Cognitive Function in Preschool Children after Epilepsy Surgery: Rationale for Early Intervention


Address correspondence and reprint requests to Dr. I. Tuxhorn at Bethel Epilepsy Center, Maraweg 17-21, D-33617 Bielefeld, Germany. E-mail:


Summary: Purpose: The detrimental effect of frequent early seizures on the cognitive potential of children is a significant clinical issue. Epilepsy surgery in childhood offers a good prognosis for seizure control and improved developmental outcome. We studied the postoperative outcome and the developmental velocity after surgery and analyzed risk factors for developmental delay in 50 consecutive preschool children treated surgically for severe epilepsy at ages 3 to 7 years.

Methods: Pre- and postoperative developmental quotients (DQs) were analyzed with analysis of variance; stepwise linear regressions were performed on preoperative DQs and on a difference score between post- and preoperative DQs to determine risk factors for preoperative development and factors influencing postoperative development.

Results: Of the 50 patients, 70% were retarded, with IQ < 70; 16% were of average intelligence, with IQ ranging from 85 to 115. Age at seizure onset and extent of lesion were predictive variables for preoperative cognitive development. Six to 12 months after surgery (early postoperative phase), 66% were seizure free (Engel outcome class I), 26% had substantial to worthwhile seizure reduction (classes II and III), and 8% were unchanged (class IV). Forty-one (82%) children showed stable velocity of development; three children showed gains of ≥15 IQ points; three had developmental decline (loss of ≥10 IQ points), which was transient in two children; and three children moved from not assessable to assessable. At last follow-up (6 months to 10 years after surgery), 11 children showed IQ/DQ gains of ≥15 IQ points. Gains in IQ were observed only in seizure-free children and were stable over time. Shorter duration of epilepsy was significantly associated with a postoperative increase in DQ.

Conclusions: (a) Substantial global mental delay is common in young children treated for epilepsy with surgery; (b) In most patients, postoperative development proceeded at a stable velocity; (c) Catch-up development may occur but only in seizure-free patients; (d) Substantial cognitive losses were noted in only one child. and (e) Early seizure control stabilized developmental velocity in this patient cohort.

Epilepsy is a chronic condition with a high risk of co-morbidity. Children with refractory epilepsy are at considerable risk for cognitive impairment (1–8) as well as school failure (9–12), behavior and mental health problems (6,13–19), and overall compromised quality of life (20,21). Risk factors for comorbidity discussed in the literature are numerous and comprise the mere presence of seizures (19,22,23), seizure-related variables such as age at onset (24–28), frequency and/or severity (10,29), duration (27,30), as well as underlying pathology (31,32), antiepileptic medication (33–36), and various psychosocial factors, often referred to as “burden of epilepsy” (37).

Epilepsy surgery has an excellent prognosis in children with well-defined foci with regard to seizure control, and the rate of seizure freedom achieved is similar to that reported in the adult population (38). A further goal of surgery in children is to improve the overall cognitive development by reducing the cognitive and psychosocial sequelae of chronic epilepsy.

Outcome studies of pediatric epilepsy surgery in young children have reported dramatic improvement in postoperative cognitive development (39), although quantitative pre- and postoperative neuropsychological assessments were not always performed. Studies reporting on the cognitive outcome of children after surgery have included children with catastrophic epilepsy with infantile spasms (31,48), children after hemispherotomy (49,50), and school-age children with temporal lobe epilepsy (40–47). The cognitive outcome after surgery has been studied in various patient groups with differing epilepsy syndromes and various age ranges from infancy and early childhood to late adolescence. Some limitations of the studies include small sample size, the use of a wide range of intelligence tests, no comparative pre- and postoperative data, and variable follow-up periods (51–53).

The aim of the present study was to investigate the preoperative cognitive function and postoperative cognitive outcome of a cohort of preschool children between ages 3 and 7 years at the time of surgery and to determine predictive factors influencing cognitive development. In particular, we were interested in determining the risk for decline in development after surgery, as well as the chance for improved postoperative development compared with the preoperative developmental baseline. Evidence for postoperative catch-up of development would provide a strong rationale for early timing of surgical intervention.



From 1990 to 2001, 58 children between ages 3 and 7 years underwent resective surgery for intractable symptomatic focal epilepsy at the Bethel Epilepsy Center. Presurgical evaluation included continuous scalp video-EEG monitoring, high-resolution magnetic resonance imaging (MRI), and neuropsychological assessment. Eight children were excluded from the study because of missing neuropsychological data. The children in this study had at least one follow-up at 6 to 12 months after surgery, including neurologic examination, EEG, MRI, and neuropsychological assessment. Forty children were followed up for ≥2–3 years.

Mean age at surgery was 4.99 years (SD, 1.0); age at seizure onset ranged from 1 to 52 months (mean, 14 months; SD, 1.14); and mean duration of epilepsy before surgery was 3.77 years (SD, 1.41) (see Table 1 for patient characteristics). There were 25 girls and 25 boys. Preoperative seizure frequency was less than daily seizures in eight children, at least one seizure per day in 31 children, and 11 children had more than 20 seizures per day.

Table 1. Patient characteristics

Etiology (no.)
Site and side of
resection (no.; l/r)
Age at onset
Mean (SD)
Age at surgery
Mean (SD)
Seizure outcome
Engel class: no.
  1. Age is expressed in years.

  2. FCD, focal cortical dysplasia; no., patient numbers; l, left; r, right.

FCD (22)Frontal (6; 1/5)0.7 (1.1)5.2 (1.3)I: 4; II & III: 2
 Temporal (2; 1/1)0.7 (0.5)5.75 (1.2) I: 2
 Multilob/hemi (14; 6/8)0.9 (1.0)4.75 (0.9) I: 9; II & III: 4; IV: 1
Tumors (13)Temporal (13; 9/4)1.8 (1.2)4.7 (0.9)I: 13
Encephalitis (5)Frontal (1; 0/1)3.1 (−) 4.8 (−) IV: 1
 Multilob/hemi (4; 1/3)2.0 (1.7)4.7 (0.7)I: 1; II & III: 3
Hypoxic–ischemic damage (5)Multilob/hemi (5; 4/1)1.2 (0.9)5.9 (1.5)I: 1; II & III: 2; IV: 2
Tuberous sclerosis (3)Frontal (2; 1/1)0.8 (−) 5.5 (0.0)I: 2
 Multilob/hemi (1; 0/1)0.3 (−) 6.5 (−) II & III: 1
Hippocampal sclerosis plus microdysgenesis (1)Temporal (1; 0/1)0.25 (−)  3.6 (−) I: 1
Hemimegalencephaly (1)Multilob/hemi (1; 1/0) 0.9 (−) 4.9 (−) II & III: 1

Neuropsychological assessment

Cognitive development was assessed pre- and postoperatively by a developmental neuropsychologist. Tests used for preoperative and first postoperative assessment were the Bayley Scales of Infant Development I and II (BSID I; II), the McCarthy Scales of Children's Abilities (MSCA), the Kaufman Assessment Battery for Children, and the German version of the Wechsler Preschool and Primary Scales of Intelligence (HAWIWA). At the second follow-up 2–3 years after surgery, two children who had “outgrown” the age range of the MSCA were administered the German version of the Wechsler Intelligence Scales for Children–Revised (HAWIK-R).

Statistical analysis

Cognitive ability and developmental status of most of the children in our sample could not be quantified through standard scores in developmental tests (which require an IQ of ≥50). For purposes of sample description, preoperative cognitive levels were determined. Children with IQs ≥50 were grouped according to standard deviations from the mean, and ICD-10 criteria for degree of mental retardation were used for children who scored below an IQ of 50. Statistical analyses were restricted to children for whom “developmental quotients” (DQ, developmental age/chronologic age × 100) were available in either the BSID or the MSCA (n = 34 for analysis of risk factors for preoperative development; n = 21 children for analysis of factors influencing postoperative development over a 2- to 3-year period). To analyze postoperative changes, a difference score between post- and preoperative DQs was obtained. Data were analyzed in the following ways: Comparison of pre- and postoperative DQs was made by using analysis of variance. Linear regression analyses were performed to determine predictors of preoperative and changes in postoperative cognitive function. Data were analyzed with SPSS 11.0.


Surgical procedures and histology

Lesionectomies were confined to the temporal lobe in 16 and to the frontal lobe in nine children. Eighteen children had multilobar resections; seven underwent a hemispherotomy. Children with multilobar resections and hemispherotomies were similar in outcome parameters and therefore combined for further analysis. No significant difference was found in age at seizure onset or age at surgery between these three groups (F= 0.912, p > 0.4; and F= 0.677, p > 0.5); 44% of the children had focal cortical dysplasia (FCD) on histology, and 26% had tumors. Remaining etiologies (see Table 1) were combined for further group analysis. Univariate analysis of variance showed that children with tumors had later seizure onset than did children with FCDs but did not differ significantly from children with other etiologies (F= 3.325, p < 0.5). No difference was noted in age at surgery between these three groups (F= 1.030, p > 0.3).

Seizure outcome

At first postsurgical assessment 6–12 months after surgery, 66% of the children in the cohort were seizure free (Engel outcome class I), 26% had substantial seizure reduction (Engel classes II and III), and four (8%) children were unchanged (Engel class IV; see Table 1).

Two to three years after surgery (follow-up data available for 40 children), 29 (70%) were seizure free, and five (12.5%) had substantial seizure reduction at first and second postoperative assessment. Two initially seizure-free children had recurrence of seizures before their 2-year follow-up. Two children with early substantial seizure reduction had worsening of seizures. One child with early substantial seizure reduction was seizure free at the 2-year follow-up. Seizure outcome was most favorable in the less retarded children (IQ ≥ 70), with 80% in Engel class I; of the severely retarded children, almost half became seizure free (Table 2).

Table 2. Seizure outcome and preoperative cognitive level

IQ ≥70
Seizure outcome (no.)
 Engel class I (33)615 12 
 Engel class II & III (13)562
 Engel class IV (4)211

Preoperative cognitive function

Figure 1 shows the distribution of preoperative cognitive levels of the children in this study. Sixteen percent of the children were of average intelligence. More than half of the children had a cognitive level that could not be quantified by an IQ score, and thus were at least moderately mentally retarded. The superimposed gaussian curve of normal distribution for intelligence in the population (mean IQ of 100 and standard deviation of ±15 IQ points; average intelligence ranges from IQs 85 to 115) demonstrates the shift to the left of the distribution of cognitive function in the studied patients.

Figure 1.

Preoperative cognitive function compared to normal distribution.

Stepwise linear regression analysis indicated that preoperative cognitive functioning (DQ) was significantly predicted by a combination of age at seizure onset and extent of lesion [frontal/temporal vs. multilobar/hemispheric (F= 6.298, p < 0.01)], accounting for almost 25% of the variance in preoperative DQ (adjusted R2= 0.243). For other independent variables analyzed, see Table 3.

Table 3. Preoperative cognitive levels in relation to epilepsy-specific variables

IQ ≥70
  1. FCD, focal cortical dysplasia; no., patient numbers; (a), entered in the regression as linear variable.

  2. aPredictors of preoperative cognitive functioning, F = 6.298, p < 0.01.

  3. bPredictor of postoperative cognitive change scores, F = 6.36, p < 0.03.

Age at seizure onset (no.) (a) a
 0 –12 mo (30)916 5
 >12 mo (20)4610 
Site of resection (no.) a
 Frontal (9)225
 Temporal (16)277
 Mulitlob./hemi (25)913 3
Laterality (no.)
 Left (24)413 7
 Right (26)998
Etiology (no.)
 FCD (22)611 5
 Tumors (13)247
 Other (15)573
Seizure frequency (no.)
 Less than daily (8)233
 Daily (31)812 11 
 >20/day (11)371
Duration (no.) (a)b
 <2 yr (7)133
 2–<4 yr (22)589
 ≥4 yr (21)711 3

Table 3 shows that the risk of being mentally retarded was lower for children with onset of seizures after their first birthday versus those with onset within the first 12 months of life, and for children with resections confined to the frontal or temporal lobe versus those with multilobar resections or hemispherotomy.

Postoperative changes in cognitive development

In the early postoperative phase (6 to 12 months after surgery), 41 (82%) children of our sample showed unchanged cognitive functioning compared with preoperative assessment (gains of <15 and losses of <10 DQ/IQ points). Three children who had been too retarded to be assessed before surgery obtained a DQ in the MSCA or BSID at their first follow-up. A gain of ≥15 points in IQ was observed in three seizure-free children; another three seizure-free children had a loss of ≥10 IQ points.

Two to 3 years after surgery (data available for 40 children), 29 children continued to perform at their preoperative levels (gains of <15 and losses of <10 IQ/DQ points); two of these children had a transient decline in the early postoperative phase. One child with early decline continued at the low end of the average range and was recently diagnosed with severe dyslexia. Eight children showed gains of ≥15 points in IQ or DQ over a 2- to 3-year period. Two children whose early postoperative development had been stable were assessed with a different intelligence test, which yielded a substantially lower IQ at the 2-year follow-up. This loss was transient in one child, so that of 40 children, long-term postoperative development was stable or improved in 38 children, whereas two children had a substantial long-term decline. Clinical characteristics of children with substantial gains or losses in early and long-term postoperative development are listed in Table 4.

Table 4. Characteristics of children with gains/losses in DQ/IQ
SexAAO/AASLocal./etiol.Seizure DQ/IQ outcomeDQ/IQ
pre½—1 yr post2–3 yr postLast follow- up (yr)
  1. Age is expressed in years;months.

  2. AAO, age at seizure onset; AAS, age at surgery; local site of resection; l left; r right; hemi, hemispheric; fp, frontoparietal; t, temporal; f, frontal; to, temporooccipital; FCD, focal cortical dysplasia; n.a., not assessable;

  3. aChange in IQ test accounted for by using two tests at previous assessment, with ≤5 points difference in IQs

  4. bChange in IQ test, not accounted for by double testing.

Gains of ≥15 DQ/IQ points
 m1;9/5;3r tpo FCDClass I DQn.a.58 
 f2;9/3;8l t tumorClass I DQn.a.324152 (4)
 f0;3/6;0r fp FCDClass III IQn.a.<50  6961 (5)a
 m0;6/3;11r tpo enceph.Class I IQ<50  79 
 f3;9/5;6l t tumorClass I IQ87105 116  
 f2;0/4;0r t tumorClass I IQ7388112  
 m2;6/6;9l hemi strokeClass I DQ385254 
 f0;3/3;6r t FCDClass I IQ<50  <50  6577 (10) b
 f0;11/3;2r f FCDClass I IQ798498 
 f0;4/4;11l t FCDClass I IQ<50  <50  6465 (6) b
Losses of ≥10 DQ/IQ points
 f0;11/6;6r t tumorClass I IQ100 101b89b88 (7)
 m2;10/6;3l f FCDClass II; class I IQ
6 yr after surgery
100 108 86b97 (7)
 f0;10/5;8r f TSClass II IQ938188 
 m1;6/4;9r t tumorClass I IQ61<50  66 
 m4;4/4;10l t tumorClass I IQ108 9095 

A repeated-measures analysis of variance revealed a significant postoperative increase in DQ (F= 3.863, p < 0.05), within-subjects contrasts showed that this was due to increases from preoperative to second, but not preoperative to first postoperative assessment (F= 6.675, p < 0.02; and F= 0.457, p > 0.4).

Stepwise linear regression analysis showed that duration of epilepsy was the only significant predictor of long-term cognitive change: children with shorter intervals between onset of epilepsy and surgery had greater gains in DQ (F= 6.36, p < 0.03; see Fig. 2). Cognitive gains were almost exclusively observed in seizure-free children, but so were cognitive losses. Seizure outcome did not significantly contribute to the prediction of postoperative cognitive change.

Figure 2.

Relation between gains/losses in DQ and duration of epilepsy.


Cognitive impairment in patients with epilepsy was recognized and documented in the early nineteenth century medical literature. A number of potential factors including etiology, brain injury before epilepsy onset, heredity, medication, psychological handicaps, and multiple factors relating to the epilepsy itself may affect cognitive performance in epilepsy patients. The degree of continuous or transitory neuronal dysfunction associated with epilepsy may be a critical factor contributing to the developmental potential in children with epilepsy. A number of studies in adult and pediatric patients suggest that cognitive deterioration may take place in a subset of patients who have early onset and long duration of epilepsy, suggesting a window of vulnerability for irreversible decline of cognitive potential (2–5,24–28,30). Early surgical control of seizures may therefore have a marked impact on the developmental potential of children with early-onset, severe epilepsy. The immediate and long-term developmental changes after epilepsy surgery have not been well studied in young children in a quantitative fashion.

In this retrospective and prospective study, we analyzed the presurgical cognitive function and serial developmental outcome after epilepsy surgery in an unselected sample of 50 consecutive preschool children aged between 3 and 7 years at the time of surgery.

Cognitive development was delayed in 84% of the cohort and was in the normal range in only 16%. Eight percent were profoundly retarded, 18% severely, 30% moderately retarded, and 28% in the mildly subnormal range.

These findings confirm the high incidence of developmental delay in early-onset epilepsy and the deleterious effects of early seizure onset on cognitive development, as has been reported by other authors in the literature (22,24,28,32,39). In the analysis of epilepsy-specific variables as risk factors for retardation, children with early seizure onset had a significantly higher risk of mental retardation than did those with later seizure onset.

Children with larger lesions involving more than one lobe were more likely to have preoperative cognitive levels in the retarded range than were those in whom lesions were confined to the frontal or temporal lobe. This result is in keeping with the results reported by other authors (28).

The seizure-free outcome in 66% and substantial seizure reduction in 26% of our cohort compares with that reported in other case series in the literature (38,39,53,54).

Our data show that a good seizure outcome after surgery is not restricted to relatively high-functioning children. Even among the severely and profoundly retarded children, complete seizure control was achieved in 46%, whereas another 39% experienced substantial or worthwhile seizure reduction. Profound mental retardation should therefore not be an exclusion criterion for epilepsy surgery selection.

Over a follow-up of 2–3 years, 29 (72.5%) of 40 children had stable developmental velocity relative to preoperative level. Eight (20%) children showed substantial gains of ≥15 IQ/DQ points. One child showed a substantial decline after resection of a left anterior temporal tumor that did not improve over a 3-year period. He declined from high average to average range and has since been diagnosed with severe dyslexia. Two children with stable development in the early phase did achieve substantially lower IQs at second follow-up, but they were assessed with a different test because they had “outgrown” the test used at first follow-up. When tested 5 years later with the same test, one child again scored within preoperative range, and the other remained at low-average level.

Our data suggest that developmental gains may accumulate over a longer period and do not necessarily become evident in the early postoperative months. This is supported by statistical analysis of changes in DQ in a subset of children: comparison of pre- and postoperative DQs revealed a nonsignificant increase in DQ in the early phase but a statistically significant long-term increase. Although parents often are enthusiastic about their children's early postoperative development, this is not always confirmed by neuropsychological assessment. The same observation has been reported by others (54). Conversely, we sometimes observe that young children whose extremely limited attention capacities rendered preoperative assessment impossible can undergo standardized testing at their first follow-up. We suggest that in the absence of seizures, children's attention capacities increase so that they can benefit more from environmental input and engage more in social interaction. Parents are likely to perceive this altered behavior as improved development, whereas improvement may become evident in neuropsychological assessment only after a longer period. A similar observation has been reported in a sample of school-age children after temporal lobe resections, where a longer time to follow-up predicted gains in performance IQ (46). Recent research on the cognitive effects of seizures suggests that reduced attention skills can have a cumulative negative impact on intelligence and school performance (7,8,12).

Preoperative cognitive function was not a significant predictor for postoperative cognitive change, but as Table 4 reveals, most of the children with substantial gains had rather low preoperative IQs/DQs. Postoperative developmental gains might therefore indicate restart of development after a period of developmental stagnation or regression due to severe epilepsy. Developmental decline, conversely, was only observed in a relatively high-functioning child. Only few children were without developmental delay in our sample, so that conclusions drawn from this study are applicable mainly to children functioning below average levels.

A study of the early intellectual development in hemiplegic children has clearly shown that the presence of seizures has a substantial and markedly depressing effect on intellectual functioning (4). Some studies suggest that seizure-related cognitive deterioration may occur early after epilepsy onset in a cascadic fashion (5). The question of whether earlier surgery may prevent additional developmental handicap can be answered only indirectly from our data. Nevertheless, a shorter duration of epilepsy was the one predictive factor for postoperative long-term gains in our study. Timing of surgery may therefore be critical for long-term cognitive outcome, especially as postoperative development, as our data suggest, may stabilize after seizure remission.

Further prospective longitudinal studies on well-defined patient populations (with regard to etiology and epilepsy-related variables) who are treated surgically at different times will be needed to clarify these multifactorially determined issues.


Acknowledgment:  We dedicate this article to the memory of our colleague Dr. Anke Moch (deceased, May 2003), first child psychologist in the pediatric epilepsy surgery unit Bethel, who inspired us and laid the foundation for the long-term follow-up study of our surgical patients.

This work was partially supported by a grant from the German Ministry of Health. We thank Drs. H. Holthausen, T. Pieper, S. Kloss, R. Schulz, A. Ebner, H. Pannek, and A. Karlmeier for their assistance in supplying some of the clinical patient data.