Article Text

Download PDFPDF

Long term course of childhood epilepsy following relapse after antiepileptic drug withdrawal
  1. P A D Bouma1,
  2. A C B Peters2,
  3. O F Brouwer3
  1. 1Department of Neurology, Division of Child Neurology, Leiden University Medical Center, The Netherlands
  2. 2Department of Child Neurology, University Medical Center Utrecht and Wilhelmina Children's Hospital, The Netherlands
  3. 3Department of Neurology, Division of Child Neurology, Groningen University Hospital, The Netherlands
  1. Correspondence to:
 Professor A C B Peters, Department of Child Neurology, University Medical Center Utrecht and Wilhelmina Children's Hospital, The Netherlands;


Objective: To explore the course of epilepsy following relapse after antiepileptic drug (AED) withdrawal.

Methods: Forty two patients were identified with onset of epilepsy in childhood in whom AEDs had been withdrawn after at least 2 years of seizure freedom, and in whom a relapse had occurred. Two patients were lost to follow up.

Results: Median follow up after AED withdrawal was 5.9 years (range 1.6–13.2 years). Relapse occurred in more than half of the patients within 6 months of AED withdrawal. At the end of follow up, 12 patients (30%) were seizure free for at least 1 year (mean 10.4 years) without medication; 16 (40%) were seizure free for at least 1 year (mean 5.3 years) with ongoing medication; and 12 patients (30%) were seizure free for less than 1 year with medication. No status epilepticus occurred in any patient after withdrawal. Age at onset, if over the age of 5, combined with normal intelligence were predictive of an excellent outcome; presence of a neurological disorder, and hence symptomatic aetiology, was predictive of poor outcome after a relapse.

Conclusions: Fears that premature withdrawal of AEDs might result in uncontrollable seizures were unsubstantiated in this study. The current practice of withdrawing AEDs in children who have been seizure free for 2 years can be benificial to most of these patients.

  • epilepsy
  • antiepileptic drugs
  • withdrawal
  • outcome
  • AEDs, antiepileptic drugs

Statistics from

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

It is difficult to give a precise prognosis for individual children with epilepsy at an early stage of their disease. Once antiepileptic drugs (AEDs) are started, many patients will become seizure free, and after a period the AEDs can be withdrawn with relatively good results: more than 70% of the children in whom AED treatment is withdrawn will remain seizure free.1 A young age at onset of seizures, idiopathic aetiology without neurological disorder, and absence of EEG abnormalities are associated with a lower relative risk of relapse after discontinuation, but do not guarantee lasting seizure freedom. Withdrawal of AEDs, therefore, always carries a certain risk of recurrence of seizures in an individual child, or a roughly 30% risk of relapse in a group of patients. Nevertheless, the risk of prolonged use of AEDs, due to the occurrence of cognitive, psychosocial, and physical side effects,2–5 is sufficient reason to withdraw AEDs when possible.

The risks and consequences of a relapse after AED withdrawal, after years of seizure freedom, are not yet known, as such a relapse has been the end point in most discontinuation studies. Questions concerning the risks of seizure recurrence, such as a possible evolution to refractory epilepsy after relapse, the psychosocial implications for children and their parents, the decision whether or not AED treatment should be re-started, and the effectiveness of renewed AED treatment, have remained unanswered.

This study explores the long term course of epilepsy following a relapse after AED withdrawal in the childhood period.


Outpatient records of children treated for their epilepsy at the Department of Neurology, Division of Child Neurology, Leiden University Hospital, were screened for the occurrence of a relapse after AED withdrawal between October 1975 and March 1987. During this period, 393 children with epilepsy had been systematically treated in a prospective fashion, and AEDs were withdrawn after informed consent had been obtained in 198 children after a seizure free period of at least 2 years. Methods and results have been published previously.6 All children had been referred directly by general practitioners, and all had had at least two afebrile seizures, three atypical febrile seizures and one afebrile seizure, or status epilepticus. The age at onset of seizures was 1 month to 14 years. This cohort yielded 42 patients who had had a relapse after AED withdrawal: 26 children stem from the cohort of 116 described earlier.6 An additional group of 82 children, who had been seizure free for 2 to 2.5 years on AED treatment, and who were between 14 and 16 years of age when AEDs were withdrawn, were not included in the original study. This group was also followed up after the occurrence of a relapse after AED withdrawal: this cohort yielded 16 children (fig 1).

Figure 1

Flow chart showing follow up after withdrawal of AEDs followed by relapse.

Follow up of this study ended in November 1995. Relevant clinical information was obtained from the records, from a questionnaire completed by the patients or their parents, and from a structured telephone interview.

Epilepsy was classified according to the 1989 classification of the International League against Epilepsy (ILAE).7 Clinical data were obtained concerning the presence of neurological abnormalities (present on physical examination, based on relevant history, and on neuroradiological imaging when indicated), presence of prior febrile convulsions, presence of mental retardation (objectively ascertained behavioural disturbances or IQ less than 70 determined by neuropsychological testing), family history of seizures, type and duration of seizures before AED withdrawal, type and duration of AED treatment before withdrawal; type and duration of seizures after relapse, and type and duration of AED treatment after relapse. In the case of more than one attempt to withdraw AEDs, the number of withdrawal attempts, seizure outcome, and treatment characteristics at the end of the follow up period were ascertained.

Electroencephalograms obtained less than 2 months before the start of AED treatment, before AED withdrawal, after a relapse, and the last EEG before the end of follow up, were scored by one of the authors (PB) on a semiquantitative scale for the presence of spike and wave complexes (absent, present); focal or generalised slowing and background abnormalities (absent, slight, moderate, severe); and focal or generalised sharp abnormalities (absent, sporadic, paroxysmal, with or without clinical features).

Outcome was classified according to the duration of the terminal remission (modified from Arts et al8): excellent (at least 1 year seizure free at the end of follow up without medication), good (at least 1 year seizure free at the end of follow up, with medication), and poor (less than 1 year seizure free at the end of follow up, with continued AED treatment). Statistical analysis was performed comparing the three outcome groups.

Data entry and statistical analysis were carried out with the SPSS package.9 Analysis of variance (ANOVA), Spearman's rank correlation test, and χ2 test were used when appropriate. When variables were time dependent and intervals varied in duration, results were analysed with Kaplan-Meier statistics and displayed as actuarial survival curves, and the log rank test was used to assess differences. Multivariate logistic regression analysis was performed accepting covariates by the backward conditional method.


A total of 42 patients were identified who had had a relapse after AED withdrawal.

This study cohort comprised patients originating from two groups, as stated, yielding 24 and 16 patients. Testing for differences between these two groups yielded no significant differences for sex, epilepsy type, seizure type, concomitant neurological disorder, mental retardation, or outcome (χ2), nor for age at onset of seizures, duration of active epilepsy, numbers of seizures before AED withdrawal and after relapse, or follow up from onset of seizures (ANOVA).

Two patients were lost to follow up because of unknown residence. The median follow up after diagnosis in these 40 patients was 15.6 years (range 5.2–27.0 years). Median follow up after AED withdrawal was 5.9 years (range 1.6–13.2 years). The median interval between AED withdrawal and relapse was 6.0 months (range 2 weeks-72 months) (fig 2). All but four patients restarted AEDs: they had all experienced one isolated relapse, and since that time had remained in remission (table 1).

Table 1

Clinical characteristics of patients according to outcome after relapse from AED withdrawal

Figure 2

Survival curve indicating the interval between AED withdrawal and relapse. Half of the patients had had a relapse within 1 year after AED withdrawal; all patients had had a relapse within 6 years.

Outcome was excellent in 12 patients (30.0%). This group reached terminal remission at a median age of 17.0 years (range 10.9–25.7 years). Outcome was good in 16 patients (40.0%), who reached terminal remission with medication at a median age of 16.3 years (range 5.1–24.9 years). Poor outcome was seen in 12 (30.0%) patients. Hence, 70% (28/40) of the patients attained renewed and lasting seizure freedom after relapse, half of them within 4.5 years (fig 3). All patients with excellent outcome had terminal remissions lasting for more than 5 years (table 1). Terminal remission in the group with good outcome lasted for more than 2 years in 13 (81.5%), and for more than 5 years in seven (43.8%) of the 16 patients.

Figure 3

Survival curve indicating the interval from relapse to terminal remission. After 12 years 28 patients had attained lasting seizure freedom (70% of study population); half of them within 4.5 years.

Onset of seizures after the age of 5 years, absence of concomitant neurological disorder, and short interval between AED withdrawal and relapse were associated with excellent outcome (log rank test, table 1). Testing for discriminative power yielded no significant difference in outcome between patients who relapsed within 1 year after AED withdrawal compared with patients who relapsed after more than 1 year (χ2).

The influence of mental retardation on outcome did not reach significance, but retardation combined with the presence of concomitant neurological disorder correlated significantly with poor outcome (p=0.015, χ2). None of the five children with this combination became seizure free during follow up: one had generalised tonic-clonic seizures, one complex partial seizures, and three secondary generalised seizures; all were classified aetiologically as having symptomatic epilepsy.

Type of epilepsy according to the ILAE classification had no influence on outcome after relapse (table 1, χ2). Although the figures suggest a somewhat better outcome in patients with generalised and with idiopathic partial epilepsies, the number of patients was too small to reach levels of significance. “Progression” of epilepsy was seen in eight of the 16 patients who had complex partial seizures before AED withdrawal, and who relapsed with secondary generalised seizures: five of these eight patients again became seizure free (influence on outcome non-significant: χ2). These five patients all had idiopathic partial epilepsy. Presence of EEG abnormalities at diagnosis (abnormalities in 11 patients), before AED withdrawal (abnormalities in five patients), after relapse (abnormalities in four patients), or at the end of follow up (abnormalities in nine patients) was not associated with outcome: statistical analysis by EEG pooled over outcome strata, pairwise per stratum, as well as sequentially (variations in subsequent EEG results in the same patients according to outcome) yielded no significantly different results.

After transfer of continuous variables into categories, age at onset (younger v older than 5 years), concomitant neurological disorder (absent v present), and mental retardation (absent v present) were accepted, but interval from AED withdrawal to relapse (shorter v longer than 1 year) and all other variables (table 1) were rejected by multivariate analysis of outcome determinants. Excellent outcome after relapse was correctly predicted in 70.0% by age at onset of seizures (if over 5 years of age) combined with normal intellect, at a significance level of p=0.005. Poor outcome after relapse was predicted correctly by the presence of a neurological disorder alone in 77.5%, at a significance level of p=0.003. This pertains to eight of 12 (66.6%) patients with symptomatic partial epilepsy (table 1).


About 70% of children with epilepsy who have been free of seizures for at least 2 years while on AEDs will remain so when medication is withdrawn.1 The decision to stop AEDs when an individual child has been free of seizures for 2 years or more is made, weighing this risk of recurrence against the potential side effects of chronic AED treatment. In some children who become seizure free with AED treatment, a specific epileptic syndrome warrants continuation of medication—for example, juvenile myoclonic epilepsy.10

It has been shown that prolonged AED treatment after 2 years without seizures does not guarantee lasting seizure freedom in adults.11 A recent Dutch randomised clinical trial on AED withdrawal in children after a 6 or 12 month seizure free period showed seizure recurrence in 12% of the children while still under AED treatment after 6 months of seizure freedom.12 Hence AED treatment does not necessarily imply protection against seizure activity, even after a period of “remission”. Earlier studies have shown that 86% of patients who had a relapse after discontinuation of AEDs become seizure free again when AEDs are restarted within a period of 4 years,13 and 21 of 28 children (75%) with cerebral palsy (and, hence, with remote symptomatic aetiology) became seizure free again after restarting AEDs.14 Shinnar et al found 64% of children to be seizure free 6 years after relapse: 20% were without medication and 44% were on medication, but follow up was incomplete.15

This study shows that patients who relapse after AED withdrawal at a young age also face a favourable outcome: 70.0% (28/40) were seizure free at the end of follow up (median: 16.7 years) and in 33.3% (12/36) new attempts to withdraw AED treatment were successful. Thus, children who relapse after AED withdrawal may very well remain seizure free after new attempts to stop AED treatment at a later age. Discontinuation studies where relapse was used as study end point found age at onset over 10 years16 or adult onset of seizures to be associated with a higher relative risk of relapse.1 The subsequent course of epilepsy in these patients is not known, as a relapse does not necessarily mean poor outcome: the terminal remission in these patients may very well be better than the overall outcome figures based on recurrence as the study end point suggests. This terminal remission, however, may be attained at a considerably older age. This cohort shows a median age of 17.0 years at terminal remission without medication, and a median age of 16.3 with medication. This constitutes long lasting uncertainty for the individual patient, with potential social and psychological consequences, and for the treating neurologist, because any decision regarding treatment can only be evaluated after a very long period.

In patients who relapse, AED withdrawal may have been attempted too early in the course of the seizure disorder, as the natural history of the seizure disorder is not known. In addition, the effect of AEDs in preventing chronicity is certainly doubtful. Consequently, timing of an attempt to withdraw AEDs remains a question of weighing risks and benefits that are not firmly established.1, 11, 16, 17 The failure to identify predictors of lasting remission after AED withdrawal in individual patients who have attained seizure freedom is also reflected by the present study: neither seizure type nor EEG abnormalities before or after AED withdrawal have any value in the prediction of the outcome after relapse. These results may, however, contain a bias, as the entire study population had a seizure free period of at least 2 years before withdrawal of AEDs.

Fears that withdrawing AEDs “too early” might result in uncontrollable seizures, remain unsubstantiated in this study. Of the five patients in whom complex partial seizures evolved to secondarily generalised seizures after AED withdrawal, four had an excellent outcome. Moreover, no status epilepticus occurred after AED withdrawal in this population. Withdrawal of AEDs after a seizure free period of 2 years in children seems to be safe in terms of renewed seizure control in case of recurrence. This is in accordance with other studies with shorter follow up periods.13–15 Thus, patients attaining remission after a relapse may contribute considerably to the outcome after AED withdrawal. Therefore, overall outcome of patients who are followed up for longer periods will be better than the outcomes of study cohorts in AED withdrawal studies with a relapse as study end point.

When faced with the decision to withdraw AEDs after 2 years of seizure freedom, the risk of relapse increases relatively with onset after adolescence (relative risk 1.79 in adolescents and 1.34 in adults), with remote symptomatic seizures (relative risk 1.55), and with EEG abnormalities (relative risk 1.45).1 These figures leave room for considerable uncertainty. This study shows that the occurrence of a relapse after AED withdrawal allows a fairly accurate prediction of outcome: 70.0% of children with normal intelligence and age at onset of seizures over 5 years will become seizure free and will again discontinue AEDs. By contrast, 22.5% of the children with a neurological disorder (defined as “static encephalopathy”), and hence remote symptomatic seizures, will not become seizure free again, despite restarting AED treatment.

In conclusion, the current practice of withdrawing AEDs in children who have been seizure free for 2 years can be beneficial to most of these patients.1 Prediction of individual outcome before withdrawal remains uncertain. Patients in whom relapse occurs have good chances of achieving renewed seizure freedom, although this may take several years. Patients with static encephalopathy who relapse will almost always need continued AED treatment afterwards. The occurrence of a relapse itself does not have important consequences: the patients in this study were not harmed by relapse and 70.0% became seizure free again, adding considerably to the remission rates found in the literature to date, using relapse after withdrawal as a primary study end point.


We thank Brenda Vollers-King for linguistic editing.