The cognitive effects of interictal epileptiform EEG discharges and short nonconvulsive epileptic seizures

Authors

  • Joost Nicolai,

    1. Department of Neurology, Maastricht University Medical Centre, Maastricht, The Netherlands
    Search for more papers by this author
  • Saskia Ebus,

    1. Department of Clinical Neurophysiology, Epilepsy Center Kempenhaeghe, Heeze, The Netherlands
    Search for more papers by this author
  • Danielle P. L. J. J. G. Biemans,

    1. Department of Behavioral Sciences, Epilepsy Center Kempenhaeghe, Heeze, The Netherlands
    Search for more papers by this author
  • Johan Arends,

    1. Department of Clinical Neurophysiology, Epilepsy Center Kempenhaeghe, Heeze, The Netherlands
    Search for more papers by this author
  • Jos Hendriksen,

    1. Department of Behavioral Sciences, Epilepsy Center Kempenhaeghe, Heeze, The Netherlands
    Search for more papers by this author
  • Johan S. H. Vles,

    1. Department of Neurology, Maastricht University Medical Centre, Maastricht, The Netherlands
    2. Research School Mental Health & Neurosciences Maastricht University Medical Centre, Maastricht, The Netherlands
    Search for more papers by this author
  • Albert P. Aldenkamp

    1. Department of Neurology, Maastricht University Medical Centre, Maastricht, The Netherlands
    2. Department of Behavioral Sciences, Epilepsy Center Kempenhaeghe, Heeze, The Netherlands
    3. Research School Mental Health & Neurosciences Maastricht University Medical Centre, Maastricht, The Netherlands
    Search for more papers by this author

Address correspondence to Albert P. Aldenkamp, Department of Behavioural Sciences, Epilepsy Center Kempenhaeghe, P.O. Box 61, 5590 AB Heeze, The Netherlands. E-mail: aldenkampb@kempenhaeghe.nl

Summary

Purpose:  Educational difficulties or even severe cognitive deterioration is seen in many childhood epilepsy syndromes. Many of those cognitive deficits are related directly to the brain disorder underlying the epilepsy syndrome. However, in other types of epilepsy, the epileptic seizures and/or epileptiform activity can be the dominant factor. This is especially unknown for the more “subtle” short nonconvulsive seizure types. For this reason, we analyzed a new cohort of children.

Methods:  A cross-sectional study of 188 children with epilepsy. Electroencephalography (EEG)–video recordings and cognitive testing were performed simultaneously. The results of children with short nonconvulsive seizures during a 2-h testing session were compared with all children with epilepsy without seizures during the 2-h cognitive testing session and with controls without epilepsy. In a second analysis the cognitive effects of frequency of epileptiform EEG discharges were analyzed.

Key Findings:  The cognitive effects of short nonconvulsive seizures were large, ranging from 0.5 to 1 standard deviation and concerned global cognitive function, speed of central information processing, and memory function. In children without seizures during cognitive testing, the occurrence of frequent epileptiform discharges showed more subtle effects. These effects were independent from the occurrence of short nonconvulsive seizures.

Significance:  We concluded that although the effect is less pronounced in number of areas involved and magnitude, the type of association between frequent epileptiform activity (>1% of the time) and cognitive function in children with epilepsy is comparable to the association between short nonconvulsive seizures and cognitive function.

Childhood epilepsy is commonly associated with cognitive impairment. The cognitive profiles in children with epilepsy are as diverse as the epileptic syndromes themselves (Elger et al., 2004). The cognitive sequelae of epilepsy syndromes manifesting with, for example, an epileptic encephalopathy or an electrical status epilepticus during slow-wave sleep (ESES), have shown to be detrimental (MacAllister & Schaffer, 2007). In other epilepsy syndromes a more subtle influence on cognition has been observed. The mean Full-scale IQ in children with benign childhood absence epilepsy is, for example, mildly lower compared with controls (Pavone et al., 2001; Henkin et al., 2005). In addition to these syndrome-related stable effects, there are more transient effects of paroxysmal epileptiform activity. The most subtle cognitive effects are seen when the presence of epileptiform EEG discharges without any sign of a seizure is associated with increased reaction time or increased number of missed test items (Browne et al., 1974; Aarts et al., 1984; Fisch, 2003; Aldenkamp & Arends, 2004a; Fonseca et al., 2007).

The relative contribution of each of the aforementioned influences on cognitive function in an individual child is unknown (Elger et al., 2004), and only well-controlled studies may isolate these factors.

Our focus here is on the effect of epileptiform activity in children. In analyzing the relation between epileptiform activity and cognition, the distinction between ictal activity (epileptic seizures) and interictal epileptiform activity can only be made when using EEG recording and video simultaneously (Niedermeyer & Lopes da Silva, 2005). When evaluating the effect of ictal activity we have focused on short and difficult-to-detect nonconvulsive seizures, as these are most difficult to distinguish from interictal epileptiform activity. Moreover, the cognitive effects of such seizures may be as difficult to detect as the effect of interictal activity. These seizures are defined as short (i.e., ictal activity of seconds); they have a semiology that is difficult to detect (staring, subtle movements). The seizures are nonconvulsive (i.e., without falling or continuous fierce jerking); however, subtle myoclonic rhythmic movements may be visible. We refer here to these seizures as short nonconvulsive seizures or subtle seizures. This is a descriptive label; in terms of seizure classification the seizures may range from absence seizures to complex partial seizures.

To which extent such short nonconvulsive epileptic seizures have clinical relevant effects on cognitive function is still in debate (Niemann et al., 1985; Ellenberg et al., 1989; Mandelbaum & Burack, 1997; Sirén et al., 2007). This question is of more than theoretical interest because decisions on treatment in daily clinical practice sometimes depend in part on the question of whether such short seizures diminish cognitive functioning (Stores, 1990). Subtle nonconvulsive seizures are—by definition—difficult to detect, and often these seizures present themselves as behavioral fluctuations or attention disorders (Aldenkamp et al., 2001). Therefore, they can persist unnoticed for a longer period and, consequently the cognitive effects may accumulate and lead to learning problems (Aldenkamp et al., 1999a,b). Of interest, Pressler et al. (2006) described that children who showed a significant reduction in subclinical seizures after initiation of lamotrigine showed significant improvement of behavior (Pressler et al., 2005).

Interictal epileptiform activity is defined by spikes or spike-related discharges (Niedermeyer & Lopes da Silva, 2005). In the majority of children with epilepsy, epileptiform EEG discharges may present as a complicating factor (Aarts et al., 1984; Aldenkamp et al., 2001; Binnie, 2003). In the first study that used combined EEG-video telemetry, it was demonstrated that epileptiform EEG discharges may also have cognitive effects and it may be difficult to distinguish these effects from the cognitive effect of short subtle nonconvulsive seizures (Aarts et al., 1984). This was confirmed in recent studies (Aldenkamp & Arends, 2004a). The label “transient cognitive impairment” or “transitory cognitive impairment” (TCI) was proposed for episodes with interictal epileptiform discharges associated with increased reaction time or increased number of missed test items (Aarts et al., 1984). No agreement exists about the need for treatment of such interictal discharges (Besag, 1995; Binnie, 2003).

As yet there are no studies that have systematically evaluated the differential cognitive effect of interictal epileptiform discharges versus nonconvulsive seizures of short duration in a large homogeneous sample. Our previous study evaluated the effects of type of epilepsy and seizure type versus the periictal effect of seizures and epileptiform discharges in 121 children using a multivariate analysis of variance (Aldenkamp & Arends, 2004a). As the occurrence of seizures and the frequency of interictal epileptiform activity were related, the periictal effects of seizures and the effects of interictal epileptiform discharges could have been overestimated (if the presence of both is positive related) or underestimated (if presence is related inversely). Therefore, we analyzed a new cohort of children. In this cohort, children with seizures during testing were distinguished from those without. Our main aim was to further analyze the influence of frequency of interictal activity on cognitive functions. For this reason analysis was done subsequently in those children with epilepsy but without seizures during recording.

Patients and Methods

Subject selection and design

Children, aged 6 years 0 month to 17 years 11 months, were included consecutively in a prospective cross-sectional, standardized, open, nonrandomized, and comparative study during the period 2005–2010. The primary inclusion criteria were (1) equivocal seizures and fluctuations in cognitive performance and/or (2) epileptiform EEG discharges in a recent EEG with a minimal frequency of 1 p/5 min >1 s duration or a minimal frequency of 1 each 30 s for discharges <1 s duration (in earlier EEG recordings).

Excluded were all children with malignant epilepsy syndromes with etiologies that affect cognitive function (such as Lennox-Gastaut syndrome). Children with comorbid diagnoses of Attention Deficit/Hyperactivity Disorder (ADHD) or dyslexia were excluded. Children with a Full-Scale IQ below 70 were not excluded, because a low intelligence test result can be the temporary result of frequent seizures or ongoing epileptiform activity (Mandelbaum & Burack, 1997).

Based on seizure semiology, clinical signs, and EEG recordings, children were divided into subgroups based on the occurrence of subtle epileptic seizures during the 2-h recording (seizure group) and interictal epileptiform EEG discharges without seizures (interictal group). These children were compared with controls. It is important to note that the control group did not consist of a population of school children or best friends, but of a population of children in whom epilepsy was suspected, but this diagnosis was rejected after analysis in our center. Only children without epileptiform discharges were included. These children did not have another neurologic diagnosis and did not have learning difficulties.

Table 1 shows the most important demographic characteristics for the three major groups: seizure group, interictal group, and control group. One hundred eighty-eight children with epilepsy (60 patients in the seizure group and 128 patients in the interictal group) were included as were 41 controls. The mean age was 10 years 0 month for the children with epilepsy (9 years 7 months for the seizure group and 10 years 2 months for the interictal group) and 10 years 6 months for the controls. The gender division was comparable across the groups (Table 1). There was a statistically significant difference between children with epilepsy and controls for Full-Scale IQ. Controls showed a Full-Scale IQ in the normal range (mean 95.4, standard deviation [SD] 15.1), whereas the patients with epilepsy showed a significantly lower IQ: mean 84.7, SD 17.9 (p = 0.001). This difference was attributable to the seizure group (mean 80.2, SD 17.7).

Table 1.   Demographic and clinical characteristics of the study population
 Seizure groupInterictal groupControl groupEpilepsy total
N6012841188
Mean age (years)9.7 (2.6)10.2 (2.8)10.6 (2.6)10.0 (2.8)
Wechsler full-scale IQ80.2 (17.7)87.1 (17.6)95.4 (15.1)84.9 (17.9)
Gender (% boys)525561.053.7

Table 2 shows the clinical characteristics of the children with epilepsy (total and separately for the seizure Group and the interictal Group). Most patients used antiepileptic drugs in monotherapy, with valproic acid and carbamazepine as the most commonly used drugs. As in former studies of our group (Aldenkamp & Arends, 2004a), complex partial seizures with subtle symptomatology were the dominant seizure type recorded during the 2-h recording. Those patients who had seizures had an average of 18.5 seizures during the cognitive testing session, although there was a wide range. Average duration of the seizures was 5.2 s (range 1–19 s). Most patients had epileptiform EEG discharges in 0 to <1% of the recording time. Very frequent epileptiform EEG discharges (>75% of the time) were observed only in a minority of the patients. There was a significant positive relation between frequency of interictal epileptiform discharges and the occurrence of seizures during recording (Spearman’s nonparametric coefficient = 0.617; p ≤ 0.001).

Table 2.   Clinical characteristics of the patients with epilepsy (seizure group versus interictal group)
 Seizure groupInterictal groupTotal epilepsy group
N60128188
Antiepileptic treatment (% patients)   
 No medication28.325.727
 Monotherapy33.348.443.4
 Polytherapy (2 AEDs)21.717.318.5
 Polytherapy (>2 AEDs)16.78.611.1
Most frequently used AEDs in monotherapy (% patients)   
 Valproate6.615.612.7
 Carbamazepine13.312.512.7
 Lamotrigine6.78.67.9
 Oxcarbazepine58.67.4
Epilepsy classification (% patients)   
 Idiopathic generalized1514.814.3
 Idiopathic localization-related1.711.78.5
 Cryptogenic localization-related5554.755.0
 Symptomatic generalized3.31.1
 Symptomatic localization-related2518.820.6
Percentage of time with interictal epileptiform EEG discharges during the 2-h cognitive test session (% patients)   
 No epileptiform EEG discharges13.39.0
 <1% of the time36.751.446.8
 1–9% of the time4018.825.5
 10–49% of the time13.314.113.8
 50–75% of the time6.71.63.2
 >75% of the time3.30.81.6
Seizures occurring during the 2-h cognitive test session (% patients)   
 Complex partial seizures18.0See first column
 Simple partial seizures3.2 
 Absence seizures4.8 
 Myoclonic seizures1.6 
 Atypical absence seizures4.2 
Mean number of seizures in the patients with seizures during the cognitive testing session18.5 (54.5)See first column
Mean duration of seizures in the patients with seizures during the cognitive testing session5.2 s (4.3 s) range 1.0–19 sSee first column

Instruments and set-up

All children were assessed with 32 channel EEG (Brainlab, Feldkirchen, Germany) with synchronized video-recording. Video and EEG were synchronized with the computerized cognitive Fepsy test system (Alpherts & Aldenkamp, 1990; Aldenkamp et al., 1991), using a separate software program that allows synchronization with millisecond precision. Simultaneous video-EEG and cognitive testing were always performed in the morning.

Cognitive tests were selected for their sensitivity to epileptiform EEG discharges and subtle seizures in earlier studies (Aarts et al., 1984; Aldenkamp et al., 1996, 1999a,b, 2001, 2004, 2005; Aldenkamp, 1997; Tromp et al., 2003; Aldenkamp & Arends, 2004a,b) and to cover a broad domain of cognitive function.

Before simultaneous EEG-video-cognitive recordings were performed, global cognitive function was measured using the Wechsler intelligence scale for children (Wechsler 1974). Score is the full-scale IQ with an average score of 100 (SD 15).

The following tests were presented during EEG recording:

  • Language and visual-spatial function:

    •  Vocabulary: subtest of the Wechsler Intelligence Scale for Children, measuring vocabulary (Wechsler, 1974). Score is the standard score compared to norms with an average score of 10.

    •  Block design: subtest of the Wechsler Intelligence Scale for Children, measuring visual-spatial skills (Wechsler, 1974). Score is the standard score compared to norms with an average score of 10.

  • Attention tests:

    •  Simple visual reaction-time measurement (Visual RT): reacting to simple visual stimuli (white square on the screen) that are presented at random intervals by the computer. Score is the average reaction time in milliseconds for the dominant hand (Fepsy, Alpherts & Aldenkamp, 1990).

    •  Simple auditory reaction-time measurement (Auditory RT): reacting to simple auditory stimuli (800-Hz tones) that are presented at random intervals by the computer. Score is the average reaction time in milliseconds for the dominant hand (Fepsy, Alpherts & Aldenkamp, 1990).

    •  Binary Choice Reaction Test (BCRT): The patient has to react differentially to a red square (presented on the left side of the screen) than to a green square (presented on the right side). Score is the reaction time in milliseconds (Fepsy, Alpherts & Aldenkamp, 1990).

  • Speed of central information processing:

    •  The Computerized Visual Searching Task (CVST24), an adaptation of Goldstein’s Visual Searching Task. A centered grid pattern is to be compared with 24 surrounding patterns, one of which is identical to the target pattern. The test consists of 24 trials. The score is the total average searching time in seconds (Fepsy, Alpherts & Aldenkamp, 1990).

  • Educational achievement:

    •  Reading: reading sentences during 1 min. This test is part of a Dutch short screening test for school achievement (De Vos, 1992), similar to the WRAT for English-speaking countries. Score is the number of months delayed compared to standards.

    •  Arithmetic: calculations during 1 min. This test is part of a Dutch short screening test for school achievement (De Vos, 1994), similar to the WRAT for English-speaking countries. Score is the number of months delayed compared to standards.

  • Memory function:

    •  Recognition of words: The test stimuli are presented simultaneously during a learning phase. Six words are presented with a presentation time of 1 s per item. After a delay of 2 s, the screen shows one of these words between distracters. The target item has to be recognized. The score is the number of correct responses out of 24 (Fepsy, Alpherts & Aldenkamp, 1990).

    •  Recognition of figures: Equal test-design for four figures. The score is the number of correct responses out of 24 (Fepsy, Alpherts & Aldenkamp, 1990).

    •  Corsi’s Block tapping test: This is a continuous performance task in which the children have to repeat sequences of blocks that are presented by the computer. The test measures the nonverbal memory span. The score is the total memory span (Fepsy, Alpherts & Aldenkamp, 1990).

This system was connected with an exactly synchronized video-monitoring system, allowing a computer program to relate in time with millisecond precision cognitive performance with EEG and clinical symptoms. Each epoch with epileptic EEG discharges could therefore be linked precisely with the video-recording and the test results during that epoch. The simultaneous EEG/cognitive recording lasted 2 h per patient.

EEG was recorded using the 20-10 system with Brainlab equipment. Spikes and spike-related activity was scored as interictal activity. The localization and frequency of interictal epileptiform activity was determined visually. The occurrence of seizures was recorded, and total number and mean duration of seizures were determined.

Sample size and statistical analysis

The power calculation is based on an estimate of the effect-size obtained for the CVST during cognitive testing in our previous studies (Aldenkamp et al., 1993; Aldenkamp & Arends, 2004a). The CVST has shown epilepsy-related cognitive impairment in a number of studies (Aldenkamp et al., 1996, 1999a,b). The size of the effect for the CVST is calculated as the difference in mean scores between epilepsy and controls divided by the pooled standard deviation: Xi + Xj/∑SDi/i: 27.6 − 21.1/7.4 + 11.5/2 = 0.7 SD. Type-1 (α) and type-2 (ß) errors with an effect-size index of 0.7 SD are entered in the power calculation as primary factors. The statistical power is set to 1 − β = 80%, and the sensitivity to α = 1% to be able to detect also mild effects. Power analysis according to conventions set by Cohen (1977) and Cook and Campbell (1979) sets the required number of subjects at >40. However, given the division of the study group into subgroups (see previous variables and statistical model), this requirement is valid for any subgroup. Using a maximum of four subgroups based on the frequency of epileptiform EEG discharges, a sample size of >160 patients is used.

The data were entered into a database for analysis. ANOVA was used, for comparisons over the subgroups. Only in case of statistical significance over all groups, post hoc comparisons between each group were performed, using t-tests, corrected for multiple testing using the Bonferroni correction. All calculations were performed with SPSS/PC+.

The analysis followed three steps (see Fig. 1). In the first step we compared three groups: the control group with the group of patients with subtle seizures (Seizure group) and the group without seizures (but with interictal epileptiform EEG discharges: interictal group). In the second step, only the patients of the interictal group were included and divided on the basis of frequency (<1 or >1%) of epileptiform EEG discharges. In the third step the group with >1% epileptiform EEG discharges was divided with the cut-off value of < or >10%.

Figure 1.

 Steps in the analysis procedure.

The analysis was based on a division in subgroups rather than a continuous variable as, for most patients only global discrete data were available for the frequency of epileptiform EEG discharges (as presented in Table 2).

Results

In a first step we analyzed the periictal effects of short nonconvulsive seizures. The patients were divided into the group in which no seizures were recorded (interictal group) versus the group of children in whom short subtle seizures occurred during the 2-h recording period (Seizure group). Table 3 provides the data for this comparison.

Table 3.   The periictal effects of subtle seizures on cognition
 Controls
N = 41
Group 2
Epilepsy interictal group
N = 128
Group 3
Epilepsy seizure group
N = 60
Results of statistical testing
  1. Statistical testing: ANOVA for the overall comparison of differences between the three groups. Only in case of statistical significance post hoc comparisons between each group were performed, using t-tests, corrected for multiple testing using the Bonferroni correction. Bold indicates statistical significance.

  2. Differences indicated with < show the groups with fastest/smallest/lowest scores.

  3. aHigher scores signifies a poorer result.

Wechsler full-scale IQ95.4 (15.1)87.1 (17.6)80.2 (17.7)F = 8.865 p ≤ 0.001
Group 3 < 1; 3 < 2; 2 < 1
Vocabulary (standard score with 10 as average score for age)8.7 (3.8)8.0 (3.2)7.3 (3.3)F = 2.193 p = 0.113
Block Design (standard score with 10 as average score for age)9.4 (3.7)8.3 (3.4)7.7 (3.6)F = 3.247 p = 0.041
Group 3 < 1
Visual RT (mean RT in msec)a442 (200)426 (128)484 (227)F = 2.497 p = 0.085
Auditory RT (mean RT in msec)a346 (119)345 (123)393 (153)F = 3.871 p = 0.022
Group 3 < 2
BCRT (mean RT in msec)a453 (160)480(151)520 (154)F = 2.261 p = 0.107
CVST24 (mean RT in s)a21.8 (9.9)26.3 (12.0)35.9 (19.4)F = 11.397 p ≤ 0.001
Group 3 < 1; 3 < 2
Recog Words (n correct out of 24)17.0 (3.8)15.0 (4.8)10.4 (7.2)F = 12.428 p ≤ 0.001
Group 3 < 1; 3 < 2
Recog Figures (n correct out of 24)12.5 (3.9)10.1 (3.3)7.0 (3.8)F = 20.488 p ≤ 0.001
Group 2 < 1; 3 < 1; 3 < 2
Corsi’s block tapping test (memory span)4.5 (1.0)4.4 (1.1)3.3 (1.4)F = 13.747 p ≤ 0.001
Group 3 < 1; 3 < 2
Reading (number of months delayed)a0.49 (8.6)18.1 (16.2)13.5 (14.2)F = 20.468 p ≤ 0.001
Groups 2 < 1; 3 < 1
Arithmetic (number of months delayed)a6.3 (11.2)15.9 (17.5)13.4 (13.5)F = 5.431 p = 0.005
Group 2 < 1

The interictal group showed significantly lower scores compared to controls on 3 of the 12 tests. They performed worse compared to the controls on the educational outcomes: Reading and Arithmetic, and on Recognition of figures. The interictal group did not score significantly lower on any of the tests when compared to the seizure group.

The seizure group shows more statistically significant effects (8 of the 12 tests). On two tests (Block Design and Reading) the results show worse performance only compared to the controls. On six tests there is also worse performance in comparison with the interictal group: full-scale IQ, auditory RT, CVST, recognition of words, recognition of figures, and Corsi’s block-tapping test. Results were only worse compared to the interictal group and not to controls on Auditory RT. In general, the effects of short nonconvulsive seizures are large compared to both controls and the patients without seizures during cognitive testing. The effect on full-scale IQ compared with the children with epilepsy without seizures is almost a half standard deviation. Similarly there is an effect of half a standard deviation for an attentional test: auditory RT, and for the CVST (speed of central information processing) and around 1 standard deviation for three memory tests: recognition of words, recognition of figures, Corsi’s block tapping test. Effects >0.5 standard deviation are considered to represent large and clinically relevant effects (Cohen, 1977).

To control for the frequency of epileptiform EEG discharges without the periictal effect of subtle seizures, the interictal group was divided into two groups. Children who showed epileptiform EEG discharges in <1% of the time (interictal <1% group) versus those who showed interictal epileptiform discharges in >1% of the time (interictal >1% group) (Table 4). None of these children had seizures during cognitive testing.

Table 4.   The effects of interictal epileptiform EEG discharges on cognition
 Interictal <1% group#
N = 79
Interictal >1% group
N = 44
Results of statistical testinga
  1. aStatistical testing: ANOVA for the overall comparison of differences between the two groups. Only in the case of statistical significance post hoc comparisons between each group were performed, using t-tests, corrected for multiple testing using the Bonferroni correction. Bold indicates statistical significance.

  2. bHigher scores signifies a poorer result.

Wechsler full-scale IQ90.7 (17.5)79.2 (15.2)F = 11.182 p = 0.001
Vocabulary (standard score with 10 as average score for age)8.1 (3.5)7.6 (2.8)F = 0.537 p = 0.465
Block Design (standard score with 10 as average score for age)8.9 (3.3)7.0 (3.4)F = 8.183 p = 0.005
Visual RT (mean RT in msec)b409 (110)460 (154)F = 3.553 p = 0.062
Auditory RT (mean RT in msec)b327 (91)380 (163)F = 4.743 p = 0.031
BCRT (mean RT in msec)b472 (141)498 (169)F = 0.634 p = 0.428
CVST24 (mean RT in s)b24.1 (9.9)30.9 (14.8)F = 6.392 p = 0.013
Recog Words (n correct out of 24)15.4 (4.6)13.8 (5.2)F = 1.142 p = 0.289
Recog Figures (n correct out of 24)10.3 (3.4)9.7 (3.1)F = 0.744 p = 0.391
Corsi’s block tapping test (memory span)4.7 (0.9)3.8 (1.1)F = 11.882 p = 0.001
Reading (number of months delayed)b15.1 (14.6)24.2 (17.8)F = 6.651 p = 0.012
Arithmetic (number of months delayed)b14.0 (16.6)20.1 (18.8)F = 2.759 p = 0.101

The interictal >1% group showed statistically significant lower scores compared to the interictal <1% group on 6 of the 12 tests. There is worse performance for full-scale IQ, Block design, Auditory RT, CVST, Reading, and Corsi’s block tapping. For none of the tests was performance of the interictal <1% group worse than the interictal >1% group. The magnitude of these effects was in the range of a half to 0.7 SD for all these tests.

In a last step (Table 5), we also reanalyzed the results for the interictal >1% group. Children with interictal epileptiform EEG discharges in >1% but <10% of the time (interictal 1–10% group) were compared to patients with very frequent epileptiform EEG discharges, defined as >10% of the recording time (interictal >10% group). The comparisons showed no statistically significant differences between the two groups.

Table 5.   The effect of frequency of interictal epileptiform EEG discharges on cognition
 Interictal 1–10% group
N = 23
Interictal >10% group
N = 21
Results of statistical testing (not including controls)a
  1. ns, not significant.

  2. aStatistical testing: ANOVA for the overall comparison of differences between the 4 groups. Only in case of statistical significance post-hoc comparisons between each group were performed, using t-tests, corrected for multiple testing using the Bonferroni correction.

  3. bHigher scores signifies a poorer result.

Wechsler full-scale IQ77.0 (14.3)82.4 (16.3)T = −1.038; p = 0.30
Vocabulary (standard score with 10 as average score for age)7.4 (2.9)8.1 (2.6)T = −792; p = 0.43
Block Design (standard score with 10 as average score for age)6.5 (3.6)7.8 (3.1)T = −1.118; p = 0.25
Visual RT (mean RT in msec)b452 (179)463 (121)T = −0.117; p = 0.91
Auditory RT (mean RT in msec)b376 (137)384 (194)T = −152; p = 0.88
BCRT (mean RT in msec)b499 (140)496 (202)T = 0.058; p = 0.95
CVST24 (mean RT in s)33.2 (13.5)27.1 (16.0)T = 1.289; p = 0.21
Recog Words (n correct out of 24)13.6 (4.6)13.9 (6.3)T = −0.050; p = 0.96
Recog Figures (n correct out of 24)9.5 (3.3)10.0 (3.0)T = −356; p = 0.73
Corsi’s block tapping test (memory span)3.9 (1.1)3.8 (1.3)T = −0.105; p = 0.92
Reading (number of months delayed)b27.3 (19.1)19.4 (15.0)T = 1.160; p = 0.26
Arithmetic (number of months delayed)b23.4 (21.2)14.1 (12.3)T = 1.268; p = 0.22

Discussion

Our study showed that short nonconvulsive seizures had an effect on global cognitive functioning (Wechsler full-scale IQ), speed of central information processing (CVST 24), and short-term memory function (Recognition of words, Recognition of figures, Corsi’s block tapping). The magnitude of the effect is large: in the range of 0.5 SD for global cognitive function, attention, and speed of central information processing. The effects are largest for short-term memory (1 SD). According to Cohen’s convention, effects exceeding 0.5 SD can be labeled as large and clinically relevant effects (Cohen, 1977).

Numerous studies have analyzed cognitive function in children with epilepsy. The effect on intelligence may well be secondary, a consequence when memory and information processing impairments persist over time (Aldenkamp & Arends, 2004a). The average full-scale IQ in our epilepsy study group was 84.3. This relatively low value is in line with results of a recent study describing performance on the WISC III in children with epilepsy (O’Leary et al., 2006). The effect of short nonconvulsive seizures on central information processing speed has been demonstrated in several studies (Espie et al., 1999). Especially children with generalized nonconvulsive seizures show low cognitive scores on such tasks (Mandelbaum & Murack, 1997). In children with a high frequency of nonconvulsive seizures and complete seizure remission after introduction of medication, the effects of such seizures can best be demonstrated. The epilepsy syndrome that theoretically fits these criteria best is childhood absence epilepsy. In childhood absence epilepsy, short seizures predominate. Besides, in childhood absence epilepsy the success rate of treatment with antiepileptic drugs is high. Sirén et al. (2007) treated 11 children with newly diagnosed typical absences with generalized 3-Hz spike-wave discharges. Remission of absence seizures did not affect general intelligence, but memory tasks improved significantly. These results are in line with Henkin et al. (2005) describing attention, verbal learning and memory, word fluency, and fine motor deficits in children with childhood absence epilepsy. In this study 12 children with absence seizures were included; all were using valproic acid, but only half were seizure free. Other studies describe significant deficits on global functioning in children with absence seizures (Pavone et al., 2001).

In children without seizures during cognitive testing, the occurrence of frequent epileptiform discharges had an additional effect on cognitive function in our study. We initially divided spike frequency into two categories; <1% versus >1%. Significant differences were noted on global cognitive functioning (full-scale IQ), visual-spatial functioning (Block design), an attention test (Auditory RT), speed of information processing (CVST24), memory function (Corsi’s block tapping test), and educational achievement (reading). In all cases the results were lower with higher spike frequency. The type of effect is similar to the effect of short nonconvulsive seizures, although the impact is less pronounced in number of areas involved and severity (ranging from 0.50.7 SD), in line with our previous findings (Aldenkamp & Arends, 2004a). The effect of epileptiform EEG discharges is therefore additional and more subtle than the effect of seizures. Important to note is that our study showed that the effects of epileptiform EEG activity can be characterized as a negative effect, independent of the occurrence of seizures.

There seems to be a threshold for the cognitive effects of epileptiform EEG discharges. When the group with epileptiform EEG discharges 1–10% of the time is compared to those with epileptiform EEG discharges >10% of the time, no statistically significant differences occur. Apparently the threshold for cognitive effects can be found between 1% and 10% of the time and functioning does not deteriorate further with a higher frequency of epileptiform EEG discharges. Of course in the extreme severe end of the continuum (ESES and Landau-Kleffner syndrome) we do see further deterioration.

It is unclear why epileptiform EEG discharges affect cognitive functioning, but apparently some of these discharges, although focal in the majority of our patients, can disrupt central information processing similar to subtle seizures.

The phenomenon of transient cognitive impairment (TCI) was first described by Aarts et al. (1984). TCI can be demonstrated in about 50% of patients who show epileptiform activity during testing (Binnie, 2003). Although of ultimate theoretical interest, the question of whether TCI influences global functioning and school performance remains unsolved. This suggests that the demonstration of TCI in an individual child should not automatically lead to initiation of antiepileptic medication (Binnie, 2003). When performing combined EEG/cognitive tests in order to demonstrate TCI, it is important to recognize the phenomenon of cognitive EEG activation or suppression (Matsuoka et al., 2000). Cognitive tasks can provoke epileptic discharges in idiopathic generalized epilepsies, especially in epilepsy syndromes characterized by myoclonic seizures (Matsuoka et al., 2000). An inhibitory effect on discharges is even more common. Sixty-four percent of 208 patients (children and adults) showed inhibition of epileptiform discharges in a large series by Matsuoka et al. (2000). This problem can be tackled by recording EEG and cognitive data at different occasions (Koop et al., 2005).

Disclosure

None of the authors has any conflict of interest to disclose. We confirm that we have read the Journal’s position on issues involved in ethical publication and affirm that this report is consistent with those guidelines.

Ancillary