Convulsive status epilepticus (CSE) is a potentially devastating condition which can result in significant morbidity and mortality. Studies addressing status epilepticus in children are rare and there is a paucity of large randomised controlled trials in children looking at forms of drug treatment for SE. There is consistency worldwide in guidelines for first line treatment of CSE with benzodiazepines, with slight variations in type and route of administration of agents. Second line therapy usually entails phenobarbital or phenytoin parenterally. Both repeated phenobarbital loading doses and midazolam infusions have been shown to be effective and safe in the management of established convulsive SE, but there are no prospective randomized controlled trials comparing the two in children.

Our study has been undertaken to review 2 existing, and routinely used, interventions for children presenting to our center with acute convulsive seizures. In order to permit comparable data to be collected we are randomly allocating these standard interventions prospectively. This is in order to compare the efficacy and safety of two treatment protocols (phenobarbital vs phenytoin and midazolam) both of which as stated are already part of existing standard protocols internationally and in South Africa. Parenteral phenobarbital is a safe, affordable and easy to use drug in the management of status epilepticus especially for poorly resourced communities where undertaking infusions may be unsafe, time consuming or unavailable.

We hypothesize that repeated phenobarbital loading is as effective and safe, or more so, than phenytoin followed by midazolam infusion in the management of established and refractory childhood convulsive SE. If proven, then the former would be a viable option for all health care workers with access to intravenous routes (including Day hospitals) where infusions are unsafe, time consuming or unavailable.

This prospective study was conducted at the Red Cross War Memorial Children's Hospital (RCWMCH) and recruitment ran from between March 2015 to March 2018. All patients presenting with convulsive status epilepticus (CSE) who presented to the medical emergency unit at RCWMCH and needed therapeutic intervention were entered into the study by the attending medical staff. Study data was collected using REDCap hosted by the University of Cape Town's eResearch Centre and the study was approved by the UCT Human Research Ethics Council (UCT HREC 297/2005).

Definitions The definition of CSE was defined as any convulsive seizure that lasted longer than five minutes or multiple discrete seizures between which there is no extended period of recovery between events (Trinka et al., 2015). The onset of CSE was defined as the time provided by the caregiver who accompanied the child. The time to admission and to treatment were recorded by the staff in the unit. If children were admitted multiple times, each admission was captured independently, but only data from the first admission was included in this report . The full diagnosis of CSE was described using the multiaxial classification system. However, as it was not possible to perform EEG on all patients, this axes was excluded. Febrile status epilepticus was defined .

Treatment protocols Upon entry into the study, children were randomly allocated to one of two protocols . Both these protocols are well-established treatment protocols used in the sub-Saharan African setting for the management of SE . Randomization of protocols was performed using a free online platform (Research Randomizer ©). Both protocols began with children receiving first-line benzodiazepines (either midazolam, lorazepam or diazepam) which were either administered intravenously (IV), per rectally, intranasally or sublingually. If the children did not respond to single dose of benzodiazepines, they were given a second dose 5-10 minutes after the first dose. Pre-hospital administration of benzodiazepines by emergency services were counted if administered intravenously. However, all other routes of administration were not counted due to the lack of consistency in their administration.

If CSE continued after two doses of benzodiazepines, children were then randomized to second-line agents according to the protocol allocated to them. The one protocol, termed 'Phenobarbital' (PHB), instructed the clinician to give giving an IV bolus of phenobarbital (20mg/kg ). If CSE did not terminate after 5 - 10 minutes, a second dose was given at half the dosage (10mg/kg) and a third dose (10mg/kg) was given if CSE persisted 5-10 minutes after that.

In the second protocol, termed 'Phenytoin / Midazolam infusion' (PHY/MDZ), children were given a dose (20mg/kg) of IV phenytoin mixed with 50mL of normal saline solution and administered over 30 minutes . If the patient was still in CSE 5-10 minutes after the phenytoin was given, they were then started on a midazolam infusion. This included a loading dose of IV midazolam (0.2mg/kg) followed by an infusion set at 3mg/kg into 50mL 5% dextrose water given at a rate of 1-4 mL/hour (equivalent to 1-4 mcg/kg/min ).

If a patient child did not respond to the PHB or the PHY/MDZ protocols, they were referred to the pediatric intensive care unit (PICU). Other reasons for admission to the PICU included respiratory depression following administration of the second-line agent, need for inotropic support, etiology-related concerns requiring intensive monitoring (e.g. severe electrolyte imbalances) and or prolonged state of a depressed level of consciousness.

Demographic data inclusive of age, sex, etiologies, pre-existing medications, previous medical conditions and co-morbidities were recorded.

Outcome measures In comparing the two treatment protocols, we will only focused on the short-term outcomes of the children in each treatment protocol. These include how the agents affected the children's physiology, admission to PICU, whether subsequent breakthrough seizures occurred and days admitted to hospital. In measuring the effects on the children's physiology, their we will calculate heart rate, respiratory rate and mean arterial pressure from during CSE to immediately post-ictal period. This will be done by subtracting the first measurement as the child presented to the unit from the first measurement taken immediately after the child had stopped convulsing.

Data analysis During the analysis, the investigator performing the analysis was blinded to which protocol the patients children were allocated to. Group allocations will only be unblinded after statistical analysis is completed and verified by an external party. Data will analysed using SPSS Statistics (IBM Corp. Released 2016, Version 24.0. Armonk, NY: IBM Corp). Statistical measurements will be performed using both SPSS Statistics and GraphPad Prism version 6.0 (GraphPad Software, USA). For continuous data normality will be established using the Shapiro-Wilk test and thereafter parametric (i.e. paired or unpaired student's t-tests) or nonparametric tests (i.e. Mann-Whitney U test or Wilcoxon signed-rank test) will be performed. Normally distributed data will be reported as mean standard deviation. Data that is not normally distributed will be reported as median with the interquartile range (IQR). Categorical data will be summarized in contingency tables with differences between groups being identified using the Fisher-exact or chi-squared (X2) tests and associations calculated using odds ratios (OR). Significance will be defined as a p < 0.05.

Inclusion Criteria:

• Children in convulsive status epilepticus (as defined by Trinka et al 2015)

Exclusion Criteria:

• Children not in convulsive status epilepticus

We are exploring what data would of use

The specific treatment protocol was blinded to the data analyser