Received: 2 March 2001
Accepted: 30 October 2001
Published online: 6 December 2001
© Springer-Verlag 2001
Abstract Objective: To define the
incidence of recall and dreams during
analgosedation in critically ill pa-
tients. Design: Prospective clinical
study. Setting: Anaesthesiological in-
tensive care unit (ICU) in a universi-
ty hospital. Patients and partici-
pants: Two hundred and eighty-nine
critically ill patients, who either ar-
rived intubated and sedated at the
ICU or required intubation, mechani-
cal ventilation, and sedation during
their ICU stay. Interventions: none.
Measurements and results: The pa-
tients were interviewed 48–72 h after
discharge from the ICU. By a struc-
tured interview they were asked
whether they recalled any event be-
fore they had regained consciousness
at the ICU. Moreover they were
asked for dreams. Descriptive statis-
tics: 64.7% of all patients did not re-
call any event, before they regained
consciousness. However, 17% (n=49)
of all patients indicated that they re-
membered the tracheal tube or being
on the ventilator, before they woke
up. Some patients (21.1%) reported
dreams or dreamlike sensations.
Some patients (9.3%) recalled night-
mares, while 6.6% reported halluci-
nations. Conclusions: Critically ill
patients reported a high incidence of
recall for unpleasant events, which
they thought to have taken place be-
fore they regained consciousness.
The patients, who stayed longer than
24 h at the ICU, indicated vivid
memory for nightmares and halluci-
nations. Further studies are suggested
to evaluate: 1) whether there is an
impact of the present findings on out-
come; and 2) whether clinical scores
for sedation or neurophysiological
monitoring help to define the exact
time, when recall happens, in order to
guide therapeutic intervention.
Keywords Recall · Dreams ·
Hallucinations · Intensive care ·
Critically ill patients · Sedation ·
Intensive Care Med (2002) 28:38–43
J. Schulte am Esch
Incidence of recall, nightmares,
and hallucinations during analgosedation
in intensive care
Recollection about the time spent in the intensive care
unit (ICU) in critically ill patients who require sedation
and assisted ventilation became a research focus over the
course of recent decades. Several personal observations
and an increasing number of group studies dealing with
the recollection of patients have been published. Early
reports date back to the 1960s and 1970s, when in post-
cardiotomy patients psychiatric complications and the
impact of environmental stressors were described [1, 2].
Recent studies on patients’ reactions during intensive
care management have focused on various aspects such
as ventilator-associated factors, pain, environmental
stressors or patients’ emotional response [3, 4, 5, 6, 7].
Long-term follow up studies in ICU patients indicate
that a vivid recollection of unpleasant memories (being
on the ventilator, being in pain or having nightmares)
persists in some patients and results in an impairment of
psychosocial functioning [8, 9, 10].
The level of sedation that is deemed optimal during
critical care treatment with mechanical ventilation has
I. Rundshagen (✉)
University Hospital Charité,
Department of Anaesthesiology, Campus
Mitte, Schumannstrasse 20/21,
10117 Berlin, Germany
Department of Psychology,
University of Michigan, Ann Arbor, USA
C. Wegner · J. Schulte am Esch
Department of Anaesthesiology,
University Hospital Eppendorf,
markedly changed during recent decades [3, 11, 12].
Specific guidelines are rarely used in the every day rou-
tine. Sedation and analgesia in the ICU setting generally
aim at protecting patients from the numerous stressful
and noxious stimuli, as well as to provide anxiolysis,
nocturnal sleep and, sometimes, amnesia. It is also used
to decrease sympathetic tone and oxygen consumption in
critical conditions, to indirectly stabilise haemodynam-
ics, and to treat agitation and motor activity, which are
potentially harmful. Some 20 years ago, patients had to
be deeply sedated and even paralysed in order to tolerate
mechanical ventilation . Newer ventilatory modes
are now partly controlled by the patient or mimic more
physiological breathing patterns, thus permitting a lower
level of sedation. It is not known, however, whether the
altered concept of sedation results in an increased memo-
ry for unpleasant events during intensive care .
Different questionnaires have been used to assess
memory in patients after intensive care, in short-term
and in long-term follow-up studies [6, 14, 15]. In the
present study, we focus on the question whether the pa-
tients recalled any event before they regained conscious-
ness at the ICU. From studies with anaesthetised patients
it is known that wakefulness (awareness) during general
anaesthesia with explicit recall is of great concern to the
patients. The incidence and possible sequelae of subse-
quent recall of intra-operative events have been high-
lighted [16, 17]. In ICU patients it is not known whether
they in fact experience wakefulness during sedation. The
purpose of the present prospective investigation was to
evaluate the incidence of wakefulness and dreams in crit-
ically ill patients. In order to assess the specific situation
of ICU patients adequately, we modified the standard in-
terview technique, which is used to explore awareness in
anaesthetised patients [18, 19].
Materials and methods
The study was approved by the local Ethics Committee. Patients
who had been admitted to the ICU ANITO at the Department of
Anaesthesiology of the University Hospital Eppendorf (a tertiary
care and trauma centre of the University of Hamburg) in 1998
were included in the study if they were sedated and mechanically
ventilated on arrival at the ICU, or if they required sedation and
intubation during their intensive care treatment. Only those pa-
tients were included in the study who were at least 18 years old,
and who were able to communicate in German or English, and
who had given written informed consent to participate.
Patients were excluded from consideration for the present
study, if they had a previous history of a psychiatric disorder or if
they were in a persistent state of mental confusion after their trans-
fer to a general ward.
A female medical student (5th year), who was especially
trained and not involved in the ICU treatment, interviewed the pa-
tients on the 2nd or 3rd day after they were transferred to the gen-
eral ward. During the study period the nursing and medical staff of
the ICU were not informed to avoid any bias.
The structured interview included a standard set of questions
(Table 1, modified from [18, 19]). The patients were asked wheth-
er they had any memory for events after loss of consciousness and
before mental recovery at the ICU. In case of a positive answer,
they were classified as having recall (wakefulness). In addition, all
patients were asked whether they remembered any dreams.
As independent factors, we registered demographics, diagno-
sis, and the SAPSII-score at admission to the ICU. In addition, the
duration of the ICU treatment, mechanical ventilation, and seda-
tion were recorded. Medication used for sedation and analgesia,
the use of antipsychotic drugs, a patient’s previous history of alco-
hol or drugs, or any remarks about states of confusion made in the
patient’s every day chart were documented as well.
To evaluate the incidence of explicit recall or dreams in ICU pa-
tients during analgosedation the interviewees were categorised
based on information provided. Three distinct groups were identi-
fied: 1) patients without recall; 2) patients with recall; and 3) pa-
tients without recall but dreams. Using the group membership as
independent variable, group differences in the recorded informa-
Table 1 List of questions to evaluate patients’ memory (modified from Liu et al. and Ranta et al. [18, 19])
What is the last thing you remembered before you lost consciousness (at the ICU, in the operating room or at the accident)?
What is the first thing you remember when you woke up again?
Do you remember anything in between these two periods?
Do you remember any dreams in between?
What is the worst thing you remember during your stay at the ICU?
What is the most pleasant memory of the ICU?
Have you been prepared for your treatment at the ICU?
Additional questions, if question 2 or 3 was answered with “Yes”:
1. What kind of sensory perception did you have: auditory, visual or sensory sensations, pain, paralysis?
2.Did you feel or think anything about it?
3.Did you feel anything in your mouth or throat?
4.Did you think you had dreamt?
5. Did you have any idea how long it lasted?
6. Was it day or night?
7. Did you try to contact anybody?
8. Did this event had any consequences for you?
9.Did you talk to anybody from the personnel or your relatives/friends about your experience?
10. Did this experience influence your attitude towards your treatment at the ICU?
tion was tested for significance on an exploratory basis. Due to
the heterogeneity of variance non-parametric statistics (Kruskall-
Wallis test) were applied. Stepwise logistic regression analysis
was used to identify a robust set of predictors for group member-
ship (dichotomised: 0=patients without recall, 1=patients with re-
call and/or dreams).
All kinds of surgical patients, who required postoperative inten-
sive care (with the exception of cardiosurgical or neurosurgical
patients, who were admitted to other ICUs of the hospital), were
admitted to the ICU ANITO.
The ICU is staffed by a senior doctor, one specialist of anaes-
thesiology, four residents of anaesthesiology (serving in 8-h shift
work), and one surgical resident. The ten beds are divided in two
bedrooms with big windows, and the nurse/patient ratio is 1:2.
The staff also includes two physiotherapists, an ICU technologist,
and a secretary. Regular consultations were performed by the sur-
geons and microbiologists. Neurologists, psychiatrists, and other
specialist are consulted whenever needed.
Routine patient management, including analgesic and sedative
regimes, was maintained during the study. There were no common
sedation guidelines for all patients, but usually propofol – supple-
mented by opioids – was used in the immediate postoperative
course, while midazolam in combination with sufentanil was fre-
quently used in long-term sedation. Clonidine was often used dur-
ing the weaning periods. Sedation was fine-tuned by the nurses to
the target level set three times per day at the ward rounds. In gen-
eral, a day and night rhythm was pursued.
A total number of 329 patients were approached for the
purpose of the study. Data were collected from 289 patients.
Nineteen patients refused to participate. Five patients could
not be interviewed due to mental confusion. Eleven patients
died before the scheduled interview. Five patients could not
communicate because of language problems.
Table 2 summarises demographics and the details
about the ICU stay of the 289 patients, whose date were
available for the following analyses. Two hundred and
twenty patients (76.1%) were elective admissions, while
69 (23.9%) were emergency admissions. All participants
were surgical patients except one medical, who required
intubation and sedation because of bronchial asthma. An-
other three patients did not require surgery, because two
were readmitted after surgery due to respiratory insuffi-
ciency and one patient was transferred from another hos-
pital because of a thoracic trauma. All the other patients
came via the operating theatre to the ICU except one pa-
tient, who required surgery later. While 162 patients were
transferred to a general ward within the first 24 h, 127 pa-
tients remained at the ICU. The medication used for seda-
tion and/or analgesia is shown in Figs. 1 and 2.
The vast majority of the patients (233, i.e., 80.6%) re-
membered that they had got the information about their
admittance to the ICU prior to anaesthesia (90.9% of the
patients with elective admission, 49.3% patients with
emergency admission). Fifty-six (16.3%) patients did not
remember being given any information or were uncertain.
One hundred and eighty-seven patients (64.7%) stated
that they did not remember anything of the period be-
tween loss of consciousness and regaining it during their
ICU stay. Among them 16 patients did not even remem-
ber having been at the ICU at all (5.5% of all patients).
Of all patients, 49 (17%) described some memory of ei-
ther the tracheal tube, being on the ventilator or the pro-
cedure of extubation (n=5), before they thought they re-
gained consciousness. This group of patients was classi-
fied as having experienced recall/wakefulness. Thirty-two
of them stayed less than 1 day at the ICU (i.e., 11% of all
Table 2 Demographics and parameters of the ICU treatment (me-
ICU stay (days)
122 female /167 male
Fig. 1 Medicaments given for sedation for patients with an ICU
stay of less or equal 1 day (n=167): numbers indicate the number
Fig. 2 Medicaments given for sedation for the patients with an
ICU stay of more than 1 day (n=127): numbers indicate the num-
ber of patients. Numbers in the bars indicate the duration (days;
median/range) of continuous i.v. drug application
patients or 19.8% of the patient, who stayed less than one
day at the ICU, respectively; Table 3). Thirteen of them
did not receive any sedatives or analgesics at the ICU be-
fore they were extubated, after being transferred from the
operation theatre. Only eight of the patients with wake-
fulness reported that they remembered dreams, too.
Some of the patients [21.1% (n=61)] indicated that they
remembered dreams (Table 3). Twenty-seven (9.9%) re-
ported unpleasant dreams and classified them as night-
mares. In addition, 19 patients (6.6%) reported unpleas-
ant hallucinations. Among them, nine patients experi-
enced very frightening true-to-life delusions. Either
somebody was trying to kill or kidnap them or they were
involved in terrifying military actions. An example of a
paranoid delusion, which one patient recalled, is given.
However, at the time of the interview most of the pa-
tients indicated that they were not sure whether these
events really had happened or not. They were reluctant
to talk about these experiences and looked anxious being
reminded of them. Eight patients reported remembrance
of pleasant dreams, while another seven patients did re-
member emotionally indifferent dreams.
A 59-year-old male patient suffered from peritonitis that
required 67 days of intensive care treatment. He reported
on a doctor giving him an i.v. injection in order to kill
him. His wife performed a tracheotomy to save his life.
At the time of the interview, he was still convinced that
this happened in reality. In spite of this experience he ex-
pressed his satisfaction with the ICU treatment.
Results of non-parametric statistics
Based on their reports patients were divided into three
distinct groups: group 1=no recall (n=187), group 2=ex-
plicit recall (n=49), group 3=dreams/hallucinations
(n=53). Patients with dreams and recall (n=8) were allo-
cated to the group 2.
The Kruskal-Wallis tests revealed significant differ-
ences between the three groups for age (P<0.001), SAPS
score (P=0.014), alcohol abuse (P=0.024), elective or
emergency admission (P=0.033), and ICU-treatment-
related factors [ICU stay (P<0.001), duration of assisted
ventilation (P<0.001) and sedation (P<0.001), duration of
sufentanil (P<0.001), propofol (P<0.001) or clonidine ap-
plication (P<0.001); Table 4]. Patients, who had experi-
enced wakefulness were younger than the patients of the
two other groups. Patients who remembered dreams, on
the other hand, stayed longer at the ICU and needed lon-
ger assisted ventilation and sedation than the patients of
the other two groups. Their SAPS scores at admittance
tended to be higher than the ones in the other patients.
A stepwise logistic regression analysis revealed that the
duration of the ICU treatment was the best predictor for
the occurrence of recall or dream. Compared to the over-
all correct prediction rate of 63.2% by chance the predic-
tion is improved by 7.2% (70.4%), when ICU days are
taken into account. The longer the patients needed inten-
sive care treatment the higher the risk of having recall or
dreams was. Predictability was slightly further increased
when age was included as a second predictor (72.2%
correct prediction). As a tendency, younger patients
seem to be more prone to experience recall or dreams.
The design of the present study differed from previous
studies in its systematic exploration of patients’ experi-
ence using a standard questionnaire for evaluating pa-
tients after general anaesthesia. Recall (wakefulness)
was defined as the ability of the patients to recall any
event which they believe occurred during the period be-
tween loss and recovery of consciousness. We docu-
mented in 17% of the patients symptoms of wakefulness.
However, this estimate is not based on objective criteria
but on subjective patients’ reports. Given the content of
their memory, events during the weaning phase from the
respirator are most likely to be recalled, when sedation
Table 3 Incidence of recall and dreams
ICU duration<1 day (n=162) >1 day (n=127)
Table 4 Demographic and ICU related factors which differ among
patients without recall (group 1), with recall (group 2), and dreams
(group 3). Median; minimum and maximum are given
ICU stay (days)
Elective admissions 148
either was stopped or adjusted to a lower level. While
most of the patients only recalled being mechanically
ventilated, five patients recalled the extubation and de-
scribed it as an extremely unpleasant experience. This is
in line with the findings of Puntillo et al. who reported
on two patients, for whom the removal of the chest tube
was extremely painful, “it felt like pulling my guts out”
. Novaes et al. indicated in their study that pain, the
impossibility to sleep, and having tubes in the nose or
the mouth were considered the most severe physical
stressors in critically ill patients .
There are previous studies evaluating general aspects
of memory in ICU patients. Turner et al. documented in
a group of 100 medical ICU patients that the most fre-
quently reported unpleasant experience was arterial gas
sampling (48% of the patients) and tracheal suctioning
(30 of 68 patients) . However, 36 of the patients
did not receive any drugs for sedation or analgesia.
Bergbohm-Engberg et al. documented a high incidence
(37–59%) of recall for the respirator treatment in 304
medical, surgical, and trauma ICU patients . Unfortu-
nately, there was no information about the regimen of se-
dation. From their subsequent studies they concluded
that nursing care might be more efficient in helping pa-
tients to cope with the stressful respirator treatment than
heavy i.v. sedation or analgesia . This is in line with
the study of Russell, who pointed out that psychological
problems after discharge from the ICU might have been
prevented by improved communication between staff
and patients . The patients of the present study did
not report spontaneously that they remembered any com-
munication with the ICU staff at the time when they ex-
perienced wakefulness. Therefor it remains unclear
whether they would benefit from a better communica-
tion. In general, at the present time it remains speculative
whether experiencing wakefulness, as demonstrated in
this study, has an impact on outcome or not.
While explicit recall was more frequent in the patients
who stayed less than 1 day at the ICU, long-term critical-
ly ill patients reported nightmares, hallucinations, and/or
paranoid delusions. This observation is in agreement
with previous studies which indicated that states of con-
fusion are a common problem in long-term ICU patients.
Pouchard et al. documented delirium or hallucinations in
33 of 43 ventilated medical ICU patients during the
weaning phase . The incidence was significantly
higher in the patients sedated more than 5 days. Daffurn
et al. interviewed 54 ICU patients, either requiring inten-
sive care due to multi-trauma or major abdominal or tho-
racic surgery . Three months after discharge seven pa-
tients remembered nightmares or hallucinations. Jones et
al. found a high incidence of memory for hallucinations
and nightmares in 159 patients, who stayed longer than
24 h at the ICU [9, 23]. Almost all of their patients were
emergency admissions and presented a vivid memory
even 2 months after discharge from the ICU. In long-
term survivors of the acute respiratory distress syndrome
(ARDS), an incidence between 63% and 75% for night-
mares was documented [10, 24]. Puntillo reported on
two patients who remembered hallucinations among 22
patients from a medical ICU . However, there are
some studies about the patients’ recollection about their
ICU stay, in which nightmares, hallucinations, and delu-
sional memory were not documented [6, 20, 25]. Cer-
tainly this is due to differences in the study designs.
Some studies were sampling different ICU patient popu-
lations, others used a less precise interview technique.
Recently, Jones et al. published promising data about the
validation of a new tool to assess memory in ICU pa-
tients (ICUM tool), which certainly will be helpful to
standardise future research .
There are several reasons why patients become prone
to psychological disturbances during their ICU treat-
ment. Despite the significant changes in the set-up of
ICU environments, the risk of physical or psychoreactive
disturbances persists [1, 26]. Early reports date back to
1965 when in postcardiotomy patients psychiatric com-
plications were described . In subsequent studies a
new term was created, the ICU psychosis or the ICU
syndrome . There is no recent study indicating the
incidence of mental disturbances in surgical ICU patients
based on a strict psychiatric evaluation. Therefore the
true incidence is not known. As early as 1972, Blacher
showed in eight of 12 apparently normal patients psychi-
atric disturbances after cardiotomy which he described
as the “hidden psychosis” . He stressed the impor-
tance of a careful psychiatric examination. However, our
findings were based on an the patients’ reports only.
Even more, we had to exclude the patients who died and
the patients who were still confused 3 days after dis-
charge from the ICU. Despite this, we demonstrated a
high incidence of patients’ recollection of nightmares,
hallucinations, and paranoid delusions. Thus the real in-
cidence of psychiatric disorders might even be higher.
Long-term follow-up studies of critically ill patients
indicated that the quality of life can be impaired by the
development of a post-traumatic stress disorder (PTSD).
Schelling et al. documented in survivors of the acute res-
piratory distress syndrome, that patients who reported
multiple traumatic episodes (pain, nightmares, respirato-
ry distress) during intensive care, were likely to show the
lowest quality of life with maximal impairment in psy-
chosocial functioning . They suggested that the
recollection of traumatic ICU experiences might result in
PTSD. Subsequently Jones et al. examined the relation-
ship between memory for the ICU and the levels of anxi-
ety after ICU discharge in 45 patients in detail .
They suggested that even relatively unpleasant memories
of real events during critical illness give some protection
from developing anxiety and PTSD-related symptoms.
They proposed that the development of PTSD may be re-
lated more to recall of delusions alone. However, wheth-