Design and phenomenology of the FEBSTAT study


  • FEBSTAT study team members are in Appendix.

Address correspondence to Dale C. Hesdorffer, GH Sergievsky Center, Columbia University, 630 West 168th Street, P & S Unit 16, New York, NY 10032, U.S.A. E-mail:


Purpose:  Febrile status epilepticus (FSE) has been associated with hippocampal injury and subsequent hippocampal sclerosis (HS) and temporal lobe epilepsy. The FEBSTAT study was designed to prospectively examine the association between prolonged febrile seizures and development of HS and associated temporal lobe epilepsy, one of the most controversial issues in epilepsy. We report on the baseline phenomenology of the final cohorts as well as detailed aims and methodology.

Methods:  The “Consequences of Prolonged Febrile Seizures in Childhood” (FEBSTAT) study is a prospective, multicenter study. Enrolled are children, aged 1 month to 6 years of age, presenting with a febrile seizure lasting 30 min or longer based on ambulance, emergency department, and hospital records, and parental interview. At baseline, procedures included a magnetic resonance imaging (MRI) study and electroencephalography (EEG) recording done within 72 h of FSE, and a detailed history and neurologic examination. Baseline development and behavior are assessed at 1 month. The baseline assessment is repeated, with age-appropriate developmental testing at 1 and 5 years after enrollment as well as at the development of epilepsy and 1 year after that. Telephone calls every 3 months document additional seizures. Two other groups of children are included: a “control” group consisting of children with a first febrile seizure ascertained at Columbia University and with almost identical baseline and 1-year follow-up examinations and a pilot cohort of FSE from Duke University.

Key Findings:  The FEBSTAT cohort consists of 199 children with a median age at baseline of 16.0 months (interquartile range [IQR] 12.0–24.0) and a median duration of FSE of 70.0 min (IQR 47.0–110.0). Seizures were continuous in 57.3% and behaviorally intermittent (without recovery in between) in 31.2%; most were partial (2.0%) or secondary generalized (65.8%), and almost all (98.0%) culminated in a generalized tonic–clonic seizure. Of the 199 children, 86.4% had normal development and 20% had prior febrile seizures. In one third of cases, FSE was unrecognized in the emergency department. The Duke existing cohort consists of 23 children with a median age of FSE onset of 18.0 months (IQR 14.0–28.0) and median duration of FSE of 90.0 min (IQR 50.0–170.0). The Columbia control cohort consists of 159 children with a first febrile seizure who received almost the same workup as the FEBSTAT cohort at baseline and at 1 year. They were followed by telephone every 4 months for a median of 42 months. Among the control cohort, 64.2% had a first simple FS, 26.4% had a first complex FS that was not FSE, and 9.4% had FSE. Among the 15 with FSE, the median age at onset was 14.0 months (IQR 12.0–20.0) and the median duration of FSE was 43.0 min (IQR 35.0–75.0).

Significance:  The FEBSTAT study presents an opportunity to prospectively study the relationship between FSE and acute hippocampal damage, the development of mesial temporal sclerosis, epilepsy (particularly temporal lobe epilepsy), and impaired hippocampal function in a large cohort. It is hoped that this study may illuminate a major mystery in clinical epilepsy today, and permit the development of interventions designed to prevent the sequelae of FSE.

Intractable mesial temporal lobe epilepsy (MTLE) is a severe condition associated with increased morbidity and poor psychosocial outcome. Prolonged febrile seizures (FS) appear to be associated with MTLE, as retrospective studies from tertiary epilepsy surgery centers report a history of prolonged FS during childhood in patients with MTLE to a greater extent than would be expected by chance (Abou-Khalil et al., 1993; Cendes et al., 1993a,b; French et al., 1993). Prospective studies of FS have, however, failed to confirm this association (Annegers et al., 1987; Camfield et al., 1994; Berg et al., 1999). Among the population-based studies, children with febrile status epilepticus (FSE) had a greatly increased risk for developing epilepsy, although cases were few in number after a follow-up of 20 years or more (Annegers et al., 1987; Camfield et al., 1994). In support of the possible association between FSE and MTLE, studies in rats show that hippocampal injury and temporal lobe epilepsy occur following status epilepticus induced by high temperature (Dube et al., 2006) and such FSE in immature rats can boost hippocampal excitability into adulthood (Dube et al., 2000). Consistent with these findings, a study of children with prolonged FS and magnetic resonance imaging (MRI) studies performed soon after FSE showed acute hippocampal injury in some children with seizures lasting >90 min (VanLandingham et al., 1998; Provenzale et al., 2008). The acute injury has been shown to later evolve to mesial temporal sclerosis (MTS) in some children (Lewis et al., 2002) over a short follow-up. However, the mean latency to develop MTLE after FS is 8–11 years in retrospective studies of refractory MTLE (French et al., 1993; Mathern et al., 1995a,b).

Single case studies of children with FSE exist (Morimoto et al., 2002; Sokol et al., 2003). Aside from the Duke FSE study, there are three other small studies that followed children with FSE imaged at baseline (Scott et al., 2003) or children with FSE and afebrile SE (Grunewald et al., 2001; Farrow et al., 2006). Because FSE occurs in <10% of children with FS (Berg et al., 1995; Berg & Shinnar, 1996; Hesdorffer et al., 2011, 2008), large prospective studies are needed limited to children with FSE who could be followed long enough to have the statistical power to address whether FSE is associated with MTS and subsequent MTLE. The “Consequences of Prolonged Febrile Seizures in Childhood” (FEBSTAT) study was designed to address this question. This report describes the three FEBSTAT cohorts as well as the detailed phenomenology and the early clinical outcomes to date.


The FEBSTAT study is a prospective multicenter study designed to address the relationship between FSE, subsequent MTS, and MTLE. This study consists of three cohorts, each described below: the FEBSTAT cohort; the existing cohort; and the control cohort.

FEBSTAT study cohorts

FEBSTAT cohort

Children, aged 1 month through 5 years, were included if they presented with an episode of FSE, defined as a seizure lasting a total of 30 min or more without fully regaining consciousness (Commission on Epidemiology and Prognosis of the International League Against Epilepsy, 1993) that also met the definition of a FS (National Institutes of Health, 1981; Commission on Epidemiology and Prognosis of the International League Against Epilepsy, 1993). A FS was defined as a provoked seizure where the sole acute provocation was fever (temperature >38.4°C, 101.0°F) without a prior history of afebrile seizures and with no evidence of an acute central nervous system (CNS) infection or insult. Those children with known severe neurologic disability before entry were excluded. Five sites recruited children with FSE from June 2003 through January 2010: Montefiore Medical Center and Jacobi Hospital; Children’s Memorial Hospital; Duke University Medical Center; Virginia Commonwealth University Hospital; and Eastern Virginia Medical School.

Recruitment.  At each site, emergency department and hospitalization records were reviewed on a daily basis to identify potential subjects with a FS reportedly lasting >15 min and screen them to determine if they had FSE, defined as a seizure duration of >30 min as judged by the study team at each site. Only these children were eligible. These families were approached for consent to enroll their child in the study. Procedures were approved by the institutional review boards for the protection of human subjects at all the participating institutions.

Procedures.  Eligible children with FSE whose families provided written informed consent received an MRI within 72 h of the FSE whenever possible or shortly thereafter. The MRIs included T2-weighted coronal sequences for assessment of hippocampal morphology and signal and T1-weighted coronal volumetric whole-brain sequences for quantitative measurements of volume as well as diffusion-weighted and diffusion tensor sequences for measurement of apparent diffusion constants (ADCs) and fractional anisotropy (FA). MRI geometry was monitored through the use of phantoms at each site. Visual readings are undertaken by two neuroradiologists (JAB and SC) examining hippocampal volume, hippocampal T2 signal, and other hippocampal and extrahippocampal abnormalities. The team at Duke University undertook the assessment of hippocampal volumes, ADCs, and FA maps. Geometry phantoms and T2 standards were imaged with each child allowing for quantitative T2 and to correct volumes for any changes among machines or over time. For hippocampal volume the primary outcome was the quantitative volumetric measurement, whereas for T2 signal it was the visual reading. An EEG examination was also undertaken within 72 h of the FSE for the FEBSTAT cohort. All EEGs were done according to the standards of the American Clinical Neurophysiological Society for the recording of pediatric EEGs (American Clinical Neurophysiology Society, 2006) and lasted at least 30 min. EEG studies were recorded using digital machines, and the studies were archived onto CDs. Studies were performed with children awake and asleep when possible. Two teams (DRN and SLM) interpreted the EEG recordings and assessed them for slowing and spikes as well as any other abnormalities.

Members of the imaging core and the EEG core were blinded to all clinical details except the age of the subject as well as to the results of the other cores.

Seizure phenomenology was classified based upon a detailed history of the circumstances and nature of the seizure. The ambulance call sheets, emergency department records, and interviews were used for this classification. For FSE, the following were evaluated: convulsion duration, intermittent or continuous FSE, focality and lateralization, seizure type, prior febrile seizures and FSE, prior development, and recognition of FSE by nonstudy clinicians. FS in controls were classified as simple or complex, and features of complex febrile seizures were categorized; duration of seizure and prior development were also determined. The phenomenology core was blinded to results of the EEG and the MRI. The phenomenology results for the first 119 children of the FEBSTAT cohort and portions of the phenomenology of the Columbia controls have been reported previously (Shinnar et al., 2008; Hesdorffer et al., 2011).

Identical procedures were followed for interpretation of all subsequent MRIs, EEGs, and seizures.

As part of the diagnostic evaluation during the initial hospitalization the following procedures were performed: detailed history of the circumstances of the seizure with a concentration on duration and focality; a structured interview including developmental milestones, past medical history including prenatal and perinatal, family history of FS and epilepsy in first-degree relatives, and characteristics of the prolonged FS; neurologic and physical examination; laboratory data summarized including cerebrospinal fluid (CSF) results with >20 white blood cells (WBCs) in the CSF considered an exclusionary criterion. Because lumbar punctures were done for clinical indication, the absence of a lumbar puncture was not an exclusionary criterion. Serum specimens to assay acute human herpes virus (HHV)-6 and HHV-7 were collected and assayed centrally to examine the possible role of HHV-6 and HHV-7 in the pathogenesis of FSE and subsequent temporal lobe epilepsy (TLE) (Theodore et al., 2008); serum samples for genetic studies were sent to the National Institute of Neurological Disorders and Stroke (NINDS) repository at the Coriell Institute for Medical Research ( for future analysis. Genomic samples are being collected and sent to the Genomics repository at Cincinnati Children’s Hospital (Sabin et al., 2010).

One month evaluation.  One month after the FSE, children were evaluated with an age-appropriate developmental battery by the psychometrician at each site (Table 1). All developmental battery results were double scored by the neuropsychology team at Montefiore Medical Center. Parents also completed several scales, including the Vineland Adaptive Behavior Scales (Sparrow & Cicchetti, 1985), the Child Behavior Checklist for children who were at least 2 years of age (Achenbach et al., 1987), and the Parenting Stress Index (Abidin, 1995).

Table 1.   Procedures at baseline, and 1, 5, and 10 years for the FEBSTAT, Duke, and Columbia cohorts
ProceduresFEBSTAT cohortDuke cohort with FSEColumbia cohort with first febrile seizures
 Developmental testingYesNoYes
 Parental interviewsYesNoYes
One-year follow-up   
 Developmental/neuropsychological testingYesNoYes
 Parental interviewsYesNoYes
Five-year follow-up   
 Developmental/neuropsychological testingYesYesNA
 Psychiatric interview (DISC)YesNoNA
 Parental interviewsYesYesNA
Ten-year follow-up   
 MRINot yetYesNA
 EEGNot yetYesNA
 Developmental/neuropsychological testingNot yetYesNA
 Psychiatric interview (DISC)Not yetYesNA
 Parental interviewsNot yetYesNA
 GeneticsNot yetYesNA

Follow-up assessment.  One year and 5 years after the initial assessment, the MRI, EEG, and developmental testing were repeated for the FEBSTAT cohort. The testing battery changed as the child aged, moving from the Bayley Tests of Infant Development to IQ and memory tests available for older children.

Quarterly telephone calls document the occurrence of further seizures of any kind. If new episodes of FSE or a first unprovoked seizure was noted, the child was brought in for a repeat of the baseline assessment.

Duke existing cohort

Ongoing pilot data for the newly identified FEBSTAT cohort comes from a cohort of 39 children with FSE prospectively enrolled at Duke from 1994 to 1997 (VanLandingham et al., 1998) and 22 children with FSE prospectively enrolled at Duke (Provenzale et al., 2008) from 1997 to 2001 were eligible for follow-up in FEBSTAT. Twenty-three of the children with FSE in the combined original cohort agreed to participate in FEBSTAT (one child prospectively ascertained from the earlier cohort and the remaining from the later cohort), and enrolled after signing informed consent. Like the FEBSTAT cohort, these children were all imaged within 72 h of their index FSE and have continued to be followed throughout the FEBSTAT study. Although the procedures used in this cohort were very similar for MRI, there was no baseline neuropsychological or developmental testing, and samples for virology titers were not obtained. Baseline EEGs were done for clinical care in about 50% of subjects, and efforts are being made to obtain the digital files.

During the first 5 years of the grant, the Duke FSE cohort underwent evaluations that were 5–10 years after their index FSE. During the second 5-year cycle, they are being evaluated 10–15 years after their index FSE. They will continue to be followed to generate hypotheses concerning what to expect over time from the larger FEBSTAT cohort. They will also contribute to analysis of each time period when methodology is overlapping for MRI endpoints, further occurrence of FSE or occurrence of new-onset epilepsy, and for cognitive outcomes at 5 years onward. EEG studies were performed during the 5–10 year follow-up and continue to be done at the 10–15 year follow-up. All MRI examinations and seizure phenomenology at baseline have been reevaluated by consensus reviewers.

Columbia control cohort

One hundred fifty-nine children, aged 6 months to 5 years, with a first FS, including some with FSE, were enrolled in a prospective cohort study: the Columbia Study of Febrile Seizures (Hesdorffer et al., 2008). They were identified at The Morgan Stanley Children’s Hospital of New York-Presbyterian Pediatric Emergency Department between March 1999 and April 2004 through active daily screening of all febrile seizures whether first or recurrent. Children with prior neonatal seizure were included. Enrolled children received an MRI within 1 week of the FS. Within 1 month, parents were interviewed regarding the child’s medical history, demographics, and family history of FS and epilepsy in first-degree relatives, using the same interview administered to the FEBSTAT parents at baseline. At the time of the parent interview, children received a neurologic examination and developmental testing (Table 1). The baseline examination was repeated at 1 year. Children were followed through quarterly telephone calls for a median of 42 months to assess the occurrence of further FS or new onset unprovoked seizures. An EEG was not performed and virology was not virology assessed. MRI examinations and seizure phenomenology have been reevaluated by consensus reviewers as part of the FEBSTAT study (Hesdorffer et al., 2011).

Consensus review of seizure phenomenology, MRI, and EEG

MRI visual readings, EEG readings, and seizure phenomenology were evaluated by the central cores previously described, each of which reviewed the data independently. When disagreements occurred, the experts reviewed the differences together to arrive at a mutually agreed upon classification. MRI and EEG consensus were performed blinded to all other information about the child except age. For subsequent EEG and MRI studies, readers were blinded to the nature of the study and to the results of prior studies. In addition, the quantitative MRI core was blinded to the readings of the visual MRI reading core, although any differences were subsequently resolved. The cores have interpreted all data from each of the three cohorts (FEBSTAT, Duke existing cohort, and Columbia controls) including all the MRIs across cohorts, all the EEGs from FEBSTAT and the Duke existing cohort, and the phenomenology of seizures. For neuropsychological and behavioral testing of the FEBSTAT and Duke existing cohort, double scoring by a central neuropsychologist was used for all tests that were not computer administered to ensure consistency. Double scoring was also used for the Columbia cohort developmental testing and parental questionnaire like the Vineland Adaptive Behavior Scale.


FEBSTAT cohort

The FEBSTAT cohort consists of 199 children with FSE (Table 2), after excluding one enrolled child whose LP later revealed the presence of a central nervous infection. The median age at FSE was 16.0 months (Table 2). Development was normal in 86.4% and clearly abnormal in only 7.0%. Approximately 20% had a history of febrile seizures and 7% had a history of possible or definite FSE. The mean temperature in the emergency room was 102.2°F (standard deviation [SD] 1.8).

Table 2.   Characteristics of febrile status epilepticus in the FEBSTAT cohort, Duke cohort, and Columbia control cohort
FactorsFEBSTAT cohort (N = 199)Duke cohort (N = 23)Columbia control cohort (N = 15)
  1. a Limited to children with definite focal febrile status epilepticus.

  2. b One hundred ninety-one had MRI examinations.

  3. c Among the 12 with MRI examinations.

  4. d One hundred sixty-nine had bloods for HHV-6/HHV-7 determination.

Mean age (months) (SD)20.5 ± 12.920.91 ± 9.517.20 ± 9.2
Median age (months) (IQR)16.0 (12.0–24.0)18.0 (14.0–28.0)14.0 (12.0–20.0)
Sex (%)   
 Male105 (52.7)10 (43.5)7 (46.7)
 Female94 (47.2)13 (56.5)8 (53.3)
Duration of febrile seizure (%)   
 30–59 min81 (40.7)7 (30.4)10 (66.7)
 60–89 min49 (24.6)3 (13.0)2 (13.3)
 90–119 min21 (10.6)5 (21.7)2 (13.3)
 120–239 min41 (20.6)6 (26.1)1 (6.7)
 ≥240 min7 (3.5)2 (8.7)0 (0)
Mean duration of febrile seizure (min)89.8 ± 71.7123.0 ± 113.859.9 ± 32.4
Median duration of febrile seizure (IQR) (min)70.0 (47.0–110.0)90.0 (50.0–170.0)43.0 (35.0–75.0)
Type of seizure (%)   
 Continuous114 (57.3)10 (43.5)7 (46.7)
 Intermittent without full recovery62 (31.2)10 (43.5)7 (46.7)
 Intermittent with drug therapy23 (11.6)3 (13.0)1 (6.7)
Focal seizure (%)   
 Definitely101 (50.3)17 (73.9)12 (80.0)
 Probably34 (17)5 (21.7)1 (6.7)
 Maybe18 (9)0 (0)0 (0)
 No (Generalized)46 (23.1)1 (4.4)2 (13.3)
Definite cerebral lateralizationa (%)   
 Left33 (32.7)9 (52.9)5 (41.7)
 Right27 (26.7)4 (23.5)2 (16.7)
 No definite cerebral lateralization41 (40.6)4 (23.5)5 (41.7)
Seizure classification (%)   
 Generalized tonic–clonic64 (32.2)1 (4.4)2 (13.3)
 Complex partial only4 (2.0)0 (0)0 (0)
 Partial with secondary generalization131 (65.8)22 (95.7)13 (86.7)
Was status recognized by nonstudy clinician? (%)   
 Definitely104 (52.3)8 (34.8)4 (26.7)
 Probably29 (14.6)3 (13.0)2 (13.3)
 Maybe27 (13.6)7 (30.4)4 (26.7)
 No39 (19.6)5 (21.7)5 (33.3)
Prior development (%)   
 Normal172 (86.4)20 (87.0)0 (0.0)
 Suspect13 (6.5)2 (8.7)11 (73.3)
 Abnormal14 (7.0)1 (4.4)4 (26.7)
Baseline MRI findings (%)   
 Any definite abnormality46 (24.1)b7 (30.4)5 (41.7)c
 Increased T2 signal in the hippocampus17 (9.0)5 (21.7)1 (8.3)b
Baseline EEG findings (%)   
 Any abnormality90 (45.5)NANA
 Epileptiform abnormality13 (6.5)NANA
Baseline HHV-6/HHV-7 (%)   
 HHV-6/HHV-7 viremia58 (34.3)dNANA

The characteristics of the initial episode of FSE are summarized in Table 2. The median duration was 70 min, although the entry criteria specified >30 min and the criteria to screen for eligibility was >15 min. A substantial proportion if the children had seizures lasting >2 h (24.1%).

Among the 199 subjects, 46 did not activate emergency medical services (EMS) and 153 did. Overall, 20 subjects (10.1%) did not receive a medication to stop the seizure: four (9.0%) of those not transported by EMS and 16 (10.5%) of those transported by EMS. In the great majority of cases (87.9%), seizures did not stop spontaneously, ceasing only after the administration of a benzodiazepine class drug and, in some children, a second drug, mostly phenytoin/fosphenytoin (Fig . 1). Of the 175 treated subjects, 29 (16.6%) had termination of FSE with one dose. Sixteen (9.1%) of 175 treated subjects had termination of FSE with EMS treatment only; five had multiple doses of treatment, and 11 had one dose. Despite the relatively long duration of the seizures and the active intervention required to stop them, the emergency department clinicians, while providing appropriate therapy in most cases, consistently underestimated the duration of the seizure and failed to recognize SE in 33.2% of the cases, including many children with continuous seizures.

Figure 1.

Acute treatment of the 199 FEBSTAT subjects.

Since FEBSTAT enrollment, 23 children (11.6%) have experienced one or more additional episodes of FSE. At least one unprovoked seizure has developed in 22 children (11.1%).

Dravet syndrome has been identified in three children. Prediction of Dravet syndrome was not possible at the time of FSE, although one child with epileptiform abnormalities was, in hindsight, at increased risk. In addition, three children have died. One died during the initial hospitalization several weeks after the episode of FSE due to respiratory failure of unknown cause. The remaining two had developed epilepsy and were neurologically impaired and died of presumed sudden unexpected death in epilepsy (SUDEP); one of the SUDEP cases had Dravet syndrome.

Duke cohort

The Duke cohort consists of the 23 children with FSE whose families signed consent forms to participate in FEBSTAT. Baseline seizure semiology has been evaluated in consensus. The median age at FSE was 18 months. Development was normal in 87.0% and clearly abnormal in only 4.4%. Approximately 22% had a history of febrile seizures and none had a history of possible or definite FSE. The mean temperature in the emergency room was 103.4°F (SD 1.2).

The median seizure duration was long at 90 min (Table 2), with a substantial proportion having seizures that lasted >2 h (34.8%). Despite the relatively long duration of the seizures and the active intervention required to stop them, the emergency department (ED) clinicians, while providing appropriate therapy in most cases, consistently underestimated the duration of the seizure and failed to recognize SE in more than one fifth of the cases, including many children with continuous seizures. Although duration of FSE appears longer in the Duke than in the FEBSTAT cohort, the differences are not statistically significant.

No children experienced a subsequent FSE. Epilepsy has developed in seven children (30.4%) of whom two, both in the Duke pilot cohort (which has had longer follow-up), have had TLE and lobectomies showing hippocampal sclerosis and mild temporal focal cortical dysplasia.

Columbia control cohort

The control cohort consists of 159 children enrolled at their first febrile seizure; 15 (9.4%) had FSE.

In the full control cohort of 159 children, median age at FSE was 18 months and the mean temperature in the emergency department was 103.4°F (SD 1.6; Table 3). First simple febrile seizures occurred in 64.2%. Among the 57 children with a first complex FS, 63.2% experienced focal seizures, 46.6% experienced seizures longer than 15 min, and 31.6% experienced multiple seizures. These percentages are not independent. Development was normal in 96.60% and clearly abnormal in only 4.4%.

Table 3.   Characteristics of Columbia controls with a first febrile seizure
FactorsColumbia controls (N = 159)
  1. a Missing one case with simple febrile seizure where exact duration could not be determined.

  2. b 133 with MRI scans read.

Mean age (months) (SD)20.4 ± 10.7
Median age (months) (IQR)18.0 (14.0–24.0)
Sex (%) 
 Male87 (54.7)
 Female72 (45.3)
Mean duration of febrile seizure in minutesa10.2 ± 19.3
Median duration of febrile seizure in minutes (IQR)a4.0 (2.0–8.0)
Febrile seizure type (%) 
 Simple FS (<15 min)102 (64.2)
 Complex FS (≥15 min)57 (38.9)
Features of complex febrile seizures (%) 
  Yes26 (63.2)
  No21 (36.8)
 Prolonged (>15 min) (%) 
  Yes26 (46.6)
  No31 (54.4)
 Multiple (%) 
  Yes18 (31.6)
  No39 (68.4)
 Status (%) 
  Yes15 (26.3)
  No42 (73.7)
Prior development (%) 
 Normal152 (96.6)
 Abnormal7 (4.4)
Baseline MRI findings in simple FS and complex FS that is not FSEb (%) 
 Any definite abnormality14 (10.5)
 Increased T2 signal in the hippocampus0 (0)

Fifteen children (9.4%) in the control cohort experienced FSE (Table 2). Among these, the median duration was 43.0 min. FSE lasted >1 h in 23.3%, and >2 h in 6.7%. FSE did not stop spontaneously in 6.7% and required administration of a benzodiazepine class drug. The emergency department clinicians underestimated the duration of the seizure and failed to recognize FSE in one third of the cases. Children with FSE from the Columbia cohort are included in some analyses of FSE in which case they are excluded as controls.

Over a median of 42 months of follow-up, 52 (32.7%) of the 159 Columbia controls experienced a second febrile seizure, of which 4 (7.7%) were FSE; epilepsy developed in 9 children (5.7%).


We originally reported on phenomenology in the 119 children with FSE ascertained in FEBSTAT (Shinnar et al., 2008). In Table 2, we show the phenomenology for all three cohorts with FSE. The results are similar, but more information is provided from Duke and the Columbia controls with FSE. It is notable that the Columbia controls and Duke FSE cases were reclassified by the same reviewers used for FEBSTAT. Duration of FSE was similar for the Duke FSE and FEBSTAT cases, but duration of FSE in the Columbia FSE cases was shorter than in the FEBSTAT cases (p = 0.005).

We used the Kappa statistic to compare classification by each of the phenomenology reviewers to the final consensus for both the 237 with FSE and the 159 Columbia controls. Among those with FSE, Kappa was good to excellent. As expected, the lowest Kappas were found for determination of focal seizures (Kappa ranging from 0.45 to 0.68). On the measure of total seizure duration derived from the history and medical records, as video-EEG was not done, the intraclass correlation coefficients (ICCs) were excellent: 0.80 for overall duration of FSE and 0.99 for overall duration among Columbia controls. Among the Columbia controls, all Kappas were good or excellent (Kappa ranging from 0.69 to 0.96).

Summary of baseline HHV-6/HHV-7 titers, EEG findings, and MRI

Viremia with HHV-6/HHV-7 occurred in 34.3% of children in FEBSTAT (Table 2) and was not associated with MRI abnormality (Epstein et al., 2012). Among FEBSTAT children, any EEG abnormality occurred in 90 (45.5%) of children and epileptiform EEG recordings were seen in 13 (6.5%). Definite baseline MRI abnormalities were seen in 58 (25.7%) of the 226 baseline MRIs across the three cohorts (191 FEBSTAT, 23 Duke, and 12 controls with FSE and imaging), including definitely increased hippocampal T2 signal in 23 (10.2%) cases (Shinnar et al., 2012).


The FEBSTAT study is a large-scale prospective study that intensively investigates the aftermath of FSE. The Columbia cohort adds controls for baseline and 1-year MRI and cognition, as well as seizure occurrence up to 42 months of follow-up. The Duke cohort, which contributes pilot data, has the advantage of being 5 years older than the FEBSTAT cohort at each assessment, thus defining endpoints that can be studied later in the FEBSTAT cohort using the same methodology. These data provide useful information on what to expect in the FEBSTAT children and to better formulate next steps in FEBSTAT.

Good reliability of the features of FSE is important when studying the consequences of FSE, particularly when examining associations between MRI or EEG findings, for example, and features of FSE. Reliability between each reviewer and the consensus determination was good to excellent in almost all cases, except for focality. Determination of focal features was good for two of three readers and fair for one. The same was not seen for the Columbia controls, where Kappa for focal features was good to excellent for all reviewers compared to consensus, although still lower than that for other features of complex febrile seizures. This is consistent with a prior report of children with first febrile seizures, (Berg et al., 1992) which showed that Kappas were lower for focality than for other features of complex febrile seizures. Reasons were similar and related to whether a stare at the beginning or eyes deviating to one side were considered as evidence of focality by the raters (Shinnar et al., 2008). When lateral eye deviation at onset preceded the seizure, we considered this evidence of probable focal onset without the ability to lateralize the seizure, because it could have arisen from the ipsilateral or contralateral side. When eyes were deviated upward and to the side, we did not consider this evidence of focal onset.

A central tenet of the FEBSTAT study design has been achieving as great as possible consistency of procedures across the three FEBSTAT cohorts. To this end, the phenomenology core classified all seizures in the three cohorts and the MRI core classified all MRIs, even though these had been assessed previously for the Duke (Lewis et al., 2002) and the Columbia cohort (Hesdorffer et al., 2008). The EEG core similarly classified all EEG recordings from the FEBSTAT and the Duke cohort; no EEGs were available for the Columbia cohort. The same parental questionnaire used in the Columbia study was used in the FEBSTAT cohort; however, it was impossible to assess the same information in the Duke cohort, as the acute period was at least 5 years before enrollment into FEBSTAT. Where possible, some of the information obtained in the interview was abstracted from medical records. Most developmental tests used in the Columbia study were also used in FEBSTAT cohort at baseline and 1 year, and additional tests of memory were included for FEBSTAT. The Duke cohort has had the same follow-up intervals as the FEBSTAT cohort but is 5 years ahead, allowing for the collection of pilot data to inform hypotheses to be tested on the FEBSTAT cohort and for inclusion of the Duke FSE cases with the FEBSTAT cases when the time intervals are overlapping.

Children in the Columbia cohort were ascertained because they had a first febrile seizure and not because they had FSE. The consequence is that the entire spectrum of febrile seizure duration was included in this cohort. Consensus used to determine which children had FSE, whether or not these children would have come to attention in studies focusing on FSE. This likely explains the lower seizure duration in Columbia FSE compared to FEBSTAT. Duration did not differ between Duke FSE and FEBSTAT.

The consistency of information collected across cohorts permits several aggregations of cohorts for analysis. The FEBSTAT cohort can be compared to the Columbia cohort children with first simple febrile seizures at baseline and 1 year, and assessment of epilepsy can be compared for up to 42 months of follow-up. The FEBSTAT and Duke cohorts can be combined to study long-term outcomes of FSE, since the most methodology is parallel and sample size is increased. For analyses of baseline and 1-year imaging and developmental testing, the Columbia cohort FSE cases can be combined with the FEBSTAT and Duke cohorts and compared to Columbia controls with simple febrile seizures or complex febrile seizures that are not FSE.

Due to the young age at which children experience FSE (a median of 14.0 to 18.0 months across cohorts), memory tests were administered to few children at baseline. Memory testing begins at age 2.5 years with the McCarthy Scales of Children’s Abilities, and children older than age 5 receive the Wide Range Assessment of Memory and Learning. Therefore, it is possible to examine the association between hippocampal abnormalities and memory beginning as early as 2.5 years.

MRI examinations revealed similar findings across the FEBSTAT, Duke, and Columbia FSE cases. It is notable that the Columbia children with simple or complex febrile seizures that were not FSE did not show increased hippocampal T2 signal, a finding seen exclusively in FSE cases.

The majority of children with FSE are doing well to date and have not experienced further FSE or onset of epilepsy. This is consistent with the data from retrospective studies that report a mean of 8–11 years between the episode of FSE and development of TLE (French et al., 1993; Mathern et al., 1995b). Following a mostly well cohort is always challenging because parents may not wish to have their child undergo further testing. Fortunately, most FEBSTAT coordinators who began the study are still following the cohort and fostering relationships with study families through the quarterly calls, thereby optimizing follow-up. This is particularly important because of the long latency to the onset of epilepsy after FSE; therefore, we will have to wait some time before the main study question (i.e., do FSE cause MTS and TLE?) can be answered definitively.

In the initial report of 121 FEBSTAT cases, we found a high proportion in which, despite good management in the emergency department, clinicians failed to recognize FSE (Shinnar et al., 2008). This finding has been extended to include the full FEBSTAT cohort, as well as the Duke cohort and the Columbia cohort. Although the clinicians often did not recognize that the seizure was SE, using the 30-min definition met by all children in this study, they treated them with medications. The clinicians usually recognized ongoing seizure activity and treated it even if they did not know how long the seizure actually lasted. Whether or not they were treated by EMS was primarily a matter of the custom and practice of each ambulance service. As was the case in the Prehospital Treatment of Status Epilepticus (PHTSE) study (Lowenstein et al., 2001), it remains the case that many jurisdictions do not give any AEDs in the prehospital setting, even though the AEDs are clearly effective (Lowenstein et al., 2001; Silbergleit et al., 2012). In most cases, FSE stopped following treatment with benzodiazepines, which were the first drug administered in almost all children. It is difficult to compare the response rates to those seen in the randomized clinical trials. In those trials, prehospital treatment with lorazepam was associated with a 59.1% response rate in PHTSE (Lowenstein et al., 2001) and with 63.4% in Rapid Anticonvulsant medication Prior to Arrival Trial (RAMPART) (Silbergleit et al., 2012). However, there are two major differences. First, the failure criteria were very stringent in the trials, whereas many of the FEBSTAT children received multiple doses of the same benzodiazepine and would have therefore been considered nonresponders in the trials. Second, the trials were limited to adults and included all types of SE, including a substantial proportion of acute symptomatic SE associated with acute neurologic insults. Response to treatment is partly a function of etiology, and acute symptomatic SE tends to respond less well to treatment (Maytal et al., 1989). Therefore, it is unclear from these data whether FSE is more sensitive to benzodiazepine therapy than other forms of SE.

The combined cohorts of the FEBSTAT study allow us to examine the relationship between FSE and subsequent MTS and TLE. Controls with simple febrile seizures and complex febrile seizures that are not FSE permit comparison for early outcomes. The Duke existing cohort contributes valuable pilot data to inform hypotheses for the next wave of the FEBSTAT cohort follow-up and contributes to analyses of FEBSTAT when outcomes overlap. Long-term outcome studies are in progress, and the FEBSTAT cohort is undergoing the 5-year follow-up examinations. We are only now beginning to see TLE in the Duke existing cohort where two children have undergone lobectomy >10 years after their FSE. Early data suggest that initial MRI findings can predict development of MTS (Provenzale et al., 2008), implying that MRI may be a biomarker for the development of subsequent TLE, thus identifying high-risk children who may be candidates for antiepileptogenesis trials (Gomes & Shinnar, 2011). The cohort is adequately powered to examine the role of HHV-6 in the development of TLE as well (Theodore et al., 2008). Given the known long latency periods between FSE and the development of clinical TLE, it will be some time before the FEBSTAT final answers are known. The FEBSTAT study has a sufficiently large cohort to answer whether there is a link between prolonged febrile seizures and subsequent TLE and MTS.


Supported by NINDS grant NS43209 (P.I. S. Shinnar, M.D., Ph.D.) and NICHD grant 36867 (PI: D.C. Hesdorffer Ph.D.).


None of the authors has any conflict of interest to disclose. We confirm that we have read the Journal’s position on issues involved in ethical publication and affirm that this report is consistent with the guidelines of the Committee on Publication Ethics (COPE) guidelines for ethical publication (


FEBSTAT study team

Montefiore Medical Center and Jacobi Medical Center, Albert Einstein College of Medicine, Bronx, NY: Shlomo Shinnar M.D. Ph.D. (PI), Jennifer Ayala B.A., Jacqueline Bello M.D., Mootoo Chunasamy R.T. M.S., Patricia Clements, Ronda L. Facchini Ph.D., James Hannigan R.T., George Lantos M.D., Ann Mancini M.A., David Masur Ph.D., Solomon L. Moshé M.D., Ruth C. Shinnar R.N. M.S.N., Maryana Sigalova M.A., Yoshimi Sogawa, M.D., Erica Weiss M.A.

Children’s Memorial Hospital, Northwestern University Feinberg School of Medicine, Chicago, IL: Douglas Nordli M.D. (PI), John Curran M.D., Leon G. Epstein M.D., Aaliyah Hamidullah M.S., Andrew Kim M.D., Julie Renaldi Ph.D.

Columbia University School of Medicine, New York, NY: Dale C. Hesdorffer Ph.D. (PI). Emilia Bagiella Ph.D., Stephen Chan M.D., Veronica J. Hinton Ph.D., Claire Litherland, Yuxin Zhang M.S.

Duke University Medical Center, Durham, NC: Darrell Lewis M.D. (PI), Melanie Bonner Ph.D., William Gallentine DO, James MacFall Ph.D., James Provenzale M.D., Elizabeth Rende R.N. M.S.N. C.P.N.P., James Voyvodic, Ph.D., Allen Song Ph.D., Yuan Xu B.S.

Eastern Virginia Medical School, Norfolk, VA: L. Matthew Frank M.D. (PI), Joanne Andy R.T., Terrie Conklin R.N., Susan Grasso M.D., Diane James R-EEG T, David Kushner M.D., Susan Landers R.T., Virginia Van de Water Ph.D.

Virginia Commonwealth University, Richmond, VA: John M. Pellock M.D. (PI), Tanya Bazemore REEG-T, James Culbert Ph.D., Kathryn O’Hara R.N., Syndi Seinfeld M.D., Jean Snow RT-R.

International Epilepsy Consortium at Virginia Commonwealth University, Richmond VA: Shumei Sun Ph.D., Brian J Bush MSMIT, Sreedevi Chandrasekaran, Lori L. Davis, John M. Pellock M.D., Christiane Rogers, Cynthia Shier Sabo M.S., Helen Wang. Collaborators: Joan Conry M.D., Children’s National Medical Center, Washington DC – Safety.

Monitor: Tracy Glauser M.D., Cincinnati Children’s Hospital, Cincinnati Ohio – Genomics substudy. Jeffrey L. Noebels M.D. Ph.D., Baylor College of Medicine, Houston, Texas – Genetics substudy.