Newly Diagnosed Epilepsy in Children: Presentation at Diagnosis

Transcription

1 Epilrpsia, ():5-.5, I999 Lippincott Williams & Wilkins, Inc., Philadelphia International League Against Epilepsy Clinical Research Newly Diagnosed Epilepsy in Children: Presentation at Diagnosis Anne T. Berg, *Shlomo Shinnar,?Susan R. Levy, and?francine M. Testa Department of Biological Sciences, Northern Illinois University, DeKalb, Illinois; *Departments of Neurology and Pediatrics, Albert Einstein College of Medicine, Bronx, New Yoi-k; and fdepartrnents of Pediatrics and Neurology, Yale University, New Haven, Connecticut, U.S.A. Summary: Purpose: The current understanding of epilepsy has changed significantly in the past decades. This report presents a description of newly diagnosed childhood-onset epilepsy, with a special emphasis on epilepsy syndromes, in a large, prospectively ascertained community-based cohort evaluated and diagnosed in the mid-99s. Methods: Children, aged through 5 years at the time of the first seizure, were prospectively identified at the time of diagnosis of epilepsy through the practices of 6 of the 7 child neurologists in Connecticut as well as five adult neurologists and seven pediatricians from January 993 through December 997. Parents were interviewed, and all relevant medical records were reviewed. classification of seizures and of epilepsy syndromes was done for each child by each of three pediatric neurologists, Discrepancies were resolved in conference. Results: A total of 63 children was recruited into the study. The median age at time of the first seizure was 5.3 years. Half the cohort was boys. Eighteen percent had a remote symptomatic etiology. Epilepsy syndromes were classifiable in all but four children, although some syndromes are, by definition, relatively nonspecific. In this childhood-onset cohort, 58.6% of the syndromes were localization related, 9.% generalized, and.% undetermined as to whether focal or generalized. Benign rolandic epilepsy occurred in % of the cohort. Primarily generalized syndromes accounted for.6%, with childhood absence being the single most common syndrome in this subgroup (.% of the cohort). Secondarily generalized syndromes accounted for 8.5% of the total, with infantile spasms being the most common in this grouping (3.9% of the cohort). Conclusions: This study presents a description of childhoodand adolescent-onset epilepsy as it is diagnosed and evaluated in the 99s in one state in the United State and based on current classification guidelines. The results should be generalizable to the rest of the country. The prognostic value of early identification of epilepsy syndromes will be determined through subsequent follow-up of this cohort. Key Words: Childhood-onset epilepsy-epilepsy syndromes-classification-epidemiology. Past studies showed that roughly half of all epilepsy begins during childhood and adolescence (,). In the last to years, substantial changes in the diagnosis and treatment of epilepsy have occurred. These include the introduction and continued development of the International Classification of the Epilepsies, the advent of modern brain-imaging techniques, advances in EEG technology, and increased awareness of disorders that previously may have been mistaken for and treated as epilepsy (3). In addition, there have been some changes in the epidemiology of some factors associated with symptomatic epilepsy. We present baseline descriptive data from a cohort of Accepted October 5, 998. Address correspondence and reprint requests to Dr. A. T. Berg at Department of Biological Sciences, Northern Illinois University, DeKalh, L 6 5, U.S.A. children with newly diagnosed epilepsy. Participants were recruited from January 993 through December 997 throughout the state of Connecticut. The report focuses on characteristics of the cohort at the time of initial diagnosis, with a special emphasis on the use of the International Classification of Epileptic Syndromes (). METHODS Children were recruited from the practices of 6 of the 7 child neurologists in the State of Connecticut (including the practice of two of the authors, S.R.L. and F.M.T.). Recruitment began gradually at one office in January 993, and by May 993, all offices were involved. In addition, pediatricians and adult neurologists across the state were canvassed to determine whether potential study participants might also be recruited 5

2 6 A. T. BERG ET AL. through these sources. Seven pediatricians and five adult neurologists indicated that they would occasionally care for a child with epilepsy without referral to a pediatric neurologist. These physicians also agreed to refer prospective patients to the study. Physicians were asked to identify and refer to the study any child who met the epidemiologic criteria for epilepsy, two or more unprovoked seizures (5). Multiple seizures in a -h period were considered a single episode. When in doubt, physicians were encouraged to refer and let the study team determine eligibility. The study focused on newly diagnosed epilepsy, not all of which was necessarily of recent onset. The following inclusion and exclusion criteria were applied. Inclusion. Initial diagnosis of epilepsy (not a first seizure) by a participating physician between January 993 and December First unprovoked seizure between the ages of month (8 days) and 5 years. To be consistent with epidemiologic definitions, we considered that infants with neonatal seizures that persisted without interruption after the first 8 days of life to have had the first unprovoked seizure on day 9. Exclusion Initial diagnosis by a nonparticipating physician >3 months before seeing a participating physician without any initial plan to consult a child neurologist. A child diagnosed by a nonparticipating physician within 3 months of seeing a participating physician was still eligible. This requirement was set to avoid the bias from including patients referred for second opinions. Single unprovoked seizure or only provoked seizures or seizures confined to the neonatal period. If, after a consensus meeting, all three child neurologists agreed that it was highly doubtful that a particular child had epilepsy. No adult family member who could communicate in English. In addition, families that could not be reached by phone or who lived out of state and >lo miles from the main study center in New Haven, Connecticut were not eligible. Recruitment and consent Physicians initially discussed the study with the family and requested permission for an interviewer to contact the family or sent an invitation to participate to the family. If the family agreed, an interviewer obtained written informed consent. Verbal or written assent was obtained from the child as appropriate. All procedures were ap- proved by the institutional review boards (IRBs) of all institutions involved in the study. A trained interviewer completed a structured interview with the parent or guardian. The interview included details of the child s unprovoked seizures, history of prior provoked seizures, and the child s medical and developmental history. All baseline medical records were reviewed and information extracted and coded for computer analysis. Baseline records refer to any information known before the diagnosis of epilepsy as well as the results of any tests that were ordered at the time of diagnosis, even though there might be several weeks to a few months delay in completing all tests. Information that was coded included the neurologic examination, presence of mental retardation, neuroimaging findings, specific underlying etiology, history of prior provoked seizures, presence of other medical conditions, and the EEG findings. Seizure types Seizures were classified according to the International Classification of Epileptic Seizures (6), based on their clinical manifestations and not on other data from the EEG or magnetic resonance imaging (MRI). Thus a patient might be seen with what clinically appear to be generalized onset tonic-clonic seizures and be classified as having a localization-related syndrome with a generalized seizure type. The one instance in which EEG was used to classify seizure type was in distinguishing between atypical absence seizures and complex partial seizures. Partial seizures that secondarily generalized were counted as a seizure type distinct from those that did not. Myoclonic absence was classified under the category of atypical absence. Etiology Etiology was defined according to the epidemiologic standards set forth by the ILAE (5). Idiopathic was reserved for children with no neurologic abnormalities who had a form of idiopathic (presumed genetic) epilepsy. The most common examples of these are benign rolandic epilepsy and childhood absence epilepsy. Cryptogenic is used for epilepsy in which there is no identifiable underlying etiology and the form of epilepsy is not one of the specific idiopathic syndromes. Remote symptomatic refers to epilepsy in the presence of a neurologic abnormality or a history of brain insult or a disorder associated with an increased risk of epilepsy and which was presumed to be etiologically related to the child s epilepsy. Syndromes Epilepsy syndromes and seizure types for each child were independently classified by each of three pediatric neurologists (S.R.L., F.M.T., and S.S.). In most in- Epdepsia, Vd., No., 999

3 NEWLY DIAGNOSED CHILDHOOD EPILEPSY 7 stances, the EEG report was used for this purpose. When the report was considered to be inadequate for classification purposes, we attempted to obtain the original tracing for reinterpretation. The three independent assessments of syndromes were then compared. Significant discrepancies were resolved in conference, and a final consensus classification of seizures and syndromes reached. We used published criteria for classifying syndromes (,7). Several syndromes have not always been clearly defined or consistently used, in particular, the cryptogenic localization-related (syndrome.3 in the International classification), other primary generalized (syndrome..8), and without unequivocal focal or generalized features (syndrome 3.). The following criteria were used for these syndromes. (.3) For cryptogenic localization related, the child had to have cryptogenic etiology, clearly focal seizures or a lateralized EEG finding that was not localized (e.g., unilateral hemispheric discharges or slowing) (or both). The presence of mesial temporal sclerosis on imaging was sufficient to place the child in the symptomatic (by virtue of localization) localization-related group but not, in itself, to classify the etiology as remote symptomatic. Clearly localizable abnormalities on the EEG tracing also were sufficient to place the child in the symptomatic localization-related group (again by virtue of localization) without affecting classification of etiology. (..8) For other primary generalized syndromes, the specific syndromes listed in the classification (e.g., childhood absence) had to be excluded first. The seizure type(s) and EEG findings had to be consistent with a primarily generalized disorder. We included the syndrome of random grand ma (8) in the..8 category, as it has not yet been entered into International Classification of epilepsy syndromes. (3.) Children placed in the category for without unequivocal focal or generalized features typically had normal EEGs or EEGs with diffuse or nonspecific findings and seizures that could not be clearly determined to be of either focal or generalized onset, typically either clinically generalized-onset tonic-clonic seizures or staring spells that might represent either atypical absence or a complex partial seizures. The most typical scenario for children categorized in this group was a normal EEG with a generalized tonicclonic seizure with unwitnessed or uncertain onset or generalized onset by description. There were some, however, for whom the information available to us was inadequate, although we suspected a more specific syndrome could be identified at diagnosis had the information been more precise. Family History Family history and history of prior provoked seizures was obtained primarily from the interview. If there was a significant discrepancy between the interview and the medical record, the interviewer resolved the difference with the adult who completed the interview. FOIIOW-UP The cohort is under continued follow-up to determine seizure and other outcomes. These will be the focus of later reports. RESULTS A total of 885 children was identified to the study. Of these, 5 were ineligible (3 had been previously diagnosed, had provoked seizures only, 39 had a single unprovoked seizure only, 7 did not have epilepsy or the diagnosis was too uncertain, and five did not meet other entry criteria). Of the 77 eligible patients, 3 families declined to participate. Another 3 were excluded because of difficulties with access (no phone, language, etc). In I cases, the neurologist preferred that the family not be included in the study, usually because of the stresses the family was under at the time. Interviewers were unable to contact 3 families. In all, 63 children TABLE. Demographic and other baseline characteristics of 63 children with newly diagnosed epilepsy Age at first seizure <I Age at diagnosis (yr) < Sex Male Female Characteristic No. % Race/ethnici ty White African-American Hispanic African- American-Hispanic Asian Other History of prior provoked seizures None Febrile Neonatal Other Etiology Idiopathic Cryptogenic Remote symptomatic a Four had missing history; two children had two different types of prior provoked seizures. Epikpsia, Vol., No., 999

4 8 A. T. BERG ET AL. TABLE a. Distribution of seizure types Seizure type Generalized" Tonic-clonic" Tonic Typical absence" Atypical absence Atonic" Myoclonic" Partial" Simple partial Complex partial" Partial with secondary generalization" Undetermined focal or generalized Staring spells, uncertain onset" Tonic-clonic, uncertain onset Overall (N = 63) N (%I 78 (5.) 35 (.) 7 (.) 95 (5.5) 6 (.6) (.6) 5 (8.5) 338 (55.) 55 (9.) 6 (6.) 79 (9.) 8 (.9) 9 (.5) (.) Idiopathic (n = 85) N (%I (76.) 5 (7.) (.) 9 (8.7) 7 (3.8) () 6 (8.7) 8 (5.9) (3.) (6.) (.9) (.) () (.) Cryptogenic (n = 37) N (%) (3.9) 77 (.3) 3 (.9) (.6) (.3) 7 (.) 8 (5.7) 3 (67.) (7.6) 9 (3.) 3 (35.6) (.) 8 (.5) 9 (.8) Remote symptomatic (n = ) 36 (3.) 8 (7.) (.8) 3 (.7) 5 (.5) 3 (.7) 8 (6.) 77 (69.) 7 (6.3) (36.) (39.6) (.8) (.9) I (.9) " x test on cff statistically significant at p <.5 for difference between etiologic groups. with the diagnosis of epilepsy were included in the cohort, the parents were interviewed, and the children are currently being followed. This represents 79.6% of eligible or presumed eligible children. One or more imaging studies was performed in 88 (79.6%) of the children. All but five children had EEGs. The descriptive characteristics of the cohort are presented in Table. The median age at first seizures was 5.3 years (range, month-5.6 years), and at diagnosis, 6. years (range, month-6 years). Three hundred seven (5%) of the children were boys. The distribution for race reflects the distribution reported in the 99 census for the State of Connecticut as a whole. A history of prior provoked seizures was missing for four children. In the 69 with known histories, (7.%) had prior provoked seizures. Febrile seizures were the most common and had occurred in.5% of the cohort. Neonatal seizures occurred in.5% of children, and other types of provoked seizures in another.3%. The median time between the first seizure and diagnosis of epilepsy was.7 months (range, < month-8.3 years), and between the second seizure and diagnosis,.9 month (range, days-8. years). Three fourths were diagnosed within months of having had their second seizure. Etiology Five hundred two (8.9%) children had no known underlying etiology based on the initial diagnosis and evaluation. Of these, 85 (3.%) had idiopathic syndromes and therefore idiopathic etiology, and 3 7 (5.7%) had cryptogenic etiology. The remaining (8.%) had a remote symptomatic etiology. Six children had a mild Chiari malformation noted on the MRI and no other abnormalities sufficient for the designation of remote symptomatic etiology. This was thought to be an incidental finding, and the children were therefore classified as having idiopathic or cryptogenic etiology. Seizure types Most children (8.9%) had one, 7.6% had two, and only three (.5%) had three different seizure types. Generalized seizures occurred in 5.% of the cohort, partial seizures in 55.%, and seizures of undetermined onset in.9%. There was some overlap between these three categories. Most of these had generalized tonic-clonic seizures with a clinically generalized onset as well as some clear partial seizures without secondary generalization. The remainder had a secondarily generalized syndrome. There are substantial differences in the distributions of seizures types by etiology (Table ). TABLE b. Number of diflerent seizures types (as defined in Table a) Generalized Partial seizure Seizures of All seizure types seizure types types uncertain onset Number of N n n n seizure types (63) (%) (78) (%) (338) (%) (8) (%) 5 (8.9) (87.) 85 (8.3) 5 (83.3) 8 (7.6) 35 (.6) 53 (5.7) 3 (6.7) 3 3 (.5) (.) () () Epilrpia, Vol, No, 999

5 ~ NEWLY DIAGNOSED CHILDHOOD EPILEPSY 9 TABLE 3. Distribution of epilepsy syndromes with remote symptomatic etiology Remote symptomatic etiology Syndrome N % (of 63) N 7 7 % (of row) Localization related. Idiopathic localization related.. Benign rolandic epilepsy I.. Childhood epilepsy with occipital paroxysms I. Symptomatic localization related.. w/seizures characterized by specific modes of precipitation..3 Symptomatic localization related I.3 Cryptogenic localization related. Generalized. I Primary generalized..o Primary generalized not further classified..3 Benign myoclonic epilepsy in infancy.. Childhood absence epilepsy..5 Juvenile absence epilepsy..6 Juvenile myoclonic epilepsy. I.7 Epilepsy with GTCS on awakening..8 Other primary generalized..9 w/seizures precipitated by specific modes of activation. Generalized cryptogenic or symptomatic or both.. Generalized cryptogenic/symptomatic not further classified.. West syndrome.. Lennox-Gastaut syndrome..3 Epilepsy with myoclonic-astatic seizures (Doose).. Epilepsy with myoclonic absences.3 Generalized symptomatic.3. Symptomatic generalized epilepsy (nonspecific etiology).3. Symptomatic generalized epilepsy (specific syndromes) 3. Other syndromes not determined whether focal or generalized 3. With both generalized and focal features 3.. Severe myoclonic epilepsy in infancy 3..3 Epilepsy with continuous spike-wave during slow-wave sleep 3..5 Other undetermined epilepsies not defined above 3. Epilepsies without unequivocal generalized or focal features I I I GTCS, generalized tonicxlonic seizure. Syndromes An epilepsy syndrome was assigned for 69 children, and a partial classification was possible for the remaining four children (Table 3). Factors that impeded complete classification or resulted in classification into a relatively nonspecific category (e.g.,.3 or 3.) included paucity of information about the seizures in children who had only had a few episodes, normal or nonspecific EEG traces, and inadequate EEG reports for which we could not obtain the original EEG tracing for reinterpretation. Overall, 359 (58.6%) had localization-related epilepsy; 78 (9.%), generalized epilepsy; and 76 (.%), epilepsy not determined to be exclusively focal or generalized. Most of this last group is in the category without unequivocal focal or generalized features (3.), a group that is essentially unclassified. The full distribution is presented in Table 3. Among the 9 with symptomatic localization-related epilepsy, children were classified as having localiza- tion-related symptomatic epilepsy by virtue of etiology alone, although no focus could be identified. Localization was to the temporal lobe in 9 cases (53 temporal; seven temporoparietal; centrotemporal, not consistent with benign rolandic epilepsy; 3 frontotemporal; six occipitotemporal); the frontal lobe in 3 cases; the parietal lobe in 5 (includes central); and the occipital lobe in 3 (6.7%). Four children were classified as having multifocal epilepsy. Seven children clearly had adequately localized epilepsy to meet the criteria for symptomatic localization-related epilepsy; however, from the descriptions in their EEG reports (e.g., posterior quadrant ), it was not possible to determine the lobe. Etiology versus syndrome There is a strong but not always absolute correlation between syndrome and etiology (Table 3). Two children with generalized idiopathic epilepsy also had a remote symptomatic etiology (one mild mental retardation and Epilepsia, Vol., No., 999

6 5 A. T. BERG ET AL. one mild cerebral palsy). The occurrence of such cases was noted before (7). The greatest discrepancy is in the symptomatic localization-related group. This contains patients who are symptomatic by virtue of either localization or etiology. DISCUSSION The current understanding of epilepsy has greatly changed over the last decades. This has been largely due to advances in EEG and imaging technology as well as the introduction and development of the concept of epilepsy syndromes. This study represents the most recent of only a few worldwide attempts to gather clinically detailed information about newly diagnosed epilepsy in children on a near-population basis. There are many considerations in determining how best to identify and recruit a representative series of patients for a study such as this. The trade-off is ultimately between quality of diagnosis and representativeness with respect to the population as a whole. Because accurate diagnosis not just of epilepsy overall but of a specific epilepsy syndrome requires accurate and complete evaluation by a qualified specialist; identification of potential study patients through child neurologists was considered the optimal approach. This impression was confirmed by a survey performed over a 9-month period in the private practices of two of the authors (S.R.L. and F.M.T.) in which we found that -3% of children referred by pediatricians because of seizures or epilepsy either had a different disorder or the diagnosis was not confirmed. Delivery of health care in the United States is very different from, for instance, that throughout much of Europe where general practitioners are the primary care providers, pediatricians are considered specialists, and child neurologists represent a tertiary level of referral. In the United States, pediatricians are primary care providers, and child neurologists represent the secondary as well as tertiary levels of care. To put this in perspective, in Connecticut, a state of 3.3 million, there are 7 child neurologists. By contrast, in the United Kingdom, a country of -5 million, there are 38 child neurologists (9). Thus recruitment from child neurology practices is appropriate in the United States, whereas it might not be in other countries. The child neurologists participating in this study were largely from community-based practices and did not represent only university hospital-based practice. To ensure that large pockets of patients were not missed, all adult neurologists and a large segment of the pediatricians in the state were contacted. The few who did report occasionally caring for pediatric patients agreed to refer pediatric patients to the study. In the end, despite a high level of enthusiasm and cooperation, relatively few referrals came from those sources, further validating the primary strategy of recruiting from child neurology practices. Of note, our approach of recruiting primarily from pediatric neurologists and then carefully reviewing each case is comparable to that used in a recent similar study from the Netherlands (,l I ). Such an approach ensures the exclusion of children with other disorders that may be difficult to distinguish from epilepsy. The strong similarities between our study and other population-based studies with respect to basic demographic and clinical features suggests that no significant biases affected recruitment of the cohort [(), Nova Scotia, Canada (), the Netherlands (lo), and Finland (3)]. A key focus of this study is the epilepsy syndrome. Unlike seizure types, which are simply symptoms, epileptic syndromes are essentially specific disease diagnoses (). Although seizure type contributes to the diagnosis of syndrome, it is only one piece of information. EEG characteristics, age at onset, presence of underlying neurologic abnormalities (including those detected by imaging), and in some instances, the diurnal pattern of seizure occurrence are also keys to determining the syndrome. Syndromes are acknowledged to be essential to optimal diagnosis and treatment of patients with epilepsy (5). Other studies reported on the classification of syndromes in series of patients with varying results in the distribution of syndromes and in the proportion who are unclassifiable (,6-; M. Sillanpaa et al., unpublished observations). There are several possible reasons for this. Age distribution: Many of the syndromes in children are highly specific and do not appear as adultonset epilepsy. If they are present in adults, they are, by necessity, cases of long-standing, often. refractory epilepsy. Thus comparisons of distributions must be limited to epilepsy in groups of the same age. Incident versus prevalent samples: Many forms of childhood-onset epilepsy resolve by adolescence or adulthood. To capture these relatively short-lived forms of epilepsy, patients must be included from the time of diagnosis. Samples of prevalent cases will tend to underrepresent forms of epilepsy (or any disease) that rapidly resoive and overrepresent more refractory cases. Timing of classification: One of the hoped-for advantages of the syndromic classification is that it will provide valuable information about prognosis from the time of diagnosis. In prevalent case series, the syndrome is often based, in part, on the prognosis. In such a scenario, the outcome contributes to the identification of the syndrome and not the syndrome to the outcome. In addition, even in an incident cohort, if syndromes are classified only Eplepiu, Vol., No., 999

7 NEWLY DIAGNOSED CHILDHOOD EPILEPSY 5 I after a few years of information have been allowed to accumulate, that additional information may allow more complete and accurate identification of syndromes. This was the case in the cohort from Finland (; M. Sillanpaa et al., unpublished observations) and may explain some of the difference between that study and the Connecticut cohort. A recent study from France () examined syndromic classification at initial diagnosis and then years later and found some significant and substantial changes had occurred in the classification after years had elapsed. Preliminary data from the Dutch study (3) also indicated that the classification of syndrome at diagnosis was changed after years in 59 (%) of 9 children.. Criteria for less well-defined syndromes: Despite strong similarities between the Connecticut cohort and other similarly conducted studies (,,; M. Sillanpaa et al., unpublished observations) in the distributions of basic demographic features, as well as some of the better recognized syndromes (e.g., benign rolandic epilepsy), there are some striking differences for some of the less-well-defined syndromes, in particular, the cryptogenic localization related (.3), the other primary generalized (..8), and the without unequivocal focal or generalized features (3.). This last syndrome is often critiqued in the literature as providing a false sense of accuracy in diagnosis ( 6,). It consists largely of patients with generalized tonic-clonic seizures or staring spells who have normal EEGs. A possible explanation for differences in the distributions of these syndromes across studies may have to do with variability in the criteria for classifying the syndromes. The distribution of epilepsies is notable for the majority being localization-related syndromes. This was seen in the Finish cohort (M. Sillanpaa et al., unpublished observations). Of those with nondiopathic localizationrelated epilepsy, there is a preponderance of temporal lobe involvement. The proportion of temporal lobe epilepsy in this new-onset cohort in children is in between what is seen in series of children who are surgical candidates, a group with predominantly extratemporal foci and surgical series in adults who have predominantly temporal foci (IS). A combination of differences in underlying etiology between children and adults and selection through natural history of who becomes intractable renders problematic direct comparisons with both adults and children with refractory epilepsy. In addition, because localization in this newly diagnosed cohort was based on standard clinical evaluations for new-onset patients and did involve extensive and invasive EEG evalu- ations, localization cannot be expected to be as precise as for patients who are candidates for epilepsy surgery (5). In summary, we presented the initial baseline characteristics of a large cohort of children with newly diagnosed epilepsy in the 99s in whom special care has been taken to classify the epilepsy syndrome based strictly on information available at diagnosis or generated as part of the initial diagnostic evaluation. The longrange goal of this ongoing study will be to determine the relative prognostic value of the syndromic classification in determining the outcomes of childhood-onset epilepsy. Acknowledgment: We are especially grateful to the parents and children who patiently and selflessly have participated in this study and to the physicians in Connecticut who referred their patients to this study: Drs. Robert Cerciello Francis Dimario, Barry Russman, Michelle Kleiman, Carol Leicher, Edward Zalneraitis, Philip Brunquel, Laura Ment, Edward Novotny, Bennet Shaywitz, S. Nallainathan, Alok Bargava, Martin Kreminitzer, Barbara Coughlin, Harriet Fellows, Jack Finkelstein, Daniel Moalli, Louise Resor, Brenda Balch, Patricia Braun, Owen Erlich, Bernard Giserman, John Monroe, Lawrence Rifkin, Lourdes Rosales, and Murray Engel. We also thank Drs. Edward Novotny and Francis DiMario for reinterpreting selected EEGs for the study. Dr. Eugene Shapiro kindly facilitated many administrative issues for us. We also thank the research associates, Susan Smith-Rapaport, Barbara Beckerman, Heather LaCoste, Lynnette Bates, Joann Gehrels, and Kris Engel, for their dedicated work on this project and Wuthikrai Uayingsak for his exceptional programing expertise. This study was funded by a grant from the National Institutes of Health, NINDS ROl-NS REFERENCES Hauser W, Annegers J, Kurland L. Incidence of epilepsy and unprovoked seizures in Rochester, Minnesota: Epilepsia 993;3: Berg AT. The epidemiology of seizures and epilepsy in children. In: Shinnar S, Amir N, Branski D, eds. Childhood seizures. Basel: S. Karger, 995:l-. Berg AT, Testa FM, Levy SR, Shinnar S. The epidemiology of epilepsy: past, present and future. Neurol Clin 996;: Commission on Classification and Terminology of the International League Against Epilepsy. Proposal for revised classification of epilepsies and epileptic syndromes. Epilepsia 989;3: Commission on Epidemiology and Prognosis, International League Against Epilepsy. Guidelines for epidemiologic studies on epilepsy. Epilepsia 993;3:59-6. Commission on Classification and Terminology of the International League Against Epilepsy. Proposal for revised clinical and electrographic classification of epileptic seizures. Epilepsia 98; :89-5. Roger J, Bureau M, Dravet C, Dreifuss FE, Perret A, Wolf P. Epileptic syndromes in infancy, childhood, and adolescence. nd ed. London: John Libbey, 99. Greenberg DA, Durner M, Resor S, Rosenbaum D, Shinnar S. The genetics of idiopathic generalized epilepsies of adolescent onset: differences between juvenile myoclonic epilepsy and epilepsy with random grand ma and with awakening grand mal. Neurology 995;5:9-6. Chadwick D, Reynolds EH. Services for epilepsy in the United Kingdom. Neurology 997;8:S3-7. Callenbach PMC, Geerts AT, Arts WFM, et al. Familial occurrence of epilepsy in children with newly diagnosed multiple sei- Epilepsia, Vol., No., I999

8 5 A. T. BERG ET AL. zures: Dutch study of epilepsy in childhood. Epilepsia 998;39: Peters AC, Brouwer OF, Geerts AT, Arts WF, Stroink H, van Donselaax CA. Randomized prospective study of early discontinuation of antiepileptic drugs in children with epilepsy. Neurology 998;5:7-3.. Camfield CS, Camfield PR, Gordon K, Wirrell E, Dooley JM. Incidence of epilepsy in childhood and adolescence: a populationbased study in Nova Scotia from 977 to 985. Epilepsia 996; 37: Sillanpaa M. Remission of seizures and prediction of intractability in long-term follow-up. Epilepsia 993;3: Benbadis SR, Luders HO. Epileptic syndromes: an underutilized concept [Editorial]. Epilepsia 996;37: Duchowny M, Harvey AS. Pediatric epilepsy syndromes: an update and critical review. Epilepsia 996;37:S6. 6. Manford M, Hart YM, Sander J, Shorvon SD. The National General Practice Study of Epilepsy: the syndromic classification of the International League Against Epilepsy applied to epilepsy in a general population. Arch Neurol 99;9: Eslava-Cobos J, Narino D. Experience with the International League Against Epilepsy proposals for classification of epileptic seizures and the epilepsies and epileptic syndromes in a pediatric outpatient epilepsy clinic. Epilepsia 989;3: Viani F, Beghi E, Atza MG, Gulotta MP. Classifications of epi- leptic syndromes: advantages and limitations for evaluation of childhood epileptic syndromes in clinical practice. Epilepsia 988; 9:-S. 9. Loiseau J, Loiseau P, Guyot M, Duche B, Dartigues JF, Aublet B. Survey of seizure disorders in the French southwest. I. incidence of epileptic syndromes. Epilepsia 99;3: Osservatorio Regionale per L'Epilessia (OREp) L. ILAE classification of epilepsies: its applicability and practical value of different diagnostic categories: Osservatorio Regionale per L'Epilessia (OREp), Lombardy. Epilepsia 996;37: Sillanpaa M, Jalava M, Kaleva, Shinnar S. Long-term prognosis of seizures with onset in childhood. NEngl JMed 998;338:75-.. Loiseau J, Picot MC, Jallon P, Dartigues JF, Loiseau P. Classification and incidence of epileptic syndromes in a prospective study: reliability and cause of change. Epilepsia 998;39(suppl 6): Arts WF, Geerts A, Brouwer, Peters A, Stroink H, van Donselaar C. Classification schemes in childhood epilepsy: reliability and causes of discrepancy. Epilepsia 997;38(suppl 3):.. Aydinli N, Caliskan M, Ozmen M, Apak S. Classification of epilepsies and epileptic syndromes in a child neurology unit. Brain Dev 996;8: Sperling MR, Shewmon DA. General principles for presurgical evaluation. In: Engel J, Pedley TA, eds. Epilepsy: I( comprehensive tent. Philadelphia: Lippincott-Raven, 997: Epilepsiu. Vol., No., 999