Intranasal midazolam for treating febrile seizures in childrenBMJ 2001; 322 doi: https://doi.org/10.1136/bmj.322.7278.107 (Published 13 January 2001) Cite this as: BMJ 2001;322:107
- Tony Johnson (firstname.lastname@example.org), statistician
- Medical Research Council Biostatistics Unit, Institute of Public Health, University of Cambridge CB2 1TN
- Neurosciences Unit, Institute of Child Health, Wolfson Centre, London WC1N 2AP
- St Piers Lingfield, Lingfield RH7 6PW
- Institute of Child Health, Wolfson Centre, London WC1N 2AP
- Royal Liverpool Children's Hospital, Alder Hey, Liverpool L2 2AP
- Department of Paediatric Neurology, Birmingham Children's Hospital, Birmingham B4 6NH
- Department of Primary Health Care, University of Oxford, Institute of Health Sciences, Oxford OX3 7LF
Caution is advised in interpreting trial conclusions
EDITOR—The importance of the study by Lahat et al1 is acknowledged both in the editorial by Koren2 and in subsequent correspondence, which recognises the need for effective and safe treatment for acute seizures in the community. But important methodological and analytical issues need to be clarified before the conclusions can be accepted.
The logistics of randomisation are not described in detail, although, firstly, apparently parents were asked to sign a consent form for enrolment in the study after seizures were controlled. The usual ethical practice in randomised controlled trials is to seek consent to randomisation before treatment; here, the order seems to have been reversed unless the controlled seizure actually preceded the seizure for which randomised treatment was allocated. Secondly, randomisation was apparently performed in advance, although this could refer to the frequent practice of drawing up a sequence of treatment allocations before the start of the trial, or it could mean in advance of the trial seizure itself, in which case we would need to know how far in advance. Thirdly, although 100 episodes of febrile seizure were randomly assigned to the two treatments, the analysis is confined to just 52. Under the principle of intention to treat, Lahat et al should report the outcome for all randomised patients' seizures and include them in their primary analysis. It is vital to know exactly what happened to the 48 randomly assigned episodes that are not mentioned further.
Lahat et al conclude that the drugs were equally effective at stopping seizures. This conclusion is drawn from the observation that out of 26 seizures treated with intravenous diazepam 24 responded, compared with 23 out of 26 treated with nasal midazolam; the difference between the percentages is small, at 3.8%. The 95% confidence interval, however, for this difference (diazepam minus midazolam) …