Dravet syndrome in children — A population-based study Epilepsy Research

Objective: The aim was to describe age at diagnosis, cumulative incidence, SCN1A variants, mortality, seizure types and treatments in children with Dravet Syndrome (DS) in Sweden. Methods: Children diagnosed with DS, born between January 1st 2000 and December 31st 2018 were included in a population-based study. Clinical data, frequency of seizure types and treatments were collected from caregivers and medical records in 42 children. Age at diagnosis, cumulative incidence and treatment were compared be- tween children born in Sweden 2000 – 2009 and 2010 – 2018. Results: We identified 55 children with DS, 53 were born in Sweden. Three children had died of definite, probable, or possible sudden unexpected death in epilepsy, one of acute anoxic brain injury and three of pneumonia or pneumonitis. Median age at death was 4.7 (range 3.3 – 11) years. In 49/53 children with known SCN1A status, a pathogenic/likely pathogenic variant of SCN1A was detected. In two a SCN1A variant of unknown significance was found. For children born in Sweden 2010 – 2018, median age at DS diagnosis was lower (1.6 vs 4.5 years, p = 0.001) and cumulative incidence higher (1/33,000 vs 1/46,000 live-born children, p = 0.03), compared to children born in 2000 – 2009. The most common seizure types were focal to bilateral tonic clonic (n = 41/42) and myoclonic (n = 35/42). Tonic seizures were reported in 25/42 children. Sodium-channel inhibitors had been used in 9/24 children born in 2010 – 2018 and 17/18 children born in 2000 – 2009 (p = 0.001). Significance: A SCN1A variant that could explain the syndrome was found in over 90% of children. Tonic seizures seem to be more frequent than earlier described. Median age at diagnosis was lower, cumulative incidence higher and use of contra-indicated sodium-channel inhibitors less common for children born in 2010 – 2018 compared with children born in 2000 – 2009. This could indicate an increased awareness of DS.


Introduction
Dravet syndrome (DS) is both an epileptic and developmental encephalopathy , characterized by early-onset intractable epilepsy typically starting in the first year of life. DS was first described in 1978 (Dravet, 1978), and is characterized by multiple seizure types often triggered by elevated body temperature. The syndrome usually presents in the first year of life in a previously healthy infant. The infant develops alternating unilateral or bilateral tonic-clonic febrile or afebrile seizures. The seizures frequently evolve Abbreviations: ASD, autism spectrum disorder; ASMs, anti-seizure medications; benzo, benzodiazepines; CIN, Cumulative incidence; DS, Dravet syndrome; GEFS+, generalized epilepsy with febrile seizures plus; ID, intellectual disability; SE, status epilepticus; stir, stiripentol; SUDEP, sudden unexpected death in epilepsy; vpa, valproate.
into status epilepticus (SE). Between the ages of 1 and 4 years, these seizure types persist, and the children develop other seizure types such as atypical absences and myoclonic seizures (Dravet, 2011). An initially normal cognitive development followed by stagnation in development at this stage is common (Dravet, 2011). Most affected individuals meet the criteria for intellectual disability (ID) (Jansson et al., 2020). Behavioral problems are common, and the frequency of autism spectrum disorder (ASD) appears to be higher than in the general population (Jansson et al., 2020). Progressive gait impairment described as crouch gait, can start in early school age and often affects mobility after puberty (Wyers et al., 2019). Sudden unexpected death in epilepsy (SUDEP) and SE are described as the main causes of high premature mortality (Cooper et al., 2016;Sakauchi et al., 2011;Skluzacek et al., 2011). In most cases, DS is caused by a dominant pathogenic variant in the SCN1A (sodium voltage-gated channel type 1 alpha subunit) gene encoding the type I voltage-gated sodium channel (Nav1.1) alpha subunit (Claes et al., 2001). Pathogenic variants in the pore region of the gene are more common in DS compared with generalized epilepsy with febrile seizures plus (GEFS+), a less severe SCN1A phenotype (Kanai et al., 2004;Zuberi et al., 2011). Variants that lead to truncating of the gene are associated with earlier onset of prolonged seizures and cognitive decline compared to missense variants (Ishii et al., 2017). DS is increasingly perceived as a single gene channelopathy and other pathogenic single gene variants are described as separate entities. However, phenotypes due to pathogenic variants of SCN1A and of other genes can be indistinguishable and there are individuals with a typical DS phenotype but without a detectable SCN1A variant (Li et al., 2021;Steel et al., 2017).
In Europe, valproate has been suggested as first-line treatment for children with DS and add on stiripentol with or without clobazam is suggested as second-line treatment. Based on randomized, controlled studies, cannabidiol and fenfluramine have also been proposed as second-line treatments (Cross et al., 2019). In an American consensus article, valproate and clobazam, alone or in combination, were the recommended first-line treatments and add-on stiripentol, topiramate or ketogenic diet were suggested as second-line treatments (Wirrell et al., 2017). Sodium-channel inhibitors can aggravate seizures in DS (Guerrini et al., 1998;Horn et al., 1986;Mueller et al., 2011). Prolonged use of these medications within the first five years after seizure-onset has also been associated with worse cognitive outcome (de Lange et al., 2018). They should generally be avoided according to both American and European guidelines (Cross et al., 2019;Wirrell et al., 2017).
The aim of this population-based study was to identify and characterize all children with DS in Sweden. In this paper, we focus on the incidence of DS, mortality, genetic profiles, epilepsy characteristics, and treatments, since there are limited population-based data in these areas.

Inclusion in the study
All identified children with DS in Sweden, born between January 1st 2000 and December 31st 2018 were included in the study. The criteria used for Dravet Syndrome (DS) were those described by Nabbout et al. (2013): a) a normal EEG and no preexisting cerebral lesion in a normal infant; b) normal development until the first seizure occurring before one year of age; c) refractory clonic or tonic-clonic seizures affecting one or both sides simultaneously or alternatively; d) exclusion of any other identified epilepsy syndrome including negative PCDH19 analysis in SCN1A negative participants (Nabbout et al., 2013).

Participants, identification, and recruitment
Of the 55 eligible children, 26 (47%) were females. Fifty-three were born in Sweden and two in Syria. Thirty-one children were identified from a previous study, of those 24 were included (Rosander and Hallbook, 2015). Seven did not fulfill our study inclusion criteria; one had an extensive deletion in chromosome band 2q, including the entire SCN1A gene and a more severe phenotype than DS. Four children had a phenotype compatible with GEFS+ , and did not fulfil study criteria as they did not have refractory clonic or tonic-clonic seizures. Genetic re-evaluation in two children who did not fulfil the study criteria revealed that one had a PPP2R5D related neurodevelopmental disorder and one had Okur-Chung syndrome due to a CSNK2A1 variant. In agreement between BB and TH, one child with otherwise typical features of DS and a pathogenic SCN1A missense variant who did not fulfill two of the inclusion criteria was included. This child had a benign infratentorial tumor and the first reported seizure was at the age of 2.3 years.
We calculated prevalence based on 48 children alive and living in Sweden and cumulative incidence (CIN), age at diagnosis and SCN1A analysis for the 53 children born in Sweden. Age at DS diagnosis, based on information from caregivers, and age at SCN1A analysis, based on medical records were compared between children born in Sweden 2000-2009 (n = 22) and 2010-2018 (n = 31). Of the 53 children born in Sweden, seven (13%) had died.
Assessment of epilepsy characteristics, ASMs and other treatments and neurodevelopment were studied in 42 of the 48 living children. Of the six who did not participate, five did not provide informed consent to participate and one could not be assessed due to Covid-19 restrictions. The identification process is illustrated in Fig. 1. In the 42 assessed children, a case report form was used to collect clinical data from the caregivers, supplemented with information from the medical records. Relevant questions from the case report form are presented in appendix A.
In Sweden there are neuropediatric departments in seven university hospitals and pediatric departments in 21 county hospitals. All children with complicated epilepsy are treated in these departments by pediatric neurologists. We contacted the heads of all pediatric and neuropediatric departments to ensure that children with a known diagnosis of DS were identified. Epilepsy nurses and all pediatric neurologists in Sweden were also approached, to inquire if they treated any patients who fulfilled our study criteria for DS. Some children were identified from an established network and a register started after an earlier study (Rosander and Hallbook, 2015). Members of the Dravet Syndrome Association Sweden were also informed about the study. Children were included in the study between October 1st 2018 and December 31st 2019 and examined between October 15th 2018 and April 3rd 2020. They were examined at the Queen Silvia Children´s Hospital in Gothenburg or at their home hospitals. One child was assessed at home.

Mortality
We describe mortality for all children born in Sweden (n = 53). Cause of mortality was derived from the medical records and postmortem records, and diagnosis of SUDEP was compared to current criteria: SUDEP was classified as 'definite' in cases with autopsy and as 'probable' in cases without autopsy, where no other causes of death had been found. It was classified as 'possible', where there were other possible causes of death (Devinsky et al., 2018).

Genetics
Variants in the SCN1A gene were described as pathogenic, likely pathogenic or of unknown significance, according to current genetic terminology (Richards et al., 2015), and delineated regarding the type of variant. Data concerning SCN1A testing were collected from medical records in 53/55 children, as described in appendix B. In 2/55 children SCN1A status was unknown, due to lack of informed consent from caregivers to retrieve this information. In 31/53 children a single gene analysis, performed between 2007 and 2018, had been used. In 21/53 children a multiple gene panel, performed between 2012 and 2019, had been used. One child was considered to have the same SCN1A variant as her twin, with a similar phenotype, without any genetic testing being undertaken.

Characteristics of epilepsy, anti-seizure treatments and neurodevelopment
Clinical data, including frequency of current and previous seizure types and treatments regarding the 42 children who participated in the assessment, were collected from caregivers, complemented with information in medical records. The seizure types were characterized, according to the current ILAE seizure classification (Fisher et al., 2017), by one of two epileptologists (BB in 34 participants and TH in eight participants), based on descriptive semiology from caregivers and complemented with information in medical records. The number of seizures per month in the last three months and the number of episodes of SE were also collected. We defined SE as a seizure lasting at least 30 min or as a series of convulsive seizures lasting at least 30 min, without the subject regaining consciousness. Occurrence of different seizure types, effects on seizure frequency (≤50% or >50% reduction in number of seizures), current and previous use of the most common anti-seizure medicines (ASMs) (valproate, benzodiazepines, levetiracetam, sodium channels inhibitors, stiripentol, and topiramate) were compared between 18/42 children born between 2000 and 2009 and 24/42 children born between 2010 and 2018. Information about ID, ASD and attention deficit hyperactivity disorder diagnoses was collected from parents and medical records.

Data analysis
Point prevalence was calculated on December 31st, 2018, as the number of affected living children (n = 48) in relation to all children in Sweden. Cumulative incidence (CIN), age at DS diagnosis and SCN1A analysis were calculated for two time periods, 2000-2009 (n = 22/53) and 2010-2018 (n = 31/53) for children born in Sweden. Children with DS born in Sweden in each period were divided by all children born in Sweden during the same period to calculate CIN. A Mann Whitney U-test was used to compare the age at DS diagnosis and age at SCN1A testing. A Chi square test was used to compare the use of ASMs and presence of different seizure types between children born in 2000-2009 (18/42) and 2010-2018 (24/42) and between current and previous use of different ASMs for 42 children. A p-value < 0.05 was considered statistically significant in all the analyses. The PASW Statistics for Windows, version 28 software (IBM Corporation, Armonk, NY, USA) was used for the statistical analyses.

Ethics
The study was approved by the Ethical Review Committee of the University of Gothenburg (Sweden) and The Swedish Ethical Review Authority (Dnr 450-10, T672-18, and 2020-03783). Written informed consent was obtained from caregivers and from participants older than 15 years if they were developmentally able to understand information about the study.

Epidemiology, age at Dravet syndrome diagnosis and mortality
The prevalence of DS was 1/45,000 (95% CI 1/35,000-1/63,000) for the 48 living children and resident in Sweden on December 31st, 2018. The median age at DS diagnosis was 2.0 (range 0.5-10) years. The cumulative incidence (CIN) was significantly higher (p = 0.03) and the median age at DS diagnosis (p = 0.001) and SCN1A analysis (p < 0.001) were lower for younger children born in 2010-2018 compared to older children born in 2000-2009 (Table 1). CIN was 1:33,000 for younger and 1:46,000 for older children. Seven (13%, five females) out of 53 children born in Sweden had died. Median age at death was 4.7 (range 3.3-11.0) years. Causes of death were definite SUDEP (n = 1), probable SUDEP with pneumonia (n = 1), possible SUDEP with pneumonitis (n = 1), acute anoxic brain injury after seizure-induced aspiration (n = 1), pneumonia (SUDEP excluded) (n = 2) and pneumonitis (SUDEP excluded) (n = 1). One child, who died of pneumonia (SUDEP excluded), had a chronic anoxic brain injury after prolonged SE. All children who died carried a pathogenic or likely pathogenic SCN1A variant.

Genetics
SCN1A analysis was available in 53 of 55 children (appendix B). In 49 of these 53 cases, a pathogenic/likely pathogenic SCN1A variant was detected. In two cases with a typical DS phenotype, SCN1A missense variants classified as of unknown significance were found. Thus, a  Table 2.

Characteristics of epilepsy, anti-seizure treatments and neurodevelopment
Clinical data, including frequency of current and previous seizure types, treatments and neurodevelopment regarding the 42 children who participated in the assessment are presented in Tables 2-5 and appendix C.

Epilepsy
The clinical characteristics of the 42 children are described in Table 2. The median age at first seizure was 5 months and at epilepsy diagnosis 10 months. All children had ongoing seizures. Current and previous seizure types are presented in Table 3. The most common seizure types seen currently or earlier were focal to bilateral tonic-clonic (n = 41, 98%) and myoclonic seizures (n = 35, 83%). The frequency of different seizure-types including tonic seizures did not differ between the two age groups. Thirty-eight participants (90%) had a history of convulsive SE and 14 (33%) had a history of non-convulsive SE.

Anti-seizure treatment
Current and previous anti-seizure treatments are shown in Tables 2, 4, 5 and appendix C. The children were currently being treated with a median of three (range 1-4) ASMs and had used a median of six (range 1-20) ASMs. The most common currently or previously used ASMs were valproate, (n = 42, 100%); benzodiazepines (n = 40, 95%), levetiracetam, (n = 35, 83%) and sodium-channel inhibitors, (n = 26, 62%). Sodium-channel inhibitors (lamotrigine, carbamazepine, oxcarbazepine, phenytoin, lacosamide and/or rufinamide) had been used by 9/24 There were no significant differences in the use of any other ASMs between the two age groups (Table 5). Sodiumchannel inhibitors were currently used by four children: one used lacosamide with > 50% seizure reduction, one used lacosamide with ≤ 50% seizure reduction of unknown reason, one had been prescribed  carbamazepine abroad against the advice of the treating neurologist in Sweden and one had recently been diagnosed with DS and oxcarbazepine was tapered down.

Neurodevelopment
Twenty-eight out of 42 (67%) children had a diagnosis of ID, 19/42 (45%) a diagnosis of ASD, and 9/42 (21%) had neither of these two diagnoses. Forty out of 42 (95%) had a present or previous contact with the habilitation interdisciplinary service for children with ID and for pre-school children with ASD, and four (9.5%) children went to school/ preschool without extra support (Table 2).

Discussion
This is one of the first comprehensive, population-based studies with a focus on epidemiology, mortality, genetics, seizure characteristics and treatment in children with Dravet Syndrome (DS). The cumulative incidence (CIN) was higher and the median age at DS diagnosis was lower in younger children. Of 53 children born in Sweden, seven (13%) had died at a median age of 4.7 years. Genetic testing did not show a SCN1A variant nor any other genetic variant in two (3.8%) individuals out of 53. Caregivers reported tonic seizures in more than half of participants. The use of sodium-channel inhibitors was lower in children born in 2010-2018, the younger age group, compared to children born in 2000-2009, the older age group.
The increased CIN and reduced age at diagnosis in the younger age group is in line with the view that the awareness of rare diseases like DS tend to improve with time (Auvin et al., 2018). Increased awareness can be influenced by factors such as more diagnose-specific treatment guidelines and more accurate genetic testing (Auvin et al., 2018). In recent years genetic testing is used more frequently in the diagnostic workup of epileptic and developmental encephalopathies due to lower cost and possible precision-therapies (Møller et al., 2019). Increased genetic testing and a gradual increased availability of gene panels in Sweden could be one factor explaining increased CIN and earlier age at #The caregivers were asked which types of ASM that the child was treated with and had been treated with previously. They were also asked about the efficacy of present ASM regimen. ASM, anti-seizure medication, benzo benzodiazepines, stir stiripentol, vpa valproate. 1 Side-effects and seizure reduction considered a result of ongoing treatment by caregivers. 2 The caregivers did not know if there was a seizure reduction of > 50%. 3 Cognitive, n = 1, tiredness n = 2, increased appetite n = 1, ataxia n = 1, ataxia, tremor, trombocytopenia and tiredness n = 1, tremor, and tiredness n = 1, reduced concentration and aggressiveness n = 1 4 Clobazam, clonazepam or nitrazepam. 5 Aggressiveness and concentration problems. 6 Weight loss and insomnia n = 1 7 Lacosamide n = 1. 8 Lacosamide n = 1. 9 Carbamazepine (n = 1), oxcarbazepine (n = 1). 10 Carbamazepine (n = 3), lacosamide (n = 2), lamotrigine (n = 14), oxcarbazepine (n = 15), phenytoin (n = 1), rufinamide (n = 1). One or several sodium-channel inhibitors had been used by each individual. 11 Acetazolamide, ethosuximide, perampanel, prednisolone, sulthiame, or vigabatrine. 12 Self-treatment. 13 Less than 50% seizure reduction. diagnosis in the younger age group. The CIN for children born in 2000-2009 (1/46,000) in our study is comparable with CINs of 1:41,000 and 1:46,000 reported in two large studies with SCN1A-positive DS participants from referral centers for SCN1A testing (Brunklaus et al., 2012;Ishii et al., 2017). The CIN for children born in 2010-2018 (1:33, 000) is closer to, albeit not as high as the CINs reported from smaller single and multicenter studies and one prospective study. In these studies, where closer contact with all children with epilepsy, easier access to medical records and more comprehensive surveillance might simplify the work to find and diagnose all children with DS, the CIN varied between 1:12,200 and 1:30,000 (Bayat et al., 2015;Symonds et al., 2021Symonds et al., , 2019Wu et al., 2015;Yakoub et al., 1992). Inclusion criteria mandating a pathogenic SCN1A variant, cognitive slowing, refractory seizures and/or tonic-clonic seizures as requirements for DS diagnosis vary between studies and makes it difficult to compare incidences.
The prevalence rate of DS in our study (1:45,000) is comparable with that of 1:46,000 in a previous Swedish study from 2015 (Rosander and Hallbook, 2015), even though our inclusion criteria were stricter excluding seven of the participants from that study, suggesting better identification of a narrower phenotype. In our study one participant had seizure onset at 2.3 years. He was included in the study since he had a typical DS phenotype and a pathogenic SCN1A variant. Age at onset can occur later than the typical < 12 months in some children with DS (Li et al., 2021;Wirrell et al., 2017).
A SCN1A variant that could explain the syndrome was detected in 51/53 (96%) of the children where data was available. This can be compared to an estimated frequency of 88% in the previous Swedish study (Rosander and Hallbook, 2015), and around 80% in studies published before 2013 (Brunklaus et al., 2012;Harkin et al., 2007;Marini et al., 2009). New molecular technologies (next generation sequencing) have improved the diagnostic yield (Djemie et al., 2016), and the definition of DS has been stricter leading to exclusion of individuals with pathogenic variants in other genes with a phenotype with traits of DS (Mei et al., 2019). Comparisons with more recent studies is difficult as the presence of a pathogenic SCN1A variant has often been one of the inclusion criteria (Bayat et al., 2015;Ishii et al., 2017;Li et al., 2021;Symonds et al., 2021Symonds et al., , 2019. It must be noted that in a recent study from Scotland all 11 children who met criteria for DS had an SCN1A variant (Symonds et al., 2021). The most common pathogenic variants in our participants were truncating 23/51 (45%) and missense variants 21/51 (41%). Other studies have reported similar frequencies of 37-42% for truncating and 36-46% for missense variants (Brunklaus et al., 2012;Depienne et al., 2009).
Of 53 children born in Sweden seven (13%) had died at a median age of 4.7 years. In 3/53 (7.5%) the death was due to definite/probable/ possible SUDEP. These data are in line with other studies showing a mortality rate of 3.7-17%, a SUDEP rate of 2.3-10% and a median age at death of 4.0-7.0 years (Cooper et al., 2016;Sakauchi et al., 2011;Skluzacek et al., 2011). There was no mortality due to status epilepticus (SE) in our study, as compared to 1.0-8.2% in these studies. Mortality due to SE might have decreased because of earlier diagnosis and better treatment (Gataullina and Dulac, 2017). Surprisingly, 3/53 (5.7%) of individuals died of non-SUDEP related pulmonary infections in our study, as compared to no registered death due to infection in three other studies with a mortality of 111/1556 individuals (Cooper et al., 2016;Sakauchi et al., 2011;Skluzacek et al., 2011). Three out of seven (43%) deaths were due to pulmonary infections without SUDEP, and another two (29%) due to probable/possible SUDEP associated with pulmonary infection, in our study, as compared to nine out of 177 (5.1%) deaths due to infection in a review on mortality in DS (Shmuely et al., 2016).
All 42 children included in the assessment had ongoing seizures. Febrile seizures had been observed in all children and convulsive SE in 90%. Focal to bilateral tonic-clonic seizures and myoclonic seizures were the most common seizure types. These results are in line with previous studies (Brunklaus et al., 2012;Dravet et al., 2011). Caregivers reported ongoing or previous tonic seizures in 25/42 (60%) children, even after a thorough description and defined distinction from tonic-clonic seizures. This seizure type was as common in the younger as in the older age group. Tonic seizures have previously been considered unusual in children with DS Gataullina and Dulac, 2017). In recent studies tonic seizures have been reported by caregivers in 26-35% of individuals with DS (Li et al., 2021;Madan Cohen et al., 2021), with a median age at onset of 3 years (Li et al., 2021). Diagnosis of tonic seizures is challenging without ictal EEG studies, as tonic seizures often occur in sleep and might be very subtle (Li et al., 2021).
Most of the children were treated with anti-seizure medications recommended for DS (Cross et al., 2019;Wirrell et al., 2017). The usage of sodium-channel inhibitors was lower in the younger age group possibly reflecting a lower age at DS diagnosis and suggesting a better awareness of treatment recommendations. This is in line with a Norwegian study where the use of sodium-channel inhibitors in DS declined between 2008 and 2018 (Heger et al., 2020). However, four children (9.5%) in our study were still treated with sodium-channel blockers, and a seizure reduction of > 50% was reported in one child treated with lacosamide. It has been reported that some individuals with DS may be responsive to sodium-channel blockers, especially lamotrigine, with aggravation of seizures seen on a wean of the medication. Therefore, careful consideration needs to be given prior to an automatic wean without a definitive history of worsening of seizures (Dalic et al., 2015). In our study use of topiramate seems to be more common in the older age group. Possible explanations could be that side-effects are common and that topiramate is considered a second line ASM in Swedish treatment recommendations, and thus is tested at an older age after first line ASMs have been tried.
Twenty-two (52%) children had been using sodium-channel inhibitors previously and 28 (67%) had been using levetiracetam, compared to a lower proportion reporting current use: four (9.5%) children currently using sodium-channel inhibitors and seven (17%) currently using levetiracetam. Sodium-channel inhibitors are generally contra-indicated for DS but are a common choice for focal seizures before the DS diagnosis is made (Rosander and Hallbook, 2015). Levetiracetam is probably not continued in older children due to less effectiveness in children with DS (Chhun et al., 2011;Dressler et al., 2015). Only three children in the study had tried cannabidiol. One experienced > 50% and two ≤ 50% seizure reduction. Epidyolex had not been subsidized in the Swedish public health system at the time of data collection. Caregivers of 3/11 (27%) children treated with a ketogenic diet reported a reduction of number of seizures of > 50%. In two other studies 4/10 (40%) participants had > 50% reduction in number of seizures at long time follow up (Dressler et al., 2015) respectively 5/15 (33%) participants 12 months after starting a ketogenic diet (Nabbout et al., 2011).

Strengths and limitations of the study
Strengths of the study are that the data are population-based, that we have performed a thorough investigation to find all known cases in Sweden and that all assessed participants have been investigated by one of two pediatric neurologists, BB or TH.
There are some limitations. It has not been possible to employ universal screening/surveillance as in other studies (Bayat et al., 2015;Symonds et al., 2021Symonds et al., , 2019Wu et al., 2015), and thus there might be children who fulfill criteria for DS who have not yet been diagnosed. However, all Swedish children with a complex epilepsy are followed by a pediatric neurologist and there is a low threshold to refer such children to a university hospital. There are no local registries for DS and thus we cannot exclude that we have failed to identify some individuals with a known DS diagnosis. The data on mortality is retrospective. We cannot exclude that there were individuals, unknown to us, who have died with/without having a diagnosis of DS, which could affect both the mortality rate and the incidence in the two age-groups. Seizure diaries were not systematically used and data on seizure types and frequency might be affected by recall bias. Video EEGs were not used to distinguish the different seizure types, and home-videos of seizures were not systematically used. Only 18/42 (43%) children had undergone a formal neuropsychological evaluation and 33/42 (79%) had a diagnosis of ID and/or ASD. A high proportion of children who had contact with the habilitation interdisciplinary service (n = 40, 95%) and few children who went to ordinary school without extra support (n = 4/42, 9.5%) indicate that the proportion with ID and/or ASD might be underestimated.

Conclusions
This population-based study focuses on epilepsy, epidemiology, mortality, genetics and treatment of epilepsy in children with Dravet Syndrome (DS). Tonic seizures were common in all ages, this finding is new and must be confirmed in larger studies. The cumulative incidence was higher, the median age at DS diagnosis lower and use of contraindicated sodium-channel inhibitors lower in younger children, indicating improved diagnostic and therapeutic procedures and increased awareness of the disease. The high mortality rate is in line with previous studies. Mortality due to pulmonary infection was more common than in previous studies and there was no mortality due to SE.