[1] 19w Convulsive Status Epilepticus in a Quaternary Hospital Paediatric Intensive Care Unit (PICU) in South Africa: an 8 year review

[1] 31w  Clinical profile, aetiology, neuroimaging findings and outcome of CSE in children  Seventy two percent of patients had refractory CSE  There was a high mortality rate of 24% (n=18)

[1] 94w We identified 76 children with CSE out of a total of 2452 admissions to the PICU over an 8 year period from January 2007 to December 2014. Patient characteristics and seizure semiology on admission are detailed in Table 1. The aetiology of CSE including classification and details of infectious causes can be found in Table 2 and 3. Meningoencephalitis comprised the majority of the infectious aetiology in 33 (83%) patients. The causative organisms identified in the meningoencephalitis group were the following: herpes simplex virus (n=3), pneumococcus (n=2), haemophilus (n=1) and Group B streptococcus (n=1).

[2] 106w Fifty two (68%) patients had comorbid medical conditions diagnosed in PICU. Of these patients, 22 (42%) had bronchopneumonia, 6 (11%) had multiple organ dysfunction syndrome, 5 (10%) had sepsis, 3 (6%) had hydrocephalus and 16 (31%) with other co-morbid medical conditions. Fifty five (72%) patients had refractory CSE requiring more than 2 antiepileptic agents to suppress seizure activity prior to referral to PICU. Twenty six (34%) were initially treated with diazepam and 25 (32%) with lorazepam. The most common secondary agent administered was phenytoin used in 64 patients (84%). Midazolam infusion was administered in 53 (71%) patients and thiopentone infusion in 3 (4%) patients. [Table 4]

[3] 100w The most frequently used antiepileptic drugs for CSE in PICU were midazolam (73/76, 96%), sodium valproate (48/76, 63%), phenytoin (43/76, 57%), phenobarbitone (37/76, 49%), thiopentone (17/76, 22%), topiramate (13/76, 17%) and intermittent lorazepam or diazepam (35/76, 46%) [Table 4]. The mean dose of midazolam required for seizure control was 390mg/kg/hr (SD 283.2, range 60-1200mg/kg/hr). The mean dose of thiopentone infusion required for seizure control was 3.2mg/kg/hour (SD 2.3, range 1-8.5mg/kg/hour). Seven (9%) patients required more than 5 antiepileptic drugs to gain seizure control in PICU. The main drugs prescribed on discharge from PICU were sodium valproate in 42 (70%) patients.

[4] 33w [ Table 4] The mean duration of time from onset of seizures to admission in PICU was 2.5 days (range 1 hour-21 days). Forty patients (53%) had documented seizures after admission to PICU.

[5] 25w The mean time taken to gain seizure control of these patients for at least 48 hours was 2 days (SD 3.2, range 1 hour-15 days).

[6] 156w Neuroimaging was done in 70 (92%) patients during the course of their admission at either the base hospital or in PICU. 25 (36%) patients had magnetic resonance imaging (MRI) and 68 (97%) had computerized tomography (CT) scan. Both CT scan and MRI were performed in 23 (33%) patients. Neuroimaging findings were abnormal in 53 (76%) patients. There was evidence of hypoxia or hypoperfusion related changes in 17 (32%), arterial infarct in 7 (13%), leptomeningeal enhancement in 6 (11%), cerebral oedema in 6 (11%), cerebral atrophy in 5 (9%), venous thrombosis with venous infarcts in 4 (8%) and 8 (16%) had other abnormalities. Electroencephalogram (EEG) was performed in 44 (58%) patients and 39 (89%) recordings were abnormal. Encephalopathy was the predominant abnormality recorded in 23/39 (59%) patients followed by epileptiform activity in 4 (10%), low amplitude pattern in 4 (10%), electrocerebral silence in 4 (10%), burst suppression in 3 (8%) and diffuse beta activity in 1 (3%)

[7] 18w patient. An encephalopathic EEG pattern was defined as a recording with either intermittent or continuous generalised background slowing.

[8] 95w On discharge from PICU 58% of patients (n=44) had neurological impairment. Twenty four of forty four (55%) patients had spastic quadriparesis. Of the 24 patients, 9 (21%) had visual impairment and 5 (11%) had visual and auditory impairment. Other neurologic sequelae on discharge were encephalopathy in 7/44 patients (16%), hemiparesis in 4 (9%), movement disorder in 4 (9%), isolated visual impairment in 2 (5%), monoparesis in 1 (2%), aphasia in 1 (2%) and flaccid paralysis in 1 (2%). Fifty eight of 60 (97%) patients continued antiepileptic medication for seizure control after the episode of CSE.

[9] 93w Sixteen of the 76 (21%) children admitted with CSE died during admission to PICU in comparison with an average overall PICU mortality rate of 16% determined from the audit. However, a further 2 patients died within 30 days of discharge from PICU and thus the overall mortality rate was 24% (n=18). The main cause of death was CSE related diffuse cortical and brainstem injury in 13/18 (72%) patients, suspected pulmonary aspiration in 2 (11%), unexplained cardiac arrest in 1 (5%), multiple organ dysfunction syndrome in 1 (5%) and 1 was unknown (5 %).

[10] 160w Patients ventilated for more than 10 days had more than two times the risk of poor outcome compared to those ventilated for less than 3 days (RR 2.08 (1.30 -3.33), p= 0.002). A significant effect was also observed for patients ventilated for 4-5 days (RR 1.98 (1.22-3.20), p= 0.005) and 6-10 days (RR-2.06 (1.29-3.31), p=0.003). Patients with abnormalities on brain neuroimaging were 3.2 times more likely to experience a poor outcome compared to those with normal findings (RR 3.21 (1.53 -6.72), p=0.002). The number of antiepileptic drugs administered at the base hospital, in contrast to the number of drugs used in PICU, had no significant effect on the risk of adverse outcome (p=0.699). Patients who needed more than 3 antiepileptic drugs to abort seizures in PICU had a significantly higher risk of morbidity and mortality (RR 1.41 (1.12 -1.78), p=0.003). [Tables 5 and 6] Thirty five of the 58 (60%) patients discharged from PICU were assessed at follow up visits.

[11] 69w The median duration of follow up was 5 months (IQR 2-15 months). Twenty three of 35 (65%) patients followed up had significant morbidity. Sixteen of 35 (46%) patients had persistent seizures despite antiepileptic drug therapy. The neurological deficits were cerebral palsy in 15/23 (65%), encephalopathy in 2 (9%), language impairment in 2 (9%), visual impairment in 2 (9%), movement disorder in 1 (4%) and developmental delay in 1 (4%).

[12] 37w Fourteen of 35 (40%) patients had repeat neuroimaging and persistent abnormalities were present in 11/14 (79%) patients. Twenty eight of 35 (80%) patients still needed antiepileptic medication for seizure control on their last clinic visit. [Table 2]

[1] 89w Refractory status epilepticus has been documented in 10-70% of children with CSE worldwide 12 . Fifty five of 76 (72%) patients in this study were admitted to PICU because they had refractory convulsive status epilepticus. The remaining 21 (28%) patients were admitted for respiratory depression which required ventilation after administration of repeated doses of benzodiazepine or a second antiepileptic drug. The high rate of refractory CSE in our study is related to the high incidence of a symptomatic aetiology with associated delay in patient presentation to a healthcare facility.

[2] 149w AED initiation more than 1 hour after onset of seizure has been associated with a poorer outcome 13 . The retrospective nature of this study did not allow us to accurately determine the duration of status epilepticus prior to initial treatment at the base hospital. Thus, we were unable to determine the association with outcome. However, the lack of an efficient transport system results in delays in inter hospital transportation and many patients only present to hospital hours to days after the onset of status epilepticus. The mean duration of time from onset of seizures to admission to PICU in our study was 2.5 days. In contrast, a United Kingdom PICU study showed a mean duration of CSE prior to first medical treatment of only 20 minutes and a mean duration of 44 minutes prior to seizure termination. This study demonstrated no mortality and only 6% morbidity 2 .

[3] 126w Acute symptomatic CSE occurred in 80% of our patients and this proportion far exceeds that documented in studies in developed 2,10,11 and developing countries 3,7,8,9 . Recent studies documented acute symptomatic CSE rates of 65% in India 9 and 15-26% in the UK 2 and USA 11 . An infectious aetiology was identified in more than half of our patients and gastroenteritis with metabolic derangements in a quarter of them. Central nervous system infections, such as viral encephalitis and bacterial meningitis, were the predominant aetiology and were a significant marker for morbidity and mortality in a Taiwanese study that assessed 198 episodes of status epilepticus 14 . Presumed encephalitis was also the commonest aetiology in refractory status epilepticus in a 2014 study in India 9 .

[4] 121w Only 4% of the patients in our study had complex febrile seizures. This contrasts with findings in well-resourced countries where febrile status epilepticus comprises at least one third of patients with CSE and is associated with a low mortality rate of 0-2% 2,10,12 . The low mortality rates in developed countries are possibly due to rapid pre-hospital termination of seizures using intramuscular anti-epileptic drugs as supported in the RAMPART (Rapid Anticonvulsant Medication Prior to Arrival Trial) study 15 . A prospective study of the consequences of febrile status epilepticus (FEBSTAT) demonstrated that the time from seizure onset to anti-epileptic drug initiation was significantly linked to a shorter seizure duration which was related to a decreased need for ventilator support 16 .

[5] 88w A longer duration of PICU admission and mechanical ventilation was significantly linked to increased morbidity and mortality. The duration of admission and ventilation (9 and 7 days respectively) in our study was less than in an Indian study that documented a mean admission duration of 19 days and mechanical ventilation of 8 days in children 9 . A short mean duration of PICU stay of only 3.5 days documented in a UK study may be associated with a focused protocol based rapid termination of status epilepticus 2 .

[6] 71w Thirty percent of patients with status epilepticus had a cerebral abnormality revealed on neuroimaging in a 2010 US study by RK Singh et al 11 . The main findings were metabolic, vascular, cerebral oedema and dysgenesis. This contrasts with our study that demonstrated abnormalities in 53 of the 70 patients (75%) that had neuroimaging with the predominant finding of hypoxic changes. This was significantly associated with a higher morbidity and mortality.

[7] 194w The in-hospital case fatality rate of 21% (30 day post discharge mortality of 24%) and discharge morbidity of 58% is much higher than that reported in studies in developed countries (0-6% and 1.4-17%) 2,13 . However, the morbidity and mortality rates from this study are less than that in an Egyptian study that reported a mortality rate of 37.1% and morbidity of 83% 8 . The ICU study of refractory CSE in India showed a mortality rate of 19.5% with a 6month follow up morbidity of 32% 9 . The high mortality and morbidity rates in our study could be due to selection bias as this was a study conducted in the PICU with the majority of patients having refractory status epilepticus and all patients requiring ventilation had a high rate of co-morbid infections. CSE related diffuse cortical and brainstem injury in the majority of patients that demised in our study underscores the delay in presentation to hospital and the refractory nature of the seizures as well as the severe infectious aetiology in the majority of patients. Many of these patients had ventilation withdrawn or de-escalation of care after assessment by a multidisciplinary team.

[8] 114w The limitations of this study were that it was retrospective and based on clinical records which had occasional incomplete documentation. The accurate duration of status epilepticus was not documented. Patients with CSE were managed with a patient targeted protocol in the PICU. There was no formal neurodevelopmental assessment using standardized test batteries and there was a high rate of patients lost to follow up. Poor outcomes in this series might reflect more severe aetiologies like meningoencephalitis or PICU complications rather than more severe cases of status epilepticus. However, the strength of this study is that we were still able to determine important clinical information related to the profile and outcome of patients with CSE.

[1] 29w To our knowledge, this is the first study that documents the clinical profile, neuroimaging and EEG findings, and outcome of CSE in children admitted to PICU in Sub-Saharan Africa.

[2] 61w Morbidity and mortality were significantly related to the number of anticonvulsant drugs used in PICU, abnormalities on neuroimaging and the duration of mechanical ventilation. These factors are proxy indicators of the severity of refractory status epilepticus and the associated aetiologies, comorbidities and complications. The high incidence of symptomatic aetiology highlights the need for the early recognition and treatment of underlying conditions.

[3] 98w Table 1. Patient Characteristics (n = 76) Gender n[%] Male 47 (62) Female 29 (38) Race n[%] Black 68 (90) White 1 (1) Asian 7 (9) Age at Presentation Mean age in months (SD) 25 (28) Median age in months (IQR) 15 (6 -37) By category n[%] < 6 months 16 (21) > 6 months -1 year 19 (25) > 1 year -2 years 16 (21) > 2 years -5 years 18 (24) > 5 years -10 years 6 (8) > 10 years 1 (1) HIV Status n[%] Positive 6 (8) Negative 48 (63) Not documented 22 (29)

[4] 12w Documented hypoglycaemia on admission 11 (14) No hypoglycaemia on admission 65 (86)

[5] 4w Generalised tonic-clonic 40 (53)

[6] 13w Focal onset seizures evolving into convulsive status epilepticus 14 (18) Tonic 3 (4)

[7] 372w Table 2: Aetiology (n=76) n[%] Acute Symptomatic 65 (86) Infections 40 (61) Gastroenteritis with metabolic derangements 19 (29) Trauma related 3 (5) Tumours 2 (3) Intracerebral haematoma 1 (2) Remote Symptomatic 6 (8) Idiopathic Epilepsy Related 1 (1) Prolonged Febrile 3 (4) Unclassified 1 (1) Table 3: Infectious Aetiology Subgroup (n=40) n [%] Presumed viral encephalitis 18 (45) Bacterial/Viral meningitis 10 (25) Tuberculosis meningitis 5 (13) Malaria 2 (5) Sepsis 4 (10) HIV Vasculitis 1 (2) Table 4. Antiepileptic Drug Use Base Hospital (n=76) PICU (n=76) Discharge (n = 60) Follow Up (n = 28) Antiepileptic Drug (n[%]) Diazepam 26 (34) 6 (8) 0 (0) 0 (0) Lorazepam 25 (33) 29 (38) 0 (0) 0 (0) Phenytoin 64 (84) 43 (57) 22 (37) 2 (7) Phenobarbitone 44 (58) 37 (49) 21 (35) 4 (14) Sodium Valproate 30 (40) 48 (63) 42 (70) 23 (82) Duration of ventilation < 0.001 ≤ 3 days 75 (12) 16.7 (10) 4 -5 days 12.6 (2) 30 (18) 6 -10 days 6.2 (1) 25 (15) > 10 days 6.2 (1) 28.3 (17) Duration of admission 0.039 ≤ 3 days 43.8 (7) 16.7 (10) > 3 days 56.2 (9) 83.3 (50) Neuroimaging result <0.001 Normal 75.0 (12) 8.3 (5) Abnormal 18.8 (3) 83.4 (50) Not done 6.2 (1) 8.3 (5) Number of drugs: Base ≤ 3 drugs 56.2 (9) 61.7 (37) 0.777 > 3 drugs 43.8 (7) 38.3 (23) Number of drugs: ICU 0.004 ≤ 3 drugs 46.7 (28) 87.5 (14) > 3 drugs 53.3 (32) 12.5 (2) Table 6. Multivariable analysis for poor outcome Variable Category RR (95% CI) p-value Age (years) 1.02 (0.97 -1.07) 0.364 Seizure semiology Generalised tonic clonic (GTC) CSE 1 Focal onset seizures evolving into bilateral CSE 0.89 (0.62 -1.29) 0.544 Tonic status epilepticus 0.83 (0.37 -1.88) 0.661 Duration of ventilation <=3 days 1 4-5 days 1.98 (1.22 -3.20) 0.005 6 -10 days 2.06 (1.29 -3.31) 0.003 >10 days 2.08 (1.30 -3.33) 0.002 Duration of admission <=3 days 1 >3 days 1.44 (0.95 -2.18) 0.083 Neuroimaging result Normal 1 Abnormal 3.21 (1.53 -6.72) 0.002 Number of drugs at base hospital <=3 drugs 1 >3 drugs 0.95 (0.75 -1.22) 0.699 Number of drugs in PICU <=3 drugs 1 >3 drugs 1.41 (1.12 -1.78) 0.003

[1] 36w Full ethical approval was granted by the University of KwaZulu Natal (UKZN) Biomedical Research Ethics Committee BE401/15. Gatekeeper permission was granted from IALCH and from the KwaZulu Natal Provincial Health Research and Knowledge Management Committee (HRKM).

[2] 33w We conducted a retrospective electronic health record analysis of patients admitted with CSE to the Paediatric Intensive Care unit (PICU) at Inkosi Albert Luthuli Central Hospital (IALCH) from January 2007 to December 2014.

[3] 34w . The study population was identified using admission registers. A search was conducted for keywords of 'seizures', 'status epilepticus' or 'epilepsy' in the admitting diagnosis. Data was collated from existing electronic patient health records.

[4] 72w Information collected included demographic details (age and sex), HIV status, pre-existing medical diagnosis, aetiology, duration from onset of seizures to PICU admission, anticonvulsant drugs used, neuroimaging findings, electroencephalogram (EEG) results and neurological examination at time of discharge and the last follow up assessment. Sub group analysis was done on the demographics, underlying aetiology, seizure duration prior to admission, the number of antiepileptic drugs used, seizure semiology as well as morbidity and mortality.

[5] 58w Morbidity was defined as persistent seizures or a new neurological deficit such as cerebral palsy, visual or auditory impairments, hemiparesis, movement disorder, aphasia, monoparesis or flaccid paraparesis. A favorable outcome was defined as patients without documented morbidity on discharge and follow-up. Patients that died and those that had morbidity on discharge were included in the poor outcome group.

[6] 113w Data was analysed using Stata Version 13 (StataCorp, Texas, USA). Results were considered statistically significant at the 5% level of significance. Normality tests were conducted and nonparametric tests were used where necessary. Descriptive statistics were computed and tests of differences and associations were performed. Fisher's exact test was used in lieu of Pearson's chi square test where expected cell counts were less than 5. The Kruskal-Wallis test of difference was also undertaken to investigate differences in continuous variables between the subgroups of favourable and poor outcome. To assess the effect of risk factors on adverse outcome, relative risks were computed. Log binomial regression was performed to estimate unadjusted and adjusted relative risk ratios.

[1] 116w The global burden of epilepsy is estimated to be 33 million children, of whom 80% are living in resource poor countries 1 . These countries have a wide treatment gap without access to efficacious anticonvulsant therapy 1 . Approximately 70% of status epilepticus occurs in children less than 1 year of age. The incidence of convulsive status epilepticus (CSE) in the United Kingdom (UK) has been reported as 18-20/100 000 per year in children under 16 years of age 2 . The exact incidence of CSE in Sub-Saharan Africa is unknown. However, a recent study in rural Kenya documented an incidence of up to eight times higher than that reported in the UK study 2,3 .

[2] 112w The new definition proposed by the International League Against Epilepsy (ILAE) Task Force on Classification of Status Epilepticus in 2015 is that CSE is a condition resulting either from the failure of the mechanisms responsible for seizure termination or from the initiation of mechanisms, which lead to abnormally, prolonged seizures. It can have long-term consequences after 30 minutes including neuronal death, neuronal injury, and alteration of neuronal networks 4 . The duration of seizures is independently associated with a worse response to treatment and outcome 5 . Refractory status epilepticus is defined as seizure activity that persists despite the use of an initial benzodiazepine and a second appropriate antiepileptic medication 6 .

[3] 77w Status epilepticus is associated with high mortality and morbidity rates in resource poor developing countries 3,7,8,9 . There have been numerous studies conducted in developed countries 2,10,11 assessing the aetiology, outcome and management of convulsive status epilepticus but there is a paucity of data in Sub-Saharan Africa. Standard treatment guidelines are variable with limited access to intensive care facilities. Delays in the transfer of patients with impending CSE result in a higher incidence of refractory status epilepticus.

[4] 27w The primary objective of this retrospective study was to document the clinical profile, aetiology, management and outcome in children admitted to a quaternary hospital PICU with CSE.

[5] 56w All children between the ages of 1 month and 13 years with an admission diagnosis of CSE were included. The PICU at IALCH only admits children up to the age of 13 years. Children less than 1 month of age were excluded as most had perinatal complications and were managed in the Neonatal Intensive Care Unit.