Intranasal midazolam for febrile seizures

Despite their benign nature and prognosis, uncomplicated febrile seizures are extremely stressful for both families and medical staff. Most parents believe that these seizures are harmful, and during the first episode half of all parents fear that their child is dying.¹ For decades children were given long term anticonvulsant drugs to prevent the recurrence of febrile seizures. Although some of these are effective, their serious adverse effects have led to an unacceptable ratio of benefits to risks.²,³

There is, however, a broad consensus that febrile seizures should be treated promptly. Traditionally, benzodiazepines, barbiturates, or other anticonvulsants have been given intravenously, but rectally administered diazepam is now used as an efficacious alternative.

The intranasal administration of therapeutic agents is undoubtedly the centre of tremendous interest in the field of therapeutics. That there is extensive and prompt absorption of molecules through the nasal mucosa into a rich vascular bed has long been recognised by cocaine users.

In this issue of the BMJ, Lahat and colleagues (p 83) compare the standard intravenous administration of diazepam with intranasal administration of midazolam for febrile seizures.⁴ Previous experience supports the biological plausibility of this approach: intranasally administered midazolam reaches peak concentrations at a mean of five minutes in piglets and 12 minutes in children.⁵,⁶ Equally importantly, the systemic bioavailability (how much of the drug will appear in the blood when compared with the same dose given intravenously) is relatively high.^5–7 When a drug is given intranasally, a fraction of it is not absorbed through the nasal cavity but is swallowed to be absorbed through the gastrointestinal tract. In the case of midazolam, some of the swallowed drug is metabolised in the liver before reaching the systemic circulation (a phenomenon known as the first pass effect).⁷

Lahat et al found that from the time of administration, diazepam given intravenously controlled seizures earlier than midazolam administered intranasally, due to a delay in the absorption of midazolam.⁶ Yet, overall, children's seizures were controlled slightly more quickly among those treated with midazolam when calculated from the time they entered the emergency department; this is because midazolam was administered earlier and the administration of diazepam requires an intravenous line.

What is the clinical significance of a two minute difference in the efficacy of the treatment of a benign condition? There is no question that two minutes of seizures corresponds to 120 long seconds of distress for parents and medical staff. Moreover, the apparent safety of this mode of drug delivery may allow nurse practitioners, nurses, and eventually even parents to administer midazolam intranasally to children with recurrent seizures.

Because of obvious ethical limitations, the authors could not randomly assign children to a placebo group in their study. Thus, theoretically, the similarity in responses to diazepam and midazolam may merely mean that both were not more effective than placebo. It may well be that most children would have stopped seizing spontaneously sometime after 10-15 minutes, and because 10 minutes of seizures was chosen as an entry criterion, the seizures might have gradually resolved even with placebo. However, the survival curves of children treated with diazepam and midazolam were similar, except for the difference in the time of the initial response, which corresponds to the time needed to insert an intravenous line. If this was merely a placebo effect, the two curves should have overlapped completely.

The authors did not define the power of their sample to detect certain differences. Yet it is obvious that the sample size in this study does not have enough power to address the rates and severity of adverse effects. The authors did not specify how blinding was achieved: the research team had to be in the treatment room, and the lack of an intravenous line among those in the midazolam group during the first few minutes is not easy to ignore.

Although Lahat et al defined “delayed seizure control” in their methods section, they did not report the results. Lastly, the duration of a child's seizure before arriving at the study unit was inferred to have been at least 10 minutes, based on the distance of the facility from neighbouring communities; however, it was not measured directly. It could be assumed that randomisation would ensure similar distributions of durations of seizures between the two treatment arms, but randomisation often does not do justice to all confounding variables, especially when the sample size is comparatively small, as was the case here.

However, these limitations are small in view of the step forward these investigators have made in improving drug treatment for a common paediatric condition. This study should be repeated by others before intranasal administration of midazolam becomes the standard of care to address some of the questions alluded to and to ensure the safety of this treatment.

Papers p 83