Treatment of Newly Diagnosed Pediatric Epilepsy:  A Community-Based Study FREE

THE EVALUATION and treatment of childhood epilepsy has changed considerably in the last several decades due to the increased choices in antiepileptic drugs (AEDs), the advent of neuroimaging technology, and especially with the appreciation of epilepsy as a diverse set of disorders or syndromes. The earlier concerns that epilepsy was a uniformly progressive disease that required aggressive treatment to obtain a good outcome have been largely allayed, as epidemiological studies have consistently demonstrated that most often this is not the case, especially early in the course of the disorder.^1- 2 This, coupled with the increasing awareness of the cognitive, behavioral, and other adverse physiological effects of AEDs, has had a significant effect on the treatment of children as well as adults who present with a single seizure.^3- 6 The effect on treating those with newly diagnosed epilepsy, however, has not been documented in the United States. Data from a Dutch study indicated that, at least in the Netherlands, approximately 30% of children with newly diagnosed epilepsy are not initially treated, and 20% remain untreated after 1 year.⁷

Information about current practice patterns is useful in defining current standards as implemented in everyday practice. It is also essential in identifying trends toward undertreatment or overtreatment and for providing an initial starting point for developing practice parameters and identifying areas for further study.

There are few data worldwide and none from the United States that address the pattern of treatment in children at the time the diagnosis of epilepsy is first established. The following article describes the use of AED therapy at diagnosis and during the first year after diagnosis in a large community-based cohort of children recruited during the mid 1990s.

The methods for identification and recruitment have been described elsewhere.⁸ Briefly, children were recruited from all but 1 of the child neurologists practicing in Connecticut during the recruitment interval (1993-1997), as well as from some selected adult neurologists and pediatricians who indicated that they occasionally provided all neurological care for children with epilepsy without referral to a child neurologist. To be eligible, children had to have at least 2 unprovoked seizures (multiple events in a single day did not count as separate seizures),⁹ be between the ages of 1 month and 15 years inclusive at the time of the first unprovoked seizure, and have their epilepsy diagnosed during the study interval by 1 of the participating physicians. Data were collected via a standardized interview conducted with the parent or primary guardian as well as detailed review of all pertinent medical records. Epilepsy syndromes and seizures were classified according to standard criteria.^10- 11 The syndromic system for classifying epilepsy has been increasingly used in the last several years. It provides a hierarchical classification of different forms (syndromes) of epilepsy with the first distinction being among localization-related (ie, partial), generalized, and undetermined whether partial or generalized forms of epilepsy. Localization-related epilepsies can be further subdivided into idiopathic (presumed genetic), symptomatic (by virtue of localization [eg, temporal lobe] or by virtue of etiology [eg, in a child with cerebral palsy]), or cryptogenic (clearly partial, but unlocalizable and in a neurologically intact individual). Generalized epilepsies are divided into idiopathic (presumed genetic, such as childhood absence epilepsy), cryptogenic or symptomatic (such as infantile spasms and Lennox-Gastaut syndrome), and symptomatic (such as Ohtahara syndrome). The Epidemiologic Guidelines for Studies of Epilepsy from the International League Against Epilepsy were used for classifying etiology and other factors.⁹ The classification of etiology is related to the syndromes to a certain extent. Idiopathic refers to presumed genetic forms of epilepsy with age-related onset. Cryptogenic refers to nonidiopathic forms of epilepsy in individuals who do not have any underlying neurological conditions associated with an increased risk of epilepsy (such as cerebral palsy or tuberous sclerosis). Note that the syndrome "symptomatic localization-related" may have either a cryptogenic or symptomatic etiology. Remote symptomatic is reserved for epilepsy that occurs in association with conditions associated with an increased risk of epilepsy. Follow-up calls were placed with the parent every 3 months, and additional medical records were reviewed for information about medication changes, seizures, and other information related to the child's prognosis, diagnosis, and treatment.

We recorded whether treatment was initiated prior to the diagnosis of epilepsy (eg, after a first seizure), at the time the diagnosis of epilepsy was established, or whether the decision to initiate treatment was deliberately deferred at the time of diagnosis. A minor delay until baseline blood work or a diagnostic electroencephalogram could be obtained or other minor delay because of a planned vacation or holiday was not considered a deferred treatment decision. We considered primarily whether the child was treated as of the time of diagnosis regardless of whether treatment had been initiated earlier.

A total of 613 children were recruited into the study. The median age at the time of the first seizure was 5.3 years. There were 307 boys (50%) and 306 girls (50%). Prior provoked seizures occurred in 104 children (of 609 with known histories; 17.1%) including 76 (12.5%) with febrile seizures, 16 (2.6%) with neonatal seizures, and 14 (2.3%) with other provoked seizures. This includes 2 children with multiple types of provoked seizures. The etiology was idiopathic in 185 (30.2%), cryptogenic in 317 (51.7%), and remote symptomatic in 111 (18.1%). Two hundred ninety-five children (48.1%) came to medical attention at the time of the first unprovoked seizure. This includes 6 who had neonatal seizures that continued without a break into the postneonatal period. Fifteen children were already receiving an AED for prior provoked seizures at the time of the first unprovoked seizure (4 for febrile seizures, 6 for neonatal seizures, and 5 for other provoked seizures). Treatment was initiated for a single seizure, prior to diagnosis, in 56 children (9.1%) and after more than 1 seizure but prior to diagnosis in another 50 children (8.2%).

At the time of diagnosis, 482 children (78.6%) were prescribed and took (including those who continued) an AED. By 6 months after diagnosis, an additional 63 (10.3%) were treated, and by 12 months another 7 (1.1%) were treated, for a total of 552 (90%) treated within 1 year of diagnosis. The medications initially prescribed at diagnosis and through the first year after diagnosis are listed in Table 1. Carbamazepine was the most commonly prescribed first drug during the first year (n = 238, 38.8% of the entire cohort). Sodium valproate was the next most commonly prescribed AED (n = 113, 18.4%). Ethosuximide was prescribed for 52 (8.5%), phenobarbital for 60 (9.8%), and phenytoin for 65 (10.6%). The remaining children received either adrenocorticotropin hormone (ACTH) (n = 15 [2.4%]), clonazepam (n = 7 [1.1%]) prednisone (n = 1 [0.2%]), or vigabatrin (n = 1 [0.2%]) as their first AED. Sixty-one children (10%) received no treatment during the first year after diagnosis.

There was no substantial difference in the age at first seizure in those who were vs those who were not treated at diagnosis (5.8 vs 5.9 years, P = .78). At diagnosis, comparable proportions of boys and girls were treated (80% vs 77%, P = .36). There was no especially large difference in initial treatment by etiology; 76.8% of those with idiopathic etiology, 79.5% of those with cryptogenic etiology, and 79.3% of those with remote symptomatic etiology were treated at the time of diagnosis (P = .76). By 6 months clear differences had emerged. In the idiopathic group 84.3% were treated, compared with 89.6% in the cryptogenic group and 94.6% in the remote symptomatic group (P = .02); by 12 months the figures were 84.3%, 91.2%, and 96.4%, respectively (P = .002) (Table 2).

There were substantial differences in treatment at diagnosis by underlying epilepsy syndrome (Table 3). Only half of the children with benign focal epilepsies (almost all benign rolandic epilepsy) were treated at diagnosis, compared with 70% to 90% of the other groups. Most of these differences persisted after 12 months (Table 3).

There were expected differences in the most commonly used AEDs for each syndrome, consistent with recommended guidelines for the differential use of various drugs. Carbamazepine was the preferred drug for localization-related epilepsies, valproate for the primary generalized epilepsies, and ACTH for infantile spasms. The 1 child who received vigabatrin as the first AED had infantile spasms and went outside of the United States for the drug.

Roughly 20% of children with newly diagnosed epilepsy were not treated immediately on the confirmation of the diagnosis of epilepsy. This is somewhat less than the figure recently reported from a comparable study in the Netherlands, where 29% of children were not treated immediately and, after a year, 20% were still not treated. Nonetheless, both studies point out the fact that a substantial proportion of children, both in the United States and the Netherlands and, presumably, in other similar settings, are not automatically treated at the time the diagnosis of epilepsy is initially established. This is most likely quite different from what a similar study might have shown several decades ago, although there are no comparable data from then with which to compare our findings.

The approach to a first seizure has been greatly modified from one in which all patients were automatically treated^12- 13 to one in which treatment is now generally not considered necessary for most patients.^5,14- 15 In fact, only 17% of children (excluding those with neonatal seizures) who came to medical attention at the time of their first unprovoked seizure were treated for that seizure. For epilepsy (recurrent unprovoked seizures), it seems that there has been some carry-over from the changes in the approach to first seizures. While there have been discussions of benign rolandic epilepsy and the need or lack of need for treatment,^4,16 we have found that, in addition to benign rolandic epilepsy, several other forms of epilepsy also were not treated. In particular, less than 80% of children with either cryptogenic and symptomatic localization-related epilepsies or with epilepsies that were undetermined to be either localization related or generalized had treatment initiated at diagnosis.

On the other hand, despite a reasonably good long-term outlook, the idiopathic generalized epilepsies, more than half of which are childhood absence epilepsy, were for the most part treated at the time of diagnosis. This makes sense in light of the disruptions to daily life that can be caused by the frequent absence seizures associated with these syndromes and what is known of the natural history of, for example, juvenile myoclonic epilepsy. Secondarily generalized syndromes were also treated immediately in most cases. This was particularly true of infantile spasms and is consistent with the concern that early effective treatment, especially of the underlying disorder that is responsible for causing the seizure, will result in an improved long-term outcome.

The children in this study were recruited during a period when several new AEDs became licensed in the United States. As can be seen, however, these new agents were not used as first-line therapy, and the AEDs used reflect standard and conservative approaches to drug therapy for epilepsy. The marked preference for carbamazepine in large part reflects the fact that half or more of children with epilepsy have localization-related epilepsy. Toward the end of the recruitment period, an established AED, valproate, also received approval for a new indication, complex partial seizures. The relatively uncommon use of phenytoin is also consistent with current practice both in the United States and Europe, where it is less frequently used in children because of its complex pharmocokinetics and adverse effects. About a third of children who received phenytoin did so at the time of the first seizure. This may reflect a preference for emergency medicine physicians to use the drug because of their familiarity with it. It may also be because of the ability to achieve therapeutic levels rapidly and provide some protection against seizures until the patient has been fully evaluated during the next week to several weeks. Ethosuximide was reserved almost exclusively for children with a form of absence epilepsy or a form of epilepsy with absence seizures and not for children with generalized tonic-clonic seizures.

There were a few apparently unusual uses of AEDs; however, it must be kept in mind that the classification of seizures and syndromes was based on review, for the purposes of this study, by 3 child neurologists with specialized training in clinical neurophysiology (S.R.L., F.M.T., and S.S.). The medication was prescribed by the child's neurologist who, in a few instances, reached a different diagnosis of the epilepsy syndrome and seizure type. In fact, in most of these "unusual" cases (eg, ethosuximide for complex partial seizures) there was substantial ambiguity in the information needed for classifying seizures and syndromes and initial disagreement among the 3 study neurologists regarding seizures and syndromes. In addition, there were 2 instances in which ACTH was prescribed for complex partial seizures. Both children had infantile spasms with multiple seizure types. In all, the use of AEDs was consistent with published guidelines.^17- 19 The use of new AEDs also seems to be appropriate for this population as limited information about the use of these newer agents in children is available and the drugs are currently approved primarily for add-on therapy in refractory epilepsy and not for first-line monotherapy. The patterns of drug use will most likely change in the future as the recently released drugs eventually find their roles as first-line therapies.

Although this is not a "population-based" cohort in that there is no mechanism available to allow identification of every single case in the state, it is community-based, and not dependent on referral from a tertiary referral center. The characteristics of the cohort with respect to age, sex, and types of epilepsy syndrome are comparable to other population-based or near population-based series reported in the literature.^20- 22 Consequently, it is unlikely that any significant degree of referral bias affected the identification of potential study participants.

In conclusion, this study provides a description of the practices of child neurologists with respect to drug treatment of newly diagnosed epilepsy in a community-based study. Such information may be used as part of an initial starting point in providing guidelines for practice. Follow-up of this cohort may also help identify which children are most likely to fail current first-line therapies and therefore be candidates for early consideration of treatment with the newer agents.

Accepted for publication May 6, 1999.

This study was funded by grant RO1 NS31146 from the National Institutes of Health, Bethesda, Md.

We are especially grateful to the parents and children who patiently and selflessly participated in this study. We also thank the other physicians in Connecticut who have referred their patients to this study: Robert Cerciello, MD; Francis DiMario, MD; Barry Russman, MD; Michelle Kleiman, MD; Carol Leicher, MD; Edwin Zalneraitis, MD; Philip Brunquell, MD; Laura Ment, MD; Edward Novotny, MD; Bennet Shaywitz, MD; S. Nallainathan, MD; Alok Bhargava, MD; Martin Kreminitzer, MD; Harriet Fellows, DO; Jack Finkelstein, MD; Daniel Moalli, MD; Bernard Giserman, MD; Lawrence Rifkin, MD; and Murray Engel, MD. We also thank Edward Novotny, MD, and Francis DiMario, MD, for reinterpreting selected electroencephalograms for the study. Eugene Shapiro, MD, kindly facilitated many administrative issues for us. We also thank the research associates, Susan Smith-Rapaport, MS, Barbara Beckerman, MS, Heather LaCoste, Lynnette Bates, Joann Gehrels, and Kris Engel, for their dedicated work on this project and Wuthikrai Uayingsak, MS, for his exceptional programming expertise.

Editor's Note: It's nice to see that community-based physicians in at least one part of the country don't automatically seize the opportunity to treat children who have epilepsy with medications.—Catherine D. DeAngelis, MD

Corresponding author: Anne T. Berg, PhD, Department of Biological Sciences, Northern Illinois University, DeKalb, IL 60115 (e-mail: atberg@niu.edu).