Journal of Clinical Pharmacy and Therapeutics (2005) 30, 479–485

ORIGINAL ARTICLE

Minimum effective dose of midazolam for sedation of

mechanically ventilated neonates
J.-M. Treluyer* MD PhD , S. Zohar
PhD , E. Rey* Pharm D , P. Hubert MD ,
F. Iserin MD , M. Jugie§ MD , R. Lenclen§ MD , S. Chevret
MD PhD and
G. Pons* MD PhD
*Pharmacologie, Hôpital Cochin-Saint-Vincent de Paul, AP-HP, Université Paris V, Paris,
Département
de Biostatistique et Informatique Médicale, Hôpital Saint-Louis, AP-HP, Université Paris VII, U717
INSERM, Paris, Réanimation Pédiatrique, Cardiologie Hôpital Necker Enfants malades, AP-HP,
Université Paris V, Paris and §Réanimation, Hôpital de Poissy-Saint Germain, France

estimated probability of success was 76Æ9% (95%

SUMMARY
credibility interval: 56Æ6–91Æ4%) for the 200 lg/kg
Objective: To determine the minimal effective loading dose. No significant adverse effect was
dose (MED) of intravenous midazolam, required observed.
for appropriate sedation in 95% of patients, 1 h Conclusions: Continual reassessment is a new
after drug administration. approach, suitable for dose-finding study in
Methods: A double-blind dose-finding study neonates. This method overcomes some of the
using the continual reassessment method, a ethical, statistical and practical problems associ-
Bayesian sequential design. Twenty-three new- ated with this population. Using this method, the
born infants hospitalized in intensive care unit MED was estimated to be the 200 lg/kg loading
participated. Inclusion criteria were: (i) post- dose of midazolam.
natal age <28 days, (ii) gestational age
>33 weeks, (iii) intubation and ventilatory sup- Keywords: continual reassessment method, dose-
port required for respiratory distress syndrome, finding study, midazolam, neonates, stopping
(iv) need for sedation (i.e. one of the six fol- rules for decision making
lowing criteria: agitation or grimacing or crying
facial expression before tracheal suctioning,
agitation or grimacing or crying facial expres- INTRODUCTION
sion during tracheal suctioning). Each neonate Proper sedation of patients is part of good clinical
was allocated to a loading dose, ranging from 75 practice in intensive care units (ICU) in children as
to 200 lg/kg, and a maintenance dose ranging well as in adults. The main goal of sedation is to
from 37Æ5 to 100 lg/kg/h. increase patients’ comfort by decreasing stress and
Results: The primary endpoint was the level of anxiety (1). Moreover, adequate sedation during
sedation 1 h after the onset of infusion. The mechanical ventilation may facilitate effective
sedation procedure was classified as a success if ventilation so that complications such as pneumo-
all the following clinical criteria were met: no thorax and intraventricular haemorrhage may be
agitation, no grimacing and no crying facial prevented (2, 3). Benzodiazepines, such as mida-
expression before as well as during tracheal zolam and lorazepam, are frequently chosen for
suctioning. Based on the 23 patients, the final sedation in the neonatal ICU. Intravenous mida-
zolam differs from other benzodiazepines by its
Received 18 March 2005, Accepted 5 July 2005 water solubility, short elimination half-life, and
Correspondence: Jean-Marc Treluyer, Pharmacologie Clinique,
generally short duration of action. Many authors
Hôpital Saint-Vincent de Paul, 82 avenue Denfert Rochereau,
75645 Paris cedex 14, France.
have studied the pharmacokinetic properties and
Tel.: 33/1 40 48 82 09; fax: 33/1 40 48 83 28; clinical efficacy of midazolam in ventilated neo-
e-mail: [email protected] nates. However, the dosage used to initiate and

2005 Blackwell Publishing Ltd 479

480 J.-M. Treluyer et al.

maintain sedation show dramatic differences. Most were abnormalities of renal, hepatic, neurologic or
reports to date have been case series of midazolam haemodynamic functions.
used in patients of diverse age groups (from 3 days Midazolam was administered intravenously at a
to 18 years of age), with doses ranging from 25 to 0Æ6 mL/kg loading dose over 1 h, followed by an
300 lg/kg administered as a bolus injection, and infusion of 0Æ3 ml/kg/h over 47 h using the same
24–400 lg/kg/h administered as a continuous infusion rate for all the dose levels. Only the con-
infusion (4–8). In a multicentre pilot study, Anand centration of the solution was modified. The con-
et al. (9), administered midazolam with 200 lg/kg centrations of the solution ranged from 6Æ2 to
loading dose followed by an infusion of 20, 40, or 16Æ7 mg/50 mL. Each neonate was double-blind-
60 lg/kg/h for infants of gestational ages 24–26, edly allocated to a loading dose of 75–200 lg/kg
27–29, or 30–33 weeks respectively. Jacqz-Aigrain and a maintenance dose of 37Æ5–100 lg/kg/h. Six
et al. (10), administered midazolam as a 60 lg/kg/h dose regimens were tested: 75, 100, 125, 150, 175,
infusion for up to 5 days in infants >33 week 200 and 37Æ5, 50, 62Æ5, 75, 87Æ5, 100 lg/kg for
gestation, or 60 lg/kg/h for 1 day followed by loading doses and maintenance doses respectively.
30 lg/kg/h for up to a total of 5 days in infants Physicians, nurses and all NICU staff were blind to
£33 week gestation without any loading dose. As the dose of midazolam administered.
with many drugs used in children, little is known The primary endpoint was the level of sedation
about midazolam in neonates. In using midazolam 1 h after the onset of the infusion (H1). The seda-
as a sedative in ventilated neonates, the dosing tion procedure was classified as a success if all the
regimen for effective and safe treatment is a key following clinical criteria were absent: agitation,
issue and no adequate evaluation has been repor- grimacing and crying facial expression before as
ted. However, in this patient group such a dose- well as during tracheal suctioning. The following
finding study meets with distinct ethical, statistical secondary endpoints were also studied: level of
and practical problems. A new approach proposed sedation at H4, H12, H18, H24, H48 using the same
and used previously by us for dose-finding of criteria used at H1, heart rate, blood pressure,
ibuprofen in patent ductus arteriosus (12) is pro- oxygen saturation measured by pulse oximetry,
posed for dose-finding of midazolam for sedation oxygenation index, and triggering of ventilator
of ventilated neonates, to overcome some of the breathing by the patients at H0, H1, H4, H12, H18,
ethical and recruitment issues. The main objective H24 and H48.
was to determine the dose of intravenous mida- Additional midazolam bolus doses or increases
zolam, required to obtain appropriate sedation in in the rate of infusion were not allowed. Sedative
95% of ventilated neonates within 1 h of adminis- drugs that may interact with the assessment of
tration. sedation (benzodiazepines, opioids, muscle relax-
ants) were not allowed. Infusion was continued for
as long as clinically necessary but no longer than
MATERIALS AND METHODS
48 h. At completion of the study, depending on the
The study was approved by the local ethics com- level of sedation achieved, physicians were
mittee (Cochin Hospital, Paris) and written allowed to prescribe midazolam or another sedat-
informed consent was obtained from the parents. ive drug according to their own practice. After the
Inclusion criteria were: (i) post-natal age <28 days, assessment of the primary endpoint, the treatment
(ii) gestational age >33 weeks, (iii) intubation and could be stopped before the end of the study if the
ventilatory support required for respiratory dis- level of sedation was considered insufficient by the
tress syndrome, (iv) need for sedation (i.e. one of physicians. Midazolam was also stopped after H1
the six following criteria: agitation or grimacing or if respiratory distress rapidly improved and extu-
crying facial expression before tracheal suctioning, bation of the child was considered.
agitation or grimacing or crying facial expression
during tracheal suctioning). We excluded babies if
Dose-finding statistical analysis
they or their mother had previously received ben-
zodiazepines or opioids during the 15 days before The continual reassessment method (CRM) (11), a
the time of inclusion. The other exclusion criteria Bayesian sequential design, was used in order to

2005 Blackwell Publishing Ltd, Journal of Clinical Pharmacy and Therapeutics, 30, 479–485
 Minimal effective dose of midazolam 481

determine the minimal effective dose (MED) of Table 1. Demographic and clinical variables of the neo-
midazolam for the sedation of 95% of newborn nates at inclusion
infants. The CRM is an iterative Bayesian method
based on a one-parameter model, which aims at Patients N (%)/Median (range)
estimating the percentile of among k distinct dose
Sex (male) 16 (69Æ6)
levels di (i = 1,…, k). Each of the six dose levels was
Post-natal age (days) 0 (0–4)
arbitrarily chosen by the investigator (according to Gestational age >33 weeks 23 (100)
his/her personal experience and literature data) Weight (g) 2650 (1335–3960)
with the following prior estimated success prob- FiO2 (%) 45 (21–80)
ability, pi (i = 1,…, 6), that is 0Æ3, 0Æ5, 0Æ7, 0Æ9, 0Æ95 Serum creatinine (lmol/l) 86Æ5 (50–122)
and 1Æ0 for the 75, 100, 125, 150, 175 and 200 lg/kg Prothrombin time (%) 59 (20–79)
loading dose respectively. Then, a one-parameter Plasma fibrine (g/l) 1Æ9 (0Æ2–5Æ5)
logistic model (the scale parameter was fixed at 3) Blood platelets (103/mm3) 198 (100–418)
was used to fit the dose–response curve, with a unit
exponential distribution for the model parameter.
The posterior response probability of each dose (200 lg/kg) from the fifth patient to the 23rd
level was re-estimated after each new patient’s patient (Table 2). The final estimated probability of
inclusion. The dose allocated to each new patient success of that dose level was 76Æ9% (95% credi-
was the dose level with the updated posterior bility interval: 56Æ6–91Æ4%) after 23 infants had been
response probability closest to 0Æ95. included (Fig. 1). The posterior success probability
In the present study, the first patient received a for the six doses 75, 100, 125, 150, 175 and 200 lg/kg
loading dose of 175 lg/kg with the prior success corresponded to 27Æ7%, 32Æ7%, 38Æ1%, 47Æ3%,
probability closest to the target (0Æ95). The MED 52Æ5% and 76Æ9% respectively (Table 2). Because
was defined as the dose level among the six chosen the target was fixed at 95%, the MED was selected
doses that had a final response probability closest to be 200 lg/kg. Failures were recorded in five
to the target. The decision to end the study was (25%) of 20 patients receiving a 200 lg/kg mida-
based on stopping criteria, in order to detect whe- zolam loading dose. Among these patients, the
ther all doses were likely to be inefficient, or a failure was assessed on 6, 4, 3, 3, 1 of the 6 failure
suitable estimation of the MED has been obtained criteria. Classification as a failure was the result of
(13). agitation (n = 2), grimacing (n = 3) and crying
facial expression (n = 3) before tracheal suctioning,
agitation (n = 3), grimacing (n = 4) and crying
RESULTS facial expression (n = 3) during tracheal suctioning.
The stopping decision has been made after the
Description of the population
analysis of the stopping criteria. The first decision
Twenty-three newborns entered the study. The was based on the probability that the posterior
demographic and clinical variables for the neonates estimated response probability of success of either
are presented in Table 1. Twenty-one (91Æ3%) dose was lower than the targeted response prob-
children stopped the study before H48, seven ability. This probability was higher than 95% after
(30Æ4%) because the level of sedation was consid- the inclusion of the 14th patient. The second stop-
ered insufficient by the physicians and 14 (60Æ9%) ping decision was associated with four stopping
because the respiratory distress rapidly improved criteria, based on the predictive mean and the
and children were extubated. The median obser- maximum gain from further patient inclusions on
vation time period was 12 h (1–60). the response probability and the width of the
associated credibility interval. These stopping cri-
teria (3 from 4) were lower than 5%, after the 20th
Dose-finding analysis
patient with a success probability at 78Æ7% (95%
The estimated success probability, updated after credibility interval: 57Æ6–93Æ3%). The response
each patient, was <95% for all dose level, and led probability associated with the 200 lg/kg dose
to allocation of the maximum loading dose level would have changed for <5%, even if more

2005 Blackwell Publishing Ltd, Journal of Clinical Pharmacy and Therapeutics, 30, 479–485
482 J.-M. Treluyer et al.

Table 2. Posterior estimated probabilities of success of the five tested doses, updated after each newly included patient

Midazolam loading dose (lg/kg)

75 100 125 150 175 200

Prior estimated probability of success (%)

30Æ0 50Æ0 70Æ0 90Æ0 95Æ0* 100

Subject Dose (lg/kg) Clinical response Posterior estimated probability of success (%)

1 175 Success 30Æ8 55Æ8 78Æ2 95Æ0 98Æ0 100

2 150 Success 31Æ4 60Æ8 84Æ0 97Æ3 99Æ1 100
3 150 Failure 28Æ5 38Æ6 49Æ7 67Æ1 75Æ3 96Æ2
4 200 Failure 27Æ5 30Æ9 34Æ6 40Æ9 44Æ5 63Æ6
5 200 Success 27Æ6 32Æ0 36Æ7 44Æ7 49Æ3 72Æ1
6 200 Success 27Æ8 33Æ0 38Æ6 48Æ3 53Æ7 78Æ6
7 200 Success 27Æ9 33Æ8 40Æ3 51Æ2 57Æ3 83Æ1
8 200 Success 28Æ0 34Æ5 41Æ6 53Æ6 60Æ2 86Æ2
9 200 Failure 27Æ7 32Æ1 36Æ9 45Æ2 49Æ8 73Æ0
10 200 Success 27Æ7 32Æ7 38Æ0 47Æ1 55Æ3 76Æ6
11 200 Success 27Æ8 33Æ1 38Æ9 48Æ8 54Æ4 79Æ5
12 200 Success 27Æ9 33Æ5 39Æ8 50Æ3 56Æ2 81Æ8
13 200 Success 27Æ9 33Æ9 40Æ5 51Æ6 57Æ7 83Æ6
14 200 Failure 27Æ7 32Æ6 37Æ9 46Æ8 51Æ9 76Æ1
15 200 Success 27Æ8 32Æ9 38Æ5 48Æ0 53Æ3 78Æ1
16 200 Success 27Æ8 33Æ2 39Æ0 49Æ0 54Æ6 79Æ8
17 200 Success 27Æ8 33Æ4 39Æ6 49Æ9 55Æ7 81Æ3
18 200 Success 27Æ9 33Æ7 40Æ0 50Æ8 56Æ8 82Æ5
19 200 Success 27Æ9 33Æ9 40Æ5 51Æ6 57Æ7 83Æ6
20 200 Failure 27Æ8 33Æ0 38Æ7 48Æ3 53Æ8 78Æ5
21 200 Success 27Æ8 33Æ2 39Æ1 49Æ1 54Æ7 79Æ9
22 200 Success 27Æ8 33Æ4 39Æ5 49Æ7 55Æ5 81Æ0
23 200 Failure 27Æ7 32Æ7 38Æ1 47Æ3 52Æ5 76Æ9

*Targeted probability of success.

Maximum predictive gain on the response probability

1 Maximum predictive gain on the width of the credibility interval
0·9 Mean predictive gain on the response probability
Mean predictive gain on the width of the credibility interval
0·8 Stopping bound
0·7
Stopping criteria

0·6
0·5
0·4
Fig. 1. Estimated posterior probab-
0·3
ility of success of the 200 lg/kg
0·2 midazolam dose and associated
0·1 95% credibility interval along the
0 trial. This dose level was adminis-
0 2 4 6 8 10 12 14 16 18 20 22 tered from the fourth patient up to
Patient number the last patient.

2005 Blackwell Publishing Ltd, Journal of Clinical Pharmacy and Therapeutics, 30, 479–485
 Minimal effective dose of midazolam 483

0·8

Probability
0·6
Fig. 2. Stopping criteria based on
the predicted gain from further 0·4
patient inclusions on the response
probability (mean and maximum) Estimated probability of success
and on the width of its credibility 0·2
Upper bound of credibility interval
interval (mean and maximum). Lower bound of credibility interval
Three criteria out of the four were 0
lower than 5% from the 20th 4 6 8 10 12 14 16 18 20 22
patient. Patient number

patients were included (Fig. 2). No significant P450 activity, midazolam clearance is reduced, and
effect of sedation on pulse oximetry, oxygenation half-life and time to reach steady-state increased in
index or triggering of ventilator breathing was neonates compared with children and adults (14–
evidenced after the onset of midazolam adminis- 20). Consequently, a specific phase II study was
tration. conducted to determine the optimal dose for
sedation of ventilated neonates. A CRM was cho-
sen for this study because of the potential advan-
Adverse events
tages of such an approach in paediatric patients:
No serious adverse event was reported during the small number of patients required (25 or less), no
study. Pneumothorax was reported in two children placebo group necessary and ethical suitability as
who had received a 200 lg/kg initial loading dose, each subject takes advantage of the previous
and were classified as ‘success’ for sedation. inclusions to receive the current estimated MED.
However, the sedation study was stopped after the As our objective was to obtain an optimal
diagnosis of pneumothorax to administer mor- sedation 1 h after the onset of midazolam
phine for prevention of pain. administration, a loading dose was used. The
Relative median variations between H0 and H1 final estimated probability of success of the
of systolic, diastolic and mean blood pressure were highest dose (i.e. 200 lg/kg for the loading dose
)2% ()18%; 27%), )7% ()43%; 19%), )6% and 100 lg/kg/h for the maintenance dose) was
()40%; 15%), respectively, but none was found 76Æ9% (95% credibility interval: 56Æ6–91Æ4%).
different from 0. A decrease of more than 30% for Based upon the estimated probabilities of success,
systolic, diastolic and mean blood pressure was the optimal dose to obtain 95% of success (the
observed during the study inclusion in none, two initial objective of our study) appears to be
(8Æ7%) and one (4Æ3%) children respectively (all higher than the maximum dose used in this
received the 200 lg/kg loading dose level). How- study. One can wonder if the criteria used to
ever, this event was very transient and no haemo- define sedation were too stringent. There is a
dynamic support was required. Relative variation paucity of tools to measure the level of sedation
of heart rate was )4% ()22%; 16%) between H0 in preterm neonates and those who require
and H1. mechanical ventilation. The scoring systems used
in previous studies concerning use of midazolam
for sedation in neonates were not validated and
DISCUSSION
were not assessed during clinical care. We used a
Midazolam is suitable for sedation of neonates as it behavioural scale adapted from the scoring sys-
is a short-acting benzodiazepine with a high tem by Barrier et al. (21) and Jacqz-Aigrain et al.
clearance and a short elimination half-life. How- (10). We modified these scores as we chose to
ever, consequent to immature hepatic cytochrome assess sedation before but also during tracheal

2005 Blackwell Publishing Ltd, Journal of Clinical Pharmacy and Therapeutics, 30, 479–485
484 J.-M. Treluyer et al.

suctioning. As tracheal suction is an acute and 2. Greenough A, Morley C, Davis J (1983) Interaction of
repetitively stressful in mechanically ventilated spontaneous respiration with artificial ventilation in
neonates, and as the main objective of sedation is preterm babies. Journal of Pediatrics, 103, 769–773.
to decrease the deleterious effects of clinical care 3. Perlman JM, Goodman S, Kreusser KL, Volpe JJ
(1985) Reduction in intraventricular hemorrhage by
for the patients, we assessed sedation during
elimination of fluctuating cerebral blood-flow velo-
tracheal suction. The other difference in the
city in preterm infants with respiratory distress
design of our study compared with previously
syndrome. New England Journal of Medicine, 312,
published studies is that sedation was assessed as 1353–1357.
early as 1 h after the onset of midazolam infusion 4. Hartwig S, Roth B, Theisohn M (1991) Clinical
to obtain an effective and very rapid sedation. experience with continuous intravenous sedation
This may explain the highest dose of midazolam using midazolam and fentanyl in the paediatric
being selected at the end of the study. intensive care unit. European Journal of Pediatrics, 150,
The aim of this study was to improve manage- 784–788.
ment of non-painful stress (anxiety). Analgesics 5. Pellier I, Monrigal JP, Le Moine P, Rod B, Rialland X,
were administered only if painful procedures were Granry JC (1999) Use of intravenous ketamine-mid-
anticipated. However, concomitant use of opioids azolam association for pain procedures in children
with cancer. A prospective study. Paediatric Anaes-
and benzodiazepines necessitates a decrease in the
thesia, 9, 61–68.
total dose of opioids and benzodiazepines (22).
6. Rosen DA, Rosen KR (1991) Midazolam for sedation
Consequently, results of the present study are
in the paediatric intensive care unit. Intensive Care
applicable only when midazolam is used singly. Medicine, 17 (Suppl. 1), S15–S19.
No severe side-effect was observed. Transient 7. Hogberg L, Nordvall M, Tjellstrom B, Stenhammar L
hypotension was noticed in two children. In the (1995) Intranasal versus intravenous administration
study by Jacqz-Aigrain et al. (10) the number of of midazolam to children undergoing small bowel
infants with haemodynamic instability was not biopsy. Acta Paediatrica, 84, 1429–1431.
significantly different between the two midazolam 8. Ng E, Taddio A, Ohlsson A (2000) Intravenous
and the placebo groups, although blood pressure midazolam infusion for sedation of infants in the
was significantly decreased in the midazolam neonatal intensive care unit. Cochrane Database of
group. In most cases, hypotension observed with Systematic Review, CD002052.
9. Anand KJS, McIntosh N, Lagercrantz H, Pelausa E,
midazolam occurred only during concomitant use
Young TE, Vasa R (1999) Analgesia and sedation in
of opioids (23, 24). Transient neurologic effects
preterm neonates who require ventilatory support –
have been associated with the use of midazolam,
results from the NOPAIN trial. Archives of Pediatrics
but no such side-effect was noticed in the present and Adolescent Medicine, 153, 331–338.
study (22). 10. Jacqz-Aigrain E, Daoud P, Burtin P, Desplanques L,
In conclusion, the probability of success for Beaufils F (1994) Placebo-controlled trial of midazo-
sedation in ventilated neonates, even during tra- lam sedation in mechanically ventilated newborn
cheal suctioning, using a 200 lg/kg loading dose of babies [see comments]. Lancet, 344, 646–650.
midazolam during 1 h followed by a 100 lg/kg/h 11. Desfrere L, Zohar S, Morville P et al. (2003) Dose-
maintenance dose, was estimated to be 76Æ9% in finding study of ibuprofen in patent ductus arterio-
the absence of analgesics. Given the small number sus using the continual assessment method. Journal
of children included, tolerability of this dose of Clinical Pharmacy and Therapeutics, 30, 121–132.
12. O’Quigley J, Pepe M, Fisher L (1990) Continual
schedule needs to be confirmed in a larger group of
reassessment method: a practical design for phase 1
neonates.
clinical trials in cancer. Biometrics, 46, 33–48.
13. Zohar S, Chevret S (2001) The continual reassess-
REFERENCES ment method: comparison of Bayesian stopping
rules for dose-ranging studies. Statistics in Medicine,
1. Quinn MW, Wild J, Dean HG, Hartley R, Rushforth
20, 2827–2843.
JA, Puntis JW, Levene MI (1993) Randomised dou-
14. de Wildt SN, Kearns GL, Hop WC, Murry DJ, Abdel-
ble-blind controlled trial of effect of morphine on
Rahman SM, van den Anker JN (2001) Pharmacoki-
catecholamine concentrations in ventilated pre-term
netics and metabolism of intravenous midazolam in
babies. Lancet, 342, 324–327.

2005 Blackwell Publishing Ltd, Journal of Clinical Pharmacy and Therapeutics, 30, 479–485
 Minimal effective dose of midazolam 485

preterm infants. Clinical Pharmacology and Therapeu- 20. Harte GJ, Gray PH, Lee TC, Steer PA, Charles BG
tics, 70, 525–531. (1997) Haemodynamic responses and population
15. Nahara MC, McMorrow J, Jones PR, Anglin D, pharmacokinetics of midazolam following adminis-
Rosenberg R (2000) Pharmacokinetics of midazolam tration to ventilated, preterm neonates. Journal of
in critically ill pediatric patients. European Journal of Paediatrics and Child Health, 33, 335–338.
Drug Metabolism and Pharmacokinetics, 25, 219–221. 21. Barrier G, Attia J, Mayer MN, Amiel-Tison C,
16. Jacqz-Aigrain E, Daoud P, Burtin P, Maherzi S, Shnider SM (1989) Measurement of post-operative
Beaufils F (1992) Pharmacokinetics of midazolam pain and narcotic administration in infants using a
during continuous infusion in critically ill neonates. new clinical scoring system. Intensive Care Medicine,
European Journal of Clinical Pharmacology, 42, 329–332. 15 (Suppl. 1), S37–S39.
17. Jacqz-Aigrain E, Wood C, Robieux I (1990) Pharma- 22. Fetus and Newborn Committee (2000) Prevention
cokinetics of midazolam in critically ill neonates. and management of pain and stress in the neonate.
European Journal of Clinical Pharmacology, 39, 191–192. American Academy of Pediatrics. Committee on
18. Burtin P, Jacqz-Aigrain E, Girard P et al. (1994) Fetus and Newborn. Committee on Drugs. Section
Population pharmacokinetics of midazolam in on Anesthesiology. Section on Surgery. Canadian
neonates. Clinical Pharmacology and Therapeutics, 56, Paediatric Society. Pediatrics, 105, 454–461.
615–625. 23. Van den Anker J, Sauer P. (1992) The use of mida-
19. Lee TC, Charles BG, Harte GJ, Gray PH, Steer PA, zolam in the preterm neonate. European Journal of
Flenady VJ (1999) Population pharmacokinetic Pediatrics, 151, 152.
modeling in very premature infants receiving mid- 24. Burtin P, Daoud P, Jacqz-Aigrain E, Mussat P,
azolam during mechanical ventilation: midazolam Moriette G (1991) Hypotension with midazolam and
neonatal pharmacokinetics. Anesthesiology, 90, 451– fentanyl in the newborn. Lancet, 337, 1545–1546.
457.

2005 Blackwell Publishing Ltd, Journal of Clinical Pharmacy and Therapeutics, 30, 479–485