In the UK, an estimated 211 000 suspected calls are made to ambulance services about seizures each year. Of these, 60 000 result in transportation the emergency department (ED), of which two-thirds (40 000) result in hospital admission (Dickson et al, 2017).
Convulsive status epilepticus (CSE) is a life-threatening condition affecting between 17 and 23 per 100 000 children each year (Chin et al, 2008). It is identified when the duration of a seizure is in excess of 30 minutes or when two or more seizures occur without a return to consciousness (Scott et al, 1998).
Convulsions in children lasting longer than 5 minutes constitute a life-threatening condition; it is therefore vital to patient outcomes that appropriate interventions are administered without undue delay (McCartney and Finnikin, 2019).
Prompt anticonvulsant treatment leads to the rapid termination of convulsions, thus helping to decrease morbidity and mortality in children (Smith et al, 2016).
Buccal midazolam (BM) and rectal diazepam (RD) are the primary drugs used to treat any acute tonic-clonic convulsion, including CSE, in children who require emergency medical care (Desai, 2016; Bacon, et al, 2021; Gonzalez-Viana, et al, 2022; National Institute for Health and Care Excellence (NICE), 2022a; 2022b; Resuscitation Council UK (RCUK), 2022).
The Royal College of Paediatrics and Child Health (RCPCH, 2022) states that, although BM is the preferred method within hospital settings, there are challenges faced by paramedics when administering RD in a prehospital setting. This is as a result of the requirement to firstly decant the required dosage from the available 5 mg and 10 mg rectal tubes in order to provide accurate dosage. Current Joint Royal Colleges Ambulance Liaison Committee guidelines (Association of Ambulance Chief Executives and Joint Royal Colleges Ambulance Liaison Committee, 2019) and the BNF for Children (Paediatric Formulary Committee, 2023a; 2023b) indicate a generally lower dose level for buccal administration across age groups. For example, for children aged 10–17 years, the initial dose is 10 mg of BM compared to 10–20 mg for RD (BNF for Children, 2022).
BM has a rapid onset and a short duration of action (Kutlu et al, 2003). Similarly, RD ensures fast absorption and is used when an intravenous injection is impracticable or undesirable and BM is not desirable as the first option. However, because of the route of administration, care is required to maintain patient dignity (Black Country Partnership NHS Trust, 2019).
A common side effect of both drugs is respiratory depression, which can have life-threatening complications, so it is important to address the risks both drugs pose with respect to this.
It is also important to examine medication efficacy and cost.
As well as monitoring cost and respiratory depression, assessing the time until seizure cessation and recurrence can help to identify whether BM is more effective than RD for children in the emergency treatment of seizures. This information will not only assist when deciding which medication can be administered safely and promptly but also be beneficial in helping to facilitate quick decision-making, which seizures warrant for safe termination, prevention and recurrence.
Aims and objectives
Research aim
Although there are evidence-based guidelines for the use of benzodiazepines in treating seizures in children, these are often compiled from research that includes adults (NICE), 2022a; 2022b). Furthermore, Advanced Paediatric Life Support (APLS) guidance (RCUK, 2021) states that there is limited high-quality evidence comparing differing benzodiazepines.
The aim of this review discussion is to provide an evidence-based context through which practitioners can evaluate the safety and efficacy of BM over RD when treating convulsions of any duration, in the prehospital setting and specifically in children. To determine this, the following will be considered:
In addition to this, the cost-effectiveness of both drugs is considered. This is deemed necessary as a means to assess clinical outcome versus cost. Each of these issues will be addressed using the related learning outcomes.
Methods
The review carried out followed the Preferred Reporting Items for Systematic Reviews and Meta-Analyses guidelines for reporting of systematic reviews in order to ensure transparency and thereby increase validity (PRISMA, 2021; Page et al, 2021). Both the Critical Appraisal Skills Programme (CASP, 2023) and Raby and McNaughton's (2021) framework were used to evaluate the evidence used in this review.
Eligibility criteria
Participants
As the purpose of this evidence review is to examine seizures in children, articles were selected where the primary focus was on participants aged from newborn up to 18 years.
Interventions and comparison
Included studies compared the efficacy and safety of BM with RD in the treatment of acute and prolonged seizures. No other comparison treatment groups were included.
Outcomes
The main outcome measures for the included studies were: cessation of visible seizure activity; seizure cessation without recurrence in the subsequent hour; and if respiratory depression resulted from drug administration.
Type of studies
As the aim of this review is to compare the efficacy of two interventions (BM versus RD), quantitative, experimental research was sought in order to directly answer the review question. Though randonised control trials are the gold standard, quasi-experimental research was also considered providing the inclusion criteria were met.
There were no limitations on time of publication or geographical location. This was to maximise the number of papers available for review. However, changes to guidelines/practice may raise questions regarding the applicability of such papers to current practice.
The inclusion and exclusion criteria are given in Table 1.
Components | Inclusion criteria | Exclusion criteria |
---|---|---|
Population | Pediatric, paediatric, child*, infant, adolescent, teenage* | Studies on animals and patients aged >18 years |
Intervention | Buccal midazolam, oromucosal midazolam | - |
Comparator | Rectal diazepam | - |
Outcome | Seizures, epilepsy, convulsion | - |
Study design | Quantitative primary studies (experimental and observational) | Qualitative studies |
Time of study | No limit on the time of study | - |
Language | English language only | - |
Location | No limit on the country where the study was conducted | - |
Search strategies
A detailed and comprehensive search was conducted on Embase, CINAHL, MEDLINE, the Cochrane Library and Google Scholar. No current reviews comparing BM to RD were found.
In accordance with Davies (2019), a focused question was developed before the search was started. This was developed using the PICO framework, which considers population, intervention, comparison and outcome (Aveyard and Sharp, 2013). Once these had been identified, the following question was formulated: What are the benefits of using buccal midazolam (I) over rectal diazepam (C) during (P) seizures (O) in children (P)?
Keywords and synonyms were then and combined using the Boolean operators OR and AND to maximise the specificity of the search. Searching was performed in November 2021 and repeated 2 months later to ensure no new results would be missed.
Study selection
A total of 460 articles were retrieved from the databases searched: CINAHL (20), MEDLINE (54), Embase (62), Google Scholar (322) and the Cochrane Library (2). The total number of retrieved articles was then reduced to 86 following the removal of 367 duplicates and seven articles that were not published in English. Using the inclusion criteria and in line with Davies (2019), the 86 remaining articles were initially filtered (as included, excluded or undecided) according to title and abstract.
Following this, a total of 14 articles were eligible for full text screening using the inclusion and exclusion criteria. Nine full-text articles were then excluded. The final review included the five remaining studies.
Throughout both stages of screening, the key reasons for exclusion were: not focused on children; exploring different types of pharmacokinetics; and exploring different routes of administration.
Figure 1 shows the screening and selection process detailing the search strategy used. The first author, KM, performed the initial search. The remaining articles were then reviewed by both authors (KM and PR). While it was agreed that any discrepancies would be resolved with reference to a third party, in the event, this was not necessary.

Data extraction and quality assessment
Data relevant to the PICO components of the review question were extracted using a bespoke data extraction form. The characteristics extracted from each study included author(s), date of publication, country, study setting, sampling approach and data collection methods.
The included studies were then appraised for methodological quality and risk of bias using the CASP (2023) quality assessment framework. According to the CASP framework, all papers were considered of high quality, with each scoring above 90%. However, when Raby and McNaughton's (2021) framework was used, some fundamental methodological issues were identified. These related to small sample sizes in two of the articles: Baysun et al (2005) (n=43); and Scott et al (1999) (n=28). Similarly, in the study by McIntyre et al (2005), consent was obtained for 219 separate episodes involving 177 patients, so certain patients were randomised and treated on multiple occasions, which affected the variability of the sample population.
In terms of the interventions administered, there were differences with regards to drug doses: Ashrafi et al (2010) used 0.3–0.5 mg/kg of BM; Baysun et al (2005) used 0.25 mg/kg of BM and 0.3–0.5 mg/kg of RD; while McIntyre et al (2005), Mpimbaza et al (2008) and Scott et al (1999) used 0.5 mg/kg for both BM and RD. This variation in dose makes direct comparisons problematic. The varied doses in both Ashrafi et al (2010) and Baysun et al (2005) may also present as a confounding variable when interpreting results.
A summary of the various methodological issues is shown in Table 2.
Fundamental flaws evident | Some minor issues/queries | Robust and applicable | |
---|---|---|---|
Contextual factors |
Baysun et al (2005)
|
Scott et al (1999) | McIntyre et al (2005) |
Non-UK based studies | Dated research | UK based | |
Sampling |
Scott et al (1999)
|
Mpimbaza et al (2008)
|
N/A |
Limited sample sizes | Wide range of ages may introduce differing drug effects | ||
Data collection methods |
Baysun et al (2005)
|
N/A | N/A |
Variable doses may have had a confounding effect | |||
Data analysis techniques | Reporting across all papers was clear and sufficient | ||
Applicable to practice |
Baysun et al (2005)
|
Scott et al (1999) | McIntyre et al (2005) |
Study characteristics
The five included studies were published between 1999 and 2010. Studies were conducted in Iran, Turkey and Uganda, with two in the UK. Caution needs to be taken with regards to generalising non-UK studies to UK practice because of cultural disparities and variance in practice.
In total, the five studies included 676 patients, covering 386 treatment episodes for BM and 383 for RD. All studies monitored respiratory rate, heart rate, blood pressure and oxygen saturations before, during and after drug administration. Any complications in patient management were recorded and monitored until discharge.
There were no significant differences between treatment groups in any of the papers with regards to sex or type of seizure. However, the variation in ages, while allowing for a broader total sample, does present as a potentially confounding variable to any potential conclusions.
Sample sizes ranged from 28 (Scott et al, 1999) to 330 patients (Mpimbaza et al, 2008) with ages ranging from 2 months to 19 years (Scott et al, 1999).
In all studies, patients were enrolled and randomised to receive either BM or RD. The main outcome measures were: seizure cessation; seizure recurrence; and respiratory depression.
Table 3 shows the characteristics of studies included in the review.
Author(s), year, country | Publication type | Study design | Duration of study | Number of patients | Age of patients | Inclusion criteria | Exclusion criteria |
---|---|---|---|---|---|---|---|
Scott et al (1999)
|
Journal | Quasi-randomised control trial | Not stated | 28 | 5–19 years |
|
|
Baysun et al (2005)
|
Journal | Prospective, quasi-randomised control trial | Not stated | 43 | 2 months –12 years |
|
|
Mclntyre et al (2005)
|
Journal | Multi-centre randomised control trial | Oct 2000–Feb 2004 | 177 | 7 months –12 years |
|
|
Mpimbaza et al (2008)
|
Journal | Single-blinded, randomised control trial | Nov 2005–June 2006 | 330 | 3 months–12 years |
|
|
Ashrafi et al (2010)
|
Journal | Randomised control trial | April 2007–April 2008 | 98 | 3 months–12 years |
|
|
Study results
Seizure cessation data (Table 4) within 5 minutes was provided only by Ashrafi et al (2010) and Baysun et al (2005), and showed that BM was more effective than RD; this amounted to 88.88% of BM patients being successfully treated within 5 minutes, compared with only 57.97% of those administered RD. Seizure cessation data within 10 minutes were provided by all five authors, again showing that BM was more successful than RD (total=64% compared with 59%). However, of the three papers reporting only cessation within 10 minutes, McIntyre et al (2005) found a statistically significant difference (BM: 65%; RD: 41%; P=0.001). Because of the success of both drugs within 5 minutes for both Ashrafi et al (2010) and Baysun et al (2005), the number of patients left to treat beyond this was limited. Nonetheless, in Baysun et al's (2005) study, all 98 patients were successfully treated within 10 minutes (both groups) and, in Ashrafi et al's (2010) study, 86.95% of BM patients (18/23) were treated successfully within 10 minutes versus 85% of RD patients (17/20).
Author(s), year, country | Number of patients | Number of buccal midazolam Treatment episodes | Number of rectal diazepam Treatment episodes | Treatment Response | P | |||
---|---|---|---|---|---|---|---|---|
Number of patients who were controlled within 5 minutes of drug administration | Number of patients who were controlled within 10 minutes of drug administartion | |||||||
Buccal midazolam n (%)a | Rectal diazepam n (%)a | Buccal midazolam n (%)a | Rectal diazepam n (%)a | |||||
Scott et al (1999) UK | 28 | 40 | 39 | NR | NR | 30 (75) | 23 (59) | 0.16 |
Baysun et al (2005) Turkey | 43 | 23 | 20 | 15 (65) | 0 | 3 (13) | 17 (85) | NR |
Mclntye et al (2005) UK | 177 | 109 | 110 | NR | NR | 71 (65) | 45 (41) | 0.001 |
Mpimbaza et al (2008) Uganda | 330 | 165 | 165 | NR | NR | 125 (76) | 114 (69) | 0.175 |
Ashrafi et al (2010) Iran | 98 | 49 | 49 | 49 (100) | 40 (82) | 0 | 9 (18) | 0.001 |
Total | 676 | 386 | 3831 | 72 (88.88)b | 40 (10)c | 229 (59)d | 208 (54)e | * |
NR: not recorded.
Three studies commented on the need to use additional benzodiazepines to stop the presenting convulsion if it had not been controlled within 10 minutes. McIntyre et al (2005) reported 36/109 (33%) of BM patients and 63/110 (57%) of RD patients required medication adjuncts. Mpimbaza et al (2008) also reported a higher requirement for augmentation within the RD group (RD: 51/165; 65%; BM: 47/165; 39%). Baysun et al (2005) reported lower figures of 5/23 (23%) for BM patients and 3/20 (15%) for RD patients.
A further factor to consider is that in the study by Mpimbaza et al (2008), 67% of the patients had severe malaria. When considering only those with malaria, the risk for treatment failure did not differ significantly between the BM (32%) and RD (36%) groups. There were eight deaths in the BM (n=165) and 12 in the RD (n=165) group (P=0.365). However, causes of death were associated with underlying illness (malaria, pneumonia and meningitis), which suggests that BM was valuable to children with convulsions that were not associated with infectious diseases (56% versus 27%).
All five studies reported on the number of patients who experienced respiratory depression (Table 5). Across the five trials, 26/386 (7%) patients in the BM group and 26/383 (7%) patients in the RD group experienced respiratory depression, although Ashrafi et al (2010) found no incidence of respiratory depression in either group.
Author(s), year, country | Total number of patients | Number of of buccal midazolam treatment episodes | Number of rectal diazepam treatment episodes | Incidence of respiratory depression: number of patients experiencing respiratory depression during drug administration | Percentage difference (95% CI) | |
---|---|---|---|---|---|---|
Buccal midazolam n (%)a | Rectal diazepam n (%)a | |||||
Scott et al (1999) UK | 28 | 40 | 39 | 1 (3) | 0 | |
Baysun et al (2005) Turkey | 43 | 23 | 20 | 18 (78) | 17 (85) | NR |
Mclntyre et al (2005) UK | 177 | 109 | 110 | 5 (5) | 7 (6) | 2 (−4 to 8) |
Mpimbaza et al (2008) Uganda | 330 | 165 | 165 | 2 (1) | 2 (1) | NR |
Ashrafi et al (2010) Iran | 98 | 49 | 49 | 0 | 0 | NE |
Total | 676 | 386 | 383 | 26 (7)b | 26 (7)c | * |
NE: not estimable; NR: not recorded;
McIntyre et al (2005) reported that five patients required intubation, two after BM and three after RD; they were subsequently admitted to intensive care for continuous management of their seizures.
Discussion
The aim of this study is to provide an evidence-based context through which practitioners can evaluate the safety and efficacy of BM over RD when treating convulsions in children.
Five randomised control trials explored the safety and efficacy of BM versus RD for the emergency treatment of seizures in children. All studies were assessed for bias, validity and overall quality of research.
A total of 676 patients were included in the clinical trials that compared BM with RD. In the BM group, the seizures of 64/386 (17%) patients were terminated within 5 minutes and 229/386 (59%) had seizures terminated within 10 minutes of drug administration. However, in the RD group, 40/383 (10%) of patients had seizures terminated within 5 minutes and 208/383 (54%) within 10 minutes of drug administration. These findings are in line with the evidence review conducted by NICE (2022b), which stressed the benefits of BM over RD in relation to seizure cessation. In contrast to the review, this study indicated that there was no difference between rate of respiratory depression between the two groups (both 7%).
Though the results suggest that BM is at least as effective as RD, there were inconsistencies between individual study results, which may reflect issues such as local variations, comorbidities and dose levels. For example, the median time between drug administration and seizure cessation were recorded as low as 3 minutes for BM (Baysun et al, 2005) and as high as 15 minutes for RD (McIntyre et al, 2005).
A similar confounding factor is the age range of the children in the five studies, which varied from <2 months to 19 years. Four studies assessed children with an age range from 2 months to approximately 5 years (Baysun et al, 2005; McIntyre et al, 2005; Mpimbaza et al, 2008; Ashrafi et al, 2010). In the study by Scott et al (1999), 14 (78%) of the 18 patients were aged ≤18 years and four were aged 19. Although 28 patients originally took part in the trial, only those aged ≤18 years and those who responded to medication within 10 minutes were included in the analysis (n=18); this could be seen as an impediment to the reliability and validity of the study's findings (Heale and Twycross, 2015).
In the UK, RD is the preferred choice of drug for the emergency treatment of children experiencing acute seizures and CSE (RCPCH, 2022). However, this review demonstrates that BM is a more expedient means to seizure cessation and demonstrates equity with regards to rates of respiratory depression.
A further issue with RD is that it is the more invasive method of administration, which can lead to embarrassment and have a detrimental effect in self-esteem, particularly for older children and teenagers (Body and Ijaz, 2005). This supports APLS guidance (RCUK, 2021), which highlights BM as being both less invasive than RD and more socially acceptable.
The substantial difference in cost between BM and RD suggests that RD is more cost-effective and it is therefore classified by NICE (2022b) as the chosen drug for seizures in children occurring outside the hospital setting. According to NICE (2022b) guidelines, 5 mg/1 ml (size 4) of BM cost £85.50 compared with 10 mg/2.58 ml (size 5) of RD, which costs £7.35 (both NHS indicative prices). However, the clinical benefits of BM observed within this review contrast with the cost-effectiveness of RD.
Limitations
All five studies in this review have shown that BM may be associated with a higher rate of seizure cessation. However, they vary regarding settings, participant characteristics and dose level. These issues make it difficult to draw definitive conclusions or establish broad generalisations, a fact further compounded by the age of the articles.
This demonstrates that more high-quality data are required to draw more substantive conclusions about the potential benefits of BM over RD.
There is also a need for evidence with regards to the potential for underlying health conditions to impact the efficacy of both BM and RD.
Conclusion
The evidence provided in this review suggests that BM is at least as effective as RD in treating acute and prolonged convulsive seizures in children. All five articles reviewed showed that seizure cessation was achieved more rapidly with BM administration. This was true for both studies that recorded seizure cessation in under 5 minutes and those presenting data solely for seizure cessation within 10 minutes.
Regarding respiratory depression, there were no differences observed between BM and RD. Although RD is a more cost-effective option, the need for expediency in limiting seizure duration accentuates the potential benefits of BM as the primary intervention for dealing with convulsions in children and young people in an emergency setting.
LEARNING OUTCOMES
After completing this module, the paramedic will be able to: