Content uploaded by Sherry H Stewart
Author content
All content in this area was uploaded by Sherry H Stewart
Content may be subject to copyright.
Psychopharmacology (2003) 168:377–386
DOI 10.1007/s00213-003-1429-7
ORIGINAL INVESTIGATION
Susan E. Buffett-Jerrott · Sherry H. Stewart ·
G. Allen Finley · Heather Lee Loughlan
Effects of benzodiazepines on explicit memory
in a paediatric surgery setting
Received: 22 March 2002 / Accepted: 6 February 2003 / Published online: 4 July 2003
Springer-Verlag 2003
Abstract Rationale: Many laboratory-based studies indi-
cate that benzodiazepines impair explicit memory perfor-
mance, increase sedation, and impair attention.
Objectives: The present study was designed to extend
prior lab-based findings to an applied setting in which the
amnestic effects of benzodiazepines may be beneficial for
users. In addition, the study extended the previous adult-
focused research by examining the cognitive effects of
benzodiazepines in children. Methods: The present study
examined the use of a specific benzodiazepine (midazo-
lam) as a premedicant among 40 children aged 4–6 years
old having ear tube (myringotomy) surgery, who were
randomly assigned to receive midazolam or placebo.
Results: Consistent with previous studies, the results
indicated that midazolam causes significant amnesia on a
cued recall task. In addition, free recall for post-drug
events were also impaired by midazolam relative to
placebo, indicating that benzodiazepine-induced amnesia
occurs even for highly salient information. Conclu-
sions: Overall, it appears that benzodiazepines do impair
memory in a pediatric population. This amnesia was not
secondary to the inattention and sedation also caused by
midazolam administration. The theoretical and clinical
implications of these findings are discussed, as are
potential future studies.
Keywords Benzodiazepine · Explicit memory ·
Midazolam · Surgery
Introduction
Benzodiazepines are a class of drugs with sedative,
anxiolytic, and muscle relaxant effects (Curran 1986). In
addition, laboratory research has consistently shown that
benzodiazepines produce anterograde amnesia—impaired
memory for information learned after drug administration
with no impairment for information learned pre-drug
(Curran 1986). The present study was designed to extend
prior lab-based findings to an applied setting where such
amnestic effects might be beneficial (DeJong and Verburg
1988). In addition, it extended previous adult-focused
research by examining the cognitive effects of benzodi-
azepines in children.
Early research indicated that surgical operations can be
a particularly stressful experience for children (Eckenhoff
1953). This anxiety contributes to negative postoperative
behavioural problems (Kain et al. 1996) and surgical
complications (Laycock and McNicol 1988). Both anaes-
thesiologists (DeJong and Verburg 1988) and parents
(Eckenhoff 1953) have noted that a drug that would
decrease anxiety and increase sedation before the induc-
tion of anaesthesia might be beneficial. Many anaesthe-
siologists (e.g. DeJong and Verburg 1988) have noted that
drug-induced amnesia could also be beneficial to children
having surgery, as it might attenuate the psychological
difficulties associated with separation from parents and/or
problematic anaesthesia induction.
When the benzodiazepine midazolam was introduced
in oral form, researchers suggested that it might be useful
as a pre-operative medicant (Smith et al. 1981). Midazo-
lam has a short-half life (approximately 2 h) which makes
it very useful in a day surgery setting, because patients are
not likely to be under the effects of the drug by the time
S. E. Buffett-Jerrott (
)
)
Bedford Sackville Community Mental Health (IWK),
70 Memory Lane, Lower Sackville, Nova Scotia, Canada B4C 2J3
e-mail: sjerrott@accesswave.ca
Tel.: +1-902-8325704
S. H. Stewart
Department of Psychology, Life Sciences Centre,
Dalhousie University,
1355 Oxford Street, Halifax, Nova Scotia, Canada B3H 4J1
G. A. Finley
Department of Pediatric Anesthesia,
IWK Health Centre,
5850 University Avenue, Halifax, Nova Scotia, Canada B3J 3G9
H. L. Loughlan
83 Vetran’s Avenue,
Salisbury, N.B., Canada E4J 2T1
they leave the hospital (Smith et al. 1981). In fact, the
half-life of midazolam is most likely shorter in children
than it is in adults, perhaps due to children’s faster
metabolism (Salonen et al. 1987).
Overall, few placebo-controlled studies have been
conducted on oral midazolam as a pre-operative medicant
in a paediatric surgery context. All studies on midazo-
lam’s sedative effects in children have found increased
observer-rated sedation (using Drug group blind ob-
servers) in participants treated with 0.50 or 0.75 mg/kg
oral midazolam, as compared to both pre-midazolam
sedation levels, and the sedation levels of placebo
participants (e.g. Feld et al. 1990; Weldon et al. 1992).
Unfortunately, only observer-rated tools have been used
to measure the sedative effects of oral midazolam in
children. Performance-based objective sedation measures
should also be used, as participants who appear sedated to
an observer may not actually show decreased perfor-
mance on an objective sedation (e.g. psychomotor speed)
task (Green et al. 1996; Buffett-Jerrott and Stewart 2002).
Several laboratory-based studies with adults have found
that benzodiazepines also significantly impair attention
(e.g. Buffett-Jerrott et al. 1998a). However, these studies
have not yet been extended to the paediatric setting or to
midazolam.
The effect of oral midazolam on memory has rarely
been studied in a paediatric surgery population. Feld et al.
(1990) showed children one picture at approximately
40 min post-midazolam (0.75 mg/kg). Approximately 2 h
later, midazolam-treated participants were less likely to
recall the picture on a free recall task than placebo
controls, indicating impaired explicit memory. Also,
fewer midazolam-treated than placebo-treated children
recalled the application of the mask, indicating an
amnestic effect of the drug on a perhaps more “ecolog-
ically valid” measure of explicit/episodic memory (i.e.
memory for a real life personal experience). In another
study, children received oral midazolam (0.40–0.60 mg/
kg) combined with atropine approximately 1 h before
surgery (Saarnivaara et al. 1988). Unfortunately, no
placebo group was used in this study. Immediately before
the administration of anaesthesia, the children were
shown two pictures and asked to name them. Two hours
later, more than 80% of children remembered the pictures
shown to them before surgery in a free recall task, failing
to provide any strong support for midazolam-induced
disruptions in explicit memory performance. However,
encoding took place 1 h after administration of midazo-
lam, which is well after the time of theoretical peak
plasma concentrations of the drug. Thus, the effects of the
drug were likely to be dissipating by the time of encoding.
In addition, it is not possible to determine what percent-
age of placebo participants would have had memory for
the pictures to evaluate whether the 80% correct recall
does in fact represent a significant reduction in explicit
memory abilities induced by midazolam.
In a well-designed study of oral midazolam in a
paediatric setting (Kain et al. 2000), 5- to 10-year-old
children were administered .5 mg/kg midazolam and
tested using a free recall and a recognition memory task.
Children encoded a series of 12 line drawings and later
asked what drawings they had been shown (free recall).
Next, they were shown 24 cards and asked to identify the
cards they had previously seen (recognition). Results
indicated that both free recall and recognition were
impaired by midazolam, as compared to the placebo
group. The researchers also asked children specific
questions about events that occurred in the operating
room and found significantly impaired memory for these
events in the midazolam-treated participants, when com-
pared with the placebo group.
Overall, the above studies indicate that oral midazolam
is an effective sedative drug, when used as a premedicant
in paediatric surgery populations. Unfortunately, little
research has been conducted on the effects of oral
midazolam on memory in this population. In addition,
there are many apparent problems with several of the
tasks that have been used to assess memory in the studies
to date. A comparison of the quality of memory tasks used
in the anaesthesia literature and psychopharmacology
journals indicated that anaesthesia studies are less likely
to use formal, validated memory assessment techniques
(Ghoneim et al. 1990). Studies in the anaesthesiology
literature are less likely to use a placebo group, to
measure memory at both pre- and post-treatment (to
ensure equivalence of groups prior to drug administra-
tion), and to use memory tests based on a theoretical
model of memory (Ghoneim et al. 1990). Although adult-
focussed studies of midazolam have found impaired
explicit memory, as have studies of other benzodiazepines
(see review by Curran 1986), only one well designed
study (Kain et al. 2000) of the memory effects associated
with oral midazolam in a paediatric surgery population
has been conducted. The present study attempts to extend
that study to a younger age group and using a memory
task that is more similar to the tasks traditionally used in
benzodiazepine studies using adults (i.e. a cued recall
task).
This study was designed to determine midazolam’s
effects on explicit memory in a placebo-controlled design
when the drug was used as a premedicant for 4- to 6-year-
old children having ear tube (myringotomy) surgery. In
the present study, the main outcome variables were a cued
recall and a free recall memory task. The explicit memory
cued recall task was actually administered to each child at
two time periods. The first testing time occurred soon
after the encoding task (between 5 and 20 min post-drug).
Therefore, children would still have been directly under
the influence of midazolam while the memory testing was
being conducted. The second testing time occurred after
the surgery was completed (approximately 130 min after
drug administration). At this time, it was expected that
children would no longer be directly under the influence
of midazolam. Therefore, if memory remained impaired
at the second testing time, it would indicate a midazolam-
induced encoding or storage difficulty, as opposed to a
difficulty with retrieval while under the influence of
midazolam. The present study used an explicit memory
378
cued recall test that used child-normed stimuli (cf.
Greenbaum and Graf 1989).
In the present study, a second explicit memory task
was also administered that was designed to be a more
ecologically-valid, real-world memory task. Children
were asked for their free recall of events that happened
to them on the day of the surgery. Children’s memories
for real-life events occurring when they were and were
not under the influence of the drug were compared.
Several ancillary analyses were also conducted in the
present study in an attempt to extend knowledge about the
additional cognitive effects of midazolam in children.
First, the attentional effects of midazolam premedication
have never been studied. Thus, the present study included
a task assessing focused attention that has been specif-
ically designed for children (Corkum et al. 1995). In
addition, both observer-rated and objective sedation were
assessed to determine the extent of the sedative effects of
midazolam. Statistical analysis was also used to deter-
mine if the postulated amnestic effects of midazolam
were secondary to the sedative and/or inattentive effects
of the drug.
Based on previous research with benzodiazepines
(Buffett-Jerrott et al. 1998a, 1998b; Kain et al. 2000), it
was predicted that midazolam would impair explicit
memory (free and cued recall). In addition, it was
hypothesized that sedation (objective and subjective)
and attention would be impaired by midazolam.
Materials and methods
Participants
Participants in the present study were 40 children aged 4–6 years
(mean age=5 years, 5 months, SD=10.32 months; 24 males, 16
females) who were scheduled for ear tube insertion (myringotomy)
but were otherwise healthy. Any history of neurological or
cognitive impairment or disease, as reported by parents, excluded
children from participation in this study (three children were
excluded for this reason). Those with a previous adverse reaction to
benzodiazepines or Tylenol as reported by parents were also
excluded from the study, as were children who were taking
medication other than antibiotics at the time of study. Participants
were instructed to refrain from eating after midnight the night
before the study (normal hospital procedure). Parents provided
written informed consent before their child participated. Each child
also provided verbal assent. This study had ethical committee
approval.
Only children undergoing purely myringotomy surgery (i.e. not
in conjunction with other surgeries) were used in the present study
because the procedure and anesthetic are simple and easily
standardized. The age range of 4–6 years was chosen because
preschool children are most likely to require ear tubes (Giebink and
Daly 1990), and to control for the poor explicit memory perfor-
mance of very young children (Greenbaum and Graf 1989;
Drummey and Newcombe 1995) and the improvement in explicit
memory performance as children mature (Naito 1990).
Experimental design and drugs
The present study used a randomized, placebo-controlled design.
Participants were randomly assigned to one of the two Drug groups
(n=20 per group) using a yoked design based on age (to ensure age
equivalency between Drug groups). An independent party con-
ducted this randomization. The active Drug group received oral
midazolam [Versed (Roche) for intravenous administration, 5 mg/
ml] at a dose of 0.50 mg/kg mixed with acetaminophen [Tylenol
(McNeil Consumer Products, Canada, Inc.) grape-flavoured sus-
pension, 32 mg/ml] at a dose of 15 mg/kg. The placebo group
received only the Tylenol, without the addition of midazolam. Drug
administration was double-blind in that the child, his/her parents,
and the researcher testing the child were not aware of the child’s
Drug group placement. Only the nurse responsible for drug
administration was aware of the child’s group assignment. In case
of emergency, medical staff could discover the child’s Drug group
status in the child’s file. There were no dropouts in the present
study will all participants completing the entire study procedure.
Tasks and procedure
When a child was scheduled for myringotomy surgery, a letter was
sent to the parents followed by a telephone call to explain the study.
Approximately 90 min before surgery, the researcher met with both
the child and the parent(s) of willing families to obtain parental
informed consent and child assent for participation.
Next, the child’s baseline level of sedation was then rated by the
experimenter using a 5-point rating scale with each number rating
having an associated descriptor to facilitate accurate observations
(Wilton et al. 1988). This observer-rated sedation measure was
chosen for use in the present study because it has been used
previously in studies of midazolam as a preoperative medicant in
children (e.g. Wilton et al. 1988). However, its psychometric
properties remain unknown. An initial examination of the observer-
rated sedation scores in the present study, indicated that at both
testing times (pre- and post-drug), all participants were rated on the
5-point Likert scale as either a “3” (calm) or a “4” (drowsy).
Therefore, the observer-rated sedation scores were analysed as a
categorical variable (rather than a continuous measure). A second,
independent observer who was also blind to each child’s Drug
group membership provided ratings for a random ten of the 40
participants (25% of the total sample). The inter-rater reliability
coefficient could not be calculated at the pre-drug testing time due
to the lack of variation in scores (all children were rated as calm by
both observers). At the second testing time, the reliability
coefficient was 1.00, indicating perfect agreement on this measure.
When the child entered the day surgery area, a nurse took his/
her body weight, which was used in determining drug dose. An
anaesthesiologist also met with the child to perform a pre-
anesthetic assessment, to ensure medical eligibility for the study,
and to order the study medications (this was not the same
anaesthesiologist who administered anaesthetic during surgery).
Before the study drug (midazolam or placebo) was adminis-
tered, children were given a variety of cognitive tasks to ensure pre-
drug equivalence between groups. The narrative memory task from
the NEPSY (Korkman et al. 1997) assessed baseline memory
performance. In this task, the child is told a short story and
immediately tested for free recall and cued recall of the story
(combined to create a total memory score). The picture deletion
task for preschoolers (Corkum et al. 1995) assessed baseline
attentional performance. In this task, after a few practice items,
children are presented with a page of 60 shapes and asked to scan
the page to find all examples of a specific target (ten diamonds or
triangles). Children use a bingo marker to mark each target. Time
taken to complete the task, as well as omission and commission
errors were measured. The motor deletion task (Corkum et al.
1995), objectively assessed baseline sedation. In this task, children
were presented with a page of 60 circles and asked to put a mark in
each circle on the page as quickly as possible, with a 5-min time
limit. Time to complete this psychomotor speed task was used as
the measure of objective sedation (maximum score=5 min). After
these baseline measures were completed, at approximately 50 min
before surgery, participants received their assigned drug.
Next, at approximately 12 min post-drug (range 5–20 min),
participants were shown a series of 12 line drawings and asked to
379
identify each of them. Children were told that they were going to be
asked about the pictures at a later time (intentional learning).
Following Greenbaum and Graf (1989), the materials were 27 line
drawings of common objects. Although we were unable to obtain
the original stimulus materials used by Greenbaum and Graf
(1989), we used pictures of the same items as those described in
their original study. Additionally, except where noted, all task
procedures were identical to those used by Greenbaum and Graf
(1989). Specifically, 24 pictures were used to form four sets of six
targets. Three additional pictures served as practice items. Each
picture was shown on a 1015 cm index card. The six items in each
set were all from the same category (kitchen, zoo, park, or
restaurant). Each child saw pictures from two of the four categories
(12 pictures): park and restaurant or zoo and kitchen. Which set of
stimulus materials a particular child encoded was counterbalanced
between the two groups. During encoding, children were shown
each picture for 3 s. If the child could not identify, or misidentified,
the picture he/she was informed of the correct picture name and
asked to repeat it.
After encoding, the experimenter again rated sedation using the
5-point rating scale. Children were then given the second PDTP
[using the target shape (diamond or triangle) that was not presented
pre-drug] and the MDT. Children were then administered the cued
recall task (Greenbaum and Graf 1989): children were “cued” with
the back of the picture cards and asked to list the pictures they had
seen from a specific category (e.g. “Can you name the pictures I
showed you of things you see in a restaurant?”) The task was
terminated when the child indicated that he/she could not name any
more objects (cf. Greenbaum and Graf 1989).
When the child had finished the cued recall task, he/she was
taken to the surgery room. All children separated from their parents
when leaving the day surgery room, as per typical hospital
procedure. For all study participants, anaesthesia was induced using
only halothane/nitrous oxide (N
2
O) via a mask.
After the child was awake and returned to the day surgery room,
he/she was given the cued recall task a second time. During this
administration, the cued recall task asked about the category of
encoded words that had not been asked about in the first cued recall
task. Finally, the child was asked for his/her free recall of the events
that had happened to him/her since arriving at the hospital (i.e.
“Can you tell me about what happened today since you came to the
hospital?”). The experimenter recorded each of the participants’
answers verbatim. The study procedure is summarized in graphic
form in Fig. 1.
Results
Participant characteristics
A series of one-way (midazolam versus placebo) chi-
square analyses were conducted to determine if the groups
differed on any of the categorical control variables. There
was no effect of Group for the number of females, the
number of children taking medication (total n=8), or the
number of children who had previously had surgery (total
n=25). Only one child had previously taken midazolam,
so this variable was not analysed statistically. In addition,
a one-way (Drug group) ANOVA indicated that the
groups did not differ significantly in age.
Pre-drug cognitive functioning
Pre-drug measures of narrative memory (total memory
score from NEPSY), objective sedation (time to complete
MDT), and attention (time to complete PDTP, as well as
number of omission and commission errors) were anal-
ysed with a series of one-way (Drug group) ANOVAs to
determine if the groups were equivalent prior to the drug
Fig. 1 Graphic representation
of study procedure
380
administration. Because omission and commission errors
are probably measuring different attentional processes,
they were scored and analysed separately for the PDTP.
There were no significant effects of Drug group on any of
the pre-drug cognitive measures. In addition, on the
observer-rated sedation measure, all 40 children were
rated by the experimenter as “calm” (i.e. mid-point of “3”
on the 5-point Likert scale; Wilton et al. 1988) before
being administered the drug. Means and SD for these pre-
drug cognitive measures are displayed in Table 1 as
functions of Drug group.
Memory effects
Cued recall
Cued recall performance was scored as the total number
of target pictures correctly remembered from the encod-
ing task. Explicit memory testing was conducted at two
testing periods in a within-subjects design: pre-surgery
and post-surgery. Cued recall data was analysed using a
22 (Drug groupTesting period) ANOVA.
Means (and standard errors) for the cued recall task
given before surgery (pre-surgery Testing period) and
after surgery (post-surgery Testing period), as a function
of Drug group, are displayed in Fig. 2. The analysis
revealed a significant main effect of Drug group
[F(1,38)=21.35, P<0.001] with the placebo group cor-
rectly remembering more target words. No other effects
were significant.
At both Testing periods (pre- and post-surgery), the
midazolam-treated participants recalled fewer target pic-
tures than the placebo-treated participants (see Fig. 2).
Overall, there was little to no decay in cued recall
performance from the pre-surgery to the post-surgery
testing time (see Fig. 2). It should be noted, however, that
there was a marginally significant Drug group difference
in the number of answers given (pictures named regard-
less of accuracy) on the cued recall task at both testing
periods [ F(1,38)=3.94, P=0.05]. At the pre-surgery testing
period the placebo group made an average of 3.70 guesses
(SD=2.13), as compared to the midazolam group’s 2.65
guesses (SD=2.13). At the post-surgery testing period the
placebo group made an average of 3.95 guesses
(SD=2.16), as compared to the midazolam group’s 2.85
guesses (SD=2.21). To ensure that the significant Drug
group difference on the cued recall task was not simply
due to the number of answers given by the participants,
the number of words given on the cued recall task was
used as a covariate in the above 22 analysis. This
ANCOVA indicated that neither the direction of nor the
significance level of the Drug group difference was
changed after controlling for word generation, per se
[F(1,36)=15.35, P<0.001].
Free recall
To analyse participants’ free recall of the events that had
happened to them in the hospital, possible hospital events
were broken down into 15 categories (e.g. getting
registered, going to the operating room) by our own
initial drug-group-blind review of the types of responses
produced by the children on this task. Children’s
responses were then scored for the number of categories
contained in their responses. The child scored 1 for each
category named, but each category was only scored once.
For example, a child naming three distinct events
occurring during the cognitive testing phase (e.g. use of
bingo marker, memory for a story, searching for triangles)
scored only one point for all three responses combined
because “cognitive tasks” was deemed a single category.
Next, the 15 categories were further broken down into
two sub-categories. The first contained ten events that
most likely occurred when the child was not under the
influence of the drug (e.g. nursing tasks). The other
Table 1 Means (and SD) on the
pre-drug measures of cognitive
functioning, as a function of
drug group. Memory,NEPSY
narrative memory task (Kork-
man et al. 1996); MDT motor
deletion task (Corkum et al.
1995); PDTP picture deletion
task for preschoolers (Corkum
et al. 1995)
Drug group
Placebo (n=20) Midazolam (n=20)
Memory: NEPSY (number of story items remembered) 10.60 (5.64) 10.85 (5.35)
Objective sedation [time (s) to complete MDT] 55.55 (21.34) 57.05 (21.86)
Attention speed [time (s) to complete PDTP] 44.75 (15.04) 47.20 (17.05)
Attentional omissions (number of omission errors on PDTP) 0.65 (1.53) 0.65 (1.23)
Attentional commissions (number of commission errors on
PDTP)
0.30 (1.13) 0.70 (2.47)
Fig. 2 Cued recall performance: mean number of pictures correctly
recalled on the cued recall test as a function of Drug group, at each
encoding time. Bars represent standard errors
381
contained five events that most likely occurred when the
child was under the influence of the drug (e.g. getting
anaesthetic). To ensure reliable scoring, a second indi-
vidual, also without knowledge of participants’ Drug
group status, again independently scored these two free
recall sub-categories for all participants. The inter-rater
reliability coefficients (Hartman 1977) ranged from 0.93
to 1.00 (across subcategories and testing times), indicat-
ing excellent inter-rater reliability. As inter-rater discrep-
ancies were rare, the first author always relied on her
response for final scoring when a discrepancy occurred.
The two subcategories of events were analysed
separately as there was no way of controlling equivalence
of events happening pre- and post-drug. A one-way (Drug
group) ANOVA revealed no significant main effect of
Drug group [F(1,38)=0.43, NS] for the number of
occurrences remembered for events likely occurring when
participants were not under the influence of the drug/
placebo. However, the midazolam-treated participants
remembered fewer occurrences of events that likely
occurred when they were under the influence of the drug
when compared with the placebo-treated participants
[F(1,38)=4.97, P<0.05]. Means and SD for “under the
influence” and “not under the influence” free recall are
displayed in Fig. 3 as a function of Drug group.
Observer-rated sedation
Since there was no variation in scores at the pre-drug
testing time [all 40 participants received scores of “3”
(calm)], only data from the post-drug testing time was
analysed using a one-way chi-square analysis. At the post-
drug testing time (immediately after memory task encod-
ing), there was a significant difference between the two
groups [c
2
(1)=4.44, P<0.05] with four of the 40 partic-
ipants rated as a “4” (drowsy) and the rest as “3” (calm).
All the participants rated as drowsy were in the midazo-
lam group.
Objective sedation
Scores on the MDT were analysed with a 22 ANOVA
[Drug groupDrug phase (pre- vs post- drug)]. The
ANOVA indicated a significant main effect of Drug
group [F(1,38=7.66, P<0.01) and Drug phase
[F(1,38)=11.10, P<0.005] which were qualified by a
significant Drug groupDrug phase interaction
[F(1,38)=11.19, P<0.005]. Simple effects tests indicated
no significant effect of Drug phase for the placebo
subjects [F(119)=0.00, NS]. For the midazolam-treated
participants, there was a main effect of Drug phase
[F(1,19)=11.39, P<0.005] with the participants taking
much longer to complete the task at the post-drug testing
time than the pre-drug testing time (see Table 2).
Attention tasks
A series of 22 [Drug groupDrug phase (pre- vs post-
drug)] ANOVAs were conducted on the time taken to
complete the attention task, as well as the number of
omission and commission errors on the task. The means
(and SD) for the attentional measures are presented in
Table 3, as a function of Drug group. The ANOVA on the
Fig. 3 Means (and SD) on the
free recall task, as a function of
drug influence and drug group
382
time taken to complete the PDTP indicated main effects
of Drug group [F(1,38)=6.25, P<0.05], and Drug phase
[F(1,38)=10.32, P<0.005] which were qualified by a
significant Drug groupDrug phase interaction
[F(1,38)=5.79, P<0.05]. Simple effects analyses indicated
that for the placebo participants there was no effect of
Drug phase [F(1,19)=2.28, NS]. However, for the mida-
zolam participants there was a significant effect of Drug
phase (see Table 3) with the participants taking longer to
complete the task after receiving midazolam
[F(1,19)=8.50, P<0.01]. The ANOVA on the attentional
omissions indicated no significant effects (see Table 3).
The ANOVA on the commission errors indicated a
main effect of Drug group [F(1,38)=4.90, P<0.05] (see
Table 3). There was no main effect of Drug phase and no
interaction. Overall, the midazolam-treated participants
had more commission errors than the placebo-treated
participants, regardless of Drug phase. A visual inspection
of the data indicated that the midazolam group at the pre-
Drug phase was making more commission errors than the
placebo group at the pre-Drug phase. To account for this
pre-drug difference a one-way (Drug group) ANCOVA
was performed on the post-drug commission scores while
covarying out the pre-drug commission scores. This
analysis indicated that the significant Drug group effect
remained [F(1,37)=4.99, P<0.05].
ANCOVAs
ANCOVAs were conducted which covaried out all other
cognitive variables affected by midazolam (i.e. post-drug
observer-rated sedation, objective sedation, attention
time, and attention commissions, respectively) from the
scores on the cued-recall task at both testing times. Drug
group effects remained significant in all cases. In similar
ANCOVAs on the free recall of “under the influence of
drug” surgery day events, Drug group effects remained
significant for all attentional ANCOVAs. Although
covariate-adjusted free recall memory means remained
in the same direction as prior to covariance (i.e. mida-
zolam<placebo), covarying out observer rated sedation
appeared to reduce the Drug group effect [F(1,37)=–3.71,
P=0.06]. Likewise, covarying objective sedation also
appeared to reduce the Drug group effect [F(1,37)=3.50,
P=0.07].
Discussion
The results of the present study lend support to previous
initial findings that benzodiazepines (specifically mida-
zolam) impair explicit memory in a paediatric population
(Kain et al. 2000). This was true even though the groups
were shown to be equivalent on all pre-drug cognitive
measures. In addition, the present results extended
previous findings by using a cued recall memory task, a
more structured free recall task and by examining the
effects of midazolam on attention and sedation.
There has been some controversy about which stage of
memory processing (e.g. acquisition, storage, and/or
retrieval) is disrupted by benzodiazepines (Buffett-Jerrott
and Stewart 2002). The cued recall task was given at two
separate testing times in the present study. By conducting
two periods of cued recall testing with the same set of
encoding stimuli (pre- and post-surgery), we were able to
examine which phase of memory processing was disrupt-
ed by midazolam. By the second cued recall testing period
(approximately 130 min after drug administration), par-
ticipants were theoretically no longer under the direct
influence of midazolam as the half-life of midazolam is
approximately 100 min (Smith et al. 1981). The fact that
the drug-induced cued recall impairment remained at the
post-surgery testing period provides further evidence that
benzodiazepine-induced memory impairments are not the
result of a retrieval deficit, but rather are the result of
impairments occurring at the encoding and/or storage
phases of memory processing (cf. Brown et al. 1982;
Gorissen et al. 1998). Because we used the same set of
Table 2 Means (and SD) on the
objective sedation task, as a
function of drug group and drug
phase.MDT motor deletion task
(Corkum et al. 1995)
Drug group
Placebo (n =20) Midazolam (n =20)
Pre-drug [time (s) to complete MDT] 55.55 (21.34) 57.05
a
(21.86)
Post-drug [time (s) to complete MDT] 55.40 (26.32) 131.70
a
(106.18)
a
Means with similar superscripts are significantly different from one another (P<0.005)
Table 3 Attentional performance: means (SDs) for attention speed,
attentional omissions and attentional commissions, as a function of
Drug group. PDTP picture deletion task for preschoolers (Corkum
et al. 1995). Note that the statistical comparison for attentional
commissions is for a Drug group effect at post-drug, after covarying
pre-drug commissions
Drug group
Placebo (n=20) Midazolam (n=20)
Attention speed
Time taken to complete PDTP
Pre-drug 44.75 (15.04) 47.20
a
(17.05)
Post-drug 52.30 (19.25) 99.85
a
(81.37)
Attentional omissions
Number of omissions on PDTP
Pre-drug 0.65 (1.53) 0.65 (1.23)
Post-drug 1.15 (2.25) 2.25 (3.68)
Attentional commissions
Number of commission errors on PDTP
Pre-drug 0.30 (1.13) 0.70 (2.47)
Post-drug 0.20
b
(0.52) 1.70
b
(3.05)
Means with similar subscripts differ significantly (
a
P<0.01,
b
P<0.05)
383
encoding stimuli for testing memory at each testing
period, there was potential for practice effects (Roediger
and McDermott 1993) by the post-surgery testing time.
Nonetheless, the finding that midazolam impairments are
still seen at the second testing time suggests that memory
impairments are not overcome by practice with the test.
Our multiple testing times also allowed us to examine
the effect of midazolam on memory decay. Interestingly,
participants did not show much deterioration or decay in
their memory over time, regardless of their Drug group
status. It would be expected that children would remem-
ber fewer pictures/events at the second testing time as
compared to the first testing time, due to the 1.5 h and the
many events which had elapsed between the two cued
recall testing times. However, memory decay was not
apparent. The virtual absence of decay of memories
observed in the present study is generally consistent with
what is known about children’s memory for medical
events. For example, in an interesting applied study of
children’s memory (Peterson 1999) children were asked
for their free recall of medical emergencies (e.g. broken
bones, stitches) at several times in the 2 years after
treatment. Overall, the children’s memory of the medical
experience was stable over a 2-year follow-up. It is
possible that even neutral objects/events that occur when
a child is in a potentially frightening medical situation
also tend to be better remembered. However, when
Peterson (1999) compared children’s memory for the
injury itself and memory for hospital events, only the
memory for the injury details remained stable, providing
little support for “enhanced traumatic memory” as the
explanation for our lack of memory decay. Further study
is clearly needed.
A further consideration in attempting to understand our
lack of memory decay involves the fact that memory on
the cued recall task actually tended to be relatively “poor”
(between 10% and 30% correct recall) at the first of the
two testing periods (see Fig. 2). This initially poor
memory may have created a “floor effect”. In other
words, there was not much information to begin with that
could decay over time. A possible explanation for our
generally low levels of recall was our use of a memory
task that was relatively taxing. The “cues” used in the
cued recall task were category names as opposed to
specific cues for each target word (like the word-stems
used in many traditional lab-based studies of benzodiaz-
epine effects) (e.g. Curran and Gorenstein 1993).
Researchers have noted that the lab-based tests of
memory that are typically used to study benzodiazepines
may lack “ecological validity”. In other words, tasks such
as the cued recall of word lists or sets of pictures may not
be representative of real-world memory requirements
(Curran 1986). Although it could be argued that our cued
recall task did not adequately simulate real-world mem-
ory requirements, midazolam-treated children also
showed impairments on a free recall task pertaining to
their memory for personal events that had happened to
them in the hospital context while under the influence of
the drug. This finding is similar to results from traditional
studies of midazolam and memory in the paediatric
surgery context that simply assessed children’s memory
for surgery events (O’Boyle 1987; Feld et al. 1990; Kain
et al. 2000). However, we extended these earlier findings
by examining multiple categories, rather than a single
event, by using a reliable coding system, and by
comparing memories of hospital events that occurred
when the child was drugged with memories of non-
medicated events. Thus, benzodiazepine explicit memory
impairments occur even for events that are highly salient
and personally relevant (e.g. having mask placed on face),
and not just for less salient, neutral items (e.g. the pictures
used in our cued recall task).
Analysis of the observer-rated and objective sedation
scores indicated an increase in sedation in the midazolam
group. An analysis of covariance indicated that each
measure of sedation made a small contribution to
benzodiazepine-induced impairments on the free recall
memory task (cf. Bishop and Curran 1995), consistent
with previous basic cognitive psychology findings that
arousal contributes to memory performance, but does not
fully account for explicit memory effects (e.g. Rabinowitz
et al. 1982; see also review by Roediger and McDermott
1993). Consistent with our previous work with adults
(Buffett-Jerrott et al. 1998b), sedation contributed to
performance on the more “ecologically valid” free recall
task, as compared with the traditional cued recall task.
One problem with the observer-rated sedation scale
used in the current study (Wilton et al. 1988) is that only
two of the five anchors were ever chosen in the study (i.e.
calm and drowsy). This suggests the need for a more fine-
grained measure within those two points to increase
sensitivity to subtle differences in sedation in this kind of
research. A second problem with the sedation scale used
in the current study (and those used in anaesthesia
research in general) is that it is likely to be confounded by
anxiety, in that “agitated” and “alert” are rated when a
child is crying or clinging to his/her parent. In the future,
a new sedation measure might be developed which adds
more fine-grained points and which makes a better
conceptual distinction between sedative and anxiolytic
drug effects. For example, a child who is calm should be
rated as “non-drowsy”, as should a child who is crying but
does not appear sleepy. It would also be helpful to
develop a child-rated sedation scale, to maintain consis-
tency with adult studies on the subjective sedative effects
of benzodiazepine administration (e.g. Block and Berchou
1984).
An analysis of the effects of midazolam on attention
led to some interesting results. Overall, on time taken to
complete the PDTP, midazolam participants were im-
paired relative to placebo. This result parallels the
findings with the MDT (objective sedation) where
midazolam participants were impaired relative to placebo.
This parallel is not surprising given that benzodiazepine-
induced “attentional” impairments on deletion tasks are
likely to be confounded with psychomotor speed impair-
ments (Curran et al. 1993). The fact that we also found
commission errors to be more common in midazolam-
384
treated participants provides further evidence of genuine
attentional impairments, as opposed to simple psychomo-
tor slowing. Interestingly, increased commission errors
due to benzodiazepines have not been previously reported
in the adult literature using deletion tasks of attention (e.g.
Buffett-Jerrott et al. 1998b). Most likely, the PDTP was
more difficult for children than the focused attention tasks
used in the adult literature, causing errors of commission
to be observable. Thus, the PDTP may be a more sensitive
index of benzodiazepine attentional impairments in
children. Also consistent with previous findings (Buf-
fett-Jerrott et al. 1998b), analysis of covariance indicated
no contribution of attentional impairments to benzodiaz-
epine-induced memory impairments.
Overall, it appears that midazolam leads to impair-
ments in children’s explicit memory, as well as to
inattention and increased sedation. It should be noted
that within the construct of long-term memory, re-
searchers commonly distinguish between two main
memory processes, implicit and explicit memory (e.g.
Graf and Schacter 1985). Explicit memory occurs when a
participant is consciously and effortfully attempting to
remember a previous experience. In contrast, implicit
memory occurs when a participant’s ability to perform a
task is facilitated by a previous experience. Implicit
memory occurs without conscious awareness (Graf and
Schacter 1985). In future studies, the effects of midazo-
lam on implicit memory in children should also be
studied.
Overall, more research is needed in the area of
benzodiazepines in conjunction with paediatric surgery.
It will be important for the relative time course of
midazolam’s effects on both implicit and explicit memory
to be determined to aid anaesthesiologists in determining
the most effective time to administer the drug before the
surgery begins. In addition, research is needed to
determine which children would actually benefit from
midazolam pre-medication, or whether alternatively mi-
dazolam-induced memory impairments might be counter-
productive in the longer term (cf. Wilhelm and Roth
1997). Finally, research is needed to determine how
midazolam pre-medication affects anxiety about future
surgery or other medical procedures, and how precisely
anxiety and memory may be inter-related.
Acknowledgements This research was supported by a generous
grant from the IWK Health Centre. We also wish to acknowledge
the support and cooperation of the surgeons of the Division of
Otolaryngology and of the nursing staff in the Day Surgery unit,
Operating Room, and Recovery Room of the IWK Health Centre.
References
Bishop KI, Curran HV (1995) Psychopharmacological analysis of
implicit and explicit memory: a study with lorazepam and the
benzodiazepine antagonist flumazenil. Psychopharmacology
121:267–278
Block, RI, Berchou R (1984) Alprazolam and lorazepam effects on
memory acquisition and retrieval processes. Pharmacol Bio-
chem Behav 20:233–241
Brown J, Lewis V, Brown MW, Horn G, Bowes JB (1982) A
comparison between transient amnesias induced by two drugs
(diazepam and lorazepam) and amnesia of organic origin.
Neuropsychologia 20:55–70
Buffett-Jerrott SE, Stewart SH (2002) Cognitive and sedative
effects of benzodiazepine use. Curr Pharm Design 8:45–58
Buffett-Jerrott SE, Stewart SH, Bird S, Teehan MD (1998a) An
examination of differences in the time course of oxazepam’s
effects of implicit vs explicit memory. J Psychopharmacol
12:338–347
Buffett-Jerrott SE, Stewart SH, Teehan MD (1998b) A further
examination of the time-dependent effects of oxazepam and
lorazepam on implicit and explicit memory. Psychopharmacol-
ogy 138:344–353
Corkum V, Byrne JM, Ellsworth C (1995) Clinical assessment of
sustained attention in preschoolers. Child Neuropsychol 1:3–18
Curran HV (1986) Tranquilizing memories: a review of the effects
of benzodiazepines on human memory. Biol Psychol 23:179–
213
Curran HV, Gorenstein C (1993) Differential effects of lorazepam
and oxazepam on priming. Int Clin Psychopharmacol 8:37–42
Curran HV, Gorenstein C, Lader M (1993) Comparative amnesic
and sedative effects of lorazepam and oxazepam in healthy
volunteers. J Psychopharmacol 7:249–256
DeJong PC, Verburg MP (1988) Comparison of rectal to intramus-
cular administration of midazolam and atropine for premedi-
cation of children. Acta Anaesthesiol Scand 32:485–489
Drummey AB, Newcombe N (1995) Remembering versus knowing
the past: children’s explicit and implicit memories for pictures.
J Exp Child Psychol 59:549–565
Eckenhoff JE (1953) Relationship of anesthesia to postoperative
personality changes in children. Am J Dis Child 86:587–591
Feld LH, Negus JB, White P (1990) Oral midazolam preanesthetic
medication in pediatric outpatients. Anesthesiology 73:831–834
Ghoneim MM, Ali MA, Block RI (1990) Appraisal of the quality of
assessment of memory in anesthesia and psychopharmacology
literature. Anesthesiology 73:815–820
Giebink GS, Daly K (1990) Epidemiology and management of
otitis media in children. Topics Lang Disord 11:1–10
Gorissen MEE, Curran HV, Eling PATM (1998) Proactive
interference and temporal context encoding after diazepam
intake. Psychopharmacology 138:334–343
Graf P, Schacter DL (1985) Implicit and explicit memory for new
associations in normal and amnesic subjects. J Exp Psychol
11:501–518
Green JF, McElholm A, King DJ (1996) A comparison of the
sedative and amnestic effects of chlorpromazine and loraze-
pam. Psychopharmacology 128:67–73
Greenbaum JL, Graf P (1989) Preschool period development of
implicit and explicit remembering. Bull Psychon Soc 27:417–
420
Hartman DP (1977) Considerations in the choice of interobserver
reliability estimates. J Appl Behav Anal 10:103–116
Kain ZN, Hofstadter MB, Mayes LC, Krivutza DM, Alexander G,
Wang S, Reznick JS (2000) Midazolam: effects on amnesia and
anxiety in children. Anesthesiology 93:676–684
Kain ZN, Mayes LC, O’Connor TZ, Cicchetti DV (1996)
Preoperative anxiety in children: predictors and outcomes.
Arch Pediatr Adolesc Med 150:1238–1245
Korkman M, Kirk U, Kemp S (1997) Manual for the NEPSY: a
developmental neuropsychological assessment. The Psycholog-
ical Corporation, New York
Laycock GJA, McNicol LR (1988) Hypoxamia during induction of
anaesthesia—an audit of children who underwent general
anaesthesia for routine elective surgery. Anaesthesia 43:981–
984
Naito M (1990) Repetition priming in children and adults: age-
related dissociation between implicit and explicit memory. J
Exp Child Psychol 50:462–484
O’Boyle CA, Barry H, Fox E, Harris D, McCreary C (1987)
Benzodiazepine-induced event amnesia following a stressful
surgical procedure. Psychopharmacology 91:244–247
385
Peterson C (1999) Children’s memory for medical emergencies: 2
years later. Dev Psychol 35:1493–1506
Rabinowitz JC, Craik FI, Ackerman BP (1982) A processing
resource account of age differences in recall. Can J Psychol
36:325–344
Roediger HL III, McDermott KB (1993) Implicit memory in
normal human subjects. In: Spinnler H, Boller F (eds)
Handbook of neuropsychology. Elsevier, Amsterdam, pp 63–
131
Saarnivaara L, Lindgren L, Klemola UM (1988) Comparison of
chloral hydrate and midazolam by mouth as premedicants in
children undergoing otolaryngological surgery. Br J Anaesth
61:390–396
Salonen M, Kanto J, Iisalo E, Himberg JJ (1987) Midazolam as an
induction agent in children: a pharmacokinetics and clinical
study. Anesth Analg 66:619–624
Smith MT, Eadie MJ, O’Rourke-Brophy T (1981) The pharmaco-
kinetics of midazolam in man. Eur J Clin Pharmacol 19:271–
278
Weldon BC, Watcha MF, White PF (1992) Oral midazolam in
children: effect of time and adjunctive therapy. Anesth Analg
75:51–55
Wilhelm FH, Roth WT (1997) Acute and delayed effects of
alprazolam on flight phobics during exposure. Behav Res Ther
35:831–841
Wilton NCT, Leigh J, Rosen DR, Pandit UA (1988) Preanesthetic
sedation of preschool children using intranasal midazolam.
Anesthesiology 69:972–975
386