ADHD, Neurological Correlates and Health-related Quality of Life in Severe Pediatric Epilepsy

Authors


Address correspondence and reprint requests to Dr. Elisabeth M.S. Sherman, Neurosciences Program, Alberta Children's Hospital, 2888 Shaganappi Trail NW, Calgary, AB, Canada T3B 6A8. elisabeth.sherman@calgaryhealthregion.ca

Abstract

Summary: Purpose: ADHD is reported as a frequent comorbidity in pediatric epilepsy. We aimed to clarify the prevalence of ADHD, its neurological correlates and the role of ADHD in health-related quality of life (HRQOL) in children with severe epilepsy.

Method: Data from the ADHD Rating Scale-IV (ADHD-RS-IV) from 203 children (mean age = 11.8, SD = 3.8) from a tertiary center serving children with severe epilepsy were reviewed.

Results: Inattention was frequently elevated in the sample (40% vs. 18% for hyperactivity-impulsivity). Age of onset, epilepsy duration, and seizure frequency were not related to severity of inattention or hyperactivity-impulsivity. Over 60% of children met screening criteria for ADHD-Inattentive subtype (ADHD-I) or ADHD-Combined Inattentive/Hyperactive-Impulsive subtype (ADHD-C). Compared to ADHD-I, ADHD-C was associated with earlier onset of seizures, generalized epilepsy, lower adaptive level, and in normally developing children, a higher degree of intractability compared to ADHD-I. ADHD-I was more prevalent in localization-related epilepsy, and there was a trend for a higher use of AEDs with cognitive side effects in this group. ADHD was associated with poor HRQOL: children with ADHD-I and ADHD-C had a two- and four-fold likelihood of low HRQOL, respectively, compared to non-ADHD children.

Conclusions: Children seen at tertiary care centers for severe epilepsy are at high risk for attention problems and ADHD, and ADHD is a significant predictor of poor HRQOL in epilepsy, particularly in the case of ADHD-C. ADHD occurring in the context of severe epilepsy appears to be associated with specific neurological characteristics, which has implications for comorbidity models of ADHD and epilepsy.

The notion that many children with epilepsy present with symptoms of attention-deficit/hyperactivity disorder (ADHD) is not new. Over 50 years ago, Ounstead (1955) described a “hyperkinetic syndrome” in children with epilepsy that closely approximates our current definition of ADHD. Recent studies indicate that attention problems are frequently reported in children with severe variants of epilepsy such as intractable seizure disorders, cryptogenic epilepsy, and symptomatic epilepsy as well as in less severe conditions such as idiopathic epilepsy, well-controlled complex partial epilepsy, absence epilepsy and benign seizure disorders (Gaggero et al., 1992; Mitchell et al., 1992, Piccirilli et al., 1994; Struniolo and Galletti, 1994; Williams et al., 1996a, 1998; Slick et al., 2006). Although less studied, problems with impulsivity and behavioral dyscontrol in children with epilepsy have also been reported (Stores et al., 1978; Struniolo and Galletti, 1994; Slick et al., 2006).

There is compelling evidence that ADHD symptoms may in some cases predate the onset of epilepsy. A history of attention problems is twice as common in children seen after their first seizure versus controls (Austin et al., 2001), and community-based studies have indicated a 2.5-fold increase in preexisting ADHD, predominantly of the inattentive subtype, among children with new-onset seizures compared to control children (Hesdorffer et al., 2004). In tertiary care centers, ADHD is reported to be the most common disorder in children with epilepsy, affecting 31% of preschoolers and 63% of school-age children (Thome-Souza et al., 2004).

Increasingly, studies suggest that there is a difference between ADHD occurring in the general population (“Primary ADHD”) and ADHD occurring in the context of epilepsy. First, the rate of ADHD Combined subtype (ADHD-C), the ADHD subtype characterized by deficits in both impulse regulation and attention, far outstrips that of ADHD Predominantly Inattentive subtype (ADHD-I) in primary ADHD (Barkley, 2006). However, the pattern is reversed in epilepsy. Community-based studies indicate that ADHD-I is found more often than ADHD-C in children with new-onset seizures (Hesdorffer et al., 2004). Similarly, studies involving tertiary care samples indicate a prevalence of 24% of ADHD-I compared to 11% for ADHD-C in epilepsy (Dunn et al., 2003). Another difference is the gender ratio. In community-based studies involving primary ADHD, the prevalence of boys with ADHD is three to seven times greater than girls; the gender disparity is even greater in clinic-referred children, were five to nine boys are seen for every girl (Barkley, 2006). In contrast, most epilepsy samples are reported to have equal representation of boys and girls with ADHD (Dunn et al., 2003; Hesdorffer et al., 2004). There are nevertheless some similarities: ADHD symptoms tend to have an age-related expression, with fewer symptoms as children age, both in primary ADHD (Barkley, 2006) and in pediatric epilepsy (Thome-Souza et al., 2004). Barkley (2006) notes, however, that this may be a measurement artifact due to reduced sensitivity of clinical instruments to persisting symptoms over time.

One of the major differences between children with primary ADHD and children with epilepsy is that children with epilepsy face several epilepsy-specific neurologically based risk factors for attentional disturbance. These include interictal, ictal, and postictal effects, medication effects, cognitive problems secondary to underlying brain abnormalities, and long-term effects of repeated seizures (Williams et al., 1996b; Espie et al., 1999; Meador, 2002; Laporte et al., 2002; Binnie et al. 2003; Helmsteadter et al., 2003; Aldenkamp and Arends, 2004; Elger et al., 2004; see also Schubert, 2005; Dunn and Kronenberger, 2006). Certain epilepsy syndromes may also predispose to ADHD-like behavior; for instance, frontal lobe epilepsy (FLE) shares behavioral features with ADHD, presenting in some patients with impulsivity, disinhibition, and excitement/irritability (Delgado-Escueta et al., 1991; Powell et al., 1997). Nevertheless, seizure type and focus of seizure discharges are not clear predictors of ADHD symptoms (Dunn et al., 2003), although frontal EEG discharges may be associated with ADHD symptoms (Sherman et al., 2000). Some researchers have posited that ADHD-I and epilepsy may share a common underlying deficit in the central norepinephrine system (Hesdorffer et al., 2004), but the exact mechanism remains elusive.

Given our access to a large tertiary care sample of children with epilepsy, we aimed to further delineate the nature of ADHD in severe epilepsy, with the goal of determining (1) the prevalence of ADHD symptoms, (2) which neurological and epilepsy-related factors are related to ADHD symptoms, and (3) whether ADHD symptoms are associated with poor health-related quality of life (HRQOL) and restrictions in daily activities.

METHOD

Participants

Data consisted of consecutive referrals of children seen for neuropsychological assessment referred from the Epilepsy Program at BC's Children's Hospital, the only tertiary medical center for pediatric epilepsy in the province of British Columbia, Canada. Children included (1) epilepsy surgery and VNS candidates from the period of 1999–2006, (2) children who had postsurgical or post-VNS neuropsychological assessments during this same time period, and (3) children with severe epilepsy being followed by epileptologists and who were not surgical candidates at the time of their assessment. The majority of the sample (76%) included children referred for routine pre or postsurgical neuropsychological assessments associated with epilepsy surgery or VNS implantation. The referrals were made on a routine basis for all testable surgical candidates for purposes of preoperative decision-making or routine postoperative follow-up. The remainder of the sample included referrals made to clarify learning or behavioral issues. Behavioral rating scales and HRQOL were routinely collected as part of standard neuropsychological assessments.

Data for 203 cases (44% girls) were retrieved from a clinical and research database containing demographic, neuropsychological, and HRQOL data for children with epilepsy in accordance with hospital and university review boards. Exclusion criteria included progressive neurological or serious medical condition other than epilepsy and lack of parental fluency in English due to the written language requirements of the standardized questionnaires.

Demographic and clinical characteristics of the sample are shown in Table 1. On the whole, the sample was representative of severe epilepsy, with a median age of onset of less than 3 years, a 7-year duration of uncontrolled seizures, a median of two current AEDs with a median of three failed AED trials. Seizure frequency was based on seizure counts for the month prior to the neuropsychological evaluation as estimated by chart review and confirmed by parents at the time of the evaluation. Seizure frequency for the group was high, with a median of three seizures in the month prior to testing. The median adaptive level was on the cutoff for adaptive delay, indicating that approximately half the sample had significant delays and half were normally developing children with regard to functional level.

Table 1. Demographic and clinical characteristics of the sample (N = 203)
 MedianMeanSDRange
Age12.1 11.8  3.8 4.1–20.2
Age at onset 2.5  4.2  4.1   0–15.5
Duration of epilepsy 7.1  7.6  4.30.23–17.9
Number of AEDs2   1.6  1.10–5
Number of failed AEDs3   4.2  3.6 0–20
Seizure frequency3 106.5345.10 to >2000
Adaptive level (SIB-R)69.5 69.3 36.8  4–155
Epilepsy typePercent of sample
  1. Age, age at onset, duration of epilepsy, and test–retest interval are shown in years. Seizure frequency refers to the number of seizures in the month preceding the evaluation. Number of prior AEDs refers to the number of AEDs tried unsuccessfully prior to the current evaluation. Adaptive level refers to SIB-R scores (normative mean of 100, SD = 15).

Localization related epilepsies74.4
Generalized epilepsies16.7
Undetermined whether focal or generalized 2.5
Special syndromes 0.5
Unclassifiable 1.5
Unspecified 4.4

International League Against Epilepsy (ILAE, 1989) classifications for the sample are presented in Table 1; most patients were classified as having localization-related epilepsy (74%). Eighty-one children (40%) were seen as part of presurgical evaluations (with 26 children having undergone epilepsy surgery as of this writing); 43 children (21%) had a history of prior epilepsy surgery, and 31 (15%) had a history of treatment with vagal nerve stimulation (VNS).

Children were on a variety of AEDs, with the majority on carbamazepine and lamotrigine. Because of their possible effects on cognition, the number of children taking topiramate, benzodiazepines or phenobarbital was tabulated: 33% were on a benzodiazepine (clobazam, clonazepam, or nitrazepam), 15% were on topiramate, and 1.5% of the total sample was on phenobarbital at the time of their evaluation. Of these, 13 children or 6.4% of the sample were taking two of these medications concurrently.

Few children had a prior ADHD diagnosis at the time they were assessed, or were taking stimulant medications. Only 14 children (6.9%) had been previously diagnosed with ADHD by a family physician, pediatrician, psychologist, or other health professional; in most cases, there was insufficient information to determine which ADHD subtype the diagnosis referred to, or what assessment procedures were used to make the diagnosis. Seven children were on stimulant medications at the time of the assessment (3.4% of the sample); four children were taking methylphenidate and three children were taking dexedrine.

Measures

ADHD ratings

The ADHD Rating Scale-IV (ADHD-RS-IV; DuPaul et al., 1998) is an 18-item parent-rated questionnaire based on the diagnostic criteria for ADHD as described in the 4th edition of the Diagnostic and Statistical Manual of Mental Disorders of the American Psychiatric Association (DSM-IV, American Psychiatric Association, 1994). The ADHD-RS-IV provides Inattention and Hyperactivity-Impulsivity subscales as well as a total score. Raw scores are converted to percentile scores based on age and gender. Scores on the ADHD-RS-IV can be used in two ways: (1) to characterize the severity of ADHD symptoms, and (2) to code a provisional DSM-IV diagnosis for either ADHD-I or ADHD-C based on empirically derived cutoffs for ADHD. The ADHD-RS-IV therefore allows both a quantitative estimation of symptom severity (percentile scores for Total, Inattention and Hyperactivity-Impulsivity domains) as well as DSM-IV ADHD screening diagnosis based on dimensional analysis of inattention and impulsivity-hyperactivity. Because teacher ratings were not available for the entire sample, and because parent ratings are commonly used in clinical settings and have been shown to have clinical utility when used alone (Barkley, 2006), only parent ratings were used in this study.

We operationalized ADHD symptoms in two ways. First, we examined severity of ADHD symptoms by examining group means and the proportion of children with scores above a clinically significant cutoff of 1.5 SD above the normative mean on each of the ADHD-RS-IV subscales. Second, because a 1.5 SD cutoff tends to miss some children with bona fide ADHD (Du Paul et al., 1998), we also examined the number of children meeting screening criteria for the diagnosis of ADHD, as defined by attaining scores above specific combined cutoffs on both the Inattention and Hyperactivity-Impulsivity scales of the ADHD-RS-IV. We chose combined cutoffs derived from prior research using a school-based sample consisting of mixed referrals and structured-interview-confirmed ADHD (Du Paul et al., 1998), which we felt might most closely approximate the moderate baserate and heterogeneous nature of our sample (as opposed to cutoffs derived from high base rate samples comprised primarily of children with primary ADHD). Cases were therefore coded as: (1) ADHD-I if they had Inattention ≥80th percentile and Hyperactivity-Impulsivity ≤80th percentile, (2) ADHD-C if they had Inattention ≥80th percentile and Hyperactivity-Impulsivity ≥85th percentile, and (3) as ADHD, Primarily Hyperactive-Impulsive subtype (ADHD-HI) if they had Hyperactivity–Impulsivity scale elevations of ≥85 without concurrent elevations on Inattention of ≥80.

Adaptive level

Unlike IQ tests that can only be administered to children who can tolerate relatively lengthy cognitive testing, adaptive behavior measures can be obtained for all children and therefore are more able to capture the functional level of the entire sample. The Scales of Independent Behavior-Revised (SIB-R; Bruininks et al., 1997) provides information on an individual's ability to function independently in the home and community. Parents rate their children in several domains, including motor skills, social interaction and communication skills, personal living skills and community living skills. The SIB-R Broad Independence score was collected for each participant as it provides an easily interpretable metric in a format akin to that used for IQs (i.e., standard scores with a mean of 100 and SD of 15, with higher scores indicating better functional independence). Reliability and validity are adequate to high (Strauss et al., 2006). SIB-R level for the sample is shown in Table 1.

Health-related quality of life (HRQOL) and restrictions in daily activities

The Impact of Childhood Illness scale (ICI; Hoare & Russell, 1995) is a 30-item parent-rated HRQOL questionnaire that is divided into four sections: (1) impact of the disorder and its treatment, (2) impact on the child's development and adjustment, (3) impact on parents, and (4) impact on the family. For each item, the parent rates how often the particular problem or situation occurs (Frequency score) as well as the amount of concern each one causes (Importance score). Scores for the two domains range from 0 to 60, with the total score ranging from 0 to 120. Higher scores reflect worse quality of life. The ICI was designed to be suitable for children with a variety of illnesses or disabilities (e.g., “Because of my child's illness, she may stop breathing”; “Because of her illness, my child has special problems with reading or math”; “My child's illness limits how often we go out as a family”). The parent rates each item according to the degree of perceived frequency or impact on the parent and/or child. Validity for the scale is good (Sherman et al., 2002).

The Hague Restrictions in Epilepsy Scale (HARCES; Carpay et al., 1997) is a 10-item parent-rated scale that measures the number of restrictions imposed because of seizures. Items reflect the frequency with which the child takes part in activities such as swimming, riding a bicycle, staying elsewhere overnight and participating in physical education. The scale focuses on the extent to which epilepsy affects a child's ability to take part in everyday childhood activities that contribute to HRQOL—activities in which developmentally critical exposures to peers and to age-appropriate activities occur. As would be expected because of seizure-related restrictions, the ability to participate in these activities is often restricted in children with epilepsy. The parent rates each item according to the degree to which the child's activity is limited by epilepsy; scores range from 10 (no disability) to 40 (severe disability). Internal reliability (r = 0.89), test–retest stability (r = 0.93; Carpay et al., 1997) and validity are all strong (Sherman et al., 2002), and the scale has shown utility in tracking recovery after epilepsy surgery (Van Empelen et al., 2004).

Statistical approach and significance testing

Because of the exploratory nature of the study and the need to balance Type I and Type II error, the alpha level for significance tests was set at 0.05 a priori; Tukey's HSD test was used for multiple simultaneous comparisons. Only correlations of 0.30 or above were interpreted as evidence of medium effect size. We used t-tests to compare continuous variables and Chi-square to compare categorical variables.

RESULTS

Severity and prevalence of Inattention and Hyperactivity-Impulsivity problems and association with neurological variables

Table 2 shows ADHD-RS-IV means and standard deviations (SDs). Inattention symptoms were significantly more severe than Hyperactivity-Impulsivity symptoms in the sample [t = 8.7, df = 202, p < 0.0001]. Overall, 40%, 18%, and 29% of the children had clinically significant impairments (scores > 1.5 SDs from the normative mean) on the Inattention, Hyperactivity-Impulsivity and Total scales, respectively. Elevations on Inattention were significantly more prevalent than on Hyperactivity-Impulsivity (McNemar χ2= 35.2, p < 0.001). In terms of children with single-domain elevations only, fully 25% of the sample had a clinically elevated Inattention score but not a clinically elevated Hyperactivity-Impulsivity score, while only 3% of the sample demonstrated the opposite pattern. These analyses indicated that both the severity and prevalence of significant attention problems was considerable in the sample.

Table 2. Means, SDs, ranges and percent of sample with elevations on the ADHD-RS-IV scale
 MeanSDRange≥1.5 SDs from the normative mean (%)
  1. Values reflect norm-based percentiles.

Inattention78.725.31–9940.4
Hyperactivity-Impulsivity63.529.71–9918.2
Total75.025.41–9928.6

On a group level, correlations between Inattention/Hyperactivity-Impulsivity and neurological characteristics (age at seizure onset, duration of epilepsy, number of AEDs, number of previous AEDs, seizure frequency) were all nonsignificant and below r = 0.10. The use of benzodiazepines, phenobarbital, or topiramate, either singly or in combination, was not related to ADHD symptom severity. There was a small correlation between ADHD symptoms and age, with fewer Hyperactive-Impulsive symptoms exhibited by older children (r =–0.23, p = 0.001). There were no age-related differences in the expression of Inattention symptoms.

ADHD subtypes and neurological variables

When the empirically derived optimal cutoff scores for ADHD-I and ADHD-C from Dupaul et al. (1998) were applied, about one-third of the sample met criteria for ADHD-I and one third met criteria for ADHD-C, with only about one third of children not meeting screening criteria for either disorder (Table 3). Only a very small fraction of the sample had elevations on only the Hyperactivity-Impulsivity scale (ADHD-HI subtype).

Table 3. Prevalence of ADHD subtypes based on optimal empirically derived screening cutoffs
ADHD subtypeNPercent
  1. ADHD-I, Inattentive subtype; ADHD-C, Combined subtype; ADHD-HI, Hyperactive-Impulsive subtype; cutoffs from DuPaul et al. (1998).

None 59 29.1
ADHD-I 70 34.5
ADHD-C 69 34.0
ADHD-HI  5  2.5
Total203100  

For the purposes of the following analyses, the ADHD-HI subtype was combined with the ADHD-C group, as suggested by Barkley (2006). The three sample subgroups (Non-ADHD, ADHD-I, and ADHD-C) were compared on a number of neurological and demographic characteristics. Data are shown in Table 4.

Table 4. Neurological characteristics and adaptive levels for ADHD subtypes
 ADHD-IADHD-CNon-ADHDI v. CI v. NC v. N
  1. I v. C, comparison between the ADHD-I and ADHD-C (Tukey HSD); I v. N, comparison between the ADHD-I and Non-ADHD (Tukey HSD); C v. N, comparison between the ADHD-C and non-ADHD (Tukey HSD).

Age12.5 (3.9) 10.8 (3.7) 12.2 (3.6)  0.021ns 0.054
Age at onset5.1 (4.6)3.1 (3.2)4.6 (4.2) 0.008ns 0.067
Duration7.4 (4.4)7.7 (4.2)7.7 (4.4) nsnsns
AEDs1.8 (1.1)1.5 (1.0)1.5 (1.1) nsnsns
Seizure frequency 87.6 (264.6) 93.2 (261.8) 144 (488.9)nsnsns
Number of prior AEDs4.3 (3.8)4.3 (3.7)4.1 (3.4) nsnsns
Adaptive level68.7 (34.6)55.1 (32.2)88.2 (37.0) 0.0280.007<0.001
Gender (% boys)57.154.157.6nsnsns

Significant differences emerged among the groups. Compared to the ADHD-C group, the ADHD-I group was older, had a later age of onset of epilepsy, and a higher level of adaptive functioning. The ADHD-C group was younger on average than the Non-ADHD group. The non-ADHD group had the highest average level of adaptive functioning, significantly better than that of either ADHD group. Although falling short of our a priori cutoff for statistical significance of p < 0.05, there was a trend for children in the ADHD-I group to be taking benzodiazepines, topiramate or phenobarbital compared to children with ADHD-C (50% vs. 36.5%; p = 0.07).

Table 4 shows proportions according to gender and epilepsy syndrome. The percentage of boys, usually overrepresented in primary ADHD samples, was similar to that of girls and did not differ across subtypes. Although the majority of children in both ADHD groups were diagnosed with localization-related epilepsy, the proportion of children diagnosed with generalized epilepsy was significantly greater in the ADHD-C group relative to the non-ADHD group (25.7% vs. 14.4%; χ2= 7.08, p = 0.008). Relative differences between proportions of samples diagnosed with localization-related vs. generalized epilepsy were not large enough to reach significance when ADHD-C and ADHD-I groups were compared and when non-ADHD and ADHD-I groups were compared. Duration, seizure frequency and intractability as indicated by number of prior failed AED trials did not differ between the two ADHD groups.

Correlational analyses indicated a significant amount of shared variance between ADHD-RS-IV scales and SIB-R Broad Independence (see Table 5) indicating that worse ADHD symptoms were associated with lower adaptive level. We therefore assessed whether ADHD group characteristics held when children with adaptive delays (SIB-R < 70) were excluded from the analyses. Results were similar, indicating that excluding children with suspected mental retardation did not significantly affect the relationship between ADHD symptoms and demographic/neurological characteristics. The only exception was that in nondelayed children, the ADHD-C group had more failed medication trials than non-ADHD children (p = 0.03).

Table 5. Relationship between ADHD symptoms and adaptive level
SIB-RADHD-RS-IV
InattentionHyperactivity-ImpulsivityADHD total
  1. Normal-range group refers to children without delays in adaptive functioning (SIB-R ≥70).

  2. *p = 0.02.

  3. **p = 0.001.

  4. ***p = 0.0001.

Entire sample−0.35***−0.27***−0.35***
Normal-range group−0.47***−0.24*  −0.44** 

ADHD and HRQOL

Correlations between Inattention, Hyperactivity-Impulsivity, Total ADHD-RS-IV and HRQOL as measured by the ICI indicated worse HRQOL with increasing ADHD symptoms (rs = 0.32, .26, and 0.33, all ps < 0.0001). Table 6 shows HRQOL levels for the ADHD subtypes. The ADHD-C subgroup showed the worst quality of life, with ADHD-I in the intermediate range, and Non-ADHD children showing the best HRQOL among the groups (see Fig. 1). Similarly, ADHD-C children had the most HRQOL-related restrictions on their daily activities as shown by the HARCES.

Table 6. HRQOL according to ADHD subtype
 ADHD-IADHD-CNon-ADHDI v. CI v. NC v. N
  1. I v. C, comparisons between the ADHD-I and ADHD-C (Tukey HSD); I v. N, comparisons between the ADHD-I and Non-ADHD (Tukey HSD); C v. N, comparisons between the ADHD-C and non-ADHD (Tukey HSD).

ICI total score47.1 (21.4)58.2 (22.3)35.7 (28.8).025.027<001
HARCES22.6 (9.4) 24.6 (9.4) 18.2 (8.3) ns.031 .001
Figure 1.

HRQOL for ADHD subtypes.

To determine the extent to which ADHD was associated with low quality of life, scores on the ICI and HARCES were dichotomized into low and high HRQOL using median split. Odds ratios indicated that the presence of ADHD (either subtype) was associated with a three times greater likelihood of low HRQOL compared to no ADHD diagnosis (ICI: odds ratio = 3.00, 95% CI = 1.57–5.74, p = 0.001; HARCES: odds ratio = 3.07, 95% CI = 1.59–6.00, p = 0.001). In terms of ADHD subtypes, children with ADHD-I were twice as likely to have low HRQOL compared to non-ADHD children (ICI: odds ratio = 2.19, 95% CI = 1.05–4.56, p = 0.04; HARCES: odds ratio = 2.47, 95% CI = 1.17–5.19, p = 0.02), and children with ADHD-C were four times as likely to have low HRQOL than non-ADHD children (ICI: odds ratio = 4.11, 95% CI = 1.95–8.62, p < 0.0001; HARCES: odds ratio = 3.83, 95% CI = 1.81–8.12, p < 0.0001).

DISCUSSION

There is a high prevalence of ADHD symptoms in children with epilepsy; this is particularly true of children seen at tertiary care centers (Thome-Souza et al., 2004). In our tertiary-center epilepsy sample characterized by early age of onset (< age 3), significant duration of epilepsy (>7 years), polytherapy, intractability (failure of > 2 AEDs), high seizure frequency (three per month), and low adaptive level, this was quite evident. Over 40% of children were rated as having significant attention deficits, and the proportion of children meeting screening criteria for either ADHD-C or ADHD-I was over 60%. This is consistent with previous tertiary-care sample estimates (Thome-Souza et al., 2004), far exceeding the rate reported in children at first seizure (Hesdorffer et al., 2004). Further, ADHD was associated with a two- to four-fold increased likelihood of poor HRQOL, indicating that ADHD symptoms have significant, real-world adverse implications for children with epilepsy and their families. These findings mirror reports of poor HRQOL in primary ADHD without epilepsy (Graetz et al., 2001; Klassen et al., 2004; Matza et al., 2004; Topolski et al., 2004; Escobar et al., 2005).

We found specific subgroup differences when comparing ADHD-I versus ADHD-C. Age at seizure onset was later in children with ADHD-I than in ADHD-C, and there was a trend for more ADHD-I children to be taking medications with possible cognitive side effects (i.e., benzodiazepines, topiramate, and phenobarbital) than children with ADHD-C. ADHD-I children were more likely to have localization-related epilepsy than generalized epilepsy, which was more common in ADHD-C. Children with ADHD-C were younger than those with ADHD-I, and when low functioning children were excluded, had more failed medication trials. Overall, ADHD-C was characterized by earlier onset of seizures, generalized epilepsy, lower adaptive level, and in developmentally normal children, a higher degree of intractability compared to ADHD-I. These findings suggest that ADHD-C in epilepsy may be a marker for severe epilepsy and/or severe brain dysfunction.

Notably, both ADHD subgroups had lower adaptive levels than children without ADHD, an association that is also reported in children with primary ADHD (Barkley, 2006). It is possible that the lower adaptive levels of children with ADHD and epilepsy reflect the adverse effect of inattention and impulsivity on acquiring the skills of daily living, or that ADHD symptoms and lower adaptive levels are manifestations of a common underlying brain dysfunction.

We found several discrepancies between the characteristics of children meeting criteria for ADHD in our epilepsy sample and the known characteristics of children with primary ADHD, recently reviewed and compiled by Barkley (2006; see Table 7). Most striking was the prominence of inattention as a core symptom in our epilepsy sample compared to longstanding research indicating that impulsivity and dyscontrol are core symptoms of primary ADHD, found more frequently and consistently than inattention symptoms and prominent even in children with mental retardation (Barkley, 2006; Hastings et al., 2005). Other differences such as gender distribution and subtype prevalence are listed in Table 7, including results from the small number of stimulant response studies in epilepsy (e.g., Gross-Tur et al., 1997; Gucuyener et al., 2003, and reviews by Dunn and Kronenberger, 2006, and Schubert, 2005).

Table 7. Summary of core characteristics of primary ADHD and ADHD in severe epilepsy
 Primary ADHDADHD in epilepsy
  1. Information on primary ADHD adapted from Barkley (2006).

  2. aMay reflect reduced sensitivity of clinical instruments to persisting symptoms.

Age at onset of ADHDPrior to age 7; likely to manifest in preschoolUnknown, but likely prior to seizure onset
Gender distributionLarge preponderance of boysEqual representation of boys and girls
Symptom characteristicsImpulsivity/Hyperactivity is core symptomInattention more severe and frequent than impulsivity/hyperactivity
SubtypesMajority ADHD-C; ADHD-I less frequentEqual representation of ADHD-C and ADHD-I or higher prevalence of ADHD-I
Cognition (IQ)Slightly below average or averageLow (tertiary care centers)
Age-related changesSymptoms become less severe as children ageaHyperactivity-Impulsivity dimension decreases with agea
Stimulant medication responseGood in ADHD-C; moderate in ADHD-IModerate; not well studied
Underlying mechanismsUnknown; likely geneticUnknown; likely multifactorial

The findings raise interesting questions about (1) the underlying mechanisms of ADHD in the context of epilepsy, and (2) the degree to which ADHD occurring in epilepsy is the same diagnostic entity as primary ADHD as defined by the Diagnostic and Statistical Manual of Mental Disorders IV (DSM-IV; American Psychiatric Association, 1994). Comorbidity is defined as two or more diseases with distinct etiologies, pathogenesis, and pathophysiology occurring in the same individual for a defined time period (Wittchen, 1996; Vella et al., 2000). While the characteristics of a disorder occurring alone and in the presence of a second, unrelated disorder may differ due to influence of the second disorder on the expression of the first disorder, one would not expect a disorder's phenotype to be radically changed when occurring in the context of a comorbidity. Our findings suggest that the characteristics of ADHD occurring in the context of severe epilepsy deviate from those expected in primary ADHD (Table 7). This raises an important question: does ADHD occurring in the context of severe epilepsy arise from the comorbidity of two etiologically distinct disorders, or is there something about severe epilepsy that predisposes to ADHD? One possibility is that ADHD occurring in the context of severe epilepsy is a neurologically based disorder arising from epilepsy-related factors, and is diagnostically distinct from primary ADHD (epilepsy-specific model). Alternatively, epilepsy and ADHD may be comorbid, etiologically distinct conditions, but severe epilepsy might change the expression, severity or clinical characteristics of primary ADHD to yield some of the differences evident in Table 7. Finally, children seen at tertiary centers could include a mix of children presenting with primary ADHD (comorbid cases) and others presenting with an epilepsy-specific disorder characterized primarily by severe difficulties with attention. Others have posited that in epilepsy, ADHD-C represents a true comborbidity, while ADHD-I occurring in the context of epilepsy reflects a common nervous system dysfunction (Noeker and Haverkamp, 2003). This is a plausible scenario, but the fact that ADHD-C in epilepsy was associated with early onset, intractability and low cognitive functioning in this study indicates that the co-occurrence of ADHD-C and epilepsy is unlikely to be completely accounted for by chance association. Instead, ADHD-C also appears to relate in a fundamental way to the underlying epilepsy disorder. The question remains as to whether using the ADHD label (or even the more specific ADHD-I or ADHD-C label) accurately describes the clinical characteristics of children with severe epilepsy, given the prominence of attention symptoms and the differences between ADHD with and without epilepsy. Further research, including population-based studies of ADHD in epilepsy, will be key in clarifying these issues.

Important caveats of this study methodology and sample must be addressed. First, a comprehensive assessment including psychiatric interview is necessary for making reliable ADHD diagnoses. Rating scales are one component of such an assessment, and when used for screening for ADHD yield overestimates of the true prevalence of the syndrome of primary ADHD; they are therefore useful for determining the upper bounds of ADHD prevalence rather than its true prevalence (Barkley, 2006). Determining the lower boundaries of ADHD prevalence requires targeted evaluation of symptom duration, age of onset, and degree of impairment secondary to ADHD symptoms to supplement the information provided by rating scales. Of note is the fact that diagnostic cutoff based on categorical boundaries do not inform on the proportion of children with milder but still impairing disturbances (e.g., subclinical or borderline cases).

The second limitation involves the use of a tertiary center study sample; a related point is that the sample originated from Neuropsychology referrals rather than from random cases. As noted in Participants, the majority of referrals were made on a routine basis for surgical work-up or follow-up, with only a minority of the group referred because of specific learning or behavioral issues. Because of the potential for ascertainment bias towards more severe ADHD symptoms in this minority (the assumption being that children with cognitive or behavioral symptoms would be more likely to be referred to Neuropsychology than those without), we compared this group's ADHD-RS-IV scores to the other referral groups, where referrals occurred on a routine basis (presurgical vs. postsurgical vs. general referrals); the group means were not significantly different. Thus, while the sample was not selectively biased towards cases with behavioral impairments because of selective referral to Neuropsychology, there is no doubt that it contained a higher number of children with severe cognitive and behavioral issues than would be expected in other settings, by virtue of being a sample of refractory epilepsy cases—the majority of which had been referred for additional surgical treatment because of repeated medication trial failures. A lower prevalence of attention problems would be expected in community samples or general clinic samples, as previous studies have found. While our results cannot be extended to other settings, the results provide important information for those working in tertiary care settings such as ours, and most specifically, to those working with children who have refractory epilepsy. Ideally, future studies would include the entire spectrum of children with epilepsy, including nonreferred children.

A third limitation is the use of parent ratings rather than teacher ratings or combined parent–teacher ratings. Barkley (2006) discusses the pros and cons of using different raters in screening for ADHD, and concludes that parent ratings are sufficient, in most cases, for diagnosing ADHD. Nevertheless, it will be important to replicate our results using teacher ratings, and determine whether parent–teacher agreement on ADHD symptoms in children with epilepsy mirrors that reported in general samples.

This study has significant clinical implications. Because ADHD is a treatable behavioral syndrome with effective pharmacological and behavioral therapies, a large number of children with epilepsy might potentially benefit from screening and medical intervention. To our knowledge, systematic screening of children for ADHD is rarely employed at epilepsy centers. This practice may need to be reconsidered in light of our findings and those of others (e.g., Thome-Souza et al., 2004). Screening and treatment of ADHD in pediatric epilepsy is particularly relevant given the association between ADHD symptoms and poor HRQOL.

Several avenues for future research remain, including whether there is a specific neuropsychological profile for children with ADHD in epilepsy, whether ADHD symptoms correlate with objective measures of attention and impulsivity in children with epilepsy, and delineation of other cognitive, neuropsychological, behavioral and psychosocial correlates of ADHD in epilepsy. Most importantly, research identifying specific antecedents of ADHD occurring in epilepsy is needed, given that ADHD symptoms in children with epilepsy are multifactorial and likely influenced by factors such as ictal effects, interictal disturbances, medications, and underlying brain abnormalities. Research on etiological subtypes in children presenting with ADHD symptoms and epilepsy, and on the relative importance of different etiological factors in ADHD subtypes are avenues of future research.

Acknowledgments

Acknowledgments:  This study was supported by a grant by the British Columbia Medical Services Foundation/Vancouver Foundation to the first author, and by the Neurosciences program at Alberta Children's Hospital and the Calgary Health Region.

Ancillary