Long-term Prognosis for Symptomatic (Secondarily) Generalized Epilepsies: A Population-based Study


Address correspondence and reprint requests to Dr. P. Camfield at Department of Pediatrics, Dalhousie University, and the IWK Health Centre, 5850 University Ave, PO Box 9700, B3K 6R8, Halifax, Nova Scotia, Canada. E-mail: pcamfiel@dal.ca


Summary: Purpose: To describe the long-term outcome of childhood-onset secondarily generalized epilepsies (SGEs).

Methods: Children were identified from the Nova Scotia population-based epilepsy study (n = 692). Onset of epilepsy was between 1977 and 1985, and follow-up was mainly in 2003. SGE was defined as having a mixture of more than one generalized seizure types including myoclonus, akinetic/atonic, tonic, or atypical absence, plus an interictal EEG with generalized spike–wave (irregular or slow) and/or multifocal spikes.

Results: SGE was identified in 80 children, 11.6% (80 of 692) of all childhood epilepsy. Forty percent did not fit into a currently accepted syndrome (such as Lennox–Gastaut, myoclonic–astatic or West). Seizure onset was in the first year of life in 60%, with only 9% later free of intellectual or physical handicap. Sixty-five percent (11 of 17) with Lennox–Gastaut had preceding West syndrome. During a median follow-up of 20 years, mortality was 24% (n = 19), and 53% (n = 42) had persistently intractable seizures. Surprisingly, 22 (28%) had ≥5 years of terminal remission (West, 31%, 10 of 32; Lennox–Gastaut, 0, none of four; myoclonic–astatic, 56%, five of nine; undefined, 31%, 10 of 32). At the end of follow-up, nearly 90% of patients fell into one of three outcome categories: death, 19 (24%); alive with intractable epilepsy, 31 (39%); or in remission for ≥5 years, 21 (26%).

Conclusions: Many children with SGE have ill-defined epilepsy syndromes. SGE is characterized by early age at onset, high rates of handicap, intractability, and death, although one third achieve complete seizure control with a long terminal remission.

Symptomatic or secondarily generalized epilepsies are thought to have a notoriously bad long-term outcome. The International League Against Epilepsy (ILAE) classification lists six separate syndromes under this category (West syndrome, Lennox–Gastaut syndrome, myoclonic–astatic epilepsy, epilepsy with myoclonic absences, early myoclonic encephalopathy, early infantile encephalopathy with burst suppression, or Ohtahara syndrome). We note that this classification is somewhat ambiguous because West syndrome, Lennox–Gastaut syndrome, myoclonic–astatic epilepsy, and epilepsy with myoclonic absences are listed as either cryptogenic or symptomatic. In addition, four syndromes are listed that may or may not be generalized [neonatal seizures, severe myoclonic epilepsy of infancy (Dravet syndrome), continuous spike–wave in slow sleep, and Landau–Kleffner syndrome) (Commission on Classification and Terminology, 1989; Engel, 2001). The SGE syndromes are complex and controversial. Some of the syndromes are imprecisely defined in the literature. For example, many experts insist on nocturnal tonic seizures for a diagnosis of Lennox–Gastaut, but others do not. In a comprehensive literature review, Beaumanoir and Blume noted, “certain authors continue to emphasize one or the other of its elements constituting the syndrome—usually the slow spike–waves … . this non-respect of the inclusion criteria for LGS makes a reliable estimation of its prevalence difficult” (Beaumanoir and Blume, 2005). The inclusion of some syndromes in the SGE category has been debated. For example, myoclonic–astatic epilepsy is considered by some authorities to be an idiopathic epilepsy (Guerrini et al., 2005), even though it is listed as an SGE in the ILAE classification (Commission on Classification and Terminology, 1989). Other SGE syndromes, such as Ohtahara syndrome, are quite rare despite a voluminous literature.

Current knowledge about the long-term outcome of SGE is based mainly on case series from specialized centers; the relative frequency of SGE syndromes and their long-term outcome have not been addressed in population-based studies. We report the seizure outcome for SGE syndromes by using the Nova Scotia population-based cohort with a 20-year follow-up.


The Nova Scotia childhood-onset epilepsy study is based on a population-based cohort of children with newly diagnosed epilepsy. All cases had their first two seizures between 1977 and 1985. Details of the case-finding method have been previously published (Camfield et al., 1993). In brief, cases were identified from a central EEG facility where all pediatric EEGs for the province of Nova Scotia, Canada, are interpreted. Physicians in Nova Scotia have indicated that when they suspect epilepsy, they always request an EEG. Therefore EEG requests formed the basis of a comprehensive case-finding process. We reviewed all EEG requests for the study period. When symptoms noted on the EEG requisition raised the possibility of epilepsy, patients' medical records were reviewed, and if the diagnosis of epilepsy was unclear, the family was contacted directly for further details. At the time that the cohort was assembled, the population of Nova Scotia was ∼850,000. In 2003–2004, the medical records of children with SGE were again reviewed, and patients or their caregivers were interviewed either by telephone or face-to-face to confirm the information in the medical record and complete further details about the evolution of the patient's epilepsy. For the nine patients that we were unable to contact directly in 2003–2004, we used data available from our medical record (child neurology charts) to document their clinical course until the last contact. All but one of the families with a child who died were contacted directly. All children were assessed and followed up at some point in their illness by a child neurologist from the IWK Children's Hospital, the tertiary pediatric facility for Nova Scotia.

Children with neurodegenerative disorders or CNS tumors were excluded. We also excluded neonatal seizures, because many of these patients have only acute symptomatic seizures. We included the syndromes that were “undetermined as to whether they are focal or generalized,” and we collapsed the category of Cryptogenic/Symptomatic into the Symptomatic group (Commission on Classification and Terminology, 1989).

This study was approved by the Ethics Review Board of the IWK Health Centre, Halifax, Nova Scotia.


We followed the ILAE International classification system for epilepsy syndrome type, with some modifications for clarity (Commission on Classification and Terminology, 1989). For this article, the following criteria were applied independently by both authors with consensus discussion when disagreement occurred. West syndrome included only those with infantile spasms, age at onset of ≤12 months, and EEG near the time of seizure onset showing hypsarrhythmia or modified hypsarrhythmia. Myoclonic–astatic epilepsy (MAE) was diagnosed in children with myoclonus + drop attacks (akinetic/atonic) and other generalized seizure types, except nocturnal tonic seizures, in a previously normal child (no structural lesion on brain imaging). An EEG close to the time of diagnosis had to show interictal generalized spike-and-wave of any kind. Lennox–Gastaut syndrome (LGS) was restricted to those with nocturnal tonic seizures plus at least one other generalized seizure type plus an interictal EEG with generalized slow spike-and-wave. Secondarily generalized epilepsy, undefined was diagnosed if a child did not meet the criteria for one of the other syndromes and had more than one generalized seizure type, including one of myoclonus, akinetic/atonic, drop attacks, tonic, atypical absence, and an interictal EEG with generalized spike–wave (irregular or slow) and/or multifocal spikes. The ILAE classification scheme also includes such a broad category without more specific definition (Commission on Classification and Terminology, 1989).

Intractable epilepsy was defined as at least one seizure per 3 months in the last year of follow-up despite trials of at least three appropriate AEDs at maximal tolerated doses.

Statistical analysis used χ2 for categoric data and t tests for continuous variables (SPSS version 12.0, SPSS, Chicago, IL, U.S.A.).


In total, 692 children (ages 1 month to <16th birthday) had the onset of their epilepsy in Nova Scotia between 1977 and 1985. The 80 patients with SGE represented 11.4% of all childhood epilepsy. The remainder of the results in this article are based on these 80 cases.

Boys outnumbered girls slightly (54 vs. 46%). Age at onset averaged 22 months (range, 1–144 months), and 59% had their first seizures in the first year of life. During follow-up, 19 (24%) patients died at an average age of 139 ± 75 months (range, 17–243 months). Of those who died, 79% (n = 12) had the onset of epilepsy in the first year of life. All 19 who died had a severe neurologic handicap (n = 17) or isolated moderate mental retardation (n = 2). None of the deaths was thought be directly caused by seizures; rather they were the result of the underlying neurologic disorder.

For the 61 survivors, their age at the end of follow-up averaged 255 ± 83 months (range, 23–417 months), and the length of total follow-up from onset to last contact averaged 231 ± 78 months (range, 18–313). Only two patients had a <2-year follow-up, whereas 87% had >14 years of follow-up.

Most patients had neurologic or intellectual deficits. Neurologic deficit sufficient to interfere with activities of daily living was present at onset in 47 (59%) and 50 (63%) at the end of follow-up or at the time of death. Mental retardation was noted at the end of follow-up in 72 (90%). Most with mental retardation (71%) had a severe or profound mental handicap (mild mental retardation, n = 5; moderate, n = 16; severe/profound, n = 51). Many of these patients were too handicapped for formal psychological testing (n = 39); however, for the remainder, formal psychometric testing during follow-up documented the level of intellectual deficit in 80%. Only 34% of the sample gained the ability to walk. Normal intelligence without neurologic deficit was noted in only seven (9%).

Epilepsy syndromes at the onset were symptomatic partial, one (evolved to West syndrome); SGE undefined, 32 West, 32; Lennox–Gastaut, four; MAE, nine; and Dravet, two. By the end of follow-up, the syndrome had changed in 30 (33%). Table 1 shows the details of how syndromes changed. All patients with MAE and Dravet syndrome retained the same syndrome diagnosis; however, each of the other categories showed considerable change. We did not document the age when these syndromes changed.

Table 1. Comparison of syndrome diagnosis at onset of epilepsy and final syndrome (end of follow-up or death)
Final syndromePartialUnspecifiedWestLennox–GastautMAEDravetTotal
Onset syndrome 
 Partial  1  1
 Unspecified 28 1 3 32
 West4 9 811 32
 Lennox–Gastaut  1  3  4
 MAE 9  9
 Dravet  2 2

Lennox–Gastaut syndrome was present in only 5% of patients at onset, but by the end of follow-up, it constituted 21%. Of the 17 who ended with Lennox–Gastaut, 11 (65%) initially had West syndrome. Of the 32 with an initial diagnosis of West syndrome, only eight retained this diagnosis, whereas partial epilepsy developed in four, Lennox–Gastaut in 11; and SGE, unspecified in 9.

The long-term outcome for these SGE patients were in three major categories (death, remission, and intractable epilepsy) and one minor category (persistent seizures, not intractable) (Table 3). Twenty–four percent died, 39% survived with intractable seizures, and 28% had a terminal remission for ≥5 years (92% had also successfully discontinued AED treatment). Only 10% had an outcome of partial seizure control. Another way of considering the outcomes is that 63% had the very unfavorable outcome of death or intractability, whereas 28% had the good outcome of remission. The outcome at the end of follow-up or death by the final syndrome diagnosis is shown in Table 2.

Table 3. Remission compared by syndrome at diagnosis (including those who died)
Syndrome at diagnosisTerminal remission ≥60 mo
Not specified (n = 32)10 (31%)
West (n = 32)10 (31%)
LGS (n = 4) 1 (25%)
MAE (n = 9) 5 (56%)
Dravet (n = 2)0
Symptomatic partial (n = 1)0
Overall (n = 80)26 (33%)
Table 2. Outcome at the end of follow-up compared with syndrome at diagnosis
  1. The last three categories exclude patients who died.

Death  9 (28%) 8 (25%)1 (25%)1 (11%)019
Intractable epilepsy111 (48%)14 (58%)2 (67%)1 (13%)231
≥5 yr terminal remission08905 (56%)022
Partial seizure control041120 8

Of the 19 patients who died, 11 also had intractable seizures up to the time of their death. Four of the remainder who died had persistent, nonintractable seizures, and four had discontinued AEDs before death after an average remission of 97 months.

The rate of remission compared with the syndrome at diagnosis is shown in Table 3. These rates were not statistically different between syndromes. Rates of intractability by syndrome at diagnosis and at the end of follow-up or death are shown in Table 4. MAE was the only syndrome with a very low rate of intractability (11%, p < 0.01; when compared with all others). Lennox–Gastaut at the end of follow-up had an increased risk of intractability (94%; p < 0.001 compared with all others and p < 0.01 compared with SGE Unspecified).

Table 4. Number with intractable epilepsy by syndrome at diagnosis and syndrome at the end of follow-up or death
Syndrome at diagnosisSyndrome at the end of follow-up or death
Not specified (total n = 32) 16 (50%)Not specified (total n = 38) 19 (50%)
West (total n = 32) 19 (59%)West (total n = 10) 2 (20%)
LGS (total n = 4) 3 (75%)LGS (total n = 17) 16 (94%)
MAE (total n = 9) 1 (11%)MAE (total n = 9) 1 (11%)
Dravet (total n = 2) 2 (100%)Dravet (total n = 2) 2 (100%)
Symptomatic partial (total n = 1) 1 (100%)Symptomatic partial (total n = 4) 2 (50%)


Our study has four major findings. First, symptomatic or SGEs account for only ∼10% of childhood epilepsy, based on our experience with an entire regional population. One other population-based study found that 10% of 150 incidence cases of childhood epilepsy had SGE (Sillanpaa et al., 1998). Two other large, nearly population-based studies of the natural history of childhood epilepsy (Connecticut and the Netherlands) noted similar rates of SGE (8% and 7%) (Berg et al., 1999; Arts et al., 2004). These disorders are among the most famous in childhood epilepsy and have received a great deal of attention. For example, in one recent authoritative textbook about childhood epilepsy, they account for one third of the total text (Epileptic Syndromes in Infancy, Childhood, and Adolescence, 2005), and in another, 16% of the text related to clinical descriptions (The Epilepsies, 2005). It is reassuring to note that in our overall cohort of 692 patients, in no other patients did SGE develop after a more-benign onset (Camfield et al., 1993).

A second major finding relates to the evolution of SGE syndromes. Within the broad category of SGE, we found that >30% of patients do not fit into a currently defined more-specific epilepsy syndrome, a similar finding to the study from Turku, Finland (Sillanpaa et al., 1999) Furthermore, the syndrome at diagnosis changes during the subsequent 20 years in ∼30%. Many of the best-known SGE syndromes are extremely rare and were not noted in a single case (e.g., Ohtahara syndrome).

The evolution of West syndrome to Lennox–Gastaut syndrome has been noted previously (Weinmann, 1988). Our study suggests that about one third of children with West syndrome will evolve to Lennox–Gastaut. In a German clinic series, Weinmann noted that in 29.2% of 140 children with West syndrome, eventually Lennox–Gastaut syndrome developed (Weinmann, 1988). We note that Lennox–Gastaut syndrome is rarely the presenting syndrome in SGE (only 5% of SGE). Conversely, by 20 years after the onset of SGE epilepsy, Lennox–Gastaut syndrome becomes more frequent and encompasses ∼20% of SGE. Weinmann also noted that 17.8% of 174 children with Lennox–Gastaut syndrome had preceding West syndrome (Weinmann, 1988).

A third major finding is the seizure prognosis for SGE. The long-term outlook for various SGE syndromes has been examined in a variety of clinic samples. We believe that our study is the second (Sillanpaa et al., 1999) and largest to document this prognosis in a population-based sample. It is difficult to compare our results with the very long and comprehensive follow-up study of Sillanpaa et al., who classified ∼10% of 150 children with new-onset epilepsy as having SGE (as defined in this article). Their follow-up data includes an additional 95 prevalence cases with all epilepsy syndromes, but the seizure outcome and mortality for the incidence cases are not separately reported (Sillanpaa et al., 1998). Another study from Finland is likely population-based with a 20- to 35-year follow-up but devoted to West syndrome only (Riikonen, 2001). The author noted that of 214 patients, one third died, and about one fourth had a favorable outcome, findings very similar to those of our study. We did not have sufficient patients in any specific syndrome group to warrant statistical predictions of outcome for specific syndromes.

Across all the SGE syndromes, after a median of 20 years of follow-up, 90% of the children were in one of three outcome categories: death in 24%, intractable epilepsy in 39%, and terminal remission of ≥5 years in 28%. Only 10% had an outcome of partially controlled seizures. In essence, about two thirds have very unfavorable outcomes (death or intractability). The only syndromes with a relatively uniform prognosis were myoclonic–astatic epilepsy, with only one of eight ending with intractable epilepsy, and Dravet, with two of two with persisting epilepsy.

Our fourth finding relates to mortality. Within 20 years of diagnosis, about one fourth of children with SGE died. In a previous study, we described the details of the death of most of these children and found that none of them died as the direct result of seizures (Camfield et al., 2002). The cause of death was related to their underlying neurologic deficit, particularly bulbar difficulties with recurrent aspiration. Death was particularly associated with onset of SGE before 1 year of age, severe neurological disability, and intractability; however, the average age at death was nearly 12 years of age, so the burden of care for these families was significant. The childhood epilepsy study from Connecticut noted that of 15% of 66 children with “epileptic encephalopathy” died within 8 years of diagnosis (Berg et al., 2004)

Our study has a number of limitations. The classification system that we used for epilepsy syndromes dates from 1989, and much has been learned since then (Commission on Classification and Terminology, 1989). However, the 1989 ILAE classification is the most recent to be approved by the ILAE. The more recent proposal from 2001 remains unapproved (Engel, 2001); however, even this proposal does not provide new syndrome definitions that would have accounted for patients in our unclassified group. We based our outcome data primarily on the reports of parents and caretakers; video-EEG was rarely used in this sample of patients. It is possible that some of the patients reported to be in remission are still having short seizures, especially during sleep.

Based on this study, we suggest that as treatment modalities improve for SGE, priority should be given to a more comprehensive classification system for the one third who have no clear epilepsy syndrome. For each SGE syndrome, early and accurate predictors for remission or intractability are needed to justify more experimental treatments for a group of disorders that often but not invariably have a dismal prognosis.