure burden in children with epilepsy.
In conclusion, we retrospectively reviewed 107 children with epilepsy and ADHD to determine whether seizure burden changed significantly after initiation of stimulant, nonstimulant, or combination ADHD medication. No medication class was associated with increased seizure burden in this cohort. The study adds evidence on ADHD pharmacotherapy in children with epilepsy. Its focus on children with both diagnoses and on changes in seizure burden after treatment with each medication class provides a more sensitive and clinically relevant assessment than earlier large studies.
The results suggest that neither stimulant nor nonstimulant ADHD medications are associated with increased seizure burden, supporting treatment of ADHD symptoms in this vulnerable population. Although the retrospective design and demographic composition may limit generalizability, the detailed medical record review and inclusion of multiple medication classes strengthen the study’s applicability to clinical practice. These findings further support the view that optimizing ADHD treatment in children with epilepsy is unlikely to exacerbate seizure activity and that medication for ADHD should be considered.