Recurrence after a First Seizure

Many children are seen by a physician after a first unprovoked generalized tonic-clonic seizure, a few after a first complex partial seizure, but almost none after a single absence or myoclonic seizure. For children presenting after a single unprovoked seizure, there is an immediate question: Will it happen again? Berg and Shinnar’s meta-analysis of the recurrence risk after a first seizure concluded that, overall, about 40 percent will have a second seizure [Berg and Shinnar, 1991; Shinnar et al., 1996]. There is much less information about the further recurrence risk after a second seizure, especially if medication is withheld. Current data suggest that the risk of a third seizure after a second is at least 80 percent [Camfield et al., 1985b; Shinnar et al., 2000; Hauser et al., 1998]. These observations validate the restriction of the diagnosis of epilepsy to those with two or more unprovoked seizures, because the concept of epilepsy implies a chronic, recurring disorder.

If a child has a single seizure, a small number of clinical features are helpful in predicting the risk of a second seizure. These include remote symptomatic cause, partial seizures, presence of intellectual or mental handicap, and possibly EEG spike discharge [Berg and Shinnar, 1991]. A normal child with a first generalized tonic-clonic seizure and a normal EEG has a recurrence risk of 20–30 percent, whereas a child with mental handicap, a partial seizure, and spike discharge on the EEG has a recurrence risk of 80–90 percent [Berg and Shinnar, 1991; Camfield et al., 1985b]. The value of MRI lesions in predicting recurrence after a first seizure has not been extensively evaluated [Arthur et al., 2008]. If a second seizure is to occur, it usually happens shortly after the first. At least 75 percent of recurrences happen within 6 months of the first, and most occur within a few weeks. Few recurrences are observed after 2 years.

Very few children presenting with a first seizure develop severe epilepsy. Of 407 children with a first seizure identified by Shinnar et al. [2000], only 13 percent went on to have more than 10 seizures over the next 10 years. However, 28 percent of those with two seizures had more than 10 subsequent seizures.

Starting Medication Treatment

There is little justification for beginning daily medication after a child’s first unprovoked seizure because 60 percent of these children will never have another attack [Van Donselaar et al., 1989; Hirtz et al., 2003]. Population-based studies have found that medication prescription after the first seizure does not alter the recurrence rate, probably because of poor compliance in taking the medication [Camfield et al., 1985b; Hauser et al., 1982; Hirtz et al., 1984]. An open-label, randomized trial of medication versus no medication after a first seizure in 397 children and adults demonstrated a significant reduction in recurrences for those on medication over a 2-year treatment period [First Seizure Trial Group, 1993]. The risk of recurrence without treatment was 51 percent by 24 months, but there were still relapses in 25 percent of those randomized to medication (i.e., 2.8 times higher; 95 percent confidence interval [CI] = 1.9–4.2). When this cohort was followed for several more years, the eventual remission rate was the same for those treated or untreated after the first seizure [Musicco et al., 1997].

It has become common practice to prescribe medication after a second seizure. There is no evidence that the prescription of medication alters the long-term outlook of childhood epilepsy. There is convincing information that delaying antiepileptic drug treatment until the child has had up to 10 seizures does not alter the ease of seizure control or the long-term remission rate [Camfield et al., 1996a; Arts and Geerts, 2009]. In other words, if a child has few seizures, there is no evidence that each seizure facilitates the next. The main reasons to treat children with antiepileptic drugs are the avoidance of bodily injury from seizures and improvement in psychosocial function. Results from several studies indicate that injury is uncommon with most seizure types in children. Peters et al. [2001] described 79 children with benign rolandic epilepsy who together had more than 900 seizures over an 8.5-year period. There were no significant injuries. In a group of 59 children with generalized absence seizures followed over 15 years, Wirrell et al. [1997] observed that 16 had a serious physical injury as the result of a seizure. Kirsch et al. [2001] compared 31 cognitively normal children with epilepsy with best-friend controls and found no significant differences in injury number or severity. Hodgman et al. [1979] described 25 adolescents with generalized tonic-clonic epilepsy and observed that those with poorer seizure control were better able to communicate about seizures and had a better self-image. The “hidden handicap” for children with controlled seizures may have important effects on social adjustment. Epilepsy in children and adults has attracted social stigma in most cultures. Helping children deal with perceived and actual stigma may be an important part of treatment [Bandstra et al., 2008]. Once medication is started, most children still have more seizures. Only about 20 percent have “smooth-sailing epilepsy,” meaning that they start medication, become immediately seizure-free, and later are able to discontinue medication successfully without ever having another seizure [Camfield et al., 1993]. Only 50 percent of children continue to receive the same medication 1 year after initiating treatment [Camfield et al., 1985b; Canadian Childhood Epilepsy Study Group, 1996; Verity et al., 1995]. The decision to start medication is often not the end of the seizure problem.

Long-Term Remission

For many children, epilepsy is transient, and with maturation, the problem seems to vanish. At the time of diagnosis, it is possible to predict that at least 50 percent of children will outgrow their disorder and be able to discontinue medication [Camfield et al., 1993]. The Dutch Study of Epilepsy in Childhood followed 453 newly diagnosed children and found that 64 percent were no longer receiving medication 5 years later [Arts et al., 2004; Arts and Geerts, 2009]. The longer the follow-up, the higher the proportion with remission of symptoms. In Rochester, Minnesota, 115 children with epilepsy beginning before age 10 years were followed through the Mayo Clinic record linkage system. Ten years later, 75 percent had been seizure-free for at least 5 years, and 51 percent no longer received medication [Hauser et al., 1993]. At the end of a remarkable 30-year follow-up study of children with epilepsy from a population-based sample in Turku, Finland, Sillanpaa [1993] found that 76 percent of survivors had been seizure-free for at least 3 years.

Across most studies, factors that predict which children will outgrow their epilepsy have included normal intelligence, normal neurologic examination findings, relatively small numbers of seizures at diagnosis (which makes complex partial seizures unlikely), age of the first seizure below about 12 years, and absence of a remote symptomatic cause (an identified brain problem that preceded the onset of epilepsy). In an 8-year follow-up study of a population-based study of 504 children with epilepsy from Nova Scotia, a predictive scoring system was developed; it is outlined in Table 50-2 [Sillanpaa et al., 1995]. Those with a good prognosis had an 80 percent chance of remission (i.e., seizure-free and no longer receiving medication). For those with one or more of these adverse factors, the chance of remission was less but still about 40 percent. The Dutch Study of Epilepsy in Childhood found comparable predictive variables and supplemented these with the clinical course in the first 6 months of treatment. Rates of correct prediction were similar. Based on clinical features present at the time of diagnosis, it is possible to predict the long-term outcome with moderate accuracy. Surprisingly, most epilepsy syndromes do not have a definite outcome; however, a few specific syndromes allow an accurate prognosis [Camfield et al., 2003]. Epilepsy is always outgrown in cases of benign rolandic epilepsy, benign myoclonic epilepsy of infancy, and benign occipital epilepsy, early-onset type. Epilepsy is never outgrown by children with Rasmussen’s syndrome, Lennox–Gastaut syndrome, or Dravet’s syndrome. The Dutch and Nova Scotia studies were combined to allow study of 1055 newly diagnosed patients with at least 5 years of follow-up [Geelhoed et al., 2005]. Multivariate analysis that took into account clinical factors, syndrome diagnosis, and response to treatment in the first 6 months revealed that, at the time of diagnosis, a prognostic scoring scheme could correctly predict remission or no remission in about 70 percent. This means that, at the time of diagnosis, if parents ask “will the seizures be outgrown?”, the physician will give the wrong answer in about one-third of cases.

Table 50-2 Scoring System for Remission of Childhood Epilepsy at the Time of Diagnosis

* Add the scores from this column. If the total score is greater than 495, the child is predicted to have remission of epilepsy.

(Adapted from Sillanpaa M et al. Predicting long term outcome of childhood epilepsy in Nova Scotia, Canada, and Turku: Validation of a simple scoring system. Arch Neurol 1995;52:589.)

Stopping Medication

About 60–70 percent of children with epilepsy who have become seizure-free for 1–2 years can successfully stop medication treatment [Berg and Shinnar, 1991; Dooley et al., 1996]. The rate of success is no greater if medication is continued for up to 5 years seizure-free [Hollowach-Thurston et al., 1982]. Factors that predict successful discontinuation of medication include generalized seizures, age of onset before 10–12 years, normal neurologic examination, and in some studies, resolution of interictal EEG spike discharges. Children with no adverse factors may have an 80–90 percent success rate. Each factor has an additive effect and those with all of the adverse factors may have only a 10–20 percent success rate.

If an initial discontinuation trial is unsuccessful, medication is usually restarted. About 50 percent of children again become seizure-free for sufficient time to try to discontinue medication a second time, with a 70 percent success rate [Camfield et al., 1993]. The remission rate for juvenile myoclonic epilepsy (JME) is low, although a recent population-based study suggested that eventually 40 percent will no longer require daily medication [Camfield et al., 2009]. Still, most experts suggest that medication for JME should be continued lifelong and that further attempts to discontinue medication after an initial failure are probably not warranted [Delgado-Escueta and Enrile-Bacsal, 1984].

A Scandinavian study randomized 207 children at the time of diagnosis of epilepsy to receive 1 or 3 years of treatment [Braathen and Melander, 1997]. If the child was seizure-free for the last 6 months of study, medication was discontinued. This practice meant that some of the children in the 1-year treatment group had been seizure-free for only 6 months before medication was stopped. The success rate for those in the short-treatment group was significantly less than for the 3-year group (53 percent vs. 71 percent). A Dutch study randomized 161 children with epilepsy that was controlled within 2 months of starting treatment to 6 months or 1 year of treatment. The 6-month group had a higher relapse rate when medication was stopped; however, by 2 years later, the rates of remission were identical [Peters et al., 1998]. None the less, for a substantial number of children, epilepsy was a short-lived disorder requiring only short-term medication use. Medication treatment for benign rolandic epilepsy often is not needed [Peters et al., 2001], and further studies will likely identify other children who do not require drug treatment.

Intractability

Defining intractable epilepsy is difficult, and many definitions have been suggested [Berg et al., 2006a]. For a study of 613 children with epilepsy from Connecticut, Berg and Shinnar [2001] defined intractability as a “failure, for lack of seizure control, of more than 2 first-line antiepileptic drugs with an average of more than one seizure per month for 18 months and no more than 3 consecutive months seizure free during that interval.” There were 60 intractable cases in the first 24 months after diagnosis. The proportions of intractable patients in each major syndrome grouping were as follows: cryptogenic/symptomatic generalized, 34.6 percent; idiopathic generalized, 2.7 percent; other localization-related, 10.7 percent; and unclassified, 8.2 percent. Sillanpaa’s [1993] follow-up study from Turku, Finland, defined intractability as one or more seizures per year in the past 10 years of follow-up. After 20 years of follow-up, 22 percent of the subjects met these criteria. Predictors of intractability included poor initial response to medication, remote symptomatic cause, and status epilepticus. The Dutch Study of Childhood Epilepsy (using Berg and Shinnar’s definition of intractability), with 453 children followed for 5 years, found that 6 percent were intractable [Arts et al., 2004]. Camfield et al. [1993] defined intractability as “at least one seizure each three months for the last year of follow-up, with failure of at least 3 antiepileptic drugs.” For those with partial and generalized tonic-clonic epilepsies, 8 percent of 511 children became intractable during an average of 8 years of follow-up. After 20 years of follow-up, 51 percent of 75 patients with childhood-onset symptomatic generalized epilepsy had intractable seizures. In this study, the major predictor was severe neurologic deficit at the time of diagnosis.

An important issue in the definition of intractability is the length of follow-up. In the clinical setting, many children with intractable epilepsy eventually become well controlled. In a series of 145 children with intractible seizures (i.e., less than one per month for at least 2 years) followed over 18 years, Huttenlocker and Hapke observed that 75 percent of normally intelligent children and 30 percent with mental handicap had complete or nearly complete seizure remission (i.e., less than one seizure per year) [Huttenlocker and Hapke, 1990]. In the Nova Scotia cohort, 39 patients with partial and generalized tonic-clonic epilepsies had intractable epilepsy after 7 years’ follow-up. After 12.5 years of follow-up, 7 (18 percent) of 39 became seizure-free [Camfield et al., 1996a].

In contrast, when the history was reviewed retrospectively for adults undergoing a work-up for intractable epilepsy, it was common to find long periods of remission during childhood [Berg et al., 2006b]. Intractable epilepsy in childhood may spontaneously resolve and epilepsy in childhood may remit, only to re-emerge as intractable epilepsy in adulthood.

There is controversy about how the number of drug failures required before intractability can be declared. In a Scottish study of 470 newly treated adults, 47 percent became seizure-free with their first antiepileptic drug [Kwan and Brodie, 2000]. Only 11 percent gained control with subsequent medications. In children, the failure of a first drug is important but far less ominous [Camfield et al., 1997]. Of 417 eligible patients with partial and generalized tonic-clonic epilepsies with at least 4 years of follow-up, 83 percent were successfully treated with a single antiepileptic drug during the first year of treatment. Of these, 61 percent eventually had remission of epilepsy, and 4 percent had intractable seizures. Among the 17 percent who failed treatment with their first antiepileptic drug, 42 percent eventually had complete remission of epilepsy, although 29 percent developed intractable seizures. Children with absence epilepsies had nearly identical findings [Wirrell et al., 2001].

In summary, intractability for an individual child is difficult to predict before several years of antiepileptic drug treatment. Intractability appears to decrease with prolonged follow-up, although the burden of this wait-and-see approach may be substantial. Failure of a first antiepileptic drug is a risk factor for intractability but none the less many remit. As a general rule, consideration for epilepsy surgery should await failure of three appropriate drug treatments. Other factors influence this decision, including seizure severity, frequency, and duration of epilepsy.

Psychosocial Outcome for Children with Epilepsy

Children with epilepsy have high rates of behavior and cognitive problems that contribute to social dysfunction in childhood and in later adulthood, even if the epilepsy resolves. These issues are reviewed in more detail in Chapter 51. Austin et al. [2002] reported that, at the time of diagnosis, 40 percent of children were at significant risk of behavior problems, as judged by scores on the Child Behavior Checklist. Those with multiple recurrent seizures were at even higher risk. The Isle of Wight study found that children with epilepsy had high rates of behavioral disorders [Rutter et al., 1970], which might have had their origins in social stigmatization, comorbid cognitive disorders, or medication effects. Oostrom et al. [2003] demonstrated greater cognitive and behavioral problems and the need for special educational assistance among 51 outpatient children with idiopathic or cryptogenic epilepsy compared with their classmate controls. Trostle [1988] reported that parents indicate a reluctance to have their normal children play with a child with epilepsy. Less than one-third of 20,000 U.S. high-school students indicated that they would date a person with epilepsy [Austin et al., 2002]. These behavioral, cognitive, and social stigma problems in childhood all point to grave concerns for the social adjustment and success of children with epilepsy after they reach adulthood.

Kokkonen et al. [1997] described the social outcome (by interview or questionnaire) of 81 young, “noninstitutionalized” adults from the catchment area of Oulu, Finland, compared with 211 randomly selected controls from the same birth cohort. At least 20 percent of this sample were unable to complete 9 years of education and presumably had severe learning disorders or mental retardation. They accounted for most of the poor social outcome, including a high rate of educational failure, failure to marry, and unemployment.

In Nova Scotia and Finland, children with “epilepsy only” were followed into young adulthood [Camfield et al., 1993; Jalava et al., 1997]. “Epilepsy only” meant having normal intelligence and no other neurologic handicaps. About 30 percent of each cohort had significant social adjustment problems, with decreased rates of stable relationships, marriage, social contacts, job satisfaction, and work achievement. Rates of unemployment or underemployment were high. Social outcome was not clearly related to epilepsy remission, and in the Nova Scotia study, variables related directly to epilepsy, such as age of onset, type of medication treatment, presence or absence of seizure remission, and frequency and severity of seizures, did not appear to predict social outcome in young adulthood [Camfield et al., 1993]. The strongest predictor of poor social outcome was the presence of a learning disorder, although predictive models were inaccurate; the main reasons for the unsatisfactory outcomes were unclear.

Wakamoto et al. [2000] studied 148 normally intelligent young adults (>20 years old) living in a rural district of Japan who had childhood-onset epilepsy. This population-based study found that 72 percent attended regular classes (versus 99 percent of those without epilepsy), 66 percent (versus 97 percent) entered high school, 67 percent (versus 95 percent) had employment, and 23 percent (versus 33 percent) married. There were 49 patients with mental handicap having a less satisfactory outcome: 14 percent attended regular class, 6 percent entered high school, 20 percent were employed, and 2 percent married.

Few studies have addressed the social outcome of children with specific epilepsy syndromes. Several studies have indicated a favorable outcome for adults with previous benign rolandic epilepsy (Loiseau et al., 1983; Peters et al., 2001]. In the Nova Scotia study, 56 children with typical childhood absence epilepsy were followed to young adulthood [Wirrell et al., 1997] and compared with a similar cohort with mild juvenile rheumatoid arthritis. Those with absence had significantly greater problems with impulsive behavior, including a 34 percent risk of an unplanned pregnancy. Educational and work achievement, family and other social relationships, and alcohol abuse were more often unsatisfactory. Those with on-going seizures despite medication had greater problems; however, most of the poor social outcome was unrelated to epilepsy-specific factors. The findings in JME were similar – 24 normally intelligent patients followed for 20–30 years had an unemployment rate of 35 percent with many unplanned pregnancies and social isolation [Camfield and Camfield, 2009]. Therefore, childhood epilepsy may have a lifelong, serious effect on social function, an effect that is greater than some other chronic disorders of childhood. Intervention studies have not been undertaken; however, the role of learning disorders and inadequate education is clear. If the long-term social outcome is to be rectified, physicians must address these areas with an enthusiasm equal to the drug treatment of childhood epilepsy. Early referral to an epilepsy support group may be of benefit.

Mortality in Children with Epilepsy

Children with epilepsy have a 5.3 times higher risk of dying than the general population [Camfield, et al., 2002]. The causes that lead to this excess are direct complications of seizures (e.g., aspiration or cardiac arrhythmia), accidents caused by seizures (e.g., falling into a fire), comorbid conditions (e.g., decompensated hydrocephalus), and suicide or sudden unexpected death in epilepsy (SUDEP). Three large prospective studies have found that children with epilepsy but no neurological deficit have the same survival as the general population. [Callenbach et al., 2001; Berg et al., 2006; Camfield et al., 2002]. SUDEP is very rare (2 of 1777 newly diagnosed cases). The excessive mortality in childhood epilepsy is almost entirely accounted for by comorbid severe neurological problems, especially those that affect bulbar function. The clinician can reassure families of normal children with epilepsy that they do not face an increased risk of untimely death despite the frightening nature of seizures.