All authors contributed equally to the preparation of the review.
Address correspondence and reprint requests to Dr. Rod C. Scott, Senior Lecturer in Paediatric Neurosciences and Honorary Consultant Paediatric Neurologist, Neurosciences Unit, UCL – Institute of Child Health, The Wolfson Centre, Mecklenburgh Square, London WC1 N 2AP, United Kingdom. E-mail: email@example.com
Summary: There is ongoing debate regarding the most appropriate definition of status epilepticus. This depends upon the research question being asked. Based on the most widely used “30 min definition,” the incidence of childhood convulsive status epilepticus (CSE) in developed countries is approximately 20/100,000/year, but will vary depending, among others, on socioeconomic and ethnic characteristics of the population. Age is a main determinant of the epidemiology of CSE and, even within the pediatric population there are substantial differences between older and younger children in terms of incidence, etiology, and frequency of prior neurological abnormalities or prior seizures. Overall, incidence is highest during the first year of life, febrile CSE is the single most common cause, around 40% of children will have previous neurological abnormalities and less than 15% will have a prior history of epilepsy. Outcome is mainly a function of etiology. However, the causative role of CSE itself on mesial temporal sclerosis and subsequent epilepsy or the influence of age, duration, or treatment on outcome of CSE remains largely unknown. Future studies should aim at clarifying these issues and identifying specific ethnic, genetic, or socioeconomic factors associated with CSE to pinpoint potential targets for its primary and secondary prevention.
Epidemiological studies investigate how many people develop a disease or condition over time, describe the natural history and outcomes of the disease, the characteristics of those affected, and may generate hypotheses about the cause of the disease. These results provide essential data for the prevention, control, and treatment of the condition under study (Grimes and Schulz, 2002).
There are many shortcomings in the understanding of the epidemiology of status epilepticus in childhood primarily related to methodological problems, which include inappropriate study design, case definition, ascertainment, classification of etiologies, and techniques used in assessing outcome during follow-up (Hauser, 1983). In addition to these concerns, status epilepticus is an event that is related to an underlying condition (known or unknown) and so has to be viewed as part of the natural history of several conditions that have their own epidemiological issues. This adds a layer of complexity to understand the epidemiology of status epilepticus.
Age is a fundamental determinant of the epidemiology of status epilepticus and many of the epidemiological aspects of status epilepticus differ between adults and children. Even within the pediatric population, there are substantial differences between older and younger children in terms of incidence, etiology, frequency of prior neurological abnormalities, and prior history of unprovoked seizures (Shinnar et al., 1997). Most epidemiological studies on status epilepticus have been primarily or exclusively based on adult populations and may not reflect a reliable characterization of status epilepticus in children (Chin et al., 2004a).
Status epilepticus may be classified as convulsive (CSE) or nonconvulsive (non-CSE). CSE is the most common form of status epilepticus, but its relative frequency is difficult to document because the various types of status are not specified separately in most series. Despite controversies in diagnostic criteria for non-CSE, it is accepted that EEG is mandatory for the diagnosis (Brenner, 2002; Walker et al., 2005) and the true incidence of non-CSE (and status epilepticus in general) will remain elusive until EEG testing of all children with suspected non-CSE in the emergency department is widespread practice.
Thus, epidemiology of CSE in children will be the focus of this review. Neonatal seizures will not be considered as the specific etiologies and the definitions of status epilepticus in this setting deserves detailed separate consideration. The available data on the incidence, etiology, seizure characteristics, and outcome in non-neonatal childhood CSE will be discussed and suggestions for future research will be made.
THE DEFINITION ISSUE
Epidemiological studies should use explicit definitions of the condition of interest. Debate is ongoing regarding the most appropriate definition of status epilepticus, which is reflected in the inconsistent use of the term in published studies on the subject. The most common criterion for the diagnosis of status epilepticus has been the duration of seizure activity, which has varied considerably. Studies from the 1970s used a 1-h-duration criteria (Aicardi and Chevrie, 1970; Fujiwara et al., 1979). Subsequently, most studies, including all epidemiological studies, have used a 30-min-duration criteria. The justification for this duration is that this is the time at which status epilepticus may become self-sustaining, pharmacoresistance may have developed, and seizure-induced neuronal injury may take place (Chen and Wasterlain, 2006). On its last proposal of terminology, the International League Against Epilepsy (ILAE) defined status epilepticus as “a seizure that shows no clinical signs of arresting after a duration encompassing the great majority of seizures of that type in most patients or recurrent seizures without interictal resumption of baseline central nervous system function,” but did not set a temporal criterion as had previously been the case (Commission on Epidemiology and Prognosis, 1993; Blume et al., 2001; Engel, 2001). This definition, which may be judged as somewhat ambiguous, is in more accordance with recently proposed “operational” definitions that aim to define the time when patients should be treated as if they were in established status epilepticus; this includes seizure durations as low as 5 min (Lowenstein et al., 1999) (Fig. 1).
There is probably a need for different definitions, having different specifications for duration of seizures or recovery of consciousness, which are tailored to a particular research question. Operational definitions serve as much better guides for treatment and should be those used in clinical trials. In contrast, there are no clear reasons to modify the “traditional” 30-min-duration definition to evaluate the incidence and outcome of CSE until a better understanding of the pathophysiological and prognostic determinants of CSE has been reached.
THE CLASSIFICATION ISSUE
There is good evidence from epidemiological studies and clinical series that etiology is an important determinant of the outcome of patients with CSE, and it is therefore important to identify and classify the causes of CSE accurately (Raspall-Chaure et al., 2006).
In children, a significant proportion of CSE is associated with fever. Some children with CSE associated with fever have pre-existing neurological abnormalities, including epilepsy, while others are previously neurologically normal. The latter group includes: (1) children in whom their CSE were prolonged febrile seizures (i.e., febrile CSE), defined as CSE in a previously neurologically normal child aged between 6 months and 5 years during a febrile (temperature above 38°C) illness in the absence of defined central nervous system (CNS) infection and (2) children with acute symptomatic CSE due to CNS infections (Scott et al., 2003). Of note, the ILAE's guidelines establish that both should be classified as acute symptomatic seizures (Table 1). This classification system, which includes febrile CSE in the same category as CSE due to an identified acute neurological insult, such as acute bacterial meningitis, may be inappropriate for pediatric epidemiological or outcome studies. It can be argued that febrile CSE is clearly a separate entity with an overall favorable prognosis, and that studies that do not separate febrile CSE and acute symptomatic CSE are likely to erroneously amplify the severity of outcome of febrile CSE and conversely, dilute the severity of acute neurological insults (Chin et al., 2006).
Acute symptomatic—Status epilepticus in a previously neurologically normal child, within a week of an underlying etiology including CNS infection, prolonged febrile seizures, encephalopathy, head trauma, cerebrovascular disease, and metabolic or toxic derangements
Remote symptomatic—Status epilepticus in the absence of an identified acute insult but with a history of a pre-existing CNS abnormality more than 1 week before
Idiopathic epilepsy related—Status epilepticus that is not symptomatic and occurred in children with a prior diagnosis of idiopathic epilepsy or when the episode of status epilepticus is the second unprovoked seizure that has led to a diagnosis of idiopathic epilepsy
Cryptogenic epilepsy related—Status epilepticus that is not symptomatic and occurred in children with a prior diagnosis of cryptogenic epilepsy or when the episode of SE is the second unprovoked seizure that has led to a diagnosis of cryptogenic epilepsy
Unclassified—Status epilepticus that cannot be classified into any other group
Thus a revised classification system of the etiologies of CSE, with febrile CSE as a distinct category, may be more applicable to pediatric epidemiological studies while retaining the flexibility to present data in keeping with the current ILAE guidelines and thereby facilitate comparison with other studies. Nonetheless, such a revised classification would still have its limitations as accurate categorization of individual cases may depend on the degree of investigation, the availability of ancillary tests and clinical data.
One of the main strengths of epidemiological studies is that in principle, they are free from the referral bias that may occur in hospital based studies. The higher the degree of case ascertainment the greater the likelihood that results will be more applicable to the general population. Case ascertainment poses a problem in studies on CSE. The onset of CSE is not always witnessed and the time of seizure termination is not always noted. Prospective studies demonstrate that when a detailed history is obtained, the duration of seizure activity in children presenting at the emergency department may be underestimated and thus, cases of CSE may be missed (DeLorenzo et al., 1996; Chin et al., 2006). Even when the diagnosis of CSE is recognized, cases may still be missed because of lack of reporting cases and problems associated with coding of diagnoses (Rahi and Dezateux, 1999). It is likely therefore that epidemiological studies on CSE may miss cases unless sensitive screening methods are used and strategies need to be employed to adjust crude estimates for degree of ascertainment. Multiple sources of identification of cases are superior to a single source and active surveillance systems are better than passive systems (Knowles et al., 2006).
aRestricted to generalized convulsive status epilepticus.
bRestricted to convulsive status epilepticus.
The incidence of CSE in childhood is highest among children less than 1 year of age (51/100,000/year) compared to those aged 1–4 (29/100,000/year), 5–9 (9/100,000/year), and 10–15 years (2/100,000/year). The high incidence of CSE in children aged less than 1 year is particularly marked in children with acute symptomatic etiologies (Chin et al., 2006).
The higher frequency of CSE in very young and otherwise neurologically normal children may be related not only to the high proportion of acute symptomatic causes, but also to an increased propensity for seizures of the immature brain, which has been shown to be age dependent in several experimental models (Ben-Ari and Holmes, 2006). In addition to the lower seizure threshold and the high incidence of neurological insults that may induce CSE, there are several disorders that typically present with seizures during early childhood (i.e., congenital brain anomalies, genetic disorders, inborn errors of metabolism) (Arzimanoglou et al., 2004) and these children may have episodes of CSE as part of their epilepsy.
The role of gender, ethnicity, and socioeconomic status on the incidence of CSE
In adults, males are twice as likely as females to have an episode of CSE (Table 2) (Hesdorffer et al., 1998; Coeytaux et al., 2000; Knake et al., 2001). This gender difference may be partly due to a higher incidence of certain etiologies including cerebrovascular disease and brain trauma in males. However, it may also reflect a gender difference in seizure threshold (Moshe et al., 1995; Standley et al., 1995) or a possible role of hormonal influences in the termination of seizures, which is supported by the similar incidence observed in prepubertal boys and girls. Ethnicity may also be an important determinant of incidence of CSE given that a two- to threefold increase in nonwhite populations is reported in the California and Richmond studies (DeLorenzo et al., 1996; Wu et al., 2002). The role of socioeconomic status on incidence of CSE remains to be determined although preliminary results of the North London study support the view that this may have an important role (Chin et al., 2006b). If the relationship between CSE and socioeconomic status is confirmed then it is possible that the higher incidence of CSE observed in the Richmond study attributed to the racial composition of the population (57% black, 38% white) is confounded by the socioeconomic composition of Richmond. Indeed, Richmond, relative to the other municipal localities of Virginia has the highest percentage of people living in poverty (21.4% compared to 6.2%) (http://www.richmondgov.com/departments/budget/docs/Map_Poverty_Data.pdf).
Epidemiological studies conducted in developing countries have found higher incidence rates of epilepsy than in developed countries. Several factors such as infections (malaria, neurocysticercosis), trauma, or limited medical facilities are major determining factors for the increased incidence (de Bittencourt et al., 1996), although ethnicity and lower socioeconomic status may also play a role in the observed increased incidence.
The role of genetic factors on the incidence of CSE
The higher incidence of CSE in non-white populations suggests that differences in the incidence of CSE could be genetically determined, particularly if the effect of ethnicity is independent of socioeconomic status. This requires evaluation in future studies. The role of genetic factors is clear in some genetically determined epileptic syndromes (i.e., Dravet syndrome and CSE with fever, ring chromosome 20, and non-CSE) (Augustijn et al., 2001; Nabbout and Dulac, 2003), and twin studies suggest that there might also be some still unrecognized genetic predisposition to CSE in many other individuals (Corey et al., 1998; Corey et al., 2004).
Genetic backgrounds are being assessed for their potential as modulators of adverse outcome, particularly development of mesial temporal sclerosis (MTS), following CSE and this may clarify whether CSE is a cause or symptom of epilepsy (Haut et al., 2004). However, genetic association studies have given conflicting results. For example, homozygosity for a low-expression allele for dynorphin (an anticonvulsant peptide believed to be released during seizures) has been associated with status epilepticus and heterozygosity of this allele has been associated with the presence of familial temporal lobe epilepsy. This may imply a common genetic link between the allele, temporal lobe epilepsy, and status epilepticus (Stogmann et al., 2002), although these results have not been replicated in subsequent studies (Gambardella et al., 2003; Tilgen et al., 2003). Genetic factors may also play a role on the pathogenesis of CSE-related MTS. A higher frequency of interleukin-1ß-511 polymorphism occurrence in Japanese patients with temporal lobe epilepsy associated with MTS has been reported, with the maximum increase being observed in patients with a previous episode of febrile CSE (Kanemoto et al., 2000; Kanemoto et al., 2003). Again, subsequent studies examining the same polymorphism carried out in American, Chinese, German, and Turkish populations have failed to replicate these results (Heils et al., 2000; Buono et al., 2001; Jin et al., 2003; Ozkara et al., 2006), and it has been suggested that the initial positive association between the interleukin-1ß-511 polymorphism may have arisen by chance (Tan et al., 2004). Thus, the genetic basis underlying temporal lobe epilepsy or MTS in patients with prior history of prolonged seizures remains to be elucidated.
Place of CSE in the course of seizure disorders
Overall, between 10% and 20% of children with epilepsy will have at least one episode of CSE during the course of the disease, with most occurring in the first few years of epilepsy onset (Sillanpaa and Shinnar, 2002; Berg et al., 2004). Indeed, CSE is commonly seen as the first manifestation of seizure disorders, especially in younger children: between 62% and 88% of children with first-ever episodes of CSE in population-based studies do not have prior epilepsy (DeLorenzo et al., 1996; Chin et al., 2006), although this proportion is age dependent. In young children, the majority of cases of CSE occur in patients with no history of seizures rather than as part of an established seizure disorder and there are a higher proportion of older children with prior seizures or epilepsy. It has been speculated that, because the peak incidence of seizure disorders is in the first years of life, younger children are more likely to have CSE as their first seizure episode, whereas older children are more likely to have CSE as part of an already established seizure disorder. These differences are not only explained by the age-dependent etiological profile, as the proportion of children with prior epilepsy is also different when the analysis is restricted to children with cryptogenic or remote symptomatic epilepsy (Shinnar et al., 1997). In North London, when CSE is associated with idiopathic/cryptogenic epilepsy, 62% have a history of epilepsy or it is the second seizure (29%) or first seizure (10%) leading to the diagnosis of epilepsy (Chin et al., 2006).
A prospective study in which children with newly diagnosed epilepsy were recruited found that 9.1% of children had one or more episodes of status epilepticus by the time the diagnosis of epilepsy was made. Correlates of epilepsy differed between provoked CSE (i.e., young age at onset and nonidiopathic syndromes) and unprovoked CSE (i.e., focal seizures and craniotomy) (Berg et al., 1999b). After a median follow-up of 8 years, subsequent CSE had occurred in 7.2% of children without and in 32.1% of those with CSE at diagnosis, i.e., those children who have experienced an episode of CSE at, or soon after, diagnosis are much more likely to have another episode during subsequent follow-up (Sillanpaa and Shinnar, 2002; Berg et al., 2004). In addition, children with new onset seizures largely fall into two groups; those that have short seizures (<5 min in duration) and those that are likely to have 30 min seizures unless there is a clinical intervention (Shinnar et al., 2001a). Thus, there might be a subgroup of children with epilepsy who are predisposed to prolonged seizures.
Remote symptomatic epilepsy, age at onset (<1 year or >60 years), and focal seizures were independent predictors of CSE occurring at or after the diagnosis of epilepsy in the Rochester population-based study (Hesdorffer et al., 1995). Focusing on children in the remote symptomatic group, a retrospective case–control hospital-based study found that focal background abnormalities, secondarily generalized focal seizures, first seizure as CSE, and generalized abnormalities on neuroimaging were indicators of a higher risk of subsequent CSE (Novak et al., 1997).
Time trends in incidence of CSE
The Rochester study showed an increase in the overall incidence of status epilepticus across a 30 year period (Logroscino et al., 2001). The study included mainly adults and the reported increase in incidence was largely due to the increase in myoclonic status epilepticus after cardiac arrest although methodological issues (i.e., better recognition of more subtle forms of status epilepticus) could also have accounted for the apparently time-dependent increase of incidence. When only generalized CSE is considered, two population-based studies have shown a decline in its incidence (Logroscino et al., 2001; Wu et al., 2002).
There are no published studies specifically addressing time trends of incidence of CSE in children. There is some evidence that the incidence of epilepsy over time has decreased in children (Hauser et al., 1993; Sander and Shorvon, 1996; Everitt and Sander, 1998). Whether these results also apply to incidence of CSE in children remains to be shown. More prolonged survival of children with severe underlying neurological conditions may favor an increase in the incidence of pediatric CSE, while the improvement of perinatal and overall medical care, including earlier treatment of prolonged seizures, may have reduced the incidence of established CSE.
The effect of early treatment on incidence
In more than three-quarters of children with a first-ever episode of CSE, the episode starts in the community. Although prehospital treatment is widely recommended, in practice the use of such treatment may be suboptimal. Many children are either not treated or are treated with doses that are below those suggested in guidelines such as that produced by the APLS group based in the United Kingdom. A retrospective hospital-based study emphasized that inappropriate treatment, including no prehospital treatment or excessive administration of benzodiazepines, contributes to the need for intensive care (Chin et al., 2004b).
A population-based study conducted in Bologna, Italy, reported a mortality rate of 39% in adults with all types of status epilepticus (Vignatelli et al., 2003). The authors hypothesized that this extremely high case fatality rate might be in part secondary to inadequate management of status epilepticus, i.e., administration of prehospital treatment in only 17% of cases, lack of local hospital protocols, and use of diazepam as first-line drug (instead of lorazepam). In contrast, the population-based study conducted in Switzerland reported a much lower overall case fatality rate at 7.6%, and this was in part attributed to initiation of treatment in the prehospital setting in almost 60% of patients (Coeytaux et al., 2000). Seizure clustering is also associated with an increased risk of CSE, and thus rapid treatment of seizure clusters has also been suggested as a way of reducing the incidence of CSE (Haut et al., 2005).
It is possible that early pharmacological intervention leads to termination of seizures with smaller doses than would be required if seizures were allowed to progress (Mazarati et al., 1998). Experimental data show time-dependent loss of synaptic GABAA receptors, and thus of GABA-mediated inhibition, which correlates with the progressive pharmacological resistance to GABAergic medication observed in refractory status epilepticus (Kapur et al., 1989), further supporting the view that treatment should commence in the prehospital setting.
Several studies have reported on the efficacy and safety of transmucosal benzodiazepines for the acute management of seizures in children and it has been suggested that their use in the prehospital setting may improve the outcome (Scott et al., 1999; Lahat et al., 2000; McIntyre et al., 2005). It can be hypothesized that the improved outcome may be achieved not only by decreasing the duration of CSE and facilitating its subsequent management in the hospital setting, but also by an actual decrease in the incidence of CSE (assuming a 30 min definition). Future epidemiological studies should aim to investigate whether the extensive implementation of prehospital treatment does not only improve the outcome but also reduces the incidence of established CSE in children.
It can be concluded that the incidence of CSE is mainly a function of age, which in turn may reflect differences in maturation of the developing brain and in the etiological profile of particular age groups or geographical areas. Treatment facilities, gender, ethnicity, and genetic and socioeconomic factors also seem to influence the epidemiology of CSE. Thus, the incidence in populations will vary depending upon all the above factors. Studies to identify specific genetic or socioeconomic factors associated with CSE are required to pinpoint potential targets for primary prevention of CSE.
SEIZURE TYPE: ONSET, CHARACTER, AND DURATION
The reported frequencies of the initiating seizure type in pediatric CSE are discordant. The Richmond study reported that focal seizures, defined on clinical evaluation, were the initiating seizure type in nearly two-thirds of cases, but was the initiating seizure type in only one-third of children in the North London study (DeLorenzo et al., 1996; Chin et al., 2006). However, both studies showed a high rate of secondary generalization, with generalized tonic–clonic CSE being the most common final seizure type.
Tonic and purely clonic seizures are uncommon. The incidence of tonic CSE among children with first-ever episodes peaks in children less than a year, especially in those with acute symptomatic etiologies (Chin et al., 2006). In hospital-based studies, tonic CSE is mainly reported in children and adolescents with previous epilepsy, particularly in patients with Lennox–Gastaut syndrome (Arzimanoglou et al., 2004). Purely clonic seizures are reported in less than 5% of children in population-based studies (Coeytaux et al., 2000; Chin et al., 2006). It has been argued that there are a proportion of prolonged seizures as documented clinically that may be nonepileptic events (Stephenson, 2006). All studies assessing seizure types may therefore contain such patients.
Not all seizures are continuous in their nature, i.e., there are a proportion of seizures that appear to terminate but the patient does not recover consciousness and subsequently has further clinical events. This is defined as intermittent CSE. The incidence of intermittent and continuous CSE in children is similar amongst all age and etiological groups (Chin et al., 2006). It has been shown that continuous CSE is associated with higher mortality than intermittent CSE in adults even after controlling for CSE duration, but this may not be the case in children (Waterhouse et al., 1999).
In addition to offering prompt medical treatment, seizure duration may also depend upon etiology and age. In the North London study, 60% of children had CSE lasting longer than 1 h with no differences in the duration of CSE noted between different etiological groups (Chin et al., 2006). This finding is contrary to those from some hospital-based studies that report acute symptomatic CSE to be associated with a longer duration of CSE (Maytal et al., 1989; Eriksson and Koivikko, 1997; Tabarki et al., 2001). In the Rochester study the risk of CSE of longer duration was greatest for infants and for the elderly. Additionally, among cases with first-ever acute symptomatic seizures and unprovoked seizures, the proportion with CSE was increased in infancy and in the elderly, thus suggesting that age may also modulate duration (Hesdorffer et al., 1998).
The cause of CSE varies across age groups and there is a strong correlation between age at the time of CSE and etiology: in children younger than 2 years, febrile CSE and acute symptomatic etiologies are most common, whereas cryptogenic and remote symptomatic etiologies are more common in the older children (Shinnar et al., 1997).
Several population-based studies have reported on the etiological distribution of CSE in childhood (Table 3). Although most have been conducted after the ILAE's published guidelines for epidemiologic studies in epilepsy (Commission on Epidemiology and Prognosis, 1993; ILAE Commission Report, 1997), differences in etiological criteria have resulted in discordant results. As previously discussed, according to the ILAE's guidelines febrile seizures should be classified as acute symptomatic seizures. This was the criterion used in the Richmond and Swiss studies, in which children with febrile CSE were included within the acute symptomatic group (DeLorenzo et al., 1996; Coeytaux et al., 2000). However, in the Rochester and North London studies, children with FSE were classified and analyzed separately on both pragmatic grounds and because of the probable lack of direct CNS involvement in febrile seizures (Hesdorffer et al., 1998; Chin et al., 2006).
Table 3. Etiology of status epilepticus according to population-based incidence studies that included children (the Californian study is excluded because proportional contributions of etiologies from their study could not be determined)
Richmond (n = 29)
Rochester (n = 69)
Switzerland (n = 64)
North London (n = 176)
aFebrile CSE was included in the acute symptomatic group.
bData for febrile CSE were not provided separately but 58% of acute symptomatic seizures in children <5 years old were reported to be febrile.
Acute on remote
Febrile CSE, which occurs in 5% of patients experiencing febrile seizures (Annegers et al., 1987; Verity et al., 1993), is the most common type of CSE in childhood, accounting for at least one-third of all cases. Although by definition, febrile seizure is restricted to children younger than 6 years, the age-adjusted incidence of febrile CSE is still greater than that for each of the other causes of CSE across the whole of childhood (Chin et al., 2006).
In a subgroup analysis of 95 children from the North London study with first-ever episodes of CSE associated with fever, 12% had acute bacterial meningitis and 8% had a viral CNS infection compared with a rate of 1–2% in children with a short febrile seizure (Chin et al., 2005). The remaining children had febrile CSE (59%) or had a previous neurological abnormality with a febrile intercurrent illness (22%). Previously neurologically normal children with first-ever CSE associated with fever were seven times more likely to have an acute CNS infection compared to children with first-ever CSE associated with fever but with a preexisting neurological abnormality (Chin et al., 2006). Thus, CNS infection should be carefully ruled out in children with fever and CSE, especially in those who are previously neurologically normal (Chin et al., 2005).
Each of the other etiologies (i.e., idiopathic/cryptogenic, acute symptomatic, remote symptomatic, and progressive) account for 15–20% of total. Low antiepileptic drug level was reported as the cause of 21% of CSE in the Richmond study, but was only observed in one child in the North London study; the much lower proportion of children with prior epilepsy in the latter study may account for the difference (DeLorenzo et al., 1996; Chin et al., 2006).
More than 40% of children with first-ever CSE in the North London study were previously neurologically abnormal (i.e., abnormal neurodevelopment, history of epilepsy or neurological deficits). This figure is similar to previous studies and suggests that neurologically abnormal children are more susceptible to develop seizures in general and CSE in particular. The proportion of previously neurologically abnormal children is age dependent: 21% of children younger than 2 years are neurologically abnormal compared to 43% of children aged 2–5 years and 59% of children older than 5 years (Shinnar et al., 1997).
Much of the importance attached to CSE is based upon its morbidity and mortality. Although there is an increase in morbidity and mortality in CSE this seems to be mostly related to etiology and is less in children than in adults. The role of the epileptic discharges in generating adverse outcomes requires further investigation and there is still much controversy as to whether age, duration, or treatment modifies the outcome of CSE. Much of the controversy arises from methodological differences of studies addressing this topic (Hauser, 1983; Logroscino and Hesdorffer, 2005). The impact of nonbiological variables on reported outcomes of pediatric CSE was investigated in a recent systematic review that concluded that prospective design and overall better methodological quality were associated with better outcome (Raspall-Chaure et al., 2006).
Age and duration of CSE may also affect mortality although their effects need further clarification. The higher mortality reported in younger children in some studies may only reflect the higher proportion of acute symptomatic cases in this age group (Aicardi and Chevrie, 1970; Logroscino et al., 1997). The Richmond study, using CSE lasting <1 h as the reference group, found longer duration to be associated with a higher 30-day mortality (Towne et al., 1994). In contrast, duration of CSE was not a predictor of short-term mortality in the Rochester study, even after restriction of the analysis to acute symptomatic causes: no significant differences were observed when individuals with CSE lasting less than 2 h (reference group) were compared to those with CSE lasting 2–24 h or >24h (Logroscino et al., 1997). In a retrospective study conducted over a 11 year period in Dakar, Senegal, the mean time between the onset of symptoms and the initiation of treatment was 16.6 h. Despite the delay in the initiation of treatment, overall mortality (24.8%) was similar to that observed in Richmond or Rochester, which may suggest that outcome is not significantly influenced by the duration of the episode of CSE (Mbodj et al., 2000). However, none of these studies were carried out exclusively in pediatric populations and therefore it remains uncertain whether duration of CSE influences mortality in children.
Long-term mortality data after a first-ever episode of CSE are variable, with estimates ranging from 5.4% to 17%. At 10 years follow-up, mortality was 3% for 30-day survivors aged 1–19 years and 16% for infants <1 year in the Rochester study. As with short-term mortality, a higher mortality was seen in symptomatic CSE, but not with cryptogenic or idiopathic CSE (Logroscino et al., 2002). Another population-based study reporting 24 deaths among 150 patients with childhood-onset epilepsy after a follow-up of 30 years did not identify a higher mortality in those with prior CSE (Sillanpaa and Shinnar, 2002). These population-based studies suggest that CSE itself may not have a significant impact on long-term mortality in childhood-onset epilepsy. To some extent these results may be comparable to epidemiological studies on mortality of epilepsy showing no significant increase in mortality in people with idiopathic epilepsy (Forsgren et al., 2005).
To assess the contribution of CSE itself to mortality it is necessary to either study idiopathic–cryptogenic cases or to compare the mortality of people with first-ever CSE and an underlying condition to the mortality of individuals from the same population with the underlying condition but without CSE stratified by severity of the underlying condition (Logroscino and Hesdorffer, 2005). Although this has already been carried out in studies in adults with stroke and status epilepticus (showing that there is a synergistic effect on mortality) (Waterhouse et al., 1998; Knake et al., 2006), we are not aware of any such study in children.
In addition to epilepsy, focal neurological deficits, cognitive impairment and behavioral problems can be associated with CSE although consistent specific risk factors for each of these adverse outcomes are not reported (Raspall-Chaure et al., 2006).
Etiology is also the main determinant of morbidity. The poorest outcome is observed in acute symptomatic CSE, which is followed by new neurological dysfunction in >20% of cases. In the absence of an acute or progressive neurological disorder, morbidity of CSE is low, and <15% of children with febrile CSE and unprovoked CSE develop new neurological deficits attributable to CSE (Maytal et al., 1989; Eriksson and Koivikko, 1997; Barnard and Wirrell, 1999; Shinnar et al., 2001b). However, low power and inadequate methodology might have underestimated the incidence of minor sequela in most of the studies. This is especially true for population-based studies on CSE as none have applied formal neurocognitive assessments.
A recent hospital-based study in adults with epilepsy showed no cognitive deterioration after an episode of status epilepticus (6 CSE, 9 non-CSE), as demonstrated by the Wechsler Adult Intelligence Scale—Revised before and after the episode (Adachi et al., 2005). Despite previous evidence to the contrary (Dodrill and Wilensky, 1990) and lack of similar studies in children, these results suggest that patients with earlier epilepsy and no acute medical illnesses do not have long-term cognitive deficits from CSE itself.
Other factors that are reported to influence the outcome are longer seizure duration and younger age at onset. As with mortality, the poorer outcome observed in refractory or more prolonged CSE and in young children might be only explained by the greater incidence of acute symptomatic CSE in these groups (Raspall-Chaure et al., 2006).
A significant proportion (16%) of children with first-ever CSE will have a recurrence within a year (Chin et al., 2006). The risk is mainly determined by the etiology, and it is highest in the remote symptomatic (44%) and progressive (67%) groups (Shinnar et al., 1992). Among children with first-ever episodes of febrile CSE, 17% will have another episode within a year. This contrasts with another population-based study in which no recurrences were observed (Verity et al., 1993). Children with existing neurological abnormalities are 3–23 times more likely to have a recurrence than those who are previously neurologically normal (Shinnar et al., 1992; Chin et al., 2006). The median interval from first-ever CSE to first recurrence is 25 days (95% CI 0–78) and there is no difference in the mean interval from first-ever CSE to first recurrence between children with and without previous neurological abnormalities (Chin et al., 2006). The risk of recurrence is at its maximum during the first year after the episode of CSE, but it does not disappear afterwards. In fact, the highest risk of recurrence has been reported in two population-based studies with 10 and 30 years of follow-up (Verity et al., 1993; Sillanpaa and Shinnar, 2002). Thus, differences in reported risk of recurrence might depend not only on the etiological distribution of the series, but also on the length of follow-up.
Experimental studies suggest that prolonged febrile seizures in the immature brain produce persistent enhancement in hippocampal excitability, which may facilitate the subsequent development of epilepsy if the animal is exposed to other epileptogenic insults (Dube et al., 2000). Clinical data on epilepsy following febrile CSE are controversial and much of the variability of estimates arises from differences in inclusion criteria. Studies that include children with prior neurological abnormalities reveal that, when compared to brief febrile seizures, the risk following febrile CSE is not different in neurologically normal children but it is significantly increased (38%) in those with prior neurological abnormalities. Nelson and Ellenberg (1978) reported that 4.1% of children with first febrile seizure as febrile CSE developed epilepsy, which is significantly higher than in the normal population, but did not reach statistical significance when compared to children who had brief febrile seizures. In contrast, Verity et al. (1993) found a significantly greater risk of developing afebrile seizures in children with febrile CSE compared with children with brief febrile seizures (21% vs. 3.4%; x2= 9.77; p<0.005).
It has long been hypothesized that CSE (and in particular febrile CSE) can cause MTS and associated temporal lobe epilepsy. Retrospective studies from tertiary epilepsy centers highlight this association, with a history of febrile CSE being present in 35–63% of patients with MTS (Cendes et al., 1993; Murakami et al., 1996). However, neither population-based nor prospective hospital-based studies report a significant association between CSE in childhood and subsequent MTS (Camfield et al., 1994; Berg et al., 1999a; Tarkka et al., 2003). Although febrile CSE has been associated with increased incidence of subsequent partial seizures, the structural bases for these seizures has not been characterized (Annegers et al., 1987). There are, however, studies that show evidence for an acute hippocampal insult following febrile CSE in children and this may represent the first part of a causative pathophysiological sequence linking febrile CSE to MTS (VanLandingham et al., 1998; Scott et al., 2002; Scott et al., 2003).
The risk of seizure recurrence is highest during the first year following CSE and tends to decrease with increasing interval from the index seizure. However, the positive correlation between reported incidence of seizure recurrence and length of follow-up suggests that the risk of subsequent epilepsy may remain for many years after the initial episode of CSE (Raspall-Chaure et al., 2006). The impact of early treatment on the prevention or course of epilepsy remains largely unknown, and the current recommendation of not starting long-term treatment after a first unprovoked CSE is based on the reported low impact of CSE on the risk of recurrence.
SUMMARY AND FUTURE DIRECTIONS
The population-based studies that have been conducted in Europe and the United States have clarified the incidence of CSE in children and have identified some putative risk or modifying factors for CSE in developed countries, i.e., age, ethnicity, genetics, and socioeconomic status. However, many of the objectives for “future epidemiological studies” suggested by the ILAE in order to improve our knowledge on the epidemiology of seizure disorders have not been met yet (ILAE Commission Report, 1997). There is still a need for further studies with appropriate methodology to answer some as yet unclarified questions, i.e., (a) the magnitude of geographic variations in incidence, (b) the relative contribution of different etiological profiles, genetic background, and socioeconomic status to these variations, (c) the incidence rates specific for each type of CSE, and (d) the effect of widespread implementation of prehospital treatment for prolonged seizures on incidence rates. As recommended by the ILAE, studies should ideally be population based and prospective, and all rates should be age adjusted.
Although long-term prospective population-based and appropriately conducted hospital-based studies consistently report that etiology is the main determinant of outcome, the relationship of CSE with MTS or subtle neurocognitive dysfunction, and the effect of age at CSE, seizure duration, or treatment on outcome have not yet been clarified. Future studies controlling for the severity of the underlying etiology should aim to elucidate these points in order to determine how much resource should be put into the prevention and treatment of CSE.
Acknowledgments: Miquel Raspall-Chaure is the recipient of a research fellowship from The National Centre for Young People with Epilepsy, Lingfield, UK. Funding sources played no part in the preparation of this review or in the decision to submit it for publication.