Pre-hospital and emergency department treatment of convulsive status epilepticus in adults: an evidence synthesis

Variation in services and/or uncertainty about best practice

There is considerable uncertainty and a limited evidence base on the effectiveness of the first-line treatment options available for treating status epilepticus. As a result, treatment guidelines do not recommend any specific treatment over another and treatment decisions are frequently based on expert opinion. 4,24,27

A recent retrospective chart review conducted for patients in the USA diagnosed with status epilepticus within the last 10 years concluded that treatment for status epilepticus does not consistently follow guidelines or recommendations. In particular, underdosing of first-line benzodiazepines (i.e. lorazepam and midazolam) was identified. 28

Relevant national guidelines

The National Institute for Health and Care Excellence (NICE) guideline on the diagnosis and management of epilepsies, which was published in 2012 and subsequently updated (last updated 12 May 2020),25 provides recommendations on first-line treatment for convulsive status epilepticus in people of all ages. 25

In particular, the NICE clinical guidance25 supports the following recommendations:

• Give immediate emergency care and treatment to children, young people and adults who have prolonged (lasting ≥ 5 minutes) or repeated (three or more in an hour) convulsive seizures in the community.

• Prescribe buccal midazolam or rectal diazepam for use in the community only for children, young people and adults who have had a previous episode of prolonged or serial convulsive seizures.

• Administer buccal midazolam as first-line treatment in children, young people and adults with prolonged or repeated seizures in the community. Administer rectal diazepam if preferred or if buccal midazolam is not available. If i.v. access is already established and resuscitation facilities are available, administer i.v. lorazepam.

• Care must be taken to secure the child, young person or adult’s airway and assess their respiratory and cardiac function.

• Depending on response to treatment, the person’s situation and any personalised care plan, call an ambulance, particularly if:

  • the seizure is continuing 5 minutes after the emergency medication has been administered

  • the person has a history of frequent episodes of serial seizures or has convulsive status epilepticus, or this is the first episode requiring emergency treatment

  • there are concerns or difficulties monitoring the person’s airway, breathing, circulation or other vital signs.

Scottish Intercollegiate Guidelines Network (SIGN) 143 guideline for the diagnosis and management of epilepsy in adults, which was originally published in 2015 and subsequently revised in 2018,16 states that many status epilepticus seizures resolve themselves without treatment; emergency treatment is necessary if there are continuous seizures or serial seizures lasting for ≥ 5 minutes. The current recommended treatments are:

• 10 mg of midazolam administered either buccally or intranasally

• 4 mg of i.v. lorazepam if midazolam is unavailable

• 10 mg of diazepam administered either intravenously or rectally if both midazolam and lorazepam are not available.

The guidance further points out the need to treat status epilepticus as soon as possible as it can worsen if treatment is delayed.

In 2016, the American Epilepsy Society published an algorithm for the treatment of convulsive status epilepticus. 29 The use of a benzodiazepine is recommended for initial therapy, specifically intramuscular (i.m.) midazolam, i.v. lorazepam or i.v. diazepam. If none of these is available, then i.v. phenobarbital, rectal diazepam or intranasal midazolam is recommended as first-line treatment. Recommended second-line options are i.v. fosphenytoin, i.v. valproic acid or i.v. levetiracetam, with i.v. phenobarbital recommended if these are not available.

An older set of guidelines, published by the European Federation of Neurologists in 2010, recommends i.v. administration of 4–8 mg of lorazepam or 10 mg of diazepam directly followed by 18 mg/kg phenytoin. 30

Benzodiazepines

Benzodiazepines are currently the first treatment for status epilepticus. 25 In UK clinical practice, the most common benzodiazepines used as first-line treatment for status epilepticus are midazolam, diazepam and lorazepam. Benzodiazepines vary in their potency, onset and duration of effect, uptake, distribution, metabolism and presence or absence of active metabolites. 31,32 Lorazepam is more potent than midazolam, which in turn is more potent than diazepam. Midazolam and diazepam are more lipid soluble than lorazepam and are quicker in crossing the blood–brain barrier, resulting in a more rapid onset of action (2–10 minutes) than lorazepam (5–20 minutes). 26,33 The half-life of midazolam is 3–11 hours, compared with 8–15 hours for lorazepam and 20–120 hours for diazepam. 33,34 Midazolam and diazepam metabolites are active and they tend to accumulate with prolonged administration, especially in patients with renal dysfunction. 26,35 Lorazepam metabolites are not active and, for this reason, it is the preferred benzodiazepine in patients with renal failure. 26 Clonazepam has been investigated in the treatment of status epilepticus but is rarely used in the USA, as it is not available as an i.v. formulation. 36,37 Adverse effects of benzodiazepines include hypotension, respiratory depression, paradoxical agitation and delirium. 31,33,38

Antiepileptic drugs

The NICE guideline25 on diagnosis and management of epilepsy recommends further research into three antiepileptic drugs for the initial treatment of status epilepticus: levetiracetam, sodium valproate and phenytoin. 25 This assessment includes any worldwide study assessing the effects of these drugs as a first-line treatment of status epilepticus in adults. At present, in the UK, these drugs are mainly used as a second-line treatment when benzodiazepines have failed to stop seizures.

Identification of studies (search strategy and information sources-dates)

A sensitive search strategy to identify all relevant literature, using database index terms and text words, was developed by an information specialist. The search encompassed the facets of status epilepticus, all benzodiazepines and antiepileptics in addition to named drugs, and the pre-hospital setting. The results were limited to RCTs in accordance with the review protocol. There were no date or language restrictions. The electronic databases searched included MEDLINE, EMBASE and PsycInfo^® (all via the Ovid interface), and EBSCOhost, CINAHL and Cochrane CENTRAL. The searches were carried out in July 2020. The complete search strategies are reported in Appendix 1.

Inclusion and exclusion criteria

Data extraction strategy

Two review authors (MC and MI) independently screened all citations identified by the search strategies. Full-text versions of potentially relevant articles were retrieved and independently assessed for eligibility by the same two reviewers. Disagreements were resolved by discussion or consultation with a third review author (MB, LA or CC). A data extraction spreadsheet was developed for the purpose of this assessment, piloted and amended as necessary. Two review authors (MC and MI) independently extracted information from each study on study characteristics, participant characteristics, flow of participants through the study, details of intervention/comparators, specified outcomes, potential confounding factors, funding and declarations of interest by study investigators. Disagreements were resolved by discussion.

Critical appraisal strategy

The risk of bias of included RCTs was assessed independently by two review authors (MC and MI) using a revised Cochrane Risk of Bias tool for randomised trials (RoB 2). 56 Disagreements were resolved by discussion. For the purposes of the risk-of-bias assessments only, we categorised the specified outcomes into two categories, with advice from the review team’s clinical expert (CC): subjective (i.e. requiring judgement by the observer) or objective (i.e. not requiring judgement by the observer). It was decided to split the subjective outcome ‘respiratory depression’ to include the specified objective outcome ‘respiratory depression requiring ventilation’ to take account of the objective nature of the ventilation component. The outcomes were categorised as follows:

• subjective outcomes –

  • seizure cessation

  • recurrence of seizure

  • respiratory depression (all related outcomes with the exception of respiratory depression requiring ventilation)

  • return to baseline neurological function

  • quality of life.

• objective outcomes –

  • mortality

  • need for additional treatment

  • need for hospital admission

  • length of stay in ICU

  • number of people requiring an emergency call out

  • respiratory depression requiring ventilation.

Methods of data synthesis

Random-effects meta-analyses and subgroup analyses had been specified in the protocol but were not carried out owing to the limited number of identified studies and their heterogeneity in terms of treatment comparisons and reported outcomes. The results of each included study were tabulated and summarised narratively for each outcome.

Quantity of the evidence (studies included and excluded)

The literature searches identified 191 records, and 46 additional records were identified in the study brief for a total of 237 retrieved records. After deduplication, 113 records were screened for relevance. Of these, 73 were considered potentially relevant and selected for full-text assessment. Eight articles could not be obtained because of COVID-19 pandemic restrictions. The search also identified one ongoing study. Of the 64 articles retrieved and assessed in depth, 13 articles relating to four studies met the inclusion criteria, while 51 articles were not deemed suitable for inclusion.

A PRISMA flow diagram detailing the process of study selection is presented in Figure 1. Appendices 2–4 provide the bibliographic details of the included, excluded and ongoing studies, respectively.

FIGURE 1.

The PRISMA flow diagram of selected studies. Reproduced with permission from Cruickshank et al. 60 This is an Open Access article distributed in accordance with the terms of the Creative Commons Attribution (CC BY 4.0) license, which permits others to distribute, remix, adapt and build upon this work, for commercial use, provided the original work is properly cited. See: http://creativecommons.org/licenses/by/4.0/. The figure includes minor additions and formatting changes to the original figure.

Reproduced with permission from Cruickshank et al. 60 This is an Open Access article distributed in accordance with the terms of the Creative Commons Attribution (CC BY 4.0) license, which permits others to distribute, remix, adapt and build upon this work, for commercial use, provided the original work is properly cited. See: http://creativecommons.org/licenses/by/4.0/. The figure includes minor additions and formatting changes to the original figure.

Study characteristics

The study characteristics of the four included RCTs22,57–59 (total number of participants 1345, of whom 1234 were adults) are detailed in Appendix 5 and are summarised in Table 1. All four trials were published in full. Three trials enrolled only adults22,57,59 and one trial included a mixed population of adults (89%) and children (11%). 58 Each trial reported a different treatment comparison. The trial by Alldredge et al. ,22 with a total of 205 participants, was the only three-arm trial and compared 2 mg of i.v. lorazepam with 5 mg of i.v. diazepam and i.v. placebo; the trial by Navarro et al. ,57 with a total of 203 participants, compared 2.5 g of i.v. levetiracetam plus 1 mg of clonazepam with i.v. placebo plus 1 mg of clonazepam; the trial by Silbergleit et al. ,58 with a total of 893 participants, of whom 782 were adults, compared 10 mg of i.m. midazolam with 4 mg of i.v. lorazepam; and the remaining trial, by Shaner et al. ,59 with a total of 44 participants, assessed 100 mg/minute i.v. phenobarbital plus 40 mg/minute optional phenytoin compared with 2 mg/minute i.v. diazepam plus 40 mg/minute phenytoin.

Three trials involved the administration of study drugs by paramedics22,57,58 and one study was based in an emergency room setting. 59

Three trials were conducted in the USA22,58,59 and one trial was conducted in France. 57 Shaner et al. ,59 recruited participants from a single centre, whereas the other three trials were multicentre: one physician-based hospital and nine destination hospitals;22 13 emergency medical centres and 26 hospital departments;57 and 4314 paramedics, 33 emergency medical centres and 79 receiving hospitals. 58

Participant characteristics

The eligible population in the trial by Silbergleit et al. ,58 known in the literature as the Rapid Anticonvulsant Medication Prior to Arrival Trial (RAMPART), consisted of adults and children with body weight of at least 13 kg. Via personal correspondence with the principal investigator (Professor Robert Silbergleit, Department of Emergency Medicine, University of Michigan, MI, USA), it was established that 116 participants were aged < 16 years and that a publicly accessible data set of participant-level data was available. This data set was obtained, and relevant data for participants aged ≥ 16 years were extracted. Despite there being a comprehensive data dictionary alongside the data, it was not always possible to identify which variables were pertinent for the relevant outcomes. Therefore, only primary outcomes that were considered to be of clear origin in the data set are reported in this assessment.

Table 2 presents a summary of the characteristics of participants included in the four trials and Appendix 6 presents full details of participants’ characteristics.

The mean age was reported in all four trials and ranged from 55 years57 to 66 years22 in the active treatment arms and from 43.8 years59 to 71 years22 in the control arms. All four trials also reported the sex of participants and all involved more males than females, with around two-thirds of study populations tending to be male.

With regard to the discharge diagnosis, two trials distinguished between status epilepticus and non-epileptic psychogenic events or spells. Navarro et al. 57 reported a final diagnosis of status epilepticus for 97.1% of participants in the levetiracetam plus clonazepam group and 94.1% of participants in the placebo plus clonazepam group and a diagnosis of a non-epileptic psychogenic event for 2.9% and 5.9% of participants in each treatment group, respectively. For each treatment group (midazolam group and lorazepam group), Silbergleit et al. 58 reported a diagnosis of status epilepticus for 90% of participants, a diagnosis of a non-epileptic spell for 7% of participants and an undetermined diagnosis for 3% of participants. Shaner et al. 59 reported 55.6% and 50% participants with focal features in the phenobarbital and placebo groups, respectively.

Two trials reported time from onset of status epilepticus to drug administration. Alldredge et al. 22 reported means of 34 [standard deviation (SD) 17.8] minutes, 31.3 (SD 14.5) minutes and 46.7 (SD 38.8) minutes for the lorazepam, diazepam and placebo groups, respectively. Navarro et al. 57 reported a median of 58 (range 15–135) minutes for the levetiracetam plus clonazepam group and a median of 60 (range 20–258) minutes for the placebo plus clonazepam group.

With respect to the total dose of active drug administered (see Appendix 6), Alldredge et al. 22 reported the number of injections received by the study participants [lorazepam arm, 34 participants received one injection (2 mg) and 32 received two injections (4 mg); diazepam arm, 34 participants received one injection (2 mg) and 32 received two injections (4 mg); placebo arm, 29 participants received one injection and 42 received two injections]. Shaner et al. 59 reported a range of 5–23 mg/kg phenobarbital administered in the phenobarbital plus optional phenytoin arm. In addition, seven participants in this group received phenytoin in doses ranging from 6 to 23 mg/kg. The remaining two trials did not report the total dose of study drug. 57,58

All four included trials reported causes of status epilepticus (see Appendix 6). A subtherapeutic level of AEDs was the most common cause in three trials,22,58,59 while lesion was reported as the most frequent cause of status epilepticus in the Navarro et al. trial. 57 Other commonly reported causes were alcohol abuse22,57 or withdrawal,59 infections22,57,59 and metabolic factors. 22,57,59

None of the included trials reported the patients’ previous history of status epilepticus. The proportion of participants with a history of seizures ranged from 54.6% to 66.2% in the treatment arms of the Alldredge et al. trial22 and from 61% to 78% in the treatment arms of the Shaner et al. trial. 59 The proportion of participants with a history of epilepsy ranged from 59% to 70%57 in the treatment arms of the Navarro et al. trial57 and from 65% to 66% in the Silbergleit et al. trial. 58 None of the trials reported presence of known difficult-to-treat epilepsy, or functional dependence status at baseline.

All characteristics reported in the publications by Silbergleit et al. 58 (i.e. RAMPART) were comparable to those extracted for the adult population from the data set received by the trial investigators. 58

Risk-of-bias assessment of included studies

Each risk-of-bias domain was assessed separately for objective outcomes and for subjective outcomes for each of the four included trials. Therefore, a total of eight risk-of-bias assessments are reported. Figure 2 presents the summary of the risk-of-bias assessments for all included trials. Risk-of-bias assessments of individual trials are presented in Figure 3.

FIGURE 2.

Summary of risk of bias of all included trials.

FIGURE 3.

Risk-of-bias assessments of individual trials apart from Shaner et al. ,59 which was judged to be at a high risk of bias (due to the high risk-of-bias assessment of measurement of subjective outcomes). All other trials were considered to have a low risk of bias. 22,57,58 CLZ, clonazepam; DIZ, diazepam; LEV, levetiracetam; LOR, lorazepam; MDZ, midazolam; PBO, placebo; PHB, phenobarbital; PHT, phenytoin. Reproduced with permission from Cruickshank et al. 60 This is an Open Access article distributed in accordance with the terms of the Creative Commons Attribution (CC BY 4.0) license, which permits others to distribute, remix, adapt and build upon this work, for commercial use, provided the original work is properly cited. See: http://creativecommons.org/licenses/by/4.0/. The figure includes minor additions and formatting changes to the original figure.

Reproduced with permission from Cruickshank et al. 60 This is an Open Access article distributed in accordance with the terms of the Creative Commons Attribution (CC BY 4.0) license, which permits others to distribute, remix, adapt and build upon this work, for commercial use, provided the original work is properly cited. See: http://creativecommons.org/licenses/by/4.0/. The figure includes minor additions and formatting changes to the original figure.

For the ‘randomisation process’ domain, three trials were judged to be at a low risk of bias,22,57,58 but in the Shaner et al. 59 trial there was insufficient information on which to make a definitive judgement.

Regarding the ‘deviations from intended interventions’ domain, three trials were properly blinded and judged to be at a low risk of bias;22,57,58 the trial by Shaner et al. 59 was described as ‘non-blinded’, indicating that those delivering the intervention would have been aware of the allocated intervention. Nevertheless, it was considered unlikely that participants would have been aware of the intervention received because of the nature of the clinical condition (i.e. status epilepticus). There was insufficient information on which to make a judgement regarding deviations from the intended intervention due to the experimental context. The combination of these factors led to a judgement of ‘some concerns’ for this trial. 59

All four trials were judged to be at a low risk of bias for ‘missing outcome data’ because of the low withdrawal/discontinuation rates, which were balanced across treatment groups. 22,57–59

The ‘measurement of the outcome’ domain was judged to be at a low risk of bias for three trials for both objective and subjective outcomes. 22,57,58 As the Shaner et al. 59 trial was not blinded, the measurement of the subjective outcomes was judged to be at a high risk of bias, but objective outcomes were less likely to be affected by this bias and, therefore, were judged to be at a low risk of bias.

For the Shaner et al. 59 trial, there was insufficient information on which to make a robust judgement about the ‘selection of the reported result’. There was no evidence of selective reporting in the remaining three trials, which were considered to be at a low risk of bias for this domain. 22,57,58

Overall, three trials were assessed to be at a low risk of bias for both the subjective and the objective outcomes. 22,57,58 The trial by Shaner et al. ,59 which was published in 1988 and was not blinded, was assessed as being at a high risk of bias for subjective outcomes and with some concerns for objective outcomes.

Results of the individual studies and synthesis: primary outcomes

A summary of the primary clinical outcomes reported in the four included trials is presented in Table 3. Full details of these outcomes, including outcome definitions, are presented in Appendices 7 and 8.

All four included trials reported the number of people with cessation of seizure activity, albeit the definitions of the outcome varied across studies. Two trials22,58 defined the outcome as termination of seizures before arrival at the ED, with one of these also specifying that no rescue therapy was needed. 58 These two trials both involved i.v. administration of lorazepam in either the intervention group22 or the control group,58 with the proportion of participants in the lorazepam groups with seizure cessation being similar (59.1% and 63.4%, respectively). The trial by Alldredge et al. 22 showed that, compared with placebo, both 2 mg of i.v. lorazepam and 5 mg of i.v. diazepam were effective in terminating the episodes of status epilepticus (the proportions of participants with seizure cessation were 59.1%, 42.0% and 21.1%, in the lorazepam, diazepam and placebo groups, respectively). 22 In RAMPART, by Silbergleit et al. ,58 assessing 10 mg of i.m. midazolam versus 4 mg of i.v. lorazepam, the proportion of adults achieving seizure cessation at the time of arrival in the hospital ED was similar between intervention groups (73.9% and 62.4% in the i.m. midazolam and i.v. lorazepam groups, respectively) and akin to those reported for the overall mixed population (73.4% and 63.4% in the i.m. midazolam and i.v. lorazepam groups, respectively). 58

The remaining two trials specified that seizures had to stop within either 15 minutes57 or 10 minutes59 from onset of treatment. In general, the proportion of participants with seizure cessation was similar across these trials and across treatment groups: 73.2% of participants in the 2.5 g of i.v. levetiracetam plus 1 mg of clonazepam group and 83.8% of participants in the i.v. placebo plus 1 mg of clonazepam group of the Navarro et al. trial,57 and 72.2% of participants in the 100 mg/minute i.v. phenobarbital plus 40 mg/minute phenytoin group in the Shaner et al. 59 trial. The exception was the 2 mg/minute i.v. diazepam plus 40 mg/minute phenytoin control group in the Shaner et al. 59 trial, which reported only 33.3% of participants with seizure cessation.

All four trials reported time to seizure cessation from administration of study drug using broadly similar definitions. The median time for i.v. administration ranged from 1.6 (i.v. lorazepam group)58 to 15 minutes (i.v. diazepam plus phenytoin group). 59 The Navarro et al. 57 trial reported administration times of up to 50 minutes. For time to seizure cessation, Alldredge et al. 22 reported a HR of 2.94 [95% confidence interval (CI) 1.41 to 5.88] for the lorazepam versus the placebo comparison and a HR of 1.54 (95% CI 0.85 to 2.77) for the lorazepam versus the diazepam comparison.

Two trials reported the number of participants with recurrence of seizures. Navarro et al. 57 reported that 10.4% of the i.v. levetiracetam plus clonazepam group and 19.1% of the placebo plus clonazepam group experienced seizure recurrence during their hospital stay, whereas Silbergleit et al. 58 reported that 11.4% of participants in the i.m. midazolam group and 10.6% of participants in the i.v. lorazepam group experienced recurrence of seizures within 12 hours of arrival at the ED.

None of the four included trials reported time from seizure cessation to recurrence. Silbergleit et al. 58 reported that the mean time of seizure recurrence since ED arrival for the adult population was 4.3 (SD 9.4) minutes in the i.m. midazolam group (n = 53) and 4.5 (SD 13.6) minutes in the i.v. lorazepam group (n = 50).

Table 4 presents a summary of the primary safety outcomes reported by the four trials. Full details of these outcomes, including outcome definitions, are presented in Appendix 9.

Respiratory depression was reported in three of the four trials, with varying definitions across trials. 22,57,58 Silbergleit et al. 58 reported ‘respiratory depression’ as a serious adverse event (SAE) in 6.4% of total enrolments in the i.m. midazolam group and 10% in the i.v. lorazepam group. Alldredge et al. 22 reported ‘change in respiratory status requiring ventilation assistance by bag–valve–mask or an attempt at ventilation’ in 10.6% of the i.v. lorazepam group, 8.8% of the i.v. diazepam group and 15.5% of the placebo group. Navarro et al. 57 reported several outcomes relating to respiratory depression; these included ‘prehospital health failures: respiratory’ in 10.3% of the i.v. levetiracetam plus clonazepam group and 4.5% of the placebo plus clonazepam groups and ‘need for prehospital assistance: respiratory’ in 42.6% of the i.v. levetiracetam plus clonazepam group and 34.8% of the placebo plus clonazepam group. In addition, a small number of participants was reported to experience ‘respiratory, thoracic and mediastinal’ AEs or SAEs [i.e. hypoxia (SAE), acidosis respiratory (AE), hypoxaemia (AE), respiratory distress (AE), respiratory failure (AE)], with a maximum of two participants (2.9%) in any category. 57

The protocol for this assessment specified 30-day mortality as a primary outcome and 6-month mortality as a secondary outcome. However, none of the included studies reported mortality at these specified time points. Two trials22,57 reported the proportion of participants who died between enrolment and discharge from hospital. 22,61 The differences in mortality rates among treatment groups in each individual trial were not significant. In the trial by Alldredge et al. ,22 mortality rates ranged from 4.4% to 7.6% in the i.v. diazepam and i.v. lorazepam group, respectively, whereas the proportion of deaths in the placebo group was 15.5%. 22 In the Navarro et al. 57 trial, the proportion of people who died was 4.5% in the levetiracetam plus clonazepam group and 6.2% in the placebo plus clonazepam group. 57 Silbergleit et al. 58 reported similar mortality rates among adults treated with i.m. midazolam and those treated with i.v. lorazepam (2.8% vs. 2%, respectively).

Results of the individual studies and synthesis: secondary outcomes

A summary of the secondary outcomes reported in the four included studies is presented in Table 5. Full details of these outcomes, including outcome definitions, are presented in Appendix 10.

The need for additional drugs to stop a seizure within 12 hours was specified in the protocol for this assessment. None of the studies reporting this outcome specified a 12-hour limit, so relevant outcomes are reported here as defined by the trials’ investigators. Outcome definitions varied across studies.

Navarro et al. 57 reported that similar numbers of participants in the levetiracetam plus clonazepam group and in the placebo plus clonazepam groups required a second clonazepam injection after 5 minutes (41.8% and 43.1%, respectively) and an injection of antiepileptic drugs after 15 minutes (28.4% and 23.1%, respectively). 57 In the trial by Shaner et al. ,59 anaesthesia was needed in one participant (5.6%) who received i.v. diazepam plus phenytoin but in none of those treated with i.v. phenobarbital plus phenytoin. RAMPART, by Silbergleit et al. ,58 reported the number of participants who required rescue therapy and the number who were still having seizures on arrival at the ED. 22 The proportion of participants who received rescue therapy and whose seizures were terminated was 10.5% in the i.m. midazolam group and 12.8% in the i.v. lorazepam group; the proportion of those who received rescue therapy and whose seizures were not terminated was 4.9% in the i.m. midazolam group and 9.4% in the i.v. lorazepam group; the proportion of participants whose seizures were not terminated but did not receive rescue therapy was 11.2% in the i.m. midazolam group and 14.4 in the i.v. lorazepam group. In general, those in the i.m. midazolam group were less likely to have seizures at arrival at the ED than those in the i.v. lorazepam group (proportion of participants without seizures: 83.9% vs. 76.2% in the i.m. and i.v. groups, respectively).

Two trials reported the numbers of participants who were admitted to hospital. Alldredge et al. 22 reported that a higher proportion of participants were admitted to ICU (56.9% in the i.v. lorazepam arm, 47.8% in the i.v. diazepam arm and 63.4% in the placebo arm) than to a hospital ward (29.2%, 26.9%, 23.9%, respectively). In the Silbergleit et al. 58 trial, more participants were admitted to a hospital ward (57.6% in the i.m. midazolam group and 65.6% in the i.v. lorazepam group) than to ICU (28.6% and 36.2%, respectively).

Navarro et al. 57 reported that, in each treatment group, the median length of stay in ICU was 3 days, whereas Silbergleit et al. 58 reported a mean of 5.7 days for the i.m. midazolam group and of 4.1 days for the i.v. lorazepam group.

None of the identified trials reported 6-month mortality, health-related quality of life or the number of people requiring an emergency call-out.

Quantity of the evidence

In total, 103 results were identified through database searching. Abstract screening excluded 96 records, with the main reason for exclusion being the lack of economic evaluation. The remaining seven full-text articles were assessed for eligibility, and six were excluded because of either inappropriate study design (no economic evaluations, n = 3) or patient population (n = 3). Of the three articles excluded because of the lack of economic evaluation, two reported only costs and one was a review of evidence on the use of fosphenytoin. One record matched the inclusion criteria and was included in this review. The PRISMA flow diagram for study selection is provided in Figure 4.

FIGURE 4.

The PRISMA flow diagram for the selection of cost-effectiveness evidence.

Results of the individual studies

The included study is a retrospective case note audit comparing the efficacy, safety and cost of lorazepam (4 mg i.v., repeated up to two times) compared with diazepam (10 mg i.v., repeated up to three times) in adults with convulsive status epilepticus who received treatment in a teaching hosptial in London. 62 Cases of convulsive status epilepticus were retrospectively reviewed over two 18-month periods before and after the introduction of a new management protocol that recommended lorazepam in place of diazepam for first-line treatment of convulsive status epilepticus. Phenobarbitone was also recommended instead of phenytoin for second-line therapy. A total of 720 episodes were identified and 590 medical records were retrieved (82%). These records were examined and 90 episodes of convulsive status epilepticus were identified. Of these, 13 were excluded as no benzodiazepine was received, and a further five were excluded as neither lorazepam nor diazepam was received as part of first-line treatment. The remaining 72 episodes were used to provide a comparison of lorazepam and diazepam. The results of the analysis showed that, when given as the first benzodiazepine, lorazepam was associated with a higher likelihood of treatment success than diazepam (9/17 doses lorazepam vs. 14/55 doses diazepam; p = 0.042). The cost of lorazepam was higher (£0.78 for 4 mg compared with £0.45 for 10 mg) (cost year not specified) but, when used as the first benzodiazepine, lorazepam was significantly more likely to achieve seizure control and patients treated with lorazepam had a lower likelihood of recurrence. The cost per successful outcome was not different between the two treatments (£1.47 for lorazepam vs. £1.46 for diazepam). It was concluded that, where venous access is possible, lorazepam should be the first-line treatment for premonitory and established status epilepticus in place of diazepam in both pre-hospital and in-hospital settings.

A strength of the study is that it is UK based. However, there are some important limitations, notably the retrospective case note design and challenges with note retrieval potentially introducing bias. There were also some differences between the two groups including more patients with premonitory convulsive status epilepticus in the lorazepam group and a shorter time to first treatment, although neither was significant. The number of episodes included in the analysis is also relatively small to draw robust conclusions. The economic analysis presented in the study was a simple cost-effectiveness analysis using the outcome of cost per successful outcome, which was defined as seizure cessation without recurrence over 12 hours. No other outcomes were included in the economic analysis and costs were also limited to first-line medicine acquisition costs with no details provided on the impact on resource use or other medicine costs. Given the narrow focus and limitations identified, it is not possible to draw any robust conclusions on cost-effectiveness.

Other relevant studies

The literature search showed that there is limited evidence available on the cost-effectiveness of first-line pre-hospital or ED treatments for adults with status epilepticus. Given this lack of evidence, other studies (e.g. cost-of-illness studies) were considered with the aim of informing the design of future economic evaluations in this area. With this aim, two costing studies initially excluded during the full-text review were identified in the relevant patient population and these studies were reviewed in detail, as summarised below.

The first study is a retrospective cost-of-illness study that was conducted using data on patients with status epilepticus at three hospitals in Germany. 63 The aim of the study was to identify and characterise direct hospital costs (2014 EUR) and cost-driving factors for inpatients with status epilepticus from a German health-care perspective. In total, data from 341 admissions in 316 adult patients between 2013 and 2014 were analysed for costs and cost-driving factors. For each inpatient admission due to status epilepticus, average costs, length of stay and ventilation time were calculated. Patients were categorised according to the aetiology and onset of status epilepticus as follows: acute symptomatic status epilepticus due to acute brain injury (26.3%), new-onset remote symptomatic status epilepticus with no previous history of epilepsy or status epilepticus (28.8%), remote symptomatic status epilepticus with previous history of epilepsy or status epilepticus (38.9%), and other aetiologies (e.g. idiopathic generalised epilepsy) (6.0%). Patients were defined as refractory if they had recurrent seizures despite two appropriately selected and dosed AEDs including benzodiazepine. Super-refractory status epilepticus was defined as status epilepticus that continues or recurs ≥ 24 hours after initiation of treatments with anaesthetic AEDs including cases where seizure control is attained after induction of anaesthesia but recurs on weaning the patient off the anaesthetic agent.

The results of the analysis showed that benzodiazepines were used as first-line treatment in 137 admissions (40.2%) pre-hospital and in 308 admissions (90.3%) prior to AED treatment. Nearly all admissions required the use of electroencephalography (EEG) and cerebral imaging [computerised tomography (CT) and/or magnetic resonance imaging (MRI)]. The mean costs of hospital treatment of status epilepticus were €14,946 per patient per admission with a mean length of stay of 19 days (mean cost per day of €787). Significant cost differences were identified according to the aetiology of status epilepticus, with significantly higher mean costs in patients with acute symptomatic aetiology (€25,269) than in those with new-onset status epilepticus (€12,511), remote symptomatic status epilepticus (€11,204) or status epilepticus of other aetiologies (€10,380). Severity of status epilepticus also had a significant impact on costs, with the largest cost per admission associated with super-refractory status epilepticus (€50,488). Data on length of stay showed that severity was again a significant factor, with durations ranging from 12.1 days in non-refractory patients to 21.0 days in refractory status epilepticus patients and 37.0 days in super-refractory patients.

To identify cost-driving factors, univariate analyses of inpatient treatment costs and length of stay were conducted. This showed increased costs associated with newly diagnosed patients, acute aetiology, unfavourable status epilepticus severity score (STESS) of 4–6 and length of stay of > 14 days. Significantly higher costs and significantly longer hospitalisations were linked to no acute treatment with benzodiazepine before admission, high number of AEDs in hospital, refractory status epilepticus and super-refractory status epilepticus, anaesthesia, requirement for ventilation and unfavourable outcome on discharge (modified Rankin Scale 4–6). Multivariate analysis showed that super-refractory status epilepticus, ventilation and length of stay > 14 days were independent predictors of costs. Limitations of the study include potential lack of generalisability of results to other health-care systems, difficulty in distinguishing costs of status epilepticus from costs due to other acute illnesses (e.g. stroke) and the exclusion of potentially relevant costs, such as rehabilitiation and outpatient costs.

The second study is a retrospective analysis conducted in two hospitals in Spain in which data on patients with status epilepticus were analysed from December 2012 to July 2017. 64 The aim of the study was to identify factors that had an impact on the high cost of care in patients with status epilepticus, with focus on the timing of treatment and duration of status epilepticus. Data on seizure history were recorded, as well as whether the status epilepticus episode occurred out of hospital (64%) or in hospital (36%). Aetiology of status epilepticus was categorised into acute symptomatic (55.1%), remote symptomatic (36.4%), progressive symptomatic (7.9%) and cryptogenic (9.5%). The severity of status epilepticus was also evaluated using the modified STESS. Data were collected on the time from onset of status epilepticus to arrival at hospital, time to administration of first-line treatment, time to second-line treatment received (AEDs) and the number of AEDs used, and time to third-line treatment (anaesthetics). Length of stay data were also collected. Duration of status epilepticus was defined as the recorded time from symptom onset until EEG showed a seizure suppression pattern following treatments.

In total, 305 patients were included in the analysis that showed several factors influencing the cost of treatment. These include no previous history of epilepsy, lower level of consciousnessness, the presence of a potentially fatal aetiology, major complications, the presence of lateralised periodic discharges in EEG, a higher epidemiology-based mortality score, being refractory or super-refractory, and the duration of status epilepticus (all p-values ≤ 0.005). The study found that the cost of management is higher for patients in whom onset of status epilepticus occurs in the hospital, rather than out of hospital, meaning that the results for these groups were analysed separately. The total mean cost of in-hospital status epilepticus was €15,174, compared with €6559 for out-of-hospital status epilepticus (p < 0.001) (cost year not specified). Patients in the in-hospital group tended to have more severe status epilepticus symptoms, greater comorbidity, a higher proportion of super-refractory status epilepticus and a higher risk of complications, which together result in an increased length of stay and higher associated costs. The results also showed that the cost of out-of-hospital onset of status epilepticus is higher when the duration of the episode is > 24 hours. In the out-of-hospital group, the duration of the status epilepticus episode was a key factor in the length of hospitalisation and the cost of disease management, with the analysis showing a clear relationship between duration of episode and the time to treatment. For episodes lasting < 24 hours the median cost was €5005, compared with €8733 for episodes lasting ≥ 24 hours (p = 0.005). The authors concluded that the speed at which treatment is initiated is an important factor; therefore, early detection and prompt establishment of treatment could have an impact on both patients’ outcomes and health-care costs. However, there are some limitations with the study, including the retrospective design, meaning that it is difficult to distinguish between costs due to status epilepticus and those associated with other conditions.

Statement of principal findings

Strength and limitations of the assessment

This assessment was conducted in accordance with current methodological standards, and its methods were prespecified in a peer-reviewed research protocol (PROSPERO registration CRD42020201953) (www.crd.york.ac.uk/prospero/display_record.php?RecordID=201953; accessed 4 August 2020). In particular, we conducted comprehensive literature searches of major electronic databases and relevant websites and used transparent methods for data selection and extraction.

There are, however, some limitations to take into consideration when interpreting the findings of this assessment:

• We identified only a limited number of trials for the pre-hospital treatment of convulsive status epilepticus in the adult population.

• The included studies differed in terms of treatment comparisons and choice and definition of outcome measures. This hampered the possibility of conducting any meaningful meta-analysis.

• None of the included trials was conducted in the UK, with three of the four included trials conducted in the USA.

• Only one study the met the inclusion criteria for economic evaluation, and the study in question was limited by its reliance on retrospective observational data from a small number of cases, and a narrow focus on treatment acquisition costs in the relation to outcomes.

Uncertainties from the assessment

Implications for health care

Both 2 mg of i.v. lorazepam and 5 mg of i.v. diazepam administered by paramedics are more effective than placebo for the pre-hospital treatment of convulsive status epilepticus in adults; 10 mg of i.m. midazolam is non-inferior to 4 mg of i.v. lorazepam in the pre-hospital setting. The addition of levetiracetam to clonazepam does not offer clear advantages over clonazepam alone.

Recommendations for research

Harmonisation of outcome measures would be useful to facilitate future clinical research.

Identification of the optimal pre-hospital pharmacological treatment of adults with status epilepticus requires further RCTs assessing different benzodiazepines (and potentially other anticonvulsant medications) as well as their route of administration and acceptability.

This assessment indicates that large well-designed clinical trials are needed to assess the use of i.v. lorazepam versus i.v. diazepam and to confirm the efficacy and safety of i.m. midazolam compared with i.v. lorazepam. In the pre-hospital setting, i.m. midazolam may have the advantage of a rapid and easy administration, which counterbalances the faster effect of i.v. medications.

Given the current clinical recommendations about the use of benzodiazepines, which can be administered rectally or via the mouth by caregivers in the pre-hospital setting, and the absence of evidence in the adult population, direct head-to-head trials comparing buccal midazolam and rectal diazepam are required to establish their efficacy, safety and acceptability.

It is also necessary to establish which midazolam formulation is the best option for the first-line treatment of adults with convulsive status epilepticus. Future clinical trials comparing i.m. midazolam versus buccal midazolam and versus intranasal midazolam would provide useful information to inform the pre-hospital management of patients, especially when i.v. access is not feasible. Further trials aiming to conclusively establish whether or not i.m. midazolam is more effective than i.v. lorazepam, as suggested by the single trial in the review, would be helpful.

Future clinical studies could also aim to establish optimal doses of benzodiazepines used as first-line treatments in the pre-hospital setting.

It would be important to improve adherence to clinical guidelines on the use of currently available benzodiazepines in the pre-hospital setting. For example, a recent large trial of second-line hospital-based treatment of status epilepticus in the USA found that only 57% of patients had received benzodiazepines before arrival at the hospital and 7% breached the protocol by not receiving a predefined minimally adequate benzodiazepine dose. 44

As there is still a proportion of patients who do not respond to treatment, further research aiming to understand the underlying pathophysiology of treatment response would be useful to inform future treatment development.

High-quality economic evaluations are required to determine the value for money of different treatments for convulsive status epilepticus and their modes of administration. It would be useful if future economic evaluations could capture the full cost of managing the convulsive episode up to the time of discharge from hospital. In cases where higher-cost first-line treatments are not fully offset by subsequent management cost savings, it may be necessary to model the impact of improved short-term outcomes on longer-term health-related quality of life and future resources use.

Search strategy

1 Emergency Medical Services/use ppezv

2 emergency health service/or emergency care/use oemez

3 Emergency Services/use psyf

4 (accident adj2 emergency).tw.

5 (“emergency room” or “emergency department” or ED).tw.

6 (pre-hospital or prehospital or “out of hospital” or community).tw.

7 Allied Health Personnel/use ppezv,psyf

8 paramedical personnel/use oemez

9 (paramedic* or ambulance).tw.

10 or/1-9

11 Status Epilepticus/use ppezv,psyf

12 epileptic state/use oemez

13 Status Epilepticus.tw.

14 11 or 12 or 13

15 exp Benzodiazepines/use ppezv,psyf

16 exp benzodiazepine derivative/use oemez

17 (midazolam or diazepam or lorazepam).tw.

18 exp Anticonvulsants/use ppezv

19 exp anticonvulsive agent/use oemez

20 exp Anticonvulsive Drugs/use psyf

21 (levetiracetam or “sodium valproate” or phenytoin).tw.

22 or/15-21

23 randomized controlled trial.pt. use ppezv

24 controlled clinical trial.pt. use ppezv

25 “randomized controlled trial”/use oemez

26 “controlled clinical trial”/use oemez

27 ((randomi#ed or controlled or clinical) adj2 (trial or study)).tw.

28 or/23-27

29 10 and 14 and 22 and 28

30 remove duplicates from 29

Search strategy

S1 (MH “Emergency Medical Services”)

S2 TX accident N2 emergency

S3 TX “emergency room” or “emergency department” or ED

S4 TX pre-hospital or prehospital or “out of hospital” or community

S5 (MH “Allied Health Personnel”)

S6 TX paramedic* or ambulance

S7 (MH “Emergency Medical Technicians”)

S8 S1 OR S2 OR S3 OR S4 OR S5 OR S6 OR S7

S9 (MH “Status Epilepticus+”)

S10 TX Status Epilepticus

S11 S9 OR S10

S12 (MH “Antianxiety Agents, Benzodiazepine+”)

S13 TX midazolam or diazepam or lorazepam

S14 (MH “Anticonvulsants+”)

S15 TX levetiracetam or “sodium valproate” or phenytoin

S16 S12 OR S13 OR S14 OR S15

S17 TX (randomi#ed or controlled or clinical) N2 (trial or study)

S15 S8 AND S11 AND S16 AND S17

Search strategy

#1 MeSH descriptor: [Emergency Medical Services] explode all trees

#2 (accident Near/2 emergency):ti,ab,kw (Word variations have been searched)

#3 “emergency room” or “emergency department” or ED

#4 pre-hospital or prehospital or “out of hospital” or community

#5 MeSH descriptor: [Allied Health Personnel] explode all trees

#6 paramedic* or ambulance

#7 #1 or #2 or #3 or #4 or #5 or #6

#8 MeSH descriptor: [Status Epilepticus] explode all trees

#9 Status Epilepticus

#10 #8 or #9

#11 MeSH descriptor: [Benzodiazepines] explode all trees

#12 midazolam or diazepam or lorazepam

#13 #11 or #12

#14 #7 and #10 and #13

#15 MeSH descriptor: [Anticonvulsants] explode all trees

#16 (levetiracetam or “sodium valproate” or phenytoin):ti,ab,kw (Word variations have been searched)

#17 #11 or #12 or #15 or #16

#18 #7 and #10 and #17