Intensive Care Med (2003) 29:2230–2238
DOI 10.1007/s00134-003-2033-3
Claudia D. Spies
Hilke E. Otter
Bernd Hüske
Pranav Sinha
Tim Neumann
Jordan Rettig
Erika Lenzenhuber
Wolfgang J. Kox
Received: 10 July 2002
Accepted: 5 August 2003
Published online: 14 October 2003
© Springer-Verlag 2003
This study was sponsored in part by the
German Research Society
(DFG-SP 432/1-1 and 1-2)
Electronic Supplementary Material
Supplementary material is available in the
online version of this article
http://dx.doi.org/10.1007/s00134-2033-3.
C. D. Spies (✉) · H. E. Otter · B. Hüske
T. Neumann · W. J. Kox
Department of Anesthesiology and
Intensive Care Medicine,
Charité-Universitätsmedizin Berlin,
Campus Charité Mitte,
Schumannstrasse 20/21, 10117 Berlin,
Germany
e-mail: claudia.spies@charite.de
Tel.: +49-30-450-531052
Fax: +49-30-450-531911
P. Sinha
Institute of Clinical Chemistry,
Charité-Universitätsmedizin Berlin,
Berlin, Germany
J. Rettig
School of Medicine,
University of Connecticut, Conn., USA
E. Lenzenhuber
Department of Anesthesiology and
Intensive Care Medicine,
Charité-Universitätsmedizin Berlin,
Campus Benjamin Franklin,
Berlin, Germany
E. M. Sellers
Departments of Pharmacology, Medicine,
and Psychiatry, University of Toronto,
Toronto, Ont., Canada
ORIGINAL
Edward M. Sellers
Alcohol withdrawal severity is decreased
by symptom-orientated adjusted bolus therapy
in the ICU
Abstract Objective: To examine the ference 6 days) due to a lower inci-
effect of bolus vs. continuous infu- dence of pneumonia (26% vs. 43%).
sion adjustment on severity and du- Conclusions: We conclude that
ration of alcohol withdrawal syn- symptom-orientated bolus-titrated
drome (AWS), the medication re- therapy decreases the severity and
quirements for AWS treatment, and duration of AWS and of medication
the effect on ICU stay in surgical in- requirements, with clinically rele-
tensive care unit (ICU) patients. vant benefits such as fewer days of
Design and setting: Prospective ran- ventilation, lower incidence of pneu-
domized, double-blind controlled tri- monia, and shorter ICU stay.
al in a surgical ICU. Patients: 44 pa-
tients who developed AWS after ad- Keywords Alcohol withdrawal
mission to the ICU. Interventions: syndrome · Symptom-orientated
Patients were randomized to either therapy · Surgical intensive care
(a) a continuous infusion course of unit · Clinical Institute of
intravenous flunitrazepam (agita- Withdrawal Assessment · Critically
tion), intravenous clonidine (sympa- ill patients · Pneumonia
thetic hyperactivity), and intrave-
nous haloperidol (productive psy-
chotic symptoms) if needed (infu-
sion-titrated group), or (b) the same
medication (flunitrazepam, cloni-
dine, or haloperidol) bolus adjusted
in response to the development of
the signs and symptoms of AWS
(bolus-titrated group). Measurements
and results: The administration of
“as-needed” medication was deter-
mined using a validated measure of
the severity of AWS (Clinical Insti-
tute of Withdrawal Assessment). Al-
though the severity of AWS did not
differ between groups initially, it sig-
nificantly worsened over time in the
infusion-titrated group. This required
a higher amount of flunitrazepam,
clonidine, and haloperidol. ICU
treatment was significantly shorter in
the bolus-titrated group (median dif-

Introduction
Every fifth patient admitted to hospital proves to be posi-
tive for alcoholism [1]. Following major surgery chronic
alcoholics have an increased morbidity and mortality
rate [2, 3, 4]. In addition to infections, cardiac, and
bleeding complications, chronic alcoholics are at risk of
developing alcohol withdrawal syndrome (AWS) [3]. If
not treated prophylactically, one-half of chronic alcohol-
ics develop AWS [3, 5]. However, despite preventive
treatment approximately one-fourth of these patients de-
velop AWS postoperatively [3, 5]. Moreover, those who
develop AWS have a prolonged ICU stay [6].
The severity of AWS in ICU patients following major
surgery is higher than that in psychiatric patients as eval-
uated by the validated Clinical Withdrawal Assessment
for Alcohol, revised scale (CIWA-Ar) initially developed
for psychiatric patients [6, 7, 8] but which has also been
used in ICU settings [9]. The medication requirements for
the treatment of AWS in ICU patients are usually higher
[3]. The first-line of therapy for AWS is the administra-
tion of benzodiazepines [10]. When used for AWS, ben-
zodiazepines are generally administered on predeter-
mined dosage schedules for several days, often in gradu-
ally tapering doses. While monotherapy with benzodiaze-
pines is usually administered for AWS in psychiatric pa-
tients, 64% of ICU patients with AWS receive drug com-
binations [3, 11, 12]. The most common adjunctives to
benzodiazepines are clonidine (used for autonomic signs)
and haloperidol (for productive psychotic states such as
hallucinations) [10]. Intravenous benzodiazepines and in-
travenous clonidine are usually administered continuous-
ly, while intravenous haloperidol is added either accord-
ing to a predetermined dosing schedule or at a constant
infusion rate [3]. In contrast to these conventional ICU
infusion regimes, bolus therapy is currently available for
detoxification to decrease the amount of benzodiazepines
and the treatment period [7]. It is unclear whether bolus
therapy would also be beneficial for ICU patients. Over-
sedation may prolong ICU treatment [13] whereas lack of
sedation may put the patient at risk for, for example, dis-
location of tubes and prolonged weaning [14]. In AWS
the adjustment of continuous infusions is usually delayed,
resulting either in comatose patients with the conse-
quence of prolonged ICU treatment or in recurrent symp-
toms of alcohol withdrawal. There is a lack of random-
ized studies on sedation in the ICU [15]; to our knowl-
edge, only two randomized controlled trials have exam-
ined the treatment of AWS in ICU patients [11, 16].
The aims of this study were to investigate bolus and
continuous infusion therapy for AWS with respect to: (a)
the severity and duration of alcohol withdrawal syn-
drome (primary outcome measure), (b) benzodiazepine,
clonidine, and haloperidol requirements (secondary out-
come measure), and (c) duration of ICU treatment (ter-
tiary outcome measure).
2231
Patients and methods
The study protocol was approved by the institutional ethics com-
mittee. All patients fulfilled the criteria of the CIWA-Ar (>20) for
the requirement of pharmacotherapy [8]. The CIWA-Ar is a vali-
dated, reliable measure of the current severity of alcohol with-
drawal composed of ten items (see Electronic Supplementary Ma-
terial). Patients were included within 4 h after the onset of alcohol
withdrawal. Exclusion criteria were age under 18 years, pregnan-
cy, intubation and ventilation at the onset of AWS, continuous se-
dation at the onset of AWS, concurrent acute medical diagnosis,
i.e., infections and hypoxia, hemoglobin lower than 8 g/dl, bleed-
ing at the onset of AWS, acute angina, metabolic or endocrinologi-
cal disorders, head injury, intoxications or psychiatric illness, a
history of seizures of any cause, corticosteroid use or chemothera-
py, current use or withdrawal from clonidine, β-blockers, or halo-
peridol and withdrawal from or known misuse of opiates, benzodi-
azepines, or barbiturates. Patients who were diagnosed as being al-
cohol-dependent before the onset of AWS were excluded (Fig. 1),
receiving prophylactic treatment [3, 17]. After inclusion in the
study patients with AWS treatment failure defined as a persisting
AWS (CIWA-Ar >20) for longer than 8 h despite administration of
rescue medication were also excluded.
Protocol
In this prospective randomized (per envelope) controlled study
300 patients with trauma or gastrointestinal surgery, fulfilling cri-
teria for possible alcohol abuse (not dependence) as defined by the
Diagnostic and Statistical Manual of Mental Disorders, fourth edi-
tion (DSM-IV) [18], and with an alcohol consumption higher than
60 g/day, were included. In the case of elective surgery patients
gave their written informed consent before surgery; in patients
with severe trauma relatives or their legal representatives gave
consent on admission to the ICU. Physical examination and liver
function tests were performed, and complete cell count, mean cor-
puscular volume, and carbohydrate-deficient transferrin were as-
sessed to establish any diagnosis of alcohol abuse [3, 17, 19, 20].
The alcoholism-related questionnaire CAGE (see Electronic Sup-
plementary Material) was administered to patients (surgery cases)
or their relatives (trauma cases) and readministered after success-
ful treatment of AWS from the patients upon reorientation [3]. The
maximal value of the two CAGE assessments was taken. Prior to
surgery and on admission to the ICU all patients had only admit-
ted alcohol abuse and not dependence [3]. In the case of alcohol
dependence (according to DSM-IV) known before onset of AWS
patients were given prophylactic treatment and were excluded.
Of the original 300 patients 107 developed AWS following ad-
mission to the ICU (Fig. 1); 59 of these were not eligible due to
exclusion criteria. In this respect intubation was a crucial aspect of
this study because before being included patients had to be extu-
bated in order to exclude all differential diagnoses other than
AWS. Therefore a total of 48 patients were finally included and
randomized in either an infusion-titrated group (ITG) or a bolus-
titrated group (BTG; Fig. 1). A further four patients were later ex-
cluded, two due to contradictory study related histories (one ITG,
one BTG), one (ITG) for additional requirement of rescue medica-
tion within 8 h of treatment, and one (ITG) who died from pulmo-
nary embolism 8 h after enrollment. Finally, data from 44 patients
were evaluated (Fig. 1). Basic patient characteristics and alcohol-
ism-orientated diagnoses did not differ between groups (Table 1).
There were no significant differences in baseline AWS character-
istics between the treatment groups.
2232

Fig. 1 Decision tree of exclu-

sion/inclusion procedure.
DSM-IV Diagnostic and Statis-
tical Manual of Mental Disor-
ders, fourth edition; ICU inten-
sive care unit; AWS alcohol
withdrawal syndrome

Table 1 Baseline patient char-

acteristics and alcoholism- BTG (n=23) ITG (n=21) p
related diagnosis in the bolus-
titrated (BTG) and infusion- Age (years) 49 (41–58) 41 (33–68) 0.29
titrated (ITG) groups; absolute Sex: M/F 18/5 18/3 0.53
number (parentheses frequen- Body mass index 25 (23–26) 25 (24–30) 0.37
cy) or median (parentheses in- Carcinoma 13/23 (57%) 10/21 (48%) ≥0.99
terquartile range) (APACHE II Trauma 10/23 (44%) 11/21 (53%) ≥0.99
Acute Physiology and Chronic Smokers 19/23 (83%) 18/21 (86%) 0.72
Health Evaluation II, adm. on Coronary artery disease 2/23 (9%) 1/21 (5%) 0.76
admission to ICU) Chronic obstructive lung disease 12/23 (52%) 6/21 (29%) 0.37
Duration of anesthesia (h) 11 (6–14) 10 (6–17) 0.60
APACHE II 9 (5–14) 12 (7–16) 0.21
Ethanol (g/d) 90 (65–280) 90 (80–240) 0.64
CAGE (mg)a 3 (3–4) 3 (2–4) 0.46
Blood alcohol, adm. (mg/l) 0 (0–0.2) 0 (0–0.3) 0.93
Carbohydrate-deficient transferrin (fl)b 9.4 (5.8–11.7) 7.6 (4.4–16.7) 0.92
Mean corpuscular volume, adm. (U/l)c 93.1 (87.7–96.8) 92.8 (89.0–96.4) 0.77
γ-Glutamyl transferase, adm. (U/l)d 28 (23–62) 20 (10–75) 0.31
a Alcoholism-relatedquestionnaire (see Electronic Supplementary Material), administered before and
after AWS therapy, maximum value determined
b Cutoff 9 mg/l
c Cutoff 92 fl
d Cutoff 28 U/l

Assessment and data monitoring
Basic patient characteristics and the Acute Physiology and
Chronic Health Evaluation II were recorded. In addition to the
CIWA-Ar, the Glasgow Coma Scale (GCS) and the Ramsay Seda-
tion Scale (RSS; see Electronic Supplementary Material) [13, 21]
were recorded. The CIWA-Ar was administered by medical re-
search staff trained by videotapes from the Addiction Research
Foundation, Clinical Institute in Toronto, Ontario [8]. To monitor
the subjects’ response to therapy the vital signs, RSS, and
CIWA-Ar were assessed at baseline every 10 min in the first
60 min of AWS therapy, followed by every hour for the first 8 h
and thereafter every 4 h. Each item of the CIWA-Ar was set at 0
if determination was not possible following intubation. Patients
were monitored prospectively for the development of hallucina-
tions, seizures, and delirium tremens as soon as the diagnosis of
alcohol abuse was established. These patients usually do not re-
quire pharmacological treatment to prevent AWS [6] as they had
denied alcohol dependence prior to surgery and admission to ICU
[18].
Treatment regimens
ITG or BTG therapy (Fig. 1) was initiated according to the median
of previous studies in surgical patients [3, 10, 11] and clinical rou-
tine [22]. An initial cumulative dose of flunitrazepam boluses was
given to control AWS. After 10 min depending on symptoms addi-
tional boluses of adjunctive agents were administered, haloperidol
for hallucinations and clonidine for autonomic signs, and another
bolus of flunitrazepam in the case of agitation. This medication re-
gime was repeated in both groups until a CIWA-Ar score lower
than 20 was achieved. As we could not exclude seizures and kind-
ling mechanisms with a subsequently severe course [3, 10], a non-

Fig. 2 Treatment decision tree.
CIWA-Ar Clinical Withdrawal
Assessment for Alcohol, re-
vised scale; RSS Ramsay Seda-
tion Score; asterisk respective
drug for safety reasons
blinded continuous flunitrazepam infusion was then started for
both groups (Fig. 2). Afterwards, the two different treatment pro-
cedures were carried out. All infusions were started in a dose de-
pending on the initial necessary cumulative dose of boluses of the
respective drug (Fig. 2).
In the ITG infusions contained the respective drug, while the
syringes for the bolus administration in this group were labeled
with the drug but contained placebo (Fig. 2). According to CIWA-
Ar and RSS, the following modifications in the continuous infu-
sion were carried out: (a) the continuous infusion of each drug was
increased by 20% if symptoms persisted longer than 10 min, (b)
flunitrazepam infusion rates were decreased by 20% if the CIWA-
Ar was lower than 10 and RSS lower than 4, (c) clonidine was de-
creased by 20% in the event of bradycardia (fewer than 60
2233
beats/min) or hypotension (MAP lower than 60 mmHg), and (d)
all drugs were reduced by 10% per day if the CIWA-Ar remained
below 10 for 24 h.
In the BTG the continuous infusions were titrated in the same
way as for the ITG, containing placebo, except for the continuous
flunitrazepam infusion containing the respective drug for the safe-
ty reasons mentioned above. Additional boluses of flunitrazepam,
haloperidol, or clonidine (dose according to the initial cumulative
bolus and clinical signs) were administered immediately after a
CIWA-Ar higher than 20 was observed. The syringes for bolus ad-
ministration in the BTG were labeled with flunitrazepam, haloper-
idol, or clonidine and contained the respective drug. The infusion
rate of the placebo infusions was modified in the same way as de-
scribed for the ITG (Fig. 2).
2234

In both groups maintenance therapy was adapted to achieve a

CIWA-Ar lower than 10 (first priority) and a RSS between 2 and 4
(second priority). If attaining a CIWA-Ar below 10 was not possi-
ble without the GCS being 8 or lower, the patient was intubated. If
any symptoms were present, additional boluses were administered.
After each bolus the infusion rate of haloperidol, clonidine, and
flunitrazepam was modified (Fig. 2).

Rescue medication

As rescue medication propofol was allowed in all patients. This

medication was not blinded and was titrated as needed
(50–400 mg) to manage any symptoms that may have threatened
the patient (Fig. 2). Treatment failure was determined if AWS per-
sisted despite treatment and propofol rescue medication, i.e., if the
CIWA-Ar was not lower than 20 within 8 h. These patients were
excluded from the study.

Outcomes

The primary outcome measures were severity and duration AWS.

Severity was determined according to the CIWA-Ar score and du-
ration as the number of days on which the maximal CIWA-Ar ex-
ceeded 20. Secondary outcomes included benzodiazepine, cloni-
dine, and haloperidol requirements. Tertiary outcomes were dura-
tion of ICU treatment. Infections were determined by Centers for
Disease Control criteria [23]. In addition, all patients with pneu-
monia had a new pulmonary infiltrate as revealed by radiography
of the chest. Sepsis was defined according to the Consensus Con-
ference of the American College of Chest Physicians/Society of
Critical Care Medicine [24]. Bleeding was diagnosed if the patient
required blood transfusions or surgical intervention for persistent
bleeding. Cardiac complications included arrhythmias, congestive
heart failure, and myocardial ischemia. Drug-related complica-
tions due to clonidine (a–d) and haloperidol (d) included: (a) bra-
dycardia (heart rate <40 beats/min) requiring therapy with positive
chronotropes, (b) hypotension requiring positive inotropic/vaso-
pressor support not reversible with volume substitution and not re-
lated to other disease categories such as sepsis, (c) conduction ab-
normalities, (d) QTc prolongation more than 20% over baseline
that was not due to hypokalemia (<3.5 mmol/l), hypomagnesemia
(<2.2 mmol/l), or bradycardia. Fig. 3 a Clinical Withdrawal Assessment for Alcohol, revised
scale (CIWA-Ar) in the infusion-titrated and bolus-titrated groups
at the onset of alcohol withdrawal syndrome (AWS begin) and
Statistical analysis maximal CIWA-Ar during the course of AWS (AWS max); medi-
ans. b Daily average (means and SEM) CIWA-Ar course in the in-
Data are expressed as median and interquartile ranges. Statistical fusion-titrated group (triangles) and bolus-titrated group (x) from
comparison of groups used the Mann-Whitney U test. Intragroup onset AWS until discharge from the ICU; arrow onset of AWS
analysis used the Wilcoxon matched-pairs signed-rank test. Rank
variance analysis was used to compare the CIWA-Ar between
groups. Fisher’s test or the χ2 test was used to assess differences in
dichotomous variables. A p value of 0.05 or lower was considered flunitrazepam boluses did not differ between groups
statistically significant. (Table 2). Higher drug requirements were also observed
for clonidine and haloperidol in the ITG (Table 2), while
the number of patients requiring clonidine or haloperidol
Results did not differ between groups (Table 2). Clonidine-
induced complications included bradycardia requiring
CIWA-Ar scores increased following treatment for AWS positive chronotropic support (BTG 0/13 vs. ITG 1/16)
in the ITG but not in the BTG (Fig. 3a). The daily mean and hypotension (BTG 1/13 vs. ITG 3/16). Propofol as
CIWA-Ar remained elevated for a longer period in the rescue medication was not required more frequently in
ITG (Fig. 3b). The duration of AWS was significantly the ITG than the BTG. However, in patients who needed
longer in the ITG than the BTG (Table 2). rescue medication both the number of titrated propofol
Patients in the BTG required less flunitrazepam as the boluses and the total amount of propofol were signifi-
first-line agent (Table 2). The number of administered cantly higher in the ITG (Table 2).
 2235

Table 2 Alcohol withdrawal syndrome (AWS) in the bolus-titrated (BTG) and infusion-titrated (ITG) groups; absolute number (paren-
theses frequency) or median (parentheses interquartile range) (PaO2/FIO2 oxygenation index, RSS Ramsay Sedation Scale)

BTG (n=23) ITG (n=21) p

Period until onset of AWS (days) 2 (2–4) 2 (2–4) 0.12

Maximal systolic blood pressure (mmHg) 198 (180–227) 180 (167–201) ≤0.01
Maximal heart rate (beats/min) 119 (108–131) 130 (112–145) 0.21
PaO2/FIO2, onset of AWS (mmHg) 328 (280–441) 341 (269–487) 0.51
Maximal RSS, during AWS therapy (days) 4 (3–6) 5 (4–6) ≤0.01
Duration of AWS (days) 2 (2–4) 6 (4–10) ≤0.01
Intubated patients following onset of AWS 15 (65%) 19 (90%) 0.05
Time until intubation after onset of AWS (days) 2 (1–2) 2 (1–2) 0.13
Duration of ventilation (days) 6 (3–8) 12 (5–20) ≤0.01
Flunitrazepam 23 (100%) 21 (100%)
Initial bolus (mg) 3 (1–8) 2 (2–4) 0.19
Number of boluses, each adjustment 4 (1–11) 3 (2–4) 0.49
Infusion rate, max. (µg/kg per hour) 12.1 (0–26.7) 25.0 (17.5–87.4) ≤0.01
Total amount (mg) 69.7 (12.5–143.9) 162.0 (91.4–807.0) ≤0.01
Clonidine 13 (57%) 16 (76%) 0.17
Initial bolus (µg) 150 (150–300) 75 (37.5–150) ≤0.01
Number of boluses, each adjustment 4 (2–6) 2 (1–4) ≤0.01
Infusion rate, max. (µg/kg per hour) 0 5.5 (2.2–7.4) ≤0.01
Total amount (µg) 1,270 (1,050–4,768) 61,098 (7,188–147,384) ≤0.01
Haloperidol 13 (57%) 12 (57%) 0.97
Initial bolus (mg) 10 (15–70) 5 (3–70) 0.89
Number of boluses, each adjustment 6 (3–8) 2 (1–2) ≤0.01
Infusion rate, max. (µg/kg per hour) 0 412 (85–1310) ≤0.01
Total amount (mg) 180 (80–554) 1713 (270–3,288) ≤0.01
Propofol, rescue 8 (35%) 10 (48%) 0.39
Total number of boluses 34 (12–81) 53 (8–164) 0.01
Total amount (g) 6.10 (2.15–15.10) 8.95 (1.40–21.50) 0.03
Fentanyl, intubated
Infusion rate, max. (mg/h) 0.3 (0.2–0.5) 0.3 (0.2–0.5) 0.81
Total amount (mg) 51 (12–64) 78 (23–110) ≤0.01a
Piritramide, extubated
Total number of boluses 18 (6–41) 21 (11–36) 0.39
Dose of each bolus (mg) 5 (3–12) 5 (3–12) 0.92
Total amount (mg) 121 (41–395) 106 (39–371) 0.21
a Due to longer intubation period

The duration of ventilatory support was halved
(Table 2) and the incidence of pneumonia was signifi-
cantly lower in the BTG than the ITG (Table 3). The du-
ration of ICU treatment was significantly shorter in the
BTG than in the ITG (Table 3).
Discussion
The most important finding of this study was the de-
creased severity and duration of AWS with the bolus ap-
proach in surgical ICU patients. At the time of inclusion,
i.e., at the beginning of AWS, the same severity of AWS
was seen in both groups. The maximum CIWA-Ar scores
in the ITG were comparable to the severity in the few
previous studies in ICU patients which also used the
CIWA-Ar [9, 11, 16]. The shorter duration of AWS in
the BTG was due to the fact that the signs and symptoms
of AWS were better controlled, as shown by the different
maximal CIWA-Ar between groups. This is significant
because it implies that early and adequate care for these
patients prevent further deterioration and therefore im-
prove patient outcome.
Despite the fact that the CIWA-Ar score has not
been validated for ICU patients but only in psychiatric
settings [8], it has successfully served for AWS moni-
toring in ICU settings [3, 11]. Most other studies have
not used an international scale to quantify AWS; some
studies even fail to differentiate between autonomic
signs, hallucinations, and the delirious state [3]. Al-
though the CIWA-Ar can be easily monitored in extu-
bated ICU patients, it cannot be easily performed in
intubated ICU patients. Therefore we included only pa-
tients extubated at the onset of AWS. At the onset of
AWS the CIWA-Ar in extubated patients was still sig-
nificantly higher (median 23) than in a study with psy-
chiatric patients (median 11) [7]. Therefore for compar-
ing two treatment regimes the CIWA-Ar seems ade-
quately to reflect the severity of AWS in extubated ICU
patients.
2236

Table 3 ICU treatment and in-

tercurrent complications in the BTG (n=23) ITG (n=21) p
bolus-titrated (BTG) and infu-
sion-titrated (ITG) groups; ab- ICU treatment (days) 8 (5–10) 14 (7–25) ≤0.01
solute number (parentheses fre- Pneumonia 9/23 (39%) 15/21 (71%) ≤0.01
quency) or median (parentheses PaO2/FIO2, minimal (mmHg) 175 (105–356) 210 (94–412) 0.51
interquartile range) (PaO2/FIO2 Micro-organisms 4/9 (44%) 6/15 (40%) 0.83
oxygenation index, MS methi- Klebsiella pneumoniae 1 2
cillin sensitive) Pseudomonas aeruginosa 1 2
Stenotrophomona maltophilia 0 1
Haemophilus influenzae 1 0
Streptococcus pneumonia 1 0
Staphylococcus aureus, MS 0 1
Days until onset of pneumonia (days) 3 (1–8) 4 (1–14) 0.72
Septic shock 6/23 (26%) 9/21 (43%) 0.24
Maximal lactate (mmol/l) 2.0 (1.7–2.7) 2.1 (1.4–3.2) 0.71
Norepinephrine 3/6 (50%) 4/9 (44%) 0.84
Max. norepinephrine (µg/kg per minute) 0.35 (0.09–0.82) 0.19 (0.01–1.38) 0.40
Hemorrhage 11/23 (48%) 12/21 (57%) 0.43
Blood transfusion 17/23 (74%) 16/21 (76%) 0.93
Packed red blood cell transfusion (ml) 600 (250–2250) 1000 (250–3125) 0.58
Surgical revision 2/23 (9%) 4/21 (19%) 0.32
Cardiac disorders 11/23 (48%) 11/21 (52%) 0.77
Myocardial ischemia 4/23 (17%) 4/21 (19%) 0.88
Congestive heart failure 1/23 (4%) 3/21 (14%) 0.29
Arrhythmias 7/23 (30%) 6/21 (29%) 0.89
Death 1/23 (4%) 2/21 (10%) 0.50

After the onset of AWS 65% of patients in the BTG
vs. 90% in the ITG had to be intubated. Following intu-
bation sensations/hallucinations accounting for a maxi-
mum of 21 points of 67 on the CIWA-Ar had to be set at
0 because they are often hardly possible to diagnose.
However, despite this apparent limitation in monitoring
AWS during mechanical ventilation the patients in the
ITG, although intubated significantly more often, had a
higher daily mean CIWA-Ar score than patients in the
BTG.
Most patients had to be intubated because the GCS
was 8 or lower after controlling the AWS. Therefore 34
of 44 patients were mechanically ventilated after the on-
set of AWS, which was not considered an exclusion cri-
terion. Patient intubation and ventilation are frequent in
ICU patients when AWS is present. However, this is
rarely reported [5, 6, 11] for most treatment regimens in
AWS focus on psychiatric patients in which the severity
of AWS is relatively lower, and therefore the need for
drugs inducing respiratory depression is definitely much
less [3, 7]. The increased severity of AWS in ICU pa-
tients may be explained by a lower balance of the excita-
tory and inhibitory transmitter systems due to surgical
trauma, previous anesthesia, pain, and other postopera-
tive stress [3].
The amount of flunitrazepam, haloperidol and cloni-
dine in the ITG did not differ from previous studies in
ICU patients [11, 16, 23]. The longer acting benzodiaze-
pine flunitrazepam (half-life 16–22 h) was used to con-
tribute to an overall smoother withdrawal course with
less breakthrough and rebound symptoms [3]. The ad-
ministration of longer acting benzodiazepines is evi-
dence based in the case of AWS [10]. However, haloper-
idol was in the upper range [11, 16, 23]. In the BTG the
drug requirements were significantly less than in the
ITG. This was related to a better control of signs and
symptoms of AWS by the higher amount of symptom-
oriented bolus administration in the BTG. At least one-
third of the patients in both groups required propofol as
rescue medication, indicating that the gold standard for
AWS therapy remains to be found.
As recommended for clinical routine [3, 22], when pa-
tients’ CIWA-Ar score was adequate, the continuous infu-
sion was decreased by 10%, while bolus administration
was immediately interrupted. This may explain why a
higher total amount of drugs is required in the ITG. How-
ever, to avoid kindling with more severe withdrawal we
could not have simply stopped the infusion in the ITG if a
certain CIWA-Ar was reached because this would have
clearly endangered the patient [3, 25, 26, 27].
The incidence of postoperative pneumonia was signif-
icantly lower in the BTG (39%). An incidence of 71% in
the ITG is comparable to previously reported incidence
rates ranging of 67–82% in patients who develop AWS
in the ICU after major surgery or trauma [6, 9]. The
spectrum of micro-organisms did not differ from that re-
ported in a previous study [28]. The most important fac-
tor in the development of pneumonia may be the in-
creased need for ventilatory support in the ITG. Intuba-
tion and mechanical ventilation is reported to increase
the incidence of pneumonia four- to sevenfold [29, 30,
31]. On the other hand, since alcohol use disorders are
associated with an increased risk for infections [5, 6],
pneumonia primarily could have led to a longer ventila-
 2237

tion assistance with the need for more sedative agents.
Nevertheless, in both cases a better guided AWS therapy
with less sedative agents would reduce intubation time
and the risk of pneumonia as observed in this study, ex-
plaining the shorter ICU treatment of patients in the
BTG.
In addition to the difficulty in determining AWS in
the ICU setting, another limitation of the study was the
recruitment of patients. Those who had an alcohol use
disorder and were considered to be alcohol dependent re-
ceived prophylactic treatment and were not included. We
consider this ethically relevant because patients diag-
nosed as dependent and not receiving preventive treat-
ment have a worse outcome [3]. Secondly, many patients
were still intubated at the onset of AWS. Therefore due
to the limitation of the CIWA-Ar these patients were also
not included. It is suggested to be more relevant if an
isolated AWS is responsible for (re)intubation. Therefore
we included only patients who were not intubated at the
onset of AWS to rule out additional diagnoses with the
need for intubation [32].
In conclusion, at the onset of AWS a close monitoring
of the signs and symptoms combined with a measure of
the severity of AWS and an immediate bolus approach
reduced the severity and duration of AWS and the
amount of drug therapy required. This management ap-
proach decreased the postoperative pneumonia rate and
shortened overall ICU stay. Therefore a bolus adjusted
approach in combination with a basic benzodiazepine in-
fusion may be more beneficial both in limiting the pro-
gression of AWS and preventing the development of fur-
ther complications.
Acknowledgements We are most grateful to our statistician
Dipl.-Math Gerda Siebert, Institute of Medical Biometrics, De-
partment of Medical Statistics, Charité Medical Center, Humboldt
University of Berlin.

References
1. Moore R, Bone L, Geller G (1989)
Prevalence, detection and treatment of
alcoholism in hospitalised patients.
JAMA 261:403–407
2. Spies C, Tonnesen H, Andreasson S,
Helander A, Conigrave K (2001) Peri-
operative morbidity and mortality in
chronic alcoholics. Alcohol Clin Exp
Res 25:164–170
3. Spies C, Rommelspacher H (1999)
Novel aspects of alcohol withdrawal in
the surgical patient–prevention and
treatment. Anesth Analg 88:946–954
4. Tonnesen H, Kehlet H (1999) Preoper-
ative alcoholism and postoperative
morbidity. Br J Surg 86:869–874
5. Spies CD, Neuner B, Neumann T,
Blum S, Müller C, Rommelsbacher H,
Sanft C, Specht M, Hannemann L,
Striebel HW, Schaffartzik W (1996)
Intercurrent complications in chronic
alcoholics admitted to the intensive
care unit following trauma. Intensive
Care Med 22:286–293
6. Spies C, Nordmann A, Brummer G,
Marks C, Conrad CA, Berger G,
Runkel N, Neumann T, Müller C,
Rommelspacher H, Specht M, Hanne-
mann L, Striebel HW, Schaffartzik W
(1996) Intensive care unit stay is pro-
longed in chronic alcoholic men fol-
lowing tumor resection of the upper
digestive tract. Acta Anaesthesiol
Scand 40:649–656
7. Saitz R, Mayo-Smith M, Roberts M,
Redmond H, Bernard D, Calcins D
(1994) Individualized treatment for al-
cohol withdrawal. JAMA 272:519–523
8. Sullivan JT, Sykora K, Schneidermann
J, Naranjo CA, Sellers EM (1989)
Assessment of alcohol withdrawal: the
revised clinical institute withdrawal as-
sessment for alcohol scale (CIVAA-Ar).
Br J Addict 84:1353–1357
9. Spies C, Morciniec P, Lenzenhuber E,
Müller C, Marks C, Helling K, Runkel
N, Berger G, Blum S, Rommelspacher
H (1998) β-Carbolines in alcohol de-
pendent intensive care patients during
prophylactics and therapy of alcohol
withdrawal syndrome. Addict Biol
3:281–294
10. Mayo-Smith MF (1997) Pharmacologi-
cal management of alcohol withdrawal.
JAMA 278:144–151
11. Spies C, Dubisz N, Neumann T, Blum
S, Müller C, Rommelspacher H,
Brummer G, Specht M, Sanft C,
Hannemann L, Striebel HW,
Schaffartzik W (1996) Therapy of al-
cohol withdrawal syndrome in inten-
sive care patients following trauma:
results of a prospective, randomized
trial. Crit Care Med 24:414–422
12. Imdahl H, Imdahl A (1992) Prophylaxe
und Therapie des Alkoholentzugsdelirs
in der Chirurgie—eine Fragebogenana-
lyse. Aktulle Chir 27:139–143
13. Kress JP, Pohlmann AS, O’Connor
MF, Hall JB (2000) Daily interruption
of sedative infusions in critically III
patients undergoing mechanical venti-
lation. N Engl J Med 342:1471–
1477
14. Boulain T (1998) Unplanned extuba-
tions in the adult intensive care unit:
a prospective multicenter study. Assos-
cation des Reanimateurs du Centre-
Ouest. Am J Respir Crit Care Med
157:1131–1137
15. Ostermann ME, Keenan SP, Seiferling
RA, Sibbald WJ (2000) Sedation in the
intensive care unit: a systematic re-
view. JAMA 283:1451–1459
16. Lenzenhuber E, Müller C, Rommels-
pacher H, Spies C (1999) Gamma-
hydroxybutyrate for treatment of alco-
hol withdrawal syndrome in intensive
care patients. A comparison between
two symptom-oriented therapeutic con-
cepts. Anaesthesist 48:89–96
17. Spies C, Dubisz N, Funk W, Blum S,
Müller C, Rommelspacher H, Brummer
G, Specht M, Hannemann L, Striebel
HW, Schaffartzik W (1995) Prophyl-
axis of alcohol withdrawal syndrome in
alcohol dependent patients admitted to
the intensive care unit following tu-
mour resection. Br J Anaesth 75:734–
739
18. Task Force on DSM-IV (1994) Diag-
nostic and statistical manual of mental
disorders, fourth edition (DSM-IV).
American Psychiatric Association,
Washington, pp 173–177
19. Badawy AA, Rommelspacher H,
Morgan CJ, Bradley DM, Bonner A,
Ehlert A, Blum S, Spies C (1998)
Tryptophan metabolism in alcoholism:
tryptophan but not excitatory amino
acid availability to the brain is in-
creased before the appearance of the
alcohol-withdrawal syndrome in men.
Alcohol Alcohol 33:616–625
2238
20. Spies CD, von Winterfeld A, Müller C,
Rommelspacher H, Neumann T, Marks
C, Berger G, Conrad C, Blum S, Hanne-
mann L, Striebel HW, Schaffartzik W
(1996) Reliability of carbohydrate-
deficient transferrin to detect chronic
alcohol misuse in carcinoma patients.
Eur Addict Res 2:156–162
21. Ramsay MAE, Savege TM, Simpson
BRJ, Goodwin R (1974) Controlled se-
dation with alphaxalone-alphadolone.
BMJ 2:656–659
22. Otter H, Spies C (2003) Analgosedie-
rung und Entzugssyndrom. In: Kox
WJ, Spies D (eds) Check-up Anästhe-
siologie. Springer, Berlin Heidelberg
New York, pp 370–378
23. Garner JS, Jarvis WR, Emori TG,
Horan TC, Hughes JM (1998) CDC
definitions for nosocomial infections.
Am J Infect Control 16:128–140
24. American College of Chest Physicians/
Society of Critical Care Medicine Con-
sensus Conference (1992) Definitions
for sepsis and organ failure and guide-
lines for the use of innovative thera-
pies in sepsis. Crit Care Med 20:864–
874
25. Brown C, Albrecht R, Pettit H,
McFadden, Schermer C (2000) Opioid
and benzodiazepin withdrawal syn-
drome in adult burn patients. Am Surg
66:367–371
26. Tobias JD (2000) Tolerance, with-
drawal, and physical dependency after
long-term sedation and analgesia of
children in the pediatric intensive
care unit. Crit Care Med 28:2122–
2132
27. Atkins PM, Mion LC, Mendelson W,
Palmer RM, Slomka J, Franko T
(1997) Characteristics and outcomes of
patients who self-extubated from venti-
latory support: a case-control study
122:1317–1323
28. Rello J, Sa-Borges M, Correa H, Leal
SR, Baraibar J (1999) Variations in eti-
ology of ventilator-associated pneumo-
nia across four treatment sites: implica-
tions for antimicrobial prescribing
practices. Am J Respir Crit Care Med
160:608–613
29. Celis R, Torres A, Gatell JM, Almela
M, Rodriguez-Roisin R, Augusti-Vidal
A (1998) Nosocomial pneumonia. A
multivariate analysis of risk and prog-
nosis. Chest 93:318–342
30. Cross AS, Roup B (1981) Role of res-
piratory assistance devices in endemic
nosocomial pneumonia. Am J Med
70:681–685
31. Rodriguez JL (1994) Hospital-acquired
gram-negative pneumonia in critically
ill, injured patients. Am J Surg
165:34S-42S
32. Eggers V, Schilling A, Kox WJ, Spies
C (2003) Septic encephalopathy. Diag-
nosis und therapy. Anaesthesist
52:294–303