- Email: [email protected]
Buccal midazolam and rectal diazepam for epilepsy
CORRESPONDENCE
blood pressure we can also say that in the CAPPP study we could find no outcome differences related to the two dosing schedules. However, we emphasise that since once daily versus twice daily administration was not randomised, such observations are of limited value, which is why we did not include them in our report. With respect to André Scheen, the evaluation of all primary endpoints in the CAPPP study was made by an endpoint committee whose members were blinded to the treatment regimen. In this regard the PROBE design1 is equally reliable as a doubleblind trial. It is correct that previous antihypertensive therapy may have affected baseline blood pressure values, but there was no difference in the prevalence of previous antihypertensive therapy in the two patient groups in CAPPP. Needless to say, in the ideal world all patients in a comparative study should be as similar as possible, so that any difference noted between the therapeutic regimens could be attributed to the treatments compared. We were not aware of M I Wiggam and colleagues’ elegant study 2 o n insulin sensitivity during captopril treatment since it was not published when we submitted the CAPPP report. Why did significantly fewer patients develop diabetes during captopril treatment in CAPPP? We offered as a tentative explanation the fact that captopril improves insulin sensitivity, compared with a diuretic.3 We did state, however, “that studies with other ACE inhibitors have not confirmed this finding” and we gave three references to support this statement. Obviously the 1998 data from Wiggam et al2 are in accord with those of these three trials. The significant difference in the frequency of diabetes that we found might have been due to a worsening of insulin sensitivity caused by diuretics/blockers rather than an improvement caused by captopril, a possibility that would not be in disagreement with Pollare and colleagues’ 3 findings. An entirely different mechanism might also have been at play, including the possibility that we allude to that various groups of patients may respond quite differently to treatments. We are aware of Franz Messerli and Ehud Grossman’s views on the lack of a preventive effect of - b l o c k e r therapy against cardiovascular morbidity—a view that we do not necessarily share, partly because the studies they cite can be interpreted differently.
1796
We gratefully acknowledge J Brown and M J Brown’s positive comments. Their views on hypertension as a mixture of many different subtypes is one that we find easy to accept, not least in view of progress in genetic research in relation to hypertension. Accordingly, the response to different drugs may vary considerably between groups of patients. Their hypothesis4 that angiotensin II might have a preventive effect against stroke cannot be disregarded. However, there is much evidence showing that angiotensin II causes some deleterious cardiovascular effects. We would undoubtedly have shared François Gueyffier and colleagues’ concern about a less protective effect against stroke with ACE inhibitors— had we believed it to be a true effect. As it is, we feel that the difference between captopril and conventional treatment in this respect can be fully explained by the reason we gave in our report: previously treated patients in CAPPP (n=5740) had, almost without exception, received conventional therapy. Those who were randomised to conventional therapy thus received treatment that they had had before and which they were known to tolerate. Accordingly, they reached their target blood pressure much faster than the previously treated patients who were randomised to captopril, which was de novo therapy for all of them. The resulting difference in blood pressure between these two subgroups during the first part of the trial accounts for the difference in stroke morbidity. No difference in blood pressure response or stroke was shown in previously untreated patients (n=5245) in whom both treatments were introduced de novo. *Lennart Hansson, Claes Mörlin, Thomas Hedner, for the CAPPP investigators University of Uppsala, Department of Geriatrics, Clinical Hypertension Research, PO Box 609 75125 Uppsala, Sweden (e-mail: [email protected]) 1
Hansson L, Hedner T, Dahlöf B. Prospective randomized open blinded endpoint (PROBE) study: a novel design for intervention trials. Blood Pressure 1992; 1: 113–19. 2 Wiggam MI, Hunter SJ, Atkinson AB, et al. Captopril does not improve insulin action in essential hypertension; a double blind placebo-controlled study. J Hypertens 1998; 16: 1651–57. 3 Pollare T, Lithell H, Berne C. A comparison of the effects of hydrochlorothiazide and captopril on glucose and lipid metabolism in patients with hypertension. N Engl J Med 1989; 321: 868–73. 4 Brown MJ, Brown J. Does angiotensin-II protect against strokes? Lancet 1986; ii: 427–29.
Buccal midazolam and rectal diazepam for epilepsy Sir—In our opinion R C Scott and colleagues (Feb 20, p 263)1 have not shown that “buccal midazolam is at least as effective as rectal diazepam in the acute treatment of seizures”. The main drawback of their study is sample size. With only 40 episodes of seizure in the midazolam group and 39 in the diazepam group and with only 18 patients, this study can only be regarded as preliminary. If the time to cessation of seizure is presumed to be normal then to detect a 20% difference in time to seizure cessation would require at least 110 episodes in each group. With the size of the study, the additional flaws are even more problematic. Nearly half the episodes occurred in two patients. Thus, this study is very dependent on the response to these drugs in only two individuals and may well not be generalisable to other patients with epilepsy. The median duration of seizure was 17 min in the midazolam group and 15 min in the diazepam group, yet the median time from drug administration to end of seizure was 6 min in the midazolam group and 8 min in the diazepam group. This discrepancy might be explained by an imbalance in the duration between time of seizure onset and administration of drug between the groups. The dose of diazepam was a standard 10 mg for the 5-year-old female and the 19-year-old male, whereas the effective dose is known to be 0·5 mg/kg.2 The description of the statistical analysis is confusing and fails to identify the primary endpoint from secondary endpoints and does not indicate which analyses were done post hoc. Confidence intervals were not provided. The study should have been adequately powered, double blinded, with the appropriate primary endpoint being the time from onset to the cessation of seizure. Failure to detect a difference does not mean there is no difference. Confidence intervals are imperative in interpreting this type of study. On the evidence presented, to conclude that buccal midazolam is at least as efficacious as rectal diazepam is unsafe. *Simon J Ellis, Linda Baddely Department of Neurology, North Staffordshire Royal Infirmary, Keele University, Stoke-on-Trent ST4 7LN, UK 1
Scott RC, Besag FMC, Neville BGR. Buccal midazolam and rectal diazepam for
THE LANCET • Vol 353 • May 22, 1999
CORRESPONDENCE
2
treatment of prolonged seizures in childhood and adolescence: a randomised trial. Lancet 1999; 353: 623–26. Remy C, Jourdil N, Villemain D, Favel P, Genton P. Intrarectal diazepam in epileptic adults. Epilepsia 1992; 33: 353–58.
Authors’ reply Sir—We thank Simon Ellis and Linda Baddely for their criticisms. We believe, however, that many are unjustified. We agree that the study is preliminary, but that does not invalidate our statistical analysis and conclusions. We did not suggest that buccal midazolam is better than rectal diazepam, nor was that the aim of the study. A power calculation to identify a 20% difference between the groups would only be appropriate if the aim of the study was to show that one medication was better than the other. We were unable to identify a 5% difference, which suggests that the drugs are similar in efficacy. To prove equivalence in statistical terms would require a much larger study. We acknowledge in our analysis that many of the episodes occurred in two patients. Additional analyses suggest that there was no difference between the groups in terms of efficacy or time to clinical effect. The suggested imbalance in the duration between time of seizure onset and drug administration between the two groups is described in the paper and confirms that any imbalance is not statistically significant. Standard doses of potential prehospital medications are essential. With current delivery devices for rectal diazepam it is not possible to give 0·5 mg/kg to each patient and approximate standard doses are used. The youngest child in our study was 5 years of age; smaller doses would be required for younger children. Because midazolam and diazepam are of similar potency,1 the same dose of both drugs was used in all episodes. Median and interquartile ranges were presented in this study because the data were not normally distributed. Confidence intervals are inappropriate in this situation. A double-blind study would require both buccal and rectal drug administration during every event. This study design is unlikely to meet with ethical approval. Ellis has previously described many weaknesses of randomised controlled trials, one of which is ethical limitation.2 He seems to be suggesting that studies should not be done unless perfect methodology is possible. With this requirement very few controlled studies would ever be possible. Although we conclude that buccal
THE LANCET • Vol 353 • May 22, 1999
midazolam is at least as effective as rectal diazepam in this select group of patients, we have also been realistic in suggesting that further studies should be done before the drug is widely used. We thank Gregory Holmes for his positive Feb 20 commentary3 on our study of buccal midazolam in the acute treatment of seizures.We entirely agree that this new treatment should not yet be widely prescribed, because it has not been evaluated in young children, those who are seizure naïve, and in those who require treatment in a community setting. Smaller doses might also be appropriate in the elderly to reduce the risk of respiratory depression. We do not recommend prescription of buccal midazolam in these populations until appropriate studies have been done to confirm efficacy and safety. Treatment in the accident and emergency setting should be with intravenous medication if rapid venous access can be achieved. If substantial delays are likely then treatment via a more readily available route could offer distinct advantages. We support the argument for continuing to attempt to establish an intravenous line as soon as the buccal (or rectal) medication is administered. *R C Scott, F M C Besag, B G R Neville *Neurosciences Unit, Institute of Child Health, London WC1N 3JH, UK; and The Wolfson Centre, London 1
Pieri L. Preclinical pharmacology of midazolam. Br J Clin Pharmacol 1983; 16 (suppl 1): 17S–27S. 2 Ellis SJ, Adams RF. The cult of the doubleblind placebo controlled trial. Br J Clin Pract 1997; 51: 36–39. 3 Holmes GL. Buccal route for benzodiazepines in treatment of seizures. Lancet 1999; 353: 608–09.
Sir—R C Scott and colleagues 1 present a long overdue alternative rescue medication to rectal diazepam for prolonged or complicated seizures. Although rectal diazepam has proved effective and beneficial, for many patients and carers it has become increasingly unpopular because of its route of administration, variable absorption, and longlasting adverse effects. These drawbacks have provoked the search for an alternative treatment which is at least as effective, but more acceptable to the patient and carer. One such alternative is midazolam, a benzodiazepine with a unique structure and favourable pharmacokinetic and pharmacodynamic properties. Most studies of midazolam have been conducted in anaesthetic practice for the premedication of children; efficacy relates to sedation rather than termination of seizures. Few of these
studies have compared the buccal and intrananasal administration route; those studies that have suggested adequate absorption from either route show conflicting findings in tolerability. 2–4 The intranasal route leads to more rapid absorption and the clinically effective dose is 0·2 mg/kg, whereas buccal and rectal routes seem to have a longer period to onset of sedation and require the slightly higher dose of 0·5 mg/kg. In a pilot study at the David Lewis Centre, we used midazolam via the intranasal route as rescue medication for prolonged, recurrent, or cyanotic seizures in nine adults and three adolescents. In one patient (three doses) it was used during a seizure to prevent severe postictal psychosis rather than to terminate the seizure. Patients served as their own controls because all had responded favourably and consistently in the past to rectal diazepam or other rescue medication. Failure to respond to intranasal midazolam resulted in the patient reverting to previous medication. Rather than titrating doses according to weight, all patients over 50 kg received 10 mg whereas those below 50 kg received 5 mg. In 28 treatment episodes we have recorded only three treatment failures—in all three instances poor technique delivering an adequate volume of midazolam liquid intranasally was to blame. All three patients will receive this medication again. Intranasal midazolam is now firstline rescue medication for all patients who were successfully treated in the pilot study. Rectal diazepam has a long systemic half-life and a large volume of distribution. Repeated doses tend to accumulate and the patient is oversedated, whereas in our experience with midazolam patients are more alert and recover quickly after administratin of this rescue treatment.5 *M Scheepers, S Comish, L Cordes, P Clough, B Scheepers *Garden Wing, Delancey Hospital, Cheltenham GL53 9DU, UK; and David Lewis Centre, Alderley Edge 1
Scott RC, Besag FMC, Neville BGR. Buccal midazolam and rectal diazepam for treatment of prolonged seizures in childhood and adolescence: randomised trial. Lancet 1999; 353: 623–26. 2 Thum P, Heine J, Hollenhorst J, Leuwer M. Midazolam given as an intranasal spray in children. Br J Anaesth 1998; 81: 100–01. 3 Malinovsky J-M, Populaire C, Cozian A, Lepage J-Y, Lejus C, Pinaud M. Premedication with midazolam in children. Effect of intranasal, rectal and oral routes on plasma concentrations. Anaesthesia 1995; 50: 351–54. 4 Connors K, Terndrup TE. Nasal versus oral
1797
blood pressure we can also say that in the CAPPP study we could find no outcome differences related to the two dosing schedules. However, we emphasise that since once daily versus twice daily administration was not randomised, such observations are of limited value, which is why we did not include them in our report. With respect to André Scheen, the evaluation of all primary endpoints in the CAPPP study was made by an endpoint committee whose members were blinded to the treatment regimen. In this regard the PROBE design1 is equally reliable as a doubleblind trial. It is correct that previous antihypertensive therapy may have affected baseline blood pressure values, but there was no difference in the prevalence of previous antihypertensive therapy in the two patient groups in CAPPP. Needless to say, in the ideal world all patients in a comparative study should be as similar as possible, so that any difference noted between the therapeutic regimens could be attributed to the treatments compared. We were not aware of M I Wiggam and colleagues’ elegant study 2 o n insulin sensitivity during captopril treatment since it was not published when we submitted the CAPPP report. Why did significantly fewer patients develop diabetes during captopril treatment in CAPPP? We offered as a tentative explanation the fact that captopril improves insulin sensitivity, compared with a diuretic.3 We did state, however, “that studies with other ACE inhibitors have not confirmed this finding” and we gave three references to support this statement. Obviously the 1998 data from Wiggam et al2 are in accord with those of these three trials. The significant difference in the frequency of diabetes that we found might have been due to a worsening of insulin sensitivity caused by diuretics/blockers rather than an improvement caused by captopril, a possibility that would not be in disagreement with Pollare and colleagues’ 3 findings. An entirely different mechanism might also have been at play, including the possibility that we allude to that various groups of patients may respond quite differently to treatments. We are aware of Franz Messerli and Ehud Grossman’s views on the lack of a preventive effect of - b l o c k e r therapy against cardiovascular morbidity—a view that we do not necessarily share, partly because the studies they cite can be interpreted differently.
1796
We gratefully acknowledge J Brown and M J Brown’s positive comments. Their views on hypertension as a mixture of many different subtypes is one that we find easy to accept, not least in view of progress in genetic research in relation to hypertension. Accordingly, the response to different drugs may vary considerably between groups of patients. Their hypothesis4 that angiotensin II might have a preventive effect against stroke cannot be disregarded. However, there is much evidence showing that angiotensin II causes some deleterious cardiovascular effects. We would undoubtedly have shared François Gueyffier and colleagues’ concern about a less protective effect against stroke with ACE inhibitors— had we believed it to be a true effect. As it is, we feel that the difference between captopril and conventional treatment in this respect can be fully explained by the reason we gave in our report: previously treated patients in CAPPP (n=5740) had, almost without exception, received conventional therapy. Those who were randomised to conventional therapy thus received treatment that they had had before and which they were known to tolerate. Accordingly, they reached their target blood pressure much faster than the previously treated patients who were randomised to captopril, which was de novo therapy for all of them. The resulting difference in blood pressure between these two subgroups during the first part of the trial accounts for the difference in stroke morbidity. No difference in blood pressure response or stroke was shown in previously untreated patients (n=5245) in whom both treatments were introduced de novo. *Lennart Hansson, Claes Mörlin, Thomas Hedner, for the CAPPP investigators University of Uppsala, Department of Geriatrics, Clinical Hypertension Research, PO Box 609 75125 Uppsala, Sweden (e-mail: [email protected]) 1
Hansson L, Hedner T, Dahlöf B. Prospective randomized open blinded endpoint (PROBE) study: a novel design for intervention trials. Blood Pressure 1992; 1: 113–19. 2 Wiggam MI, Hunter SJ, Atkinson AB, et al. Captopril does not improve insulin action in essential hypertension; a double blind placebo-controlled study. J Hypertens 1998; 16: 1651–57. 3 Pollare T, Lithell H, Berne C. A comparison of the effects of hydrochlorothiazide and captopril on glucose and lipid metabolism in patients with hypertension. N Engl J Med 1989; 321: 868–73. 4 Brown MJ, Brown J. Does angiotensin-II protect against strokes? Lancet 1986; ii: 427–29.
Buccal midazolam and rectal diazepam for epilepsy Sir—In our opinion R C Scott and colleagues (Feb 20, p 263)1 have not shown that “buccal midazolam is at least as effective as rectal diazepam in the acute treatment of seizures”. The main drawback of their study is sample size. With only 40 episodes of seizure in the midazolam group and 39 in the diazepam group and with only 18 patients, this study can only be regarded as preliminary. If the time to cessation of seizure is presumed to be normal then to detect a 20% difference in time to seizure cessation would require at least 110 episodes in each group. With the size of the study, the additional flaws are even more problematic. Nearly half the episodes occurred in two patients. Thus, this study is very dependent on the response to these drugs in only two individuals and may well not be generalisable to other patients with epilepsy. The median duration of seizure was 17 min in the midazolam group and 15 min in the diazepam group, yet the median time from drug administration to end of seizure was 6 min in the midazolam group and 8 min in the diazepam group. This discrepancy might be explained by an imbalance in the duration between time of seizure onset and administration of drug between the groups. The dose of diazepam was a standard 10 mg for the 5-year-old female and the 19-year-old male, whereas the effective dose is known to be 0·5 mg/kg.2 The description of the statistical analysis is confusing and fails to identify the primary endpoint from secondary endpoints and does not indicate which analyses were done post hoc. Confidence intervals were not provided. The study should have been adequately powered, double blinded, with the appropriate primary endpoint being the time from onset to the cessation of seizure. Failure to detect a difference does not mean there is no difference. Confidence intervals are imperative in interpreting this type of study. On the evidence presented, to conclude that buccal midazolam is at least as efficacious as rectal diazepam is unsafe. *Simon J Ellis, Linda Baddely Department of Neurology, North Staffordshire Royal Infirmary, Keele University, Stoke-on-Trent ST4 7LN, UK 1
Scott RC, Besag FMC, Neville BGR. Buccal midazolam and rectal diazepam for
THE LANCET • Vol 353 • May 22, 1999
CORRESPONDENCE
2
treatment of prolonged seizures in childhood and adolescence: a randomised trial. Lancet 1999; 353: 623–26. Remy C, Jourdil N, Villemain D, Favel P, Genton P. Intrarectal diazepam in epileptic adults. Epilepsia 1992; 33: 353–58.
Authors’ reply Sir—We thank Simon Ellis and Linda Baddely for their criticisms. We believe, however, that many are unjustified. We agree that the study is preliminary, but that does not invalidate our statistical analysis and conclusions. We did not suggest that buccal midazolam is better than rectal diazepam, nor was that the aim of the study. A power calculation to identify a 20% difference between the groups would only be appropriate if the aim of the study was to show that one medication was better than the other. We were unable to identify a 5% difference, which suggests that the drugs are similar in efficacy. To prove equivalence in statistical terms would require a much larger study. We acknowledge in our analysis that many of the episodes occurred in two patients. Additional analyses suggest that there was no difference between the groups in terms of efficacy or time to clinical effect. The suggested imbalance in the duration between time of seizure onset and drug administration between the two groups is described in the paper and confirms that any imbalance is not statistically significant. Standard doses of potential prehospital medications are essential. With current delivery devices for rectal diazepam it is not possible to give 0·5 mg/kg to each patient and approximate standard doses are used. The youngest child in our study was 5 years of age; smaller doses would be required for younger children. Because midazolam and diazepam are of similar potency,1 the same dose of both drugs was used in all episodes. Median and interquartile ranges were presented in this study because the data were not normally distributed. Confidence intervals are inappropriate in this situation. A double-blind study would require both buccal and rectal drug administration during every event. This study design is unlikely to meet with ethical approval. Ellis has previously described many weaknesses of randomised controlled trials, one of which is ethical limitation.2 He seems to be suggesting that studies should not be done unless perfect methodology is possible. With this requirement very few controlled studies would ever be possible. Although we conclude that buccal
THE LANCET • Vol 353 • May 22, 1999
midazolam is at least as effective as rectal diazepam in this select group of patients, we have also been realistic in suggesting that further studies should be done before the drug is widely used. We thank Gregory Holmes for his positive Feb 20 commentary3 on our study of buccal midazolam in the acute treatment of seizures.We entirely agree that this new treatment should not yet be widely prescribed, because it has not been evaluated in young children, those who are seizure naïve, and in those who require treatment in a community setting. Smaller doses might also be appropriate in the elderly to reduce the risk of respiratory depression. We do not recommend prescription of buccal midazolam in these populations until appropriate studies have been done to confirm efficacy and safety. Treatment in the accident and emergency setting should be with intravenous medication if rapid venous access can be achieved. If substantial delays are likely then treatment via a more readily available route could offer distinct advantages. We support the argument for continuing to attempt to establish an intravenous line as soon as the buccal (or rectal) medication is administered. *R C Scott, F M C Besag, B G R Neville *Neurosciences Unit, Institute of Child Health, London WC1N 3JH, UK; and The Wolfson Centre, London 1
Pieri L. Preclinical pharmacology of midazolam. Br J Clin Pharmacol 1983; 16 (suppl 1): 17S–27S. 2 Ellis SJ, Adams RF. The cult of the doubleblind placebo controlled trial. Br J Clin Pract 1997; 51: 36–39. 3 Holmes GL. Buccal route for benzodiazepines in treatment of seizures. Lancet 1999; 353: 608–09.
Sir—R C Scott and colleagues 1 present a long overdue alternative rescue medication to rectal diazepam for prolonged or complicated seizures. Although rectal diazepam has proved effective and beneficial, for many patients and carers it has become increasingly unpopular because of its route of administration, variable absorption, and longlasting adverse effects. These drawbacks have provoked the search for an alternative treatment which is at least as effective, but more acceptable to the patient and carer. One such alternative is midazolam, a benzodiazepine with a unique structure and favourable pharmacokinetic and pharmacodynamic properties. Most studies of midazolam have been conducted in anaesthetic practice for the premedication of children; efficacy relates to sedation rather than termination of seizures. Few of these
studies have compared the buccal and intrananasal administration route; those studies that have suggested adequate absorption from either route show conflicting findings in tolerability. 2–4 The intranasal route leads to more rapid absorption and the clinically effective dose is 0·2 mg/kg, whereas buccal and rectal routes seem to have a longer period to onset of sedation and require the slightly higher dose of 0·5 mg/kg. In a pilot study at the David Lewis Centre, we used midazolam via the intranasal route as rescue medication for prolonged, recurrent, or cyanotic seizures in nine adults and three adolescents. In one patient (three doses) it was used during a seizure to prevent severe postictal psychosis rather than to terminate the seizure. Patients served as their own controls because all had responded favourably and consistently in the past to rectal diazepam or other rescue medication. Failure to respond to intranasal midazolam resulted in the patient reverting to previous medication. Rather than titrating doses according to weight, all patients over 50 kg received 10 mg whereas those below 50 kg received 5 mg. In 28 treatment episodes we have recorded only three treatment failures—in all three instances poor technique delivering an adequate volume of midazolam liquid intranasally was to blame. All three patients will receive this medication again. Intranasal midazolam is now firstline rescue medication for all patients who were successfully treated in the pilot study. Rectal diazepam has a long systemic half-life and a large volume of distribution. Repeated doses tend to accumulate and the patient is oversedated, whereas in our experience with midazolam patients are more alert and recover quickly after administratin of this rescue treatment.5 *M Scheepers, S Comish, L Cordes, P Clough, B Scheepers *Garden Wing, Delancey Hospital, Cheltenham GL53 9DU, UK; and David Lewis Centre, Alderley Edge 1
Scott RC, Besag FMC, Neville BGR. Buccal midazolam and rectal diazepam for treatment of prolonged seizures in childhood and adolescence: randomised trial. Lancet 1999; 353: 623–26. 2 Thum P, Heine J, Hollenhorst J, Leuwer M. Midazolam given as an intranasal spray in children. Br J Anaesth 1998; 81: 100–01. 3 Malinovsky J-M, Populaire C, Cozian A, Lepage J-Y, Lejus C, Pinaud M. Premedication with midazolam in children. Effect of intranasal, rectal and oral routes on plasma concentrations. Anaesthesia 1995; 50: 351–54. 4 Connors K, Terndrup TE. Nasal versus oral
1797