Pilot Executive Functioning Intervention in Epilepsy: Behavioral and Quality of Life Outcomes

Abstract

Objective

To examine changes in emotional and behavioral functioning and health-related quality of life (HRQOL) following a web-based executive functioning (EF) intervention open pilot trial (e.g., Epilepsy Journey) for adolescents with epilepsy.

Methods

Adolescents with an established diagnosis of epilepsy, EF deficits, and without developmental disorders participated in a single-arm trial of Epilepsy Journey. Epilepsy Journey is a gamified, online learning environment comprised of 10 learning modules targeting EF deficits (e.g., working memory, organization) and tailored to epilepsy with accompanying telehealth problem-solving sessions. Adolescents completed questionnaires assessing emotional and behavioral functioning and HRQOL at baseline, posttreatment, and 2 follow-ups . Longitudinal mixed models and logistic regression analyses were used for these secondary analyses.

Results

39 adolescents were recruited for Epilepsy Journey (M_age=15.3 years; 67% female; 87% White: Non-Hispanic; 39% experienced seizures in the past 3 months). Preliminary data indicate significant improvements in caregiver-reported Externalizing symptoms, Behavioral Symptom Index scores and Adaptive Skills from baseline to 5-month follow-up. Significant improvements were observed for caregiver-reported Mood/Behavior and self-reported Impact, Cognitive Functioning, Executive Functioning, and Sleep subscales of the PedsQL Epilepsy Module. Clinically significant improvements (e.g., clinical/at-risk to normative levels) in behavioral and quality of life domains were also noted.

Conclusion

Epilepsy Journey appears to contribute to changes in emotional and behavioral functioning and HRQOL in adolescents with epilepsy. Given the proof of concept trial format of this study, an important future direction is to conduct a randomized controlled trial with a larger, generalizable cohort of adolescents with epilepsy.

Epilepsy is a common neurological disorder (Institute of Medicine, 2012), affecting more than 470,000 children and adolescents in the United States (Zack & Kobau, 2017), and contributes to wide ranging neurocognitive and behavioral impairments. Specifically, adolescents with epilepsy experience elevated rates of internalizing and externalizing symptoms (Davies et al., 2007; Ekinci et al., 2009; Hoie et al., 2008; Rutter et al., 1970), as well as social and academic difficulties (Austin et al., 1999; Fastenau et al., 2008; Tse et al., 2007). Executive function (EF) deficits, problems with working memory, initiation, organization, self-regulation, planning, and problem-solving (Lezak, 2004) are particularly common in pediatric epilepsy (Modi et al., 2018) and may underlie difficulties with academics (Fastenau et al., 2004), problem behaviors (Baum et al., 2010; Hoie et al., 2008; Modi et al., 2019a), and quality of life (Love et al., 2016; Modi et al., 2019a; Schraegle & Titus, 2016; Sherman et al., 2006). Although uncontrolled seizures exacerbate impairments in EF and behavior (Kavanaugh et al., 2015), children with well controlled seizures also exhibit poorer behavioral outcomes compared with healthy controls (Austin et al., 2011).

Problem-solving interventions address a broader spectrum of deficits and have proven efficacy in improving EF and neurobehavioral comorbidities in children with other neurological conditions (e.g., traumatic brain injury; Kurowski et al., 2013, 2014; Wade et al., 2010, 2011, 2013). Evidence-based interventions to improve EF and behavior in adolescents with epilepsy could play a critical role in promoting optimal functioning; however, none currently exist. To address this gap, we developed and tested Epilepsy Journey, an online intervention integrating e-health modules with telehealth sessions with a trained therapist (Modi et al., 2019b). Epilepsy Journey was developed based on existing online problem-solving interventions (i.e., Teen Online Problem-Solving Intervention) for pediatric traumatic brain injury (Wade et al., 2010, 2012, 2017). The Epilepsy Journey program provides training in skills to address challenges in different aspects of EF (e.g., working memory) within a problem-solving framework (Nezu, 2004; Robin & Foster, 1989) over a 12 weeks period. In addition to teaching skills, content regarding the impact of seizures and treatment side effects on EF behaviors was integrated throughout the program. Using the ORBIT model of behavioral intervention development (Czajkowski et al., 2015), we conducted Phase 1 design studies to define the basic elements of the intervention through focus groups (Phase 1a) and refined the intervention through usability studies and heuristic evaluation (Phase 1b) (Glaser et al., 2017; Modi et al., 2017b). We then conducted a Phase 2a proof of concept trial in adolescents with epilepsy (ages 13–17) and documented the feasibility and acceptability of the program, as well as preliminary changes in EF behaviors (Modi et al., 2019b). Proof of concept trials are beneficial in determining whether a behavioral treatment warrants more rigorous and costly testing using a randomized design (Czajkowski et al., 2015). Consistent with primary outcome hypotheses, the study revealed significant improvements in both self- and parent-reported EF behaviors (Modi et al., 2019b) as measured by the Behavioral Rating Inventory of Executive Function (BRIEF; Gioia et al., 2000; Guy et al., 2004), with parents reporting improvements across all BRIEF subscales. Rates of at-risk/clinically elevated behaviors associated with executive dysfunction also declined along several dimensions, suggesting that improvements were both statistically significant and clinically meaningful (Modi et al., 2019b). Notably, preliminary published data (Modi et al., 2019b), as well as the current study results are preliminary in nature given the lack of a control group for the proof of concept trial.

Although EF behaviors were the primary target of the proof of concept trial, we hypothesized that the intervention content, which targeted cognitive reframing, problem solving, and management of stress, fatigue, and sleep difficulties, might more broadly impact behavioral and emotional functioning and health-related quality of life (HRQOL). Research suggests children with epilepsy with neurobehavioral comorbidities have significantly worse baseline EF and poorer cognitive trajectories over time (Hermann et al., 2008). EF deficits are a strong predictor of 3-year neurobehavioral comorbidities in children with new-onset seizures, especially externalizing problems (e.g., hyperactivity; Baum et al., 2010). Further, youth with epilepsy who demonstrated EF deficits were twice as likely to have poor QOL (Sherman et al., 2006).

Thus, this article reports analyses examining changes on secondary outcome measures, which includes both patient and caregiver-reported emotional and behavioral functioning and HRQOL, following the Epilepsy Journey open pilot trial. We hypothesized that adolescents and caregivers would report improved emotional and behavioral functioning and better HRQOL from pre to posttreatment, as well as through the 2 and 5-month follow-up periods, after controlling for key variables associated with outcomes (e.g., child race, child age, years since diagnosis, seizure type, presence of seizures in the last 3 months, and total side effects).

Materials and Methods

Participants

Participants were recruited from the Comprehensive Epilepsy Center in a large Midwestern children’s hospital. Inclusion/exclusion criteria for the open trial included adolescents: (a) having an epilepsy diagnosis, (b) being between 13–17 years of age, (c) residing within 100 miles of one of our hospital locations, d) being English-speaking, (e) having no known medical comorbidities requiring daily medications (with the exception of asthma/allergies and attention deficit hyperactivity disorder), (f) having no known significant developmental disorders (e.g., autism), and (g) having no plan to wean from epilepsy treatment in the following 6 months. In addition, to qualify for the trial, the adolescent was required to have EF deficits based on the caregiver-reported BRIEF, which was defined as ≥ 1 clinically elevated BRIEF subscale (T-score ≥65) or ≥ 2 sub-clinically elevated BRIEF subscales (e.g., ≥ 1 standard deviation; T-scores ≥ 60 < 65) based on normative data for the measure (Gioia et al., 2000; Guy et al., 2004).

Procedures

Trained research coordinators identified eligible patients via medical chart review and approached them during routine epilepsy clinic visits. Written caregiver consent and adolescent assent were obtained for all participants. To ensure adolescents met the screening criteria, caregivers completed the BRIEF to assess for deficits in EF based on the criteria noted above. If adolescents had no parent-rated EF deficits, they were not enrolled in the study. If they screened positive for EF deficits, they proceeded with the study. After determining eligibility and providing informed consent, caregivers and adolescent participants completed a battery of questionnaires. Measures were completed at baseline, posttreatment (e.g., after completion of the 12 weeks of intervention), 2- and 5-month follow-up. Families were compensated for their time and effort using reloadable gift cards. Specifically, caregivers received $20 for completion of four assessment batteries (i.e., $80 total). Adolescents were compensated using a graduated incentive schedule: $20 for Assessment 1, $30 for session 10/posttreatment assessment, $35 for 2-month follow-up assessment, and $40 for 5-month follow-up assessment. In addition, they received $20 for telehealth session 1, $10 each for sessions 2–4, $15 for session 5–7, and $20 for sessions 8 and 9 (i.e., up to $260 total). This study was approved by the hospital’s Institutional Review Board.

Epilepsy Journey Intervention

Epilepsy Journey (NCT02925663) is a 10-session, web-based problem-solving intervention with psychoeducational modules focused on memory, initiation, inhibition, and emotional control (Glaser et al., 2017; Modi et al., 2017b). Sessions also focused on sleep and stress regulation, which are known drivers of behavioral regulation. Topics and examples throughout the modules are epilepsy-specific (e.g., side effects, seizures). Adolescents independently navigated a web-based “journey” (e.g., learning environment) through 10 different “lands” (e.g., learning modules) which occurred over the course of 12 weeks. The modules had interactive components (e.g., games), videos, case scenarios, and the ability to input answers to questions. After completion of each module, the adolescent had a telehealth session with a therapist to review the module materials from the week and implement a five-step problem-solving process to develop plans for addressing the adolescent’s self-identified struggles in each domain of executive functioning. Adolescents spent ∼9 min per session reviewing online content and 30 min per session in telehealth sessions (Modi et al., 2019b).

The first session, known as the “Peninsula of Positive Thoughts,” occurred at the adolescent’s home with the therapist both to orient the adolescent to the intervention and to ensure all technology components were working properly. The second session for all participants was “Problem Solving Peak,” which focused on teaching problem-solving skills, the fundamental skill applied to all subsequent sessions. Subsequent modules (Island of Initiation, Mists of Memory, Orchards of Organization, Falls of Inhibition, Land of Nod, State of Emotional Control, and Mesa of Monitoring) were ordered based on the highest level of EF deficits indicated by the BRIEF subscale scores, as well as adolescent preference. Modules were reviewed by the adolescent independently each week and then adolescents had an accompanying telehealth session. If the adolescent had not completed the self-guided module prior to the telehealth session, the therapist rescheduled to ensure that content had been reviewed prior to their meeting. More detailed study procedures and intervention information have been previously published (Modi et al., 2019b).

Two interventionists were trained to conduct the intervention and used a treatment manual to conduct the telehealth intervention sessions. The interventionists were master’s- and doctoral-level psychology trainees and postdoctoral fellows. Weekly supervision was provided to the interventionists by the first author. All sessions were audio-recorded, and treatment fidelity was conducted on 9% of the audio tapes by the first and senior authors, with 30% of these being coded by both authors (intraclass correlation r = .70). Treatment fidelity was 96% for the interventionists.

Measures

This study focused on secondary outcome measures from the open trial of Epilepsy Journey. Findings regarding additional measures, including the BRIEF and measures of feasibility, acceptability, and satisfaction have been published (Modi et al., 2019b).

Demographics and Medical History

A background information form was completed by caregivers to obtain basic information including child age, sex, date of epilepsy diagnosis, family income, and history of comorbid illnesses. In addition, a medical chart review was conducted to assess health insurance status, seizure type and etiology, the treatment regimen, and seizure frequency. Seizure frequency was dichotomized into the absence (coded = 0) or presence (coded = 1) of seizures in the past 3 months given the heterogeneity of seizure types.

Behavior Assessment Schedule for Children-2 Parent Report and Self-Report

The Behavior Assessment Schedule for Children-2 (BASC-2) is a reliable and valid measure of emotional and behavioral difficulties, as well as adaptive functioning, in children and adolescents (Reynolds & Kamphaus, 2004). Individual raw scores were compared with normative data for children of the same age to create standardized T-scores. Scores can be used to identify children in normative (T-scores < 60), at-risk (T-scores 60–69), or clinically elevated (T-scores ≥ 70) groups. The parent report yields four composite scales, including Internalizing Problems, Externalizing Problems, a Behavioral Symptom Index, and Adaptive Skills score. The adolescent self-report yields five composite scores, including School Problems, Internalizing Problems, Emotional Symptoms Index, Inattention/Hyperactivity, and Personal Adjustment.

PedsQL™ Epilepsy Module

The PedsQL Epilepsy Module is reliable epilepsy-specific HRQOL measure for youth 2–25 years with epilepsy (Follansbee-Junger et al., 2016; Modi, et al., 2017a). Both the parent-proxy and self-report versions were administered to participants, assessing five key domains, including Impact, Cognitive Functioning, Executive Functioning, Sleep, and Mood/Behavior. Internal consistency coefficients for these scales range from 0.70 to 0.94. Scores range from 0 to 100, with higher scores representing better HRQOL. Clinical cut-offs have been established for the measure, with scores below the cut-off representing concerning HRQOL: Impact (Parent = 60.7; Child = 64.39), Cognitive (Parent = 38.11; Child = 50.97), Executive Functioning (Parent = 46.65; Child = 57.15), Sleep (Parent = 42.07; Child = 43.90), and Mood/Behavior (Parent = 54.14; Child =53.30; Junger et al., 2019). Minimally clinically important difference (MCID) scores, which assess the smallest clinically important change perceived by a caregiver or child were as follows: Impact (Parent = 8.78; Child = 8.44), Cognitive (Parent = 7.59; Child = 10.10), Executive Functioning (Parent = 8.43; Child = 10.16), Sleep (Parent = 12.61; Child = 14.68), and Mood/Behavior (Parent = 8.04; Child =10; Modi et al., 2017a).

Statistical Analyses

Means, standard deviations, and frequencies were used to examine demographic and medical variables. Bivariate correlations were conducted to test the associations among our secondary outcome measures and BRIEF subscales. To examine whether there were statistically significant improvements on the outcome measures, longitudinal mixed models (intent to treat) were conducted in Stata version 15 (StataCorp, 2017) using robust maximum likelihood estimation for missing data. All available data for each participant were used, resulting in N = 36 for adolescents and N = 37 for caregiver analyses. Each of the measures were modeled as an outcome with time as the predictor variable and adolescent race, years since diagnosis, seizure type, presence of seizures in the last 3 months, total side effects, and child age (PedsQL analyses only) as covariates based on a priori literature suggesting a relationship between these covariates and outcomes of interest. Alpha was set to .05 for all analyses and no correction was made for multiple comparisons due to the preliminary nature of the study. Each caregiver- and adolescent-reported outcome was tested for the presence of a significant non-linear trajectory (i.e., quadratic function of time) and this non-linearity was modeled if applicable. Finally, to examine changes in clinical cut-off categories on each of our outcome measures (e.g., the odds of a participant transitioning from an at-risk/clinically elevated to a normative score), longitudinal generalized linear mixed effect logistic models were employed. At each time point, participants were coded as “at risk/elevated” or “normative” on each outcome. We included both the at-risk and elevated categories together given our inclusion criteria included participants with at ≥two subclinical/at-risk BRIEF subscales. The odds of transitioning to a normative category over time were modeled for each of the outcomes, with time as the predictor variable and race, years since diagnosis, seizure type, presence of seizures in the last 3 months, and total side effects as covariates. Post hoc descriptive analyses were conducted for those who were considered clinically elevated at baseline but transitioned into the normative range at subsequent time points. Power analyses for the primary outcome indicated a sample size of N = 30 was sufficient for a single arm within subjects trial with an anticipated effect size of >0.50.

Results

Participants

Forty-seven of 69 eligible patients agreed to participate (69% recruitment) between October 2016 to January 2018. Final 5-month follow-up data collection was completed by September 2018. Eight patients screen failed based on the need for elevated parent-BRIEF subscale scores, yielding a total of 39 participants. Of the 39 participants who consented to the study, three withdrew prior to intervention and five received varying doses of the intervention, which includes both review of the module and telehealth session (one session-1 participant, two sessions-2 participants, four sessions-1 participant, and six sessions-1 participant). This resulted in a final sample of 31 participants who completed the study (retention = 79%). Due to the small sample, comparisons were not conducted between those who withdrew and those who completed. Participant characteristics of those consented are shown in Table I and a CONSORT diagram is also presented in Figure 1. No serious adverse events occurred in the trial.

Table I.

Demographic data (N = 39)

	N (%) or M + SD
Age	15.3 + 1.3 years
Sex
Females	26 (67%)
Males	13 (33%)
Race/Ethnicity
White: Non-Hispanic	34 (87%)
Black: Non-Hispanic	1 (3%)
Bi/multiracial or other	4 (10%)
Health insurance
Private	28 (72%)
Medicaid/Public	8 (21%)
Uninsured	3 (8%)
Family incomea
<$10,000	2 (5%)
$10,001–$30,000	6 (15%)
$30,001–50,000	5 (13%)
$50,001–$75,000	9 (23%)
>$75,001	16 (41%)
Seizures in the past 3 months
Yes	15 (38.5%)
No	24 (61.5%)
Seizure type and etiology
Focal	7 (18%)
Generalized	21 (54%)
Unclassified	11 (28%)
Specific Epilepsy Syndromes
None	26 (67%)
Juvenile myoclonic	8 (21%)
Childhood absence	3 (8%)
Benign rolandic	1 (3%)
Other	1 (3%)
Antiseizure medications
Monotherapy	32 (82%)
Polytherapy	7 (18%)
Years since diagnosis	1.9 + 2.0 years

^aData missing for one participant.

Figure 1.

CONSORT Diagram of Epilepsy Journey.

Bivariate correlations between BRIEF subscales and our variables of interest are presented in Table II. It is notable that caregiver-reported BRIEF subscales were highly correlated with caregiver BASC-2 subscales, but not caregiver-reported PedsQL Epilepsy Module subscales. Adolescent-reported BRIEF subscales were highly correlated with both adolescent-reported BASC and PedsQL Epilepsy Module subscales. Validity indices of the BASC-2 were also examined, with a majority of participants demonstrating “acceptable” scores, with the exception of one adolescent who demonstrated “extreme caution” on the L Index for all his/her time points. This scale represents a tendency to give extremely positive pictures of him/herself. We chose to keep this individual in the analyses as it is unlikely to influence findings.

Table II.

Baseline bivariate correlations between the Behavioral Rating Inventory of Executive Function subscales and Behavior Assessment Schedule for Children and PedsQL Epilepsy Module Subscales

Outcome	Caregiver BRIEF			Adolescent BRIEF
	GEC	BRI	MCI	GEC	BRI	MCI	Beh Shift	Cog Shift
Caregiver Behavior Assessment System for Children-2 T-scores
Externalizing symptoms	0.65**	0.56**	0.59**	0.39*	−0.41*	0.29	0.28	0.22
Internalizing symptoms	0.54**	0.58**	0.42*	0.35*	0.43**	0.22	0.25	0.16
Behavioral Symptom Index	0.85**	0.78**	0.74**	0.40*	0.41*	0.31	0.33	0.20
Adaptive skills	−0.60**	−0.51**	−0.55**	−0.21	−0.15	−0.22	−0.25	−0.21
Caregiver PedsQL-Epilepsy Module
Impact	0.006	−0.10	0.04	−0.29	−0.32	−0.22	−0.27	−0.24
Cognitive functioning	0.002	−0.08	0.03	−0.29	−0.28	−0.25	−0.25	−0.23
Executive functioning	−0.02	−0.11	0.01	−0.32	−0.33*	−0.25	−0.28	−0.25
Sleep	0.002	−0.09	0.03	−0.31	−0.32	−0.26	−0.25	−0.23
Mood/Behavior	−0.01	−0.11	0.03	−0.31	−0.35*	−0.23	−0.30	−0.25
Adolescent Behavior Assessment System for Children-2 T-scores
School Problems	0.49**	0.44**	0.44**	0.63**	0.60**	0.54**	0.28	0.42*
Internalizing problems	0.50**	0.51**	0.43**	0.64**	0.64**	0.52**	0.52**	0.52**
Emotional Symptoms Index	0.45**	0.38*	0.44**	0.65**	0.56**	0.62**	0.48**	0.56**
Inattention/ hyperactivity	0.53**	0.48**	0.46**	0.79**	0.81**	0.60**	0.66**	0.69**
Personal adjustment	−0.29	−0.18	−0.35*	−0.55**	−0.43**	−0.56**	−0.35*	−0.49**
Adolescent PedsQL-Epilepsy Module
Impact	−0.24	−0.24	−0.19	−0.30	−0.37*	−0.18	−0.43**	−0.26
Cognitive functioning	−0.25	−0.22	−0.22	−0.67**	−0.49**	−0.72**	−0.39*	−0.51**
Executive functioning	−0.27	−0.25	−0.24	−0.72**	−0.69**	−0.60**	−0.59**	−0.53**
Sleep	−0.27	−0.25	−0.23	−0.53**	−0.48**	−0.48**	−0.37*	−0.41*
Mood/behavior	−0.25	−0.26	−0.20	−0.65**	−0.72**	−0.46**	−0.65**	−0.49**

Note. GEC = Global Executive Composite; BRI = Behavioral Regulation Index; MCI = Metacognition Index; Beh shift = behavioral shift; Cog shift = cognitive shift. *p<0.05, **p<0.01.

Changes in Emotional/Behavioral Functioning and HRQOL and over Time

Emotional and Behavioral Functioning

We observed statistically significant decreases in caregiver-reported Externalizing Symptoms (b = −6.13, p = .01) and Behavioral Symptom Index scores (b = −6.23, p = .005). Additionally, statistically significant improvements were noted on caregiver-reported Adaptive skills (b = 4.65, p = .007). Both Externalizing Symptoms and Adaptive Skills scores had significant quadratic effects, indicating slight rebounding of deficits at later time points (see Table III). No statistically significant improvements were noted on any of the adolescent-reported BASC subscale scores.

Table III.

Longitudinal Mixed Model Results for Health-Related Quality of Life and Emotional and Behavioral Functioning Scores

Outcome	Baseline (N = 36 caregivers; N = 38 adolescents)	Post (N = 32 caregivers; N = 33 adolescents)	Fup1 (N = 24 caregivers; N = 25 adolescents)	Fup2 (N = 29 caregivers; N = 24 adolescents)	b a ,b	Standard error	p-value	95% CI
Caregiver: Behavior Assessment System for Children-2 T-scores
Externalizing symptoms	52.0 (10.1)	48.47 (7.6)	46.8 (7.0)	48.8 (9.7)	−6.13	2.42	.01	−10.86, −1.39
					1.05	0.49	.03	0.10, 2.01
Internalizing symptoms	57.5 (13.2)	55.8 (12.0)	52.2 (11.1)	55.1 (10.4)	−0.53	0.51	.30	−1.52, 0.46
Behavioral Symptom Index	57.7 (10.5)	54.4 (8.0)	50.7 (6.5)	53.9 (8.0)	−6.23	2.23	.005	−10.60, −1.86
					0.98	0.43	.02	0.15, 1.81
Adaptive skills	40.6 (8.4)	43.5 (8.7)	45.5 (9.3)	43.7 (9.1)	4.65	1.71	.007	1.29, 8.00
					−0.79	0.34	.019	−1.45, −0.13
Caregiver: PedsQL-Epilepsy Module
Impact	76.4 (17.3)	78.8 (16.8)	83.9 (13.7)	80.9 (17.1)	1.55	0.95	.10	−0.31, 3.41
Cognitive functioning	58.5 (25.1)	60.3 (27.3)	67.0 (23.9)	60.3 (25.5)	2.21	1.26	.78	−0.25, 4.67
Executive functioning	60.2 (21.5)	65.6 (20.0)	71.4 (18.4)	66.8 (25.2)	2.33	1.31	.08	−0.23, 4.89
Sleep	49.3 (28.6)	52.6 (26.6)	60.1 (22.5)	51.4 (25.6)	1.25	1.43	.38	−1.54, 4.04
Mood/behavior	55.3 (20.2)	57.7 (18.4)	67.3 (19.1)	57.8 (18.3)	9.76	4.67	.04	0.62, 18.91
					−1.85	0.95	.05	−3.71, 0.007
Adolescent: Behavior Assessment System for Children-2
Mood/behavior	58.6 (26.2)	57.9 (27.6)	70.8 (23.2)	64.4 (29.5)	1.34	1.24	.28	−1.09, 3.77
School problems	46.2 (11.2)	46.6 (10.5)	44.2 (9.8)	44.8 (8.9)	−0.39	0.49	.43	−1.36, 0.58
Internalizing problems	52.7 (13.2)	51.5 (12.9)	46.7 (10.9)	48.4 (11.8)	−0.74	0.57	.20	−1.86, 0.39
Emotional Symptoms Index	52.6 (13.0)	51.5 (13.5)	46.8 (9.4)	48.3 (11.1)	−0.76	0.68	.26	−2.09, 0.57
Inattention/hyperactivity	51.6 (12.5)	50.6 (12.1)	48.6 (13.7)	48.0 (11.4)	−1.01	0.58	.08	−2.15, 0.13
Personal adjustment	49.5 (10.3)	49.8 (12.1)	53.6 (8.6)	51.2 (10.6)	0.32	0.63	.61	−0.92, 1.56
Adolescent: PedsQL-Epilepsy Module
Impact	75.8 (19.7)	82.3 (17.4)	85.9 (13.7)	81.6 (20.4)	10.34	3.44	.003	3.59, 17.08
					−1.94	0.71	.006	−3.33, −0.55
Cognitive functioning	64.8 (21.6)	66.9 (22.1)	72.0 (22.5)	74.8 (22.2)	4.23	1.34	.002	1.60, 6.86
Executive functioning	65.8 (22.2)	69.8 (21.7)	77.9 (19.2)	76.9 (24.0)	4.28	1.30	.001	1.72, 6.83
Sleep	55.6 (30.9)	60.4 (27.9)	71.7 (22.9)	56.9 (28.4)	15.85	6.64	.02	2.82, 28.87
					−3.01	1.22	.01	−5.42, −0.61
Mood/Behavior	58.6 (26.2)	57.9 (27.6)	70.8 (23.2)	64.4 (29.5)	1.72	1.34	.20	−0.91, 4.35

Note. Covariates for all models included years since diagnosis, seizure type, presence of seizures in the last 3 months, and total side effects; child age was also included in the PedsQL Epilepsy Module models; Bolded outcomes represent significant changes over time. Sample sizes are noted for the timepoint; however, there was missing data for some individuals on select subscales.

^aCoefficient for time-linear model.

^bQuadratic model (when significant).

Caregivers were three times more likely to report scores in the normative range on the caregiver-reported Behavioral Symptom Index scores over time (e.g., baseline to follow-up) than clinically elevated scores (OR = 3.27, p = .003; see Table IV). Of those who were clinically elevated at baseline on caregiver-reported Behavioral Symptom Index (N = 13), 30.7% transitioned to scores in the normative range at subsequent time points, with an average T-score decrease of 12.5 points from baseline to posttreatment. Likewise, adolescents were over 2.5 times more likely to report normative scores over time on the adolescent-reported Inattention/Hyperactivity scores (OR = 2.6, p = .004) than clinically elevated scores. Of those who were clinically elevated at baseline on adolescent-reported Inattention/Hyperactivity (N = 12), 50% transitioned to scores in the normative range at subsequent time points, with an average T-score decrease of 16 points from baseline to posttreatment.

Table IV.

Longitudinal Mixed Effects Logistic Model Results of Clinical Cut-offs of Health-Related Quality of Life and Emotional and Behavioral Functioning

Outcome	Baseline	Post	Fup1	Fup2	Odds ratio	Standard error	p-value	95% CI
Caregiver-report
Behavior Assessment System for Children-2 % Over the At-Risk/Clinical Cut-off
Externalizing symptoms	25%	9%	4%	10%	1.75	0.66	.14	0.83, 3.67
Internalizing symptoms	44%	34%	29%	31%	1.58	0.72	.31	0.65, 3.84
Behavioral Symptom Index	36%	31%	8%	21%	3.27	1.31	.003	1.49, 7.18
Adaptive skills	53%	44%	38%	48%	1.16	0.27	.52	0.74, 1.82
PedsQL-Epilepsy Module-% Over the At-Risk/Clinical Cut-off
Impact	22%	16%	8%	10%	1.83	0.64	.08	0.92, 3.62
Cognitive functioning	22%	28%	8%	17%	1.88	0.74	.11	0.87, 4.1
Executive functioninga	17%	22%	8%	21%	1.04	0.39	.91	0.51, 2.15
Sleep	47%	44%	29%	52%	1.26	0.43	.49	0.65, 2.48
Mood/Behavior	47%	50%	33%	45%	1.27	0.38	.43	0.71, 2.427
Adolescent-report
Behavior Assessment System for Children-2 % Over the At-Risk/Clinical Cut-off
School problems	17%	18%	8%	4%	4.51	4.35	.12	0.68, 29.9
Internalizing problems	22%	25%	16%	20%	0.87	0.29	.67	0.46, 1.66
Emotional Symptoms Indexa	31%	22%	16%	16%	1.24	0.56	.63	0.51, 3.00
Inattention/hyperactivity	33%	18%	20%	12%	2.60	0.85	.004	1.36, 4.95
Personal adjustment	17%	24%	8%	12%	1.41	0.59	.42	0.61, 3.22
PedsQL-Epilepsy Module-% Over the At-risk/Clinical Cut-off
Impact	24%	12%	4%	17%	1.47	0.71	.42	0.57, 3.79
Cognitive functioning	26%	24%	8%	13%	2.33	0.81	.015	1.18, 4.61
Executive functioning	34%	24%	16%	25%	1.45	0.47	.25	0.77, 2.73
Sleep	45%	31%	12%	38%	1.24	0.36	.45	0.71, 2.18
Mood/behavior	42%	33%	16%	29%	1.47	0.66	.39	0.61, 3.57

Note. Covariates for all models included years since diagnosis, seizure type, child race, presence of seizures in the last 3 months, and total side effects; child age was also included in the PedsQL Epilepsy Module models. Bolded outcomes represent significant changes over time.

^aModel would not converge with total side effects (adolescent-reported Behavior Assessment Schedule for Children-2 Emotional Symptoms Index) or child age (parent-reported PedsQL Epilepsy Module Executive Functioning scale) so this covariate was removed.

Epilepsy-Specific HRQOL

Caregivers reported statistically significant improvements in Mood/Behavior PedsQL Epilepsy Module scores (b = 9.76, p = .04), though there was a significant quadratic effect, suggesting that HRQOL scores worsened with greater time since treatment. There were no significant changes over time for any other caregiver-reported PedsQL Epilepsy Module subscales (see Table III and Figure 2).

Figure 2.

PedsQL Epilepsy Module Scores Over Time (dashed lines = parent; solid lines = teen).

Adolescents reported statistically significant improvements over time on the Impact (b = 10.35, p = .003), Cognitive functioning (b = 4.23, p = .002), Executive functioning (b = 4.28, p = .001), and Sleep (b = 15.85, p = .02) subscales of the PedsQL Epilepsy Module. The Impact and Sleep subscales also had significant quadratic effects, suggesting that scores on these two scales slightly deteriorated at later time points. There was no statistically significant improvement on self-reported mood/behavior (see Table III and Figure 2).

Adolescents were more than two times more likely to move from clinically elevated to normative levels for the PedsQL Epilepsy Module Cognitive Functioning scale over time (OR = 2.18, p = .016). Of those who were clinically elevated at baseline on the adolescent-reported PedsQL Epilepsy Module Cognitive Functioning scale (N = 10), 60% transitioned to normative levels at subsequent time points, with an average increase of 29.17 points, which exceeds the PedsQL Epilepsy Module MCID for this subscale (e.g., 10.1 points). No other self- or caregiver-reported scales demonstrated statistically significant odds of moving from clinically elevated to normative levels (see Table IV).

Discussion

This study conducted secondary analyses to determine whether Epilepsy Journey, a web-based e-health program shown to improve EF behaviors in an open pilot trial, was also associated with improvements in emotional and behavioral functioning, as well as the HRQOL of adolescents with epilepsy. Overall, these data support Epilepsy Journey’s positive influence on parent-rated emotional and behavioral functioning in adolescents with epilepsy and underscore the program’s potential to affect areas of behavioral and emotional functioning that extend beyond EF. Specifically, caregivers reported significant improvements in adolescents’ externalizing symptoms, overall behavioral symptoms, and adaptive functioning, as well as the mood/behavior HRQOL subscale over time. For each of these areas of functioning, scores improved with slight rebounds or deterioration in scores at the 5-month follow-up. The pattern of findings also suggests that improvements were clinically meaningful, with reductions in the percentage of adolescents falling in the clinically impaired range on the Behavioral Symptom Index declining from 36% at baseline to 21% at the 5-month follow-up. Post hoc analyses revealed that these findings are further supported when examining only those who were elevated at baseline and show more robust improvements over time. These data are not surprising given the wealth of literature demonstrating a strong relationship between EF skills and both emotional/behavioral functioning and HRQOL (Baum et al., 2010; Hoie et al., 2008; Love et al., 2016; Modi et al., 2019a; Schraegle & Titus, 2016; Sherman et al., 2006). Interestingly, our study highlights the strong correlations between caregiver-reported BRIEF subscales and BASC-2 subscales, but not epilepsy-specific HRQOL. Overall, it stands to reason that when we address EF skills and equip adolescents with problem-solving skills to address emotional and behavioral functioning via Epilepsy Journey, we will have a simultaneous impact on multiple areas of functioning.

In contrast to caregiver reports of improved adolescent emotional/behavioral functioning, adolescents themselves failed to report improvements in behavior on the BASC-2, with the exception of the Inattention/Hyperactivity subscale. However, they reported improvements on a number of dimensions on the PedsQL Epilepsy Module, with the highest HRQOL at follow-up. Interestingly, the improvement in our primary outcomes of EF did not yield as robust a finding as these (Modi et al., 2019b), which suggests that adolescents may be able to see the influence of Epilepsy Journey on daily functioning, as measured by HRQOL. Across caregiver reports of emotional and behavioral functioning and some adolescent-reported HRQOL subscales (e.g., Impact, Sleep), ratings improved out to 2 months posttreatment and then deteriorated slightly at the 5-month follow-up period. This finding is highlighted in the adolescent PedsQL Epilepsy Module Sleep scale and may be due to several reasons. First, adolescents received the intervention and completed follow-up assessments throughout the year, resulting in some adolescents completing assessment during summer and holiday months, where sleep schedules are typically more variable. Further, adolescents only received a limited dose of intervention around sleep with only one module focused on the constructs of both stress and sleep, which may have been insufficient to produce long-term changes in sleep. Overall, these patterns suggest that future research or clinical applications consider booster sessions (e.g., every 3 months) or a more robust focus on constructs (e.g., separate modules for sleep and stress) to maintain emotional/behavioral functioning gains longer term.

The overall discrepancy between caregiver and self-report is striking and suggests that caregivers may perceive improvements in overall and externalizing behaviors that adolescents may be unaware of. These findings are consistent with results from the Teen Online Problem Solving program with adolescents with TBI (Wade et al., 2010, 2011, 2017), where parents and not adolescents reported reductions in externalizing behavior problems. Self-awareness deficits may contribute to discrepancies between parent report and adolescent self-report. These discrepancies in reporting are also similar to reports of EF behaviors, in which caregivers identify more deficits than their adolescents (Guy et al., 2004). With the exception of internalizing disorders, there is robust literature to suggest that caregivers are reliable reporters of observable externalizing behaviors that can be seen (Eiser & Varni, 2013). However, it is notable that adolescents reported significant improvements in HRQOL over time, which was also discrepant from their caregivers. This finding is novel and may reflect the ability of adolescents to report on their daily functioning in the context of their epilepsy. It may be easier to report on aspects of daily functioning that are not psychopathology based, as is the case with the BASC-2. Thus, having multi-method reporting around behavioral and EF issues is likely to yield the most comprehensive understanding of the adolescent’s functioning, especially given the complementary nature of the profile of improvements seen in this study.

The study is not without limitations, including the open pilot design with a small homogenous sample. One of the primary limitations of the study is the lack of a control group; thus, it is unknown if reported changes in behavioral/emotional functioning and HRQOL are due to participation bias. Future trials should include an attention control group, as well as a larger and more heterogeneous sample of adolescents with epilepsy, including increased racial diversity. However, in line with the ORBIT model of behavioral intervention development (Czajkowski et al., 2015), this pilot represents an important and necessary step of identifying potential signals/effects of an intervention prior to launching a larger randomized controlled clinical trial. An important next step is to test Epilepsy Journey within a randomized controlled trial design. Reliance on parent- and self-report measures of improvement that may have been influenced by social desirability biases poses an additional limitation. In addition, given the overlap of some shared features between EF behaviors, psychosocial comorbidities, and HRQOL, there is likely shared variance in these outcomes. Future research could benefit from ratings by individuals who are unaware of treatment status, such as teachers, or including more objective measures of EF (e.g., neuropsychological screening tools). Additional secondary outcome measures could have further highlighted areas of improvement, including robust measures of sleep and stress. Finally, given that some participants benefited more from Epilepsy Journey than others, an important area of future research is to examine moderators of treatment response.

These findings add important new information regarding the potential for the Epilepsy Journey program to yield improvements not only in EF, but also behavioral and emotional functioning and epilepsy-specific HRQOL. If future trials of Epilepsy Journey are successful, dissemination of this intervention could be beneficial, especially for those who lack access to both health and mental health care. Further, although Epilepsy Journey was developed for adolescents with epilepsy, the content could be adapted to address more generally EF skills and thus could be generalizable to a broader group of patients and typically developing adolescents with EF deficits. Overall, our study shows promising results that warrant further research.