Epilepsy & Behavior 8 (2006) 713719 www.elsevier.

com/locate/yebeh

Prognosis of seizure recurrence after stopping antiepileptic drugs in seizure-free patients: A long-term population-based study of childhood-onset epilepsy
M. Sillanpa a a, D. Schmidt
a

b,*

Departments of Public Health and Child Neurology, University of Turku, Turku, Finland b Epilepsy Research Group, Berlin, Germany Received 27 January 2006; revised 28 February 2006; accepted 28 February 2006 Available online 17 April 2006

Abstract The long-term outcome with respect to seizure relapse after planned discontinuation of antiepileptic drugs (AEDs) in seizure-free patients is not well known. Relapse and its treatment outcome were evaluated in a longitudinal population-based study of 148 patients from the onset of their epilepsy to an average follow-up of 37 years. During the study, AEDs were completely discontinued by 90 patients; 58 patients remained on medication. Seizure relapse after AED discontinuation was observed in 33 (37%) of 90 patients at an average follow-up of 32 years. Among 8 of the 33 patients who elected to restart AEDs, 2 achieved 5-year terminal remission (5YTR), but only 1019 years after restarting treatment. The other 6 patients never achieved 5YTR, and 2 of the 6 never entered a 5-year remission period during follow-up. Factors associated with failure to reach 5YTR after treatment of relapse were symptomatic etiology and localization-related epilepsy. In conclusion, drug discontinuation after seizure freedom results in relapse in one-third of patients. Reinstitution of a medication that worked for years fails to achieve control in one of four patients. These risks need to be considered, although there is no evidence that discontinuation is responsible for the poor prognosis for treatment of seizure recurrence. 2006 Elsevier Inc. All rights reserved.
Keywords: Epilepsy; Discontinuation of antiepileptic drugs; Seizure relapse; Pharmacoresistant epilepsy; Outcome of recurrence

1. Introduction The risk of seizure relapse under gradual withdrawal of antiepileptic drugs (AEDs) is substantial when compared with treatment, especially during the rst year [1]. However, the disadvantages of continuing treatment indenitely include the risk of teratogenicity, drug interaction with concurrent medication, adverse events, and the concern that treatment may be unnecessary. More than 70% of patients may have no relapse at 2 years according to a meta-analysis of 25 studies [2]. On the other hand, the risk of seizure relapse is also highabout 12% per year throughout the rst 2 years of follow-up for patients who

Corresponding author. Fax: +49 30 8017679. E-mail address: dbschmidt@t-online.de (D. Schmidt).

were randomized to continue treatment but did not always do it [1]. Nevertheless, a seizure relapse may have grave consequences for the individual patient, particularly adults, and may interfere with ability to drive a car for a while, or create problems with employment, and be a cause of considerable distress and anxiety [3]. One important unresolved issue for management of a patient who has relapsed is how often reinstitution of therapy promptly controls epilepsy as before. Prompted by a review of 14 observational studies on outcome of treatment of relapse in dierent settings that reported on 19% of patients with uncontrolled epilepsy [4], we examined the long-term course of epilepsy following relapse after planned discontinuation of AEDs. In a longitudinal long-term population-based study, we addressed the following questions: (1) How many patients relapse with seizures when AEDs are discontinued? (2) How many patients who relapse

1525-5050/$ - see front matter 2006 Elsevier Inc. All rights reserved. doi:10.1016/j.yebeh.2006.02.014

714

M. Sillanpa a , D. Schmidt / Epilepsy & Behavior 8 (2006) 713719 uneventful periods were documented, each subject was contacted every 5 years or up to death. Data on the deceased were collected, on permission, from close relatives and the National Death Register. The follow-up period extended to the last contact with the patient, irrespective of later survival or death. Death was considered equal to any other event resulting in discontinuation of follow-up, such as refusal, relocation, or other cause of failure to trace the patient.

regain seizure control on reinstitution of AEDs? (3) What is the time course for regaining seizure control after a relapse? (4) Can risk factors for poor seizure control after reinstitution of treatment be determined?
2. Methods

2.2. Discontinuation method 2.1. Patients

The study subjects included all children aged 15 and younger who were living in the catchment area of the University of Turku Central Hospital, Turku, Finland, at the end of 1964 and who met the criteria for epilepsy, that is, two or more unprovoked seizures [5]. Subjects were identied on the basis of hospital, institution, and primary health care records and a review of the National Health Service records, a registry of all patients residing in Finland. In total, 245 patients were identied, of whom 223 (91%) were seen in the University of Turku Central Hospital. The remaining 22 patients (9%) were seen in other hospitals, institutions, and public or private primary or outpatient care oces. In Finland in the 1960s, the rule was that children who had an epileptic seizure be referred for evaluation. Untraceable, and thus beyond the scope of this study, were three more patients who, through ongoing surveillance, were identied and met the inclusion criteria. As these three patients could never be contacted, they were excluded from the present study. Among the 245 patients, 3 children could be followed only 69 years and, so, were excluded because of the very short follow-up. Furthermore, 94 patients with a several-year history of epilepsy when rst seen (prevalent cases) were excluded. Thus, the remaining 148 patients (incident cases) who were followed more than 9 years from the onset of their epilepsy constitute the nal study cohort of children aged 015 with epilepsy. These 148 patients were ascertained and evaluated for epilepsy from 1 January 1961 to 31 December 1964. Epilepsies were classied by one of us (M.S.) as idiopathic generalized (22%), symptomatic/cryptogenic generalized (10%), idiopathic localization-related (9%), symptomatic/cryptogenic localization-related (50%), and undeterminable by localization or unclassiable (9%). The diagnosis of temporal lobe epilepsy was based on clinical descriptions of individual seizures, EEGs (later video/EEG recordings), and, because these patients were often drug-resistant, PEG, air encephalogram, and, since the 1970s, CT and other brain imaging studies. All 148 patients were examined and evaluated by one child neurologist [6] and enrolled in a prospective followup of medical and social outcomes for an additional 35 years. Follow-up included ongoing review of the medical records and comprehensive evaluation at 5-year intervals. In 1992, in addition to the structured extensive questionnaires, the evaluation included clinical examination with appropriate tests for physical tness and laboratory investigations. Compliance was evaluated by questioning the patient. Compliance was regarded as good if the patient answered yes, according to the given instructions, to the question: Have you taken your drugs regularly? The other options were: Yes, regularly, but less than instructed; I have occasionally forgotten medication; I have taken the medication irregularly; There have been longer breaks in the medication; and I have spontaneously discontinued the medication. The study design and some earlier results were reported in detail previously [710]. For the present study, a total of 148 incident patients were prospectively followed up since the onset of their epilepsy (average SD = 37 7.1 years, median = 40.0 years, range = 1142 years). Mean time elapsed from onset of treatment to rst-ever 5-year remission for all 148 patients was 9.6 5.9 years (median = 8.0, range = 535). Length of follow-up after initiation of AED withdrawal in seizure-free patients is discussed under Results. There was an uninterrupted ow of data from hospitals and institutions with the permission of the patients; from the private oce of one of us (M.S.), where many patients were seen regularly; and from the patients themselves, who were encouraged to and did inform the rst author about epilepsy-related events, if any. To be sure that even AED discontinuation was not randomized. In principle, gradual discontinuation of AEDs (with an interval of 69 months from reduction of dose to elimination of AED) was considered after individual riskbenet assessment 2 to 3 years after the last seizure in children aged <16 with generalized nonconvulsive and convulsive epilepsies and 5 years after the last seizure in all other patients of dierent ages and with dierent types of epilepsy. Patients who reduced the daily dosage or number of AEDs but did not eliminate all AEDs were assigned to the group of patients who continued taking medication. At the time, many patients were being treated with either phenobarbital or phenytoin, both of which caused signicant adverse eects and prompted patients to consider discontinuation. Subjects also sought to discontinue AEDs because they were going to take driving lessons, were being drafted for military service, or were making marriage plans. On the other hand, patients who were not experiencing adverse eects, or who were starting a new job, or who already had a drivers license opted to remain on their medication. However, withdrawal was not randomized and customarily not recommended in patients with juvenile myoclonic epilepsy, for which the relapse rate is thought to be high. In addition, the decision to discontinue AEDs was always made after discussion with the patient or guardian. In many cases, withdrawal was delayed or declined by the patient, often because of fear of relapse. For the assessment of relapse, patient histories were examined to detect any novel cause of seizures, such as head trauma or intracerebral infection.

2.3. Denitions
Epileptic syndromes, epilepsies, epileptic seizures, and etiology of seizures were dened according to the guidelines for epidemiologic research of the International League Against Epilepsy [5,11,12]. A patient was dened as having drug-resistant epilepsy if, on at least 10 years of follow-up and despite many attempts at single-drug or combined-drug therapy, he or she did not achieve a 5-year remission during continuing state-of-the-art drug treatment supervised by the rst author. Our denition is quite inclusive and would probably not be appropriate for referral to surgery or entry into a new drug trial. Epilepsy was considered to be in uctuating remission in patients who experienced at least one 5-year or longer period of seizure freedom during state-of-the-art drug therapy, according to best clinical practice. A relapse was dened as the occurrence of repeated seizures during planned withdrawal and after discontinuation of drug treatment in a seizure-free patient. A single seizure, including those prompted immediately by drug withdrawal and other occasion-related seizures, was not classied as a relapse, and patients with a single seizure continued to be included in the 5-year terminal remission (5YTR) or 5-year remission ever (5YRE) group.

2.4. Statistical analysis

For statistical analyses, Pearsons v2 test with Fishers exact test (twotail), with Yatess correction when appropriate, and the MannWhitney test were used. A P value <0.05 was considered statistically signicant. Statistical computations were done using SAS System for Windows, Release 8.02 (SAS Institute, Cary, NC, USA). Actuarial methods with KaplanMeier curves were used to assess the chance of 5-year remission and 5-year terminal remission in patients who stopped and those who continued treatment [1215]. The study design was approved by the Joint Ethics Review Committee of the University of Turku and Turku University Central Hospital.

M. Sillanpa a , D. Schmidt / Epilepsy & Behavior 8 (2006) 713719

715

3. Results During the study, 90 patients discontinued AEDs, and 58 patients remained on medication (Fig. 1). Thirty-three of ninety (37%) who were followed up, on average, 37 years from the onset of epilepsy and who withdrew completely from drug therapy experienced seizure relapse (Fig. 1). The average SD follow-up after AED withdrawal was 32 8.7 years (median = 34, range = 742). Relapse occurred within the rst year in 36%, within the rst 2 years in 46%, and within the rst 3 years in 67%. Treatment duration prior to withdrawal was shorter in patients who relapsed than in those who did not relapse (5.1 6.2, median = 3.0, range = 022 vs 9.2 9.0, median = 7.0, range = 035; P = 0.0284). The two drug-resistant patients and four of the six patients who did not enter a 5-year remission until after resuming treatment (discussed above) were incident cases. Drug therapy was discontinued by 53% of the 90 patients within 2 years and by 73% within 5 years. The last relapse recorded during follow-up occurred 28 years after AED withdrawal was initiated. We did not nd a novel cause for remanifestation of seizures. The proportions of patients entering 5-year remission and 5-year terminal remission on and o medication relative to the duration of epilepsy are illustrated in Figs. 2 and 3. Not unexpectedly, the curves substantially dier from each other, but more strikingly, one-fourth of even dicult-totreat patients on continuing medication became seizurefree and more than half experienced at least one 5-year remission. In addition, we analyzed 5-year terminal remission by epilepsy syndrome (Table 1). The data in Table 1 conrm that symptomatic (focal and generalized) epilepsies have a less favorable prognosis than idiopathic (focal or generalized) epilepsies. Post hoc analysis indicated that treatment duration prior to withdrawal was shorter in patients who relapsed (6.1 6.2 years, median = 4.0, range = 123; P = 0.0284) than in those who did not relapse (10.2 9.0 years, median = 8.0, range = 136). Based on patient preference, 25 of the 33 patients who relapsed did not restart medication (Fig. 1). After relapse, AEDs were reinstituted in 8 of 33 (24%) patients, leading to 5-year terminal remission in 2, at delays of 10 and 19 years after restarting treatment. The remaining 6 patients, however, never achieved 5-year terminal

Fig. 2. Proportion of patients (incident cases) entering 5-year terminal remission, on and o medication, relative to the duration of epilepsy.

Fig. 3. Proportion of patients (incident cases) entering 5-year remission ever (5YRE), on and o medication, relative to the duration of epilepsy.

remission despite AED treatment, and 2 were considered to have drug-resistant epilepsy according to our denition (Table 2). Among the 90 previously seizure-free patients undergoing planned discontinuation of AEDs, the rate of drug-resistant epilepsy after AED discontinuation was 2% (2/90). This corresponds to 6% (2/33) among patients

Fig. 1. Long-term epilepsy outcome in 90 seizure-free patients who elected planned discontinuation of AEDs and in 58 patients who continued AED treatment. pts, patients; w/d, planned AED discontinuation; 5YTR, 5-year terminal remission.

716

M. Sillanpa a , D. Schmidt / Epilepsy & Behavior 8 (2006) 713719

Table 1 5-Year terminal remission by syndrome in individuals with childhood-onset epilepsy followed long-term Epilepsy syndrome Localization related Idiopathic Rolandic Symptomatic Temporal lobe Frontal lobe Occipital lobe Not localizable Cryptogenic Generalized Idiopathic Childhood absence Juvenile absence Juvenile myoclonic Awakening Random generalizeda Other primary generalized Cryptogenic and/or symptomatic West syndrome LennoxGastaut syndrome Undetermined whether focal or generalized Unclassiable All 5-YTR 57 (65.5) 13 (92.9) 13 (92.9) 39 (59.1) 23 (53.5) 2 (66.7) 2 (100.0) 12 (66.6) 5 (71.4) 29 (60.4) 27 (81.8) 4 (66.7) 0 (0.0) 1 (50.0) 6 (100.0) 14 (93.3) 2 (100.0) 2 (13.3) 0 (0.0) 2 (33.3) 3 (75.0) 8 (88.9) 97 (65.5) Remission followed by relapse 11 (12.7) 0 (0.0) 0 (0.0) 11 (16.7) 8 (18.6) 0 (0.0) 0 (0.0) 3 (16.7) 0 (0.0) 7 2 0 1 0 0 1 0 5 3 2 (14.6) (6.1) (0.0) (50.0) (0.0) (0.0) (6.7) (0.0) (33.3) (33.3) (33.3) No remission 19 (21.8) 1 (7.1) 1 (7.1) 16 (24.2) 12 (27.9) 1 (33.3) 0 (0.0) 3 (16.7) 2 (28.6) 12 (25.0) 4 (12.1) 2 (33.3) 1 (50.0) 1 (50.0) 0 (0.0) 0 (0.0) 0 (0.0) 8 (53.3) 6 (66.7) 2 (33.3) 0 (0.0) 0 (0.0) 31 (21.0) Total 87 14 14 66 43 3 2 18 7 48 33 6 2 2 6 15 2 15 9 6 (100.0) (100.0) (100.0) (100.0) (100.0) (100.0) (100.0) (100.0) (100.0) (100.0) (100.0) (100.0) (100.0) (100.0) (100.0) (100.0) (100.0) (100.0) (100.0) (100.0)

1 (25.0) 1 (11.1) 20 (13.5)

4 (100.0) 9 (100.0) 148 (100.0)

a Random generalized epilepsy was dened as epilepsy with generalized tonicclonic seizures (GTCS) randomly distributed during the sleepwake cycle [16].

Table 2 Clinical features of epilepsy in six previously seizure-free patients who never achieved 5-year terminal remission after resuming AED treatment for seizure recurrence Patient gender Male Male Female Female Male Male
a

Type of epilepsy Cryptogenic TLE West syndrome, spastic diplegia Cryptogenic TLE Symptomatic TLE, ataxichypotonic cerebral palsy Symptomatic TLE, spastic hemiplegia Unclassiable posttraumatic epilepsy (near drowning)
a

IQ Normal Abnormal Normal Abnormal Normal Abnormal

Response Drug resistance Drug resistance Fluctuating remission Fluctuating remission Fluctuating remission Fluctuating remission

Temporal lobe epilepsy.

who relapsed and 25% (2/8) among patients treated for seizure relapse. Furthermore, 6 of 8 patients achieved 5-year remission only 8.7 4.5 years (median = 8.0, range = 5 14) years after restarting AEDs following the relapse. Remarkably, 2 of the 6 patients who never regained 5-year terminal remission after treatment was reinstituted had a very good early response to treatment (seizure-free within the rst year of therapy). The clinical features of the 7 patients who never achieved 5-year terminal remission (with or without AED treatment) following AED discontinuation were compared with those of the 26 patients who entered 5-year terminal remission (see Fig. 1). Symptomatic/cryptogenic seizure etiology (7/7 vs 13/26, P = 0.0266) and an IQ of 670 (4/7 vs 3/26, P = 0.0229) were more frequent in patients who did not enter a 5-year terminal remission. The frequency of temporal lobe epilepsy did not dier between the groups (3/7 vs 5/26, P = 0.3201). Reecting the policy for AED discontinuation and the absence of a randomized control

group, patients scheduled for AED discontinuation had a much better outcome than patients who were recommended to stay on medication (5-year terminal remission in 83/ 90 vs 14/58, P < 0.0001, v2) (Fig. 1). Although reinstitution of AEDs after relapse was not randomized, as pointed out earlier, the course of epilepsy in 25 patients who elected not to take medication after a relapse may be of interest (see Fig. 1). Twenty-four of twenty-ve patients who received no AEDs after relapse with one or several seizures regained 5-year terminal remission (see Fig. 1). However, it took an average of 8.2 years (median = 7.0, range = 520) to reach 5-year terminal remission after the last relapse (Fig. 4). Remarkably, 5 of the 24 patients took P10 years to regain 5-year terminal remission (Fig. 4). 4. Discussion This population-based study is unique because of its very long average follow-up of 37 years (range = 1042).

M. Sillanpa a , D. Schmidt / Epilepsy & Behavior 8 (2006) 713719

717

Fig. 4. Time course for regaining 5-year terminal remission after seizure relapse following AED discontinuation in 24 patients who elected not to restart AED treatment (see Fig. 1).

There are no published observational studies of patients with epilepsy with a comparable long-term follow-up. The main results are: 1. Of 90 seizure-free patients with childhood-onset epilepsy who discontinued AEDs, 33 (37%) relapsed at an average follow-up of 32 years after withdrawal. However, only 8 patients preferred to resume AED treatment. 2. AED treatment for relapse was associated with 5-year terminal remission in 2 of 8 patients. However, 2 patients became drug-resistant, and in another 4 patients, 5-year remission was delayed several years despite treatment of recurrence. In summary, dicultto-treat epilepsy was noted in 6 of 8 patients following treatment of relapse. In extrapolation, as many as 18% (6/33) of patients who relapse and 7% (6/90) of all seizure-free patients who stop AEDs may have a poor outcome. Drug-resistant epilepsy and delayed seizure control were observed mostly in patients with symptomatic localization-related epilepsy. These risks need to be considered, although, as discussed below, there is no evidence that discontinuation was responsible for the poor outcome in treatment of recurrence. The overall risk of relapse in our study was 37% for a mean follow-up of 32 years after withdrawal. Our relapse rate is comparable to the 36 and 37% reported for shorter follow-up periods of 56 years following AED withdrawal in two large prospective studies of childhood-onset epilepsy [17,18]. The outcome of seizure relapse following AED treatment in our study is very similar to that reported in a review of 14 observational studies in children and adults, in which a mean of 19% of those treated for relapse (95% condence interval = 14.523.9) remained uncontrolled despite treatment [4]. This review included a large followup evaluation of the Medical Research Council study, which randomized 1021 patients (mostly adults) who had been seizure-free for at least 2 years to either discontinua-

tion of AEDs or continuation of daily medication [1]. During follow-up, 90% of those who relapsed after stopping AEDs regained a 2-year remission [19]. This result is dicult to compare with our data because of dierences in design and in patient populations; for example, the outcome of patients with childhood-onset epilepsy was not reported separately. Two studies on outcome of seizure relapse after discontinuation of AEDs became available only after completion of our review [4]. In one Canadian populationbased study, 79 (30%) of 260 children who discontinued AEDs had at least one recurrence and restarted daily AEDs [20]. They were followed up, on average, 5 years, and 25 eventually became seizure-free after a second or third attempt to discontinue medication. The authors report that 3 patients (1% of 260, and 4% of those receiving AEDs for relapse) had intractable epilepsy and were evaluated for epilepsy surgery [20]. The reported frequency of intractable epilepsy of 1% in their study with a much shorter follow-up is similar to the 2% (2/ 90) in our series. In a study from The Netherlands on discontinuation of AEDs by patients with seizure-free childhood epilepsy, 25% (14/55) of patients who relapsed and restarted medication continued to experience seizure relapse during a 4-year follow-up [21]. This proportion of patients is similar to the 18% (6/33) in our long-term study. Factors associated with drug resistance in our study were symptomatic seizure etiology and low IQ. In the absence of a controlled setting, however, we cannot exclude that our data may be biased by some unknown confounders, which may explain our ndings. To explain why nearly one of ve of those treated for relapse (and 8% of all patients withdrawn) proceeded from seizure freedom to discontinuation of AED treatment to drug-resistant seizures or a several-year delay in seizure control, we consider several scenarios. One, withdrawal seizures are responsible for the observed poor treatment outcome. In this case, reintroduction of treatment would not be required; however, withdrawal seizures usually occur within weeks of discontinuing AEDs and all patients discussed here were seizure-free after more than 3 months of complete discontinuation. Two, drug resistance was observed to emerge after AED discontinuation in a subgroup of previously seizure-free patients (for review, see [22]). However, in the absence of a randomized control group, we cannot determine if AED withdrawal caused the development of drug resistance in our observational study. It is possible that we are observing the natural history of the condition. Alternatively, the epilepsy was never in long-term seizure control, and the long intervals between seizures were mistaken for lasting control of seizures. In the latter case, discontinuation would have had no signicant impact on long-term seizure outcome. It may be of interest to discuss patients who did not relapse. Among our patients, 23 (16%) never relapsed

718

M. Sillanpa a , D. Schmidt / Epilepsy & Behavior 8 (2006) 713719

during an almost 40-year follow-up. Most patients who did not relapse had idiopathic epilepsy (20% vs 16%, P = 0.0010), or random generalized epilepsy or epilepsy with generalized tonicclonic seizures randomly distributed during the sleepwake cycle as described by Janz [16] (22% vs 9%, P = 0.0291). There were no signicant differences with respect to gender, age at onset of epilepsy, generalized epilepsy, or extratemporal localization-related epilepsy. The limitations of the present population-based longterm study include the small number (8) of patients who elected treatment for relapse. It is intriguing that 75% of those who had discontinued treatment did not restart medication, even though the ultimate outcome is better in this group. Possible reasons why patients and physicians were reluctant to restart medication include AED side eects (at the time, the early 1970s, mainly phenobarbital and phenytoin were the AEDs used) and the occurrence of only a few seizures. Of the 25 patients who took no medication after relapse, 14 experienced rare tonicclonic seizures at intervals of several years, and 6 patients had short generalized or partial seizures without tonicclonic seizures. We have no data for the remaining 5 patients who did not restart medication. Furthermore, it is possible that not all seizure-free patients were oered AED discontinuation and that patients, parents, and physicians made the decision to restart AEDs; thus, assignment to discontinuation was not randomized and there may also be an intervention bias. All these limitations result from the fact that the patients, although prospectively followed more than 10 years, were treated on clinical grounds and the treatment decisions were made in collaboration with patients. In the absence of a randomized control group who continued treatment (which would be dicult to maintain over several decades), we could not assess if drug discontinuation or changes in the natural history of epilepsy were responsible for the observed poor outcome following discontinuation in this naturalistic long-term populationbased study. Although we have no systematic information on the impact of lifestyle on recurrence, we have assessed adherence to prescribed drug regimens by asking patients, as discussed under Methods. Drug compliance was good in 75, 71, and 81% of patients at 10, 20, and 25 years of follow-up, respectively. Finally, it must be reemphasized that the small size and the heterogeneity of our sample may have precluded subgroup analysis and detection of additional factors, if they exist, predictive of prognosis after relapse. In conclusion, our data provide clinical evidence that the risk of epilepsy becoming dicult to treat after drug withdrawal is moderate, that is, <10% for those who discontinue AEDs and 25% for those who are treated for relapse. This should be considered when counseling patients at the time of treatment discontinuation. In addition to the medical riskbenet balance outlined earlier, the decision to discontinue should also take into account social aspects,

for example, drivers license, job, and leisure activities, as well as emotional and personal factors and adverse eects or drug interactions with some AEDs. Ultimately, patients have to decide whether they wish to discontinue AED treatment. Acknowledgments We thank Olli Kaleva for statistical workup and Inge Wimmer for her patience with so many requests for references. References
[1] Medical Research Council Antiepileptic Drug Withdrawal Study Group. Randomised study of antiepileptic drug withdrawal in patients in remission. Lancet 1991;337:117580. [2] Berg A, Shinnar S. Relapse following discontinuation of antiepileptic drugs: meta-analysis. Neurology 1994;44:6018. [3] Jacoby A, Johnson A, Chadwick D, for the Medical Research Council Antiepileptic Drug Withdrawal Study Group. Psychosocial outcomes of antiepileptic drug discontinuation. Epilepsia 1992;33:112331. [4] Schmidt D, Lo scher W. Uncontrolled epilepsy following discontinuation of antiepileptic drugs in seizure-free patients: a review of current clinical experience. Acta Neurol Scand 2005;111:291300. [5] Commission on Classication and Terminology of the International League Against Epilepsy. Proposal for revised classication of epilepsies and epileptic syndromes. Epilepsia 1989;30:38999. [6] Sillanpa a M. Medico-social prognosis of children with epilepsy: epidemiological study and analysis of 245 patients. Acta Paediatr Scand Suppl 1973;237:1034. [7] Sillanpa a M, Jalava M, Kaleva O, Shinnar S. Long-term prognosis of seizures with onset in childhood. N Engl J Med 1998;338:171522. [8] Sillanpa a M, Jalava M, Shinnar S. Epilepsy syndromes in patients with childhood-onset seizures in Finland. Pediatr Neurol 1999;21:5337. [9] Sillanpa a M, Haataja L, Shinnar S. Perceived impact of childhoodonset epilepsy on quality of life as an adult. Epilepsia 2004;45:9717. [10] Sillanpa a M, Shinnar S. Status epilepticus in a population based cohort with childhood-onset epilepsy in Finland. Ann Neurol 2002;52:30310. [11] Commission on Classication and Terminology of the International League Against Epilepsy. Proposal for revised clinical and electroencephalographic classication of epileptic seizures. Epilepsia 1981;22:489501. [12] Commission on Epidemiology and Prognosis, International League Against Epilepsy. Guidelines for epidemiologic studies on epilepsy. Epilepsia 1993;34:5926. [13] Kaplan EL, Meier P. Nonparametric estimation from incomplete observations. J Am Stat Assoc 1958;53:45781. [14] Elandt-Johnson RC, Johnson NL. Survival models and data analysis. New York: Wiley; 1980. pp. 15080. [15] Annegers JF, Hauser WA, Elveback LR. Remission of seizures and relapse in patients with epilepsy. Epilepsia 1979;20:72937. [16] Janz D. In: Die Epilepsien. Stuttgart: Thieme; 1969. p. 456. [17] Shinnar S, Berg AT, Moshe SL, et al. Discontinuing antiepileptic drugs in children with epilepsy: a prospective study. Ann Neurol 1994;35:53445. [18] Todt H. The late prognosis of epilepsy in childhood: results of a prospective follow-up study. Epilepsia 1984;25:13744. [19] Chadwick D, Taylor J, Johnson T, for the MRC Antiepileptic Drug Withdrawal Group. Outcomes after seizure recurrence in people with

M. Sillanpa a , D. Schmidt / Epilepsy & Behavior 8 (2006) 713719 well-controlled epilepsy and the factors that inuence it. Epilepsia 1996;37:104350. [20] Cameld P, Cameld C. The frequency of intractable seizures after stopping AEDs in seizure-free children with epilepsy. Neurology 2005;64:9735.

719

[21] Geerts AT, Niermeijer JMF, Peters ACB, et al. Four-year outcome after early withdrawal of antiepileptic drugs in childhood epilepsy. Neurology 2005;64:21368. [22] Schmidt D, Lo scher W. Drug resistance: putative neurobiologic and clinical mechanisms. Epilepsia 2005;46:85877.