ALCOHOLISM: CLINICAL AND EXPERIMENTAL RESEARCH Vol. 37, No.

4
April 2013

Prevention and Therapy of Alcohol Withdrawal on Intensive

Care Units: Systematic Review of Controlled Trials
Lavinius A. Ungur, Bruno Neuner, Susanne John, Klaus Wernecke, and Claudia Spies

Background: Alcohol withdrawal syndrome (AWS) occurs in 16 to 31% of intensive care unit
(ICU) patients after cessation of sedation. There exist many preventive and therapeutic strategies, but
no systematic review (SR) has been published on this topic so far. We aimed to perform a synopsis of
all controlled trials of AWS prevention and therapy in ICU published between 1971 and 30 March
2011 following the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA)
statement.
Methods: We performed a MEDLINE search with the terms alcohol AND ICU as well as
alcohol withdrawal AND intensive care. All publications that matched our eligibility criteria were
analyzed according to our predened criteria.
Results: We identied 6 controlled trials about AWS prevention and 8 about AWS therapy in
ICUs. For AWS prevention, benzodiazepines (BZO), ethanol (EtOH), and clonidine were evaluated as
single agents, and BZO, clonidine, clomethiazol and haloperidol were studied in drug combinations.
All evaluated single agents and combinations were found to be eective for AWS prevention. Clo-
methiazol was found to be associated with a higher tracheobronchitis rate and thus disadvised for criti-
cally ill patients. For AWS therapy, BZO, gamma-hydroxybutyric acid (GHB), and clomethiazol were
evaluated in randomized controlled trials as single agents and phenobarbital, clonidine, and haloperidol
as adjuncts. All evaluated regimens were found to be eective for AWS therapy. Overall, in the ICU,
BZO were found to be superior to GHB and clomethiazol regarding safety and ecacy. Furthermore, 4
cohort trials with historical control groups evaluated the eect of the implementation of a standardized
protocol of BZO therapy for AWS in ICUs. All of these 4 studies found better outcome for the inter-
vention groups.
Conclusions: Based on the evidence of this SR, EtOH or BZO can be advised for AWS prevention
on ICU patients with alcohol dependence, but EtOH is not allowed for therapy of AWS. AWS therapy
should be standardized and based on symptom-triggered BZO administration. Alpha2-agonists and
haloperidol should be added for autonomic and productive psychotic symptoms.
Key Words: Alcohol, EtOH, Alcohol Withdrawal, Intensive Care, Delirium.

A LCOHOL WITHDRAWAL SYNDROME (AWS)

occurs in 16 to 31% of patients in all types of inten-
sive care units (ICUs) after cessation of sedation (Spies
and Rommelspacher, 1999), with most cases occurring in
trauma ICUs (Spies et al., 1996b). Mortality of untreated
AWS can reach 15% compared with only 2% if treated
(Sander et al., 2006). Prevention and therapy standards of
AWS should be part of the basic knowledge of physicians
From the Department of Anesthesiology and Intensive Care Medicine
(LAU, BN, SJ, KW, CS), Charite University Medicine of Berlin,
Berlin, Germany; and the Institute of Epidemiology and Social Medicine
(BN), Clinical Epidemiology Section, University of Muenster, Muenster,
Germany.
Received for publication June 15, 2012; accepted August 21, 2012.
Reprint requests: Claudia Spies, MD, Professor of Anesthesiology and
Intensive Care Medicine, Klinik fur Anasthesiologie und operative Inten-
sivmedizin der Charite Universitatsmedizin Berlin, Campus Virchow-
Klinikum, Augustenburger Platz 1, 13353 Berlin; Tel.: +49-30-450-55-
10-02; Fax: +49-30-450-55-19-09; E-mail: Claudia.spies@charite.de
Copyright 2012 by the Research Society on Alcoholism.
DOI: 10.1111/acer.12002
Alcohol Clin Exp Res, Vol 37, No 4, 2013: pp 675686
working on medical, surgical, or trauma ICUs. However,
to date, there are no evidence-based international guide-
lines or any systematic reviews (SRs) for the prevention
and/or therapy of AWS in ICUs. We aimed to perform an
SR of randomized controlled trials (RCTs) as well as non-
RCTs published in MEDLINE following the Preferred
Reporting Items for Systematic Reviews and Meta-Analyses
(PRISMA) statement. Our objective was to put together an
evidence-based synopsis of drugs used for AWS prevention
and therapy in ICUs.
MATERIALS AND METHODS
1. Medline search strategy (Fig. 1): We did a MEDLINE search
with the terms alcohol AND ICU as well as alcohol with-
drawal AND intensive care, including all publications in all
languages between 1971 and the end of March 2011. In all cases,
native speakers were asked to review the publications.
2. Of all articles matching with the search terms above, we included
those dealing with alcohol withdrawal.
3. To avoid a selection bias, all articles were analyzed in 2 indepen-
dent working groups simultaneously. In a consensus meeting, we
675
676 UNGUR ET AL.

Publications identified through Additional publications identified

database searching: 545 through other sources: none

Publications after duplicates removed

545

Publications screened: 545 Publications excluded: 465

(not about alcohol withdrawal)

Full-text articles assessed Full-text articles excluded: 18

for eligibility: 80 (no ICU patients or not reported
prophylactic or therapeutic agent)

Additional publications Articles included in Articles other than controlled trials

identified through qualitative synthesis: 62 (reviews, case series, case reports or
screening of 62 articles expert opinion articles): 48
references: none.

Studies included in synthesis I Studies included in synthesis II

(controlled trials about AWS (controlled trials about AWS
prevention): 6 therapy): 8

Fig. 1. PRISMA flow diagram. AWS, alcohol withdrawal syndrome.

then excluded all publications that not clearly dealt with patients
in ICUs or failed to specify the prophylactic or therapeutic
agents administered for AWS.
4. The remaining articles were divided into the groups PREVEN-
TION and THERAPY of AWS.
5. All remaining articles were individually graded according to
the Oxford Centre for Evidence-based Medicine. Thus, preli-
minary levels of evidence (LoE) were attributed to each article:
1a (SRs of RCTs, 1a- in case of worrisome heterogeneity),
1b (individual RCTs, 1b- in case of wide condence inter-
val), 2a (SRs of cohort studies), 2b (individual cohort stud-
ies including low-quality RCTs), 3 (casecontrol studies), 4
(case series or case reports), or 5 (expert opinion). Again,
this grading was performed by each of our 2 working groups
separately to minimize conrmation or expectation bias. The
preliminary LoE results were then discussed in a consensus
meeting.
6. References of all graded articles were screened for additional
controlled trials possibly missed in the MEDLINE search.
7. All articles classied as LoE 1b, 1b-, or 2b and thus describing
controlled trials were analyzed, and their main ndings were put
together in an SR.
RESULTS
1. Our MEDLINE search with the terms alcohol AND
ICU as well as alcohol withdrawal AND intensive
care proceeded on the 30th of March 2011 revealed 545
articles.
2. Of all 545 articles matching with the search terms above,
80 were found to deal with alcohol withdrawal (see Data
S1 for complete reference list).
3. We excluded 18 of the 80 publications because they did
not clearly deal with ICU patients or failed to specify the
prophylactic or therapeutic agents administered for AWS
(see Data S1 for list of excluded articles).
4. As 10 of the 62 articles tted into both PREVENTION
and THERAPY groups, we counted 19 AWS prevention
and 53 AWS therapy publications.
5. Our 2 working groups graded most of the articles equally.
Four AWS prevention and 4 AWS therapy articles were
graded dierently, which was not relevant for the selection
of our synopsis articles: The dissents were between LoE 1a-
and 2a (review articles) in 1 case, between LoE 1b- (RCT
with wide condence interval) and 2b (cohort trial or
low-quality RCT) in 1 case, and between LoE 4 (case
reports/case series) and 5 (expert opinion) in 6 cases.
6. The screening of the references of the 62 graded articles
revealed no additional controlled trials evaluating AWS
prevention or therapy in ICUs.
REVIEW: ALCOHOL WITHDRAWAL ON ICU
7. Among all articles classied as LoE 1b, 1b-, or 2b, 6 were
about AWS prevention (Table 1) and 8 about AWS ther-
apy (Table 2).
AWS Prevention
Six publications were classied as LoE 1b, 1b-, or 2b and
thus included in the AWS prevention part of this review (see
Table 1 for details). Investigated drugs were benzodiazepines
(BZO), intravenous (i.v.) ethanol (EtOH), clonidine, clo-
methiazol, and haloperidol. Even if none of the trials was
placebo-controlled, all these drugs were concluded to be
eective in AWS prevention.
Benzodiazepines. Three studies compared dierent BZO
with other agents for AWS prophylaxis and found a ten-
dency to slight advantages over EtOH and clonidine (Huber
et al., 1990; Weinberg et al., 2008) and a safety advantage
over clomethiazol (Spies et al., 1995).
A 2-armed RCT (Weinberg et al., 2008) compared EtOH
with diazepam regarding ecacy and adverse events in 50
trauma patients with long-term alcohol abuse history.
During the rst hours of treatment, the diazepam-receiving
patients were less agitated. After 48 hours, patients in both
study arms had the same high success rates (88% calm and
cooperative). A concern is that we suspect a dosage bias: 19
of the 26 patients in the EtOH group had undetectable blood
alcohol concentration (BAC) levels at all times, and only 2
had a higher BAC than 0.1 mg/ml at any time during the
EtOH infusion. Furthermore, the authors only included
patients able to give their informed consent and thus not
severely injured patients. The exclusion of strongly injured
patients might weaken the power of this trial and thus
impede the comparability of the 2 patient groups as all
included subjects had a good outcome and no severe compli-
cations. However, the lack of any severe adverse event and
the high treatment success rate after 48 hours show that
AWS prophylaxis might have a favorable eectiveness and
that both regimens are feasible and safe.
A 3-armed RCT with 52 patients (Huber et al., 1990)
compared 3 monotherapeutic regimens (EtOH i.v., midazo-
lam, and clonidine) and found a slight advantage of mi-
dazolam in symptom control, ICU length of stay (LOS),
time spent on mechanical ventilation, and mortality. How-
ever, the given dierences were small. The authors failed to
report important additional information about the study
design like the number of patients per study arm, their
AWS denition, or administered additional sedative drugs.
Also, the results might be biased by the dosage protocols
used in the dierent study arms. Thus, a critical evaluation
and a statistical verication of the ndings cannot be per-
formed properly.
A 4-armed RCT of dierent prevention regimes (prophy-
lactic EtOH i.v./unitrazepam + clonidine/clomethiazol +
haloperidol/unitrazepam + haloperidol) (Spies et al., 1995)
found that the choice of administered agents has no
677
measurable inuence on AWS incidence, major complica-
tions, and ICU LOS. Only the clomethiazol-containing ther-
apy was associated with a higher tracheobronchitis rate,
probably due to bronchial hypersecretion.
EtOH. A 2-armed RCT (Weinberg et al., 2008) found
no advantage of EtOH over diazepam regarding ecacy
or adverse events. A 3-armed RCT (Huber et al., 1990)
compared 3 monotherapeutic regimens (EtOH i.v., midazo-
lam, and clonidine) and found slight outcome advantages in
the midazolam group. In a 4-armed RCT (Spies et al., 1995),
the authors found that EtOH was equally eective and safe
as 3 other double-drug combinations except clomethiazol
being associated with more respiratory complications. See
benzodiazepine chapter above for these 3 studies and Table 1
for study details.
A study with a historical control group (Dissanaike et al.,
2006) described benets of surgical ICU patients treated with
a standardized protocol of i.v. EtOH regarding treatment
duration and failure rate. Despite some methodological
weaknesses (poor cohort description, lack of a validated
quantitative AWS score), it can be concluded that a stan-
dardized EtOH prophylaxis can be more eective in avoiding
a long-lasting AWS in ICUs and ensuring a better care of
these patients than a nonstandardized EtOH administration.
Moreover, the sample size of this trial is large (160 patients),
and the outcome dierences are clinically important (mean
treatment duration 3 vs. 7 days, failure rate 7 vs. 20%).
A small RCT with patients who reported regular EtOH
consumption and underwent neck dissection (Heil et al.,
1990) showed that in 10 patients treated with prophylactic
EtOH (2 to 4 g per hour), none developed AWS. In the non-
prophylactically treated control group, 6 of 9 patients devel-
oped AWS and had thus to be treated symptomatically with
clomethiazol and haloperidol. The authors also mentioned
that among the 31 not included patients that were at low
dependency risk following the Munich Alcoholism Test
(MALT) (Feuerlein et al., 1979), none received EtOH
prophylaxis and none developed AWS symptoms: 12 of these
patients were previously classied as dry alcoholics (absti-
nent at least since several weeks), 10 without abuse risk,
and 9 with abuse risk but not dependent according to the
number of positive MALT questionnaire items. Despite the
low number of included patients, this trial suggests that
EtOH is highly eective in preventing AWS in surgical ICU
patients and that the number needed to treat to see a benet
in AWS prevention is amazingly low if adequate screening is
performed.
Alpha2-Agonists. In a 3-armed RCT (Huber et al.,
1990), the authors concluded clonidine to be equal to EtOH
but slightly inferior to midazolam regarding outcome and
ecacy. A combination of clonidine + unitrazepam was
compared with clomethiazol + haloperidol, unitrazepam +
haloperidol, and EtOH in a 4-armed RCT (Spies et al.,
1995). This trial found no dierences in eciency and
 678

Table 1. Synthesis I: Controlled Trials About AWS Prevention

Reason Definition/
Intervention and for scoring of Additional Study Limitations/
Ref # Study design n Inclusion criteria ICU AWS AWS drugs outcome Main findings weaknesses
1 Two-armed Continuous At least 5 alcohol Non-CNS Riker narcotics Achievement -EtOH group: initially more -Many exclusion criteria
RCT EtOH (n = 26) beverage trauma Sedation (only given of a Riker deviations from a -No blinding
infusion vs. equivalents per day Agitation in morphine score of 4 Riker = 4 score (mostly -Agitation as only scored
scheduled during last Scale (1 equivalents, (calm and due to agitation) AWS symptom
bolus dosed 6 months, ability to to 7 not cooperative) -After 48 h: both groups -Rigid protocol (not
diazepam give consent points) precised) had a Riker = 4 symptom triggered, not
(n = 24) percentage of 88% body weight adjusted)
-In all patients the 48-hour-
weaning procedure was
successful (started 48 to
96 hours after treatment
initiation)
2 Controlled Protocol -Intervention group: Not DSM-IV Not reported -Duration of -Protocol group: earlier -No validated quantitative
nonrandomized implementation: history of AWS, specified definition treatment initiation and shorter AWS score
study i.v. EtOH alcohol abuse or -Incidence of duration of treatment (3 vs. -Poor description of patient
(n = 68) vs. alcohol-related AWS 7 days), lower failure rate cohort
historical CG conditions symptoms (7 vs. 20%) and higher -6 pat. secondary excluded
(i.v. EtOH -CG: i.v. EtOH -Referral to referral to substance due to protocol violation
without treatment and specialized abuse clinic (20 vs. 8%) -5 + 2 patients already had
protocol (n = 92) history of AWS/ abuse clinic -Only AE: asymptomatic AWS when EtOH infusion
alcohol abuse -AE hyponatremia (1 pat.) was started, blending of
therapy and prophylaxis
3 Controlled -Clonidine on Elective Esophag- RAMSAY Not reported Nitrogen -Less protein catabolism in -No randomization
nonrandomized POD 1 to 6 abdomino- ectomy sedation balance as clonidine group -No blinding/no placebo
study (n = 13) thoracic score surrogate for (cumulated values of POD -Low number of patients
-CG (n = 11): no esophageal protein 1 to 6): 1.5 vs. 17.6 g -CG not adequate for AWS
clonidine, cancer catabolism mean nitrogen loss (=36 evidence (pat. without no
pat. without resection vs. 422 g muscle mass alcohol history)
alcohol abuse equivalent) -No validated AWS score
history (!) -Similar Ramsay scores in
both groups
4 Four-armed Flunitrazepam + Alcohol Tumor CIWA-Ar Only in -ICU LOS -Higher tracheobronchitis -Cardiac/pulmonary risk
partially blinded clonidine dependence surgery (assessment secondary -AWS rate in clomethiazol + patients excluded due to
RTC (n = 52) vs. (DSM-III), by blinded excluded prevention haloperidol group safety concerns
clomethiazol + daily EtOH investigators), pat. -Major -No difference in AWS -No double-blinding
haloperidol intake of 60 g AWS confirmed complications incidence or severity, ICU -23 pat. secondary
(n = 49) vs. by a neurologist LOS, other major excluded (from which 10
flunitrazepam + complications or mortality due to AE; 5 to 8 due to
haloperidol additional AWS therapy)
(n = 50) vs.
EtOH i.v.
(n = 50)

Continued.
UNGUR ET AL.
 REVIEW: ALCOHOL WITHDRAWAL ON ICU 679

safety between the combinations except a higher respira-

intravenous; vs., versus; pat., patient(s); CNS, central nervous system; AE, adverse event(s); POD, postoperative day; BZO, benzodiazepines; DSM, Diagnostic and Statistical Manual of Mental Dis-
Ref # = reference number: 1. Weinberg and colleagues (2008); 2. Dissanaike and colleagues (2006); 3. Mertes and colleagues (1996); 4. Spies and colleagues (1995); 5. Huber and colleagues
(1990); 6. Heil and colleagues (1990); RCT, randomized controlled trial; n, number of patients; AWS, alcohol withdrawal syndrome; ICU, intensive care unit; CG, control group; EtOH, ethanol; i.v.,
important items not given

-High BZO dosage (0.2 to

-No validated AWS score

-Blending of therapy and
-Low number of patients
Clonidine (1 to 2 mg/d)
& EtOH 1 to 2 g/kg per
tory complication rate in clomethiazol-treated patients. See
-Number of patients per

-No statistical analysis

0.4 mg/kg/h) vs. low

study arm and other
benzodiazepine chapter above for these 2 studies and Table 1
weaknesses
Limitations/

for study details.

day) dosage
A small trial (Mertes et al., 1996) found much less protein

prophylaxis
-No blinding
catabolism in patients treated with clonidine than in non-
treated controls. In this study, patients with alcohol abuse
history were compared to patients without alcohol abuse
history; thus, no comparison of AWS prevention success can
additional sedation needed
-Midazolam group: weakest

-Clomethiazol + haloperidol
be made. Nevertheless, this trial suggests that one of the
ventilation and ICU LOS,

(significance unclear, no

incidence, shorter ICU

LOS (3 vs. 11.5 days)
-EtOH group: 0% AWS
arguments for preventing AWS with clonidine might be a
-Clonidine group: most
statistics, or p-values
Main findings

quick recovery due to less muscle atrophy.

mortality, shortest

best AWS control

group: 66% AWS

Clomethiazol. As reported in the BZO chapter above,
clomethiazol in combination with haloperidol was associated
given)

with bronchial hypersecretion and thus with more respira-

tory complications than prophylactic EtOH i.v., unitraze-
pam + clonidine, or unitrazepam + haloperidol (Spies
complications
-Incidence of
-AWS & ICU

-Respiratory

et al., 1995). It has to be mentioned that additionally, 4

outcome

incidence
Study

-ICU LOS
-Seizures
duration

-Mortality
delirium

patients initially randomized for the clomethiazol arm had to

severe

-AWS

be excluded from the study because they required tracheal

suction more than every 5 minutes, and clomethiazol ther-
apy was thus stopped.
orders; CIWA-Ar, Clinical Institute Withdrawal Assessment of Alcohol Scale, Revised; MALT, Munich Alcoholism Test.
Not reported
information
AWS drugs
Additional

sedation
additional

(no more

given)

Haloperidol. Combined with unitrazepam, haloperidol

Table 1. (Continued)

is as eective and safe as unitrazepam + clonidine or EtOH

but should not be combined with clomethiazol because of
respiratory complications. This has been shown by the only
Assessment of
Definition/
scoring of

Not reported

study that evaluated haloperidol for AWS prevention in

withdrawal
symptoms
AWS

ICUs (Spies et al., 1995), see sections above and Table 1 for
details.
Esophag-

AWS Therapy
Reason

surgery
ectomy
ICU
for

Neck

A total of 8 publications were classied as LoE 1b, 1b-,

or 2b and thus included in the AWS therapy part of our
>100g/d EtOH intake

review (see Table 2 for details). Four studies were RCTs,

Alcohol dependence
during last 2 years
Inclusion criteria

and 4 investigated the eect of the implementation of a

(MALT) neck

standardized care protocol for AWS in ICUs. Investigated

dissection

drugs were BZO, GHB, and clomethiazol as single agents

or in combinations as well as haloperidol and clonidine as
adjuncts.
Intervention and

Benzodiazepines. Flunitrazepam was evaluated in 3

(clomethiazol +
midazolam vs.

symptomatic

RCTs and found to be overall safe and eective for AWS

(n = 10) vs.

haloperidol;
EtOH i.v. vs.

division not

Prophylactic
subgroup

reported)

EtOH i.v.
clonidine

therapy in ICUs.
n

therapy
(n = 52,

n = 9)

A 3-armed RCT of double-drug combinations (Spies

et al., 1996a) found better outcome results for unitraze-
pam + haloperidol compared to unitrazepam + clonidine
Two-armed RTC

(more cardiac complications) and to clomethiazol + haloper-

Study design
Three-armed

idol (more respiratory complications, longer mechanical

ventilation, and ICU LOS). This study included 159 alcohol-
RTC

dependent trauma patients with or without surgery.

A 2-armed RCT with 42 patients who underwent gastroin-
testinal tumor surgery and reported at least 60 grams of daily
Ref #

alcohol intake compared GHB with unitrazepam (Lenzenh-

5

6
 680

Table 2. Synthesis II: Controlled Trials About AWS Therapy

Intervention and Inclusion Reason for Definition/scoring Additional AWS

Ref # Study design n criteria ICU of AWS drugs Outcome and findings Limitations/weaknesses
1 Two-armed i.v. GHB (n = 14) Alcohol Severe Scoring of Clonidine, -Better initial symptom control with -Low number of pat.
RCT versus oral addicted AWS or -Tremor haloperidol, GHB -Unclear inclusion criteria
clomethiazol pat. (no other -Sweating, diazepam -No difference in outcome or ICU -Clomethiazol low dosed
(n = 12) for further internal -Nausea LOS (250 mg every 4 hours) vs.
severe AWS information) medicine -Restlessness high-dose continuous i.v.
disease (each 0 to 3 GHB
points) -No blinding
-No validated ICU-AWS score
2 Controlled not Protocol CAGE > 0 or Head and AWS Type Not reported -Analyzed items: days of restraint -Inclusion of 3 late enrollees
randomized implementation: reported neck Indicator use, various AWS symptoms and not matching with inclusion
study symptom daily alcohol surgery (qualitative complications; drug quantities; criteria (AWS risk factors) but
adjusted BZO/ intake; assessment of costs assumed postoperative AWS
clonidine/ age > 18; different AWS -Protocol pat. had less delirium -Numerous outcome items
haloperidol and squamous symptoms) symptoms. No differences in (fishing for p-values)
standardized cell CA of complications, drug dose, costs, -No CAGE or any other AWS
nutritional upper LOS, or mortality. scoring in CG
supplementation aerodigestive -Only partially ICU pat. (no
(n = 24) vs. tract; sufficient further information)
historical CG -CG: AWS
(n = 14) documented
in discharge
record
3 Controlled Protocol AWS as only Severe DSM-IV, Ryker Propofol (23%), -Analyzed items: various clinical -Retrospective data collection
nonrandomized implementation: ICU AWS Sedation haloperidol (4%) and demographic data -Various BZO used (diazepam,
study escalating transmission Analgesia -Protocol group: higher drug dose, chlordiazepoxide, lorazepam)
doses of BZO reason Scale reduced need for mechanical and reported only in diazepam
phenobarbital of > 4 points ventilation (22 vs. 47%), trend equivalents
(n = 41) vs. to reduced pneumonia rate and -All pat. were males
historical CG ICU LOS
(n = 54;
intermittent
boluses of same
drugs, no
protocol)
4 Controlled Protocol alcohol Severe Locally Other BZO -Analyzed items: time to symptom -Retrospective data collection
nonrandomized implementation: withdrawal AWS developed (midazolam, control, BZO dose and infusion -Intervention group treated with
study symptom driven delirium with symptom scale lorazepam, time, ICU and hospital LOS, different (longer acting) BZO
lorazepam ICU (Minnesota diazepam, complications than CG; most pat. received
(n = 24) vs. transmission Detoxification chlordiazepoxide) -Protocol group benefit: faster different BZO during the same
historical CG Scale) symptom control, less total BZO episode
(n = 16; dose and infusion time; trend to -4 readmissions of pat. to ICU
continuous less complications (intubations) counted as new cases
midazolam shorter ICU and hospital LOS -No validated AWS score
infusion)

Continued.
UNGUR ET AL.
 Table 2. (Continued)

Intervention and Inclusion Reason for Definition/scoring Additional AWS

Ref # Study design n criteria ICU of AWS drugs Outcome and findings Limitations/weaknesses
5 Two-armed Flunitrazepam + ICU pat. with Surgery CIWA-Ar Propofol (41%), -Analyzed outcomes: AWS -Exclusion of all patients with
double-blind clonidine (if possible or trauma piritramide severity and duration, drug dose, known alcohol dependency (in
RCT sympathetic alcohol abuse ICU LOS that case: prophylactic
hyperactivity) + but no -Bolus group: less total BZO dose, treatment provided)
haloperidol (if dependence faster symptom control, shorter -Secondary exclusions due to
productive (DSM-IV) and ICU LOS ( 6 days) due to less contradictory med. history (2
psychotic daily EtOH pneumonia (26 vs. 43%) and pat.), lethal complication (1
symptoms) as consumption less mechanical ventilation (65 pat.), treatment failure (1 pat.)
continuous of 60 g, not vs. 90%);
REVIEW: ALCOHOL WITHDRAWAL ON ICU

infusion (n = 21) intubated -Continuous infusion group: initial

vs. protocol at AWS worsening of AWS
based bolus onset
administration
(n = 23)
6 Two-armed Flunitrazepam ICU pat. with Gastro- CIWA-Ar Haloperidol and/or -Analyzed outcomes: AWS -One drop-out due to persistent
RCT (n = 21) vs. alcohol intestinal clonidine symptoms, drug dose, AEs hypernatremia
GHB (n = 21), dependence tumor -GHB group: higher need for -No blinding
both as loading (DSM-III and surgery haloperidol/more hallucinations,
dose + CAGE and more AEs (hypernatremia,
continuous 60 g daily myocloni, metabolic alkalosis)
infusion EtOH intake) -Flunitrazepam group: more
and AWS vegetative symptoms/higher need
for clonidine
-No difference in AWS severity or
duration, ICU LOS, severe
complications or mortality
7 Three-armed Flunitrazepam + Multiple Trauma CIWA-Ar None (if needed: -Analyzed outcomes: ventilation -More nicotine abuse among
RTC clonidine (n = 54) vs. injured surgery study exclusion, 4 time, ICU LOS, complications flunitrazepam + clonidine
clomethiazol + alcohol- pat.) -Flunitrazepam + clonidine group: group
haloperidol dependent more cardiac complications -No blinding
(n = 50) vs. pat. from -Clomethiazol + haloperidol group: -Secondary exclusions due to
flunitrazepam + emergency higher pneumonia rate, longer safety issues (14 pat.),
haloperidol room mechanical ventilation adjunctive antipsychotic
(n = 55) -Flunitrazepam + haloperidol: medication (4 pat.), discharge
authors recommendation for to other hospital (2 pat.),
cardiac/pulmonary risk pat. consent withdrawal (1 pat.)
8 Controlled Protocol Hospital Not CIWA-Ar Diazepam, -Analyzed items: AWS symptoms -Poor description of CG
nonrandomized implementation admission for reported midazolam, and duration, BZO dose, BZO -No information about reason
study for AWS AWS, BZO haloperidol, beta- costs, additive medication, ICU for longer ICU stay in CG
treatment: BZO therapy for blocker (not LOS, oversedation -Numerous outcome items
(n = 50, the disorder precised), -Protocol group: shorter ICU stay (fishing for p values)
lorazepam or morphine and less medication cost due to -No AWS assessment after
chlordiazepoxid) use of cheaper BZO start of therapy
vs. historical CG
(n = 57)

Ref # = reference number: 1. Elsing and colleagues (2009); 2. Lansford and colleagues (2008); 3. Gold and colleagues (2007); 4. DeCarolis and colleagues (2007); 5. Spies and colleagues
(2003); 6. Lenzenhuber and colleagues (1999); 7. Spies and colleagues (1996a); 8. Hoey and colleagues (1994); RCT, randomized controlled trial; n, number of patients; AWS, alcohol withdrawal
syndrome; LOS, length of stay; ICU, intensive care unit; GHB, gamma-hydroxybutyric acid; CG, control group; CA, carcinoma; EtOH, ethanol; i.v., intravenous; vs., versus; BZO, benzodiazepines;
pat., patient(s); CIWA-Ar, Clinical Institute Withdrawal Assessment of Alcohol Scale, Revised; DSM IV, Diagnostic and Statistic Manual of Mental Disorders, 4th edition; AE, adverse event.
681
682
uber et al., 1999), both arms allowing adjunctive haloperidol
for hallucinations and clonidine for vegetative symptoms.
The authors concluded that GHB is stronger for the treat-
ment of vegetative symptoms but weaker against psychotic
symptoms. More metabolic adverse events were seen in the
GHB arm. All other outcome parameters (AWS severity and
duration, ICU LOS, severe complications, mortality) were
equal in both arms.
The only double-blinded RCT for AWS therapy (Spies
et al., 2003) compared bolus administration of unitrazepam
with continuous infusion and found much lower intubation
rates for the bolus group and thus less ventilator associated
pneumonia and a shorter ICU LOS.
The eect of a standardized benzodiazepine protocol
implementation on dierent outcomes was analyzed in 4
studies (DeCarolis et al., 2007; Gold et al., 2007; Hoey et al.,
1994; Lansford et al., 2008). Despite some methodological
weaknessesall 4 studies had historical control groups and
were thus highly susceptible for experimenters biasit has
to be summarized that all 4 trials found clear outcome
advantages for the intervention groups (better and faster
symptom control, partly less complications, and a shorter
ICU LOS). The ndings about the total drug doses were con-
tradictory, possibly due to dierent agents used in the con-
trol and protocol groups. Other weaknesses of these 4 studies
were the partially small control group samples of 14
(Lansford et al., 2008) and 16 (DeCarolis et al., 2007)
patients, too many outcome items, poor reporting of addi-
tional AWS medication, and poor control group descrip-
tions.
Gamma-Hydroxybutyric Acid. A 2-armed RCT with 26
patients classied as alcohol addicted and being admitted as ecient as unitrazepam + haloperidol or clomethiazol +
in ICUs because of AWS or other internal medicine disease
(Elsing et al., 2009) compared GHB with clomethiazol while
clonidine was permitted as adjunct at any time and BZO and
haloperidol after the seventh hour. As more than 60% of the
patients in both study arms received additional AWS medi-
cation and the sample of patients in this trial was quite small,
it is dicult to draw conclusions of the outcome results.
Another weakness of this trial is the relatively small dosage
of oral clomethiazol (250 mg every 4 hours, not body weight
adjusted) compared with high-dose i.v. GHB (initially
30 mg/kg, then 15 mg/kg). However, apart from less AWS
symptoms during the rst 7 hours in the GHB group, possi-
bly explained by the dosage bias, the authors described no
signicant dierences between the 2 cohorts.
Another RCT already described in the BZO chapter above
compared GHB with unitrazepam (Lenzenhuber et al.,
1999), both arms allowing adjunctive haloperidol and cloni-
dine. The authors concluded that GHB is stronger for the
treatment of vegetative symptoms but weaker against psy-
chotic symptoms and associated with more metabolic
adverse events. All other outcome parameters (AWS severity
and duration, ICU LOS, severe complications, mortality)
were equal in both arms.
UNGUR ET AL.
Clomethiazol. A larger 3-armed RCT of double-drug
combinations (Spies et al., 1996a) compared clomethiazol +
haloperidol with benzodiazepine-based regimes (unitraze-
pam + haloperidol and unitrazepam + clonidine) for trauma
ICU patients. Clomethiazol-treated patients were found to
have a higher pneumonia rate and a longer time spent
on mechanical ventilation. Thus, the authors disadvised
clomethiazol for critically ill patients.
A small RCT (Elsing et al., 2009) compared GHB with
clomethiazol. Despite the methodological weaknesses of this
trial mentioned in the GHB chapter above, it has to be high-
lighted that clomethiazol was less ecient while it had no
advantages compared to the GHB arm.
Haloperidol. The authors of a 3-armed RCT of double-
drug combinations described in the paragraphs above and in
Table 2 (Spies et al., 1996a) advise the combination of unit-
razepam + haloperidol especially for patients with cardiac
risk because of reported cardiac adverse events in the cloni-
dine-containing arm and the higher pneumonia rate in the
clomethiazol-based arm.
In the other analyzed AWS therapy trials, haloperidol was
not tested against other agents but used as adjunct for psy-
chotic AWS symptoms within the study protocol (Lansford
et al., 2008; Spies et al., 2003) or reported as part of addi-
tional AWS therapy in case of insucient ecacy of the pro-
tocol agents (Elsing et al., 2009; Gold et al., 2007; Hoey
et al., 1994; Lenzenhuber et al., 1999).
Alpha2-Agonists. In the 3-armed RCT of double-drug
combinations described previously (Spies et al., 1996a),
clonidine was combined with unitrazepam and found to be
haloperidol. However, the authors did not recommend this
combination for cardiac risk patients due to more cardiac
complications seen in the clonidine arm.
In the other analyzed AWS therapy trials, clonidine was
not tested against other agents but used as adjunct for vege-
tative AWS symptoms within the study protocol (Lansford
et al., 2008; Spies et al., 2003) or reported as part of addi-
tional o-protocol AWS therapy (Elsing et al., 2009; Len-
zenhuber et al., 1999).
DISCUSSION
AWS Prevention
There is sucient evidence proving that BZO are eec-
tive for AWS prevention in single administration (Huber
et al., 1990; Weinberg et al., 2008) and in combination with
haloperidol or clonidine (Huber et al., 1990; Spies et al.,
1995). Despite the diculty to dierentiate paradox eects
of BZO from AWS symptoms, clinicians should be aware
of this potential risk. However, this adverse eect has not
been reported in any of our analyzed articles.
REVIEW: ALCOHOL WITHDRAWAL ON ICU
The alpha2-agonist clonidine as single agent might be infe-
rior compared to EtOH or midazolam, but this has only been
investigated in 1 trial with insuciently reported details on
statistics and patient samples (Huber et al., 1990).
Combinations containing clomethiazol should not be
preferred in critically ill patients because of bronchial
hypersecretion (Spies et al., 1995). However, this trial failed
to show statistical signicant dierences in ICU LOS or mor-
tality. On the other hand, as 1 other trial showed longer ven-
tilation times and higher pneumonia rates in critically ill
patients receiving clomethiazol for AWS therapy (Spies
et al., 1996a), further investigation of this drug in ICUs
appears ethically questionable.
Moderately dosed i.v. EtOH (2 to 4 g/h) seems to be safe
and as eective as the drugs above in preventing AWS
(Dissanaike et al., 2006; Heil et al., 1990; Spies et al., 1995).
This stands in contrast to the critical appraisal of some other
authors regarding i.v. EtOH administration (Sar and Gold,
2010; Vagts et al., 2003). This critical standpoint might be
explained by blending of prophylaxis and therapy in litera-
ture assessment. Furthermore, Sar bases his skepticism on
a trial with 32 treated patients (Eggers et al., 2002) without a
control group. In this trial, patients received a quite high-
loading dose of EtOH reaching a BAC of 0.7 mg/ml after
45 min. Then, they received a continuous EtOH infusion
dosed upon their individual anamnestic drinking habit
reports that lead to a BAC of around 0.1 mg/ml at day 2.
The AWS-enduring patients (40%, called failure group)
had a BAC of closely 0.0 mg/ml in contrast to 0.15 to
0.3 mg/ml in the success group. This suggests a bias due to
inadequate dosing and supports the need of a standardized
body weightadjusted EtOH infusion protocol. By imple-
menting such a protocol (Dissanaike et al., 2006), the failure
rate was dropped from 20% to 8%, and the mean treatment
was shortened from 7 to 3 days in 160 patients comprising
trial. Numerous previous publications see i.v. EtOH as an
acceptable agent for AWS prophylaxis (Engelhardt, 1996;
Kork et al., 2010; Sander et al., 2006; Spies and Rommelsp-
acher, 1999; Vagts and Noldge-Schomburg, 2002). More
powerful multicenter trials are required to answer the ques-
tion if any of the above-discussed agents or combinations is
superior to the others.
There are many other potential substances or methods to
prevent AWS: some authors stress the importance of detect-
ing patients at risk by screening tools as well as general
measures like nutrition, volume resuscitation, and preopera-
tive abstinence (Kork et al., 2010; Sar and Gold, 2010).
Two general review articles conclude that GHB can also be
eective to prevent AWS in ICUs (Kleinschmidt and
Mertzlut, 1995; Meyer et al., 2005), but there are no
published controlled trials that investigated GHB for AWS
prevention. BZO seem to represent the rst choice substance
class to prevent AWS following previous literature recom-
mendations (Spies and Rommelspacher, 1999). Anticonvul-
sants in contrary are not ecient in AWS prevention,
following a review of Al-Sanouri and colleagues (2005) based
683
on ambulant patients trials, but, however, this has not yet
been studied in ICU patients.
AWS Therapy
Clomethiazol, for many years a common and widely
recommended agent against AWS (Batel and Lariviere,
2000; Croissant and Mann, 2000; Dittmar, 1991, 1994;
Engelhardt, 1996; Horstmann et al., 1989; Naber et al.,
1991; Spies and Rommelspacher, 1999; Tiecks and Einhaupl,
1994; Vagts et al., 2003), was shown to be inferior to regi-
mens based on BZO especially due to adverse events related
to bronchial hypersecretion (Spies et al., 1996a) and inferior
to GHB regarding initial symptom control (Elsing et al.,
2009). Despite many case reports showing the ecacy of clo-
methiazol against AWS, it should be used with caution in
critically ill patients if they are in danger of needing mechani-
cal ventilation.
GHB was inferior to a benzodiazepine regimen regard-
ing ecacy in hallucination control and metabolic side
eects (Lenzenhuber et al., 1999) but still more potent
than low-dosed clomethiazol following Elsings small RCT
(Elsing et al., 2009). Even if some experts see it as an eec-
tive agent for AWS treatment in ICUs (Kleinschmidt and
Mertzlut, 1995; Meyer et al., 2005; Vagts et al., 2003)
and our analyzed studies showed a sucient eect espe-
cially for vegetative AWS symptoms, it has to be seen criti-
cally because of its toxicity (Lenzenhuber et al., 1999) and
its hallucinatory side eect. Another important concern
about GHB is its strong dependence potential, a reason
why it is declared as a controlled substance in some coun-
tries, and it is not approved at all in some other countries.
This might also be a concern about a general recommen-
dation for this drug. On the other hand, GHB is approved
in a few European countries to treat alcohol withdrawal
and dependence.
The only larger RCT investigating AWS therapy (n = 159;
Spies et al., 1996a) prefers a combination of a benzodiaze-
pine (unitrazepam) and haloperidol for trauma or surgical
ICU patients because of worse outcome in clomethiazol-trea-
ted patients and cardiac complications associated with cloni-
dine. As shown in another double-blinded trial (Spies et al.,
2003), BZO should be administered symptom driven instead
of continuously. This was shown to be more eective, safer,
and less expensive. Clear outcome advantages of a standard-
ized protocol for benzodiazepine administration against
AWS in ICUs have been shown in all of our 4 analyzed
cohort trials (DeCarolis et al., 2007; Gold et al., 2007; Hoey
et al., 1994; Lansford et al., 2008). This conforms with
numerous recent review articles, case reports, and experts
recommendations where BZO are seen as standard treatment
for AWS in ICUs (Al-Sanouri et al., 2005; Bard et al., 2006;
Batel and Lariviere, 2000; Croissant and Mann, 2000; Dar-
rouj et al., 2008; Hayner et al., 2009; van Klei et al., 2000;
McCowan and Marik, 2000; Petignat, 2005; Phillips et al.,
2006; Powell, 1999; Remennik et al., 2005; Sander et al.,
684
2006; Sar and Gold, 2010; Skrobik, 2007, 2009; Spies and
Rommelspacher, 1999; Verstraete et al., 2008; de Wit et al.,
2007).
Clonidine and haloperidol were used as adjuncts for auto-
nomic respectively productive-psychotic symptoms, and its
need was seen as surrogate markers for the potency or weak-
ness of the backbone AWS drug in some studies (Elsing
et al., 2009; Gold et al., 2007; Hoey et al., 1994; Lansford
et al., 2008; Lenzenhuber et al., 1999; Spies et al., 1996a,
2003). No controlled trial investigated these substances as
single agents for AWS therapy in ICUs. Even if none of our
analyzed studies reported haloperidol-induced seizures as
adverse events, clinicians should be aware of the potential
epileptogenic side eect.
Publications discussing EtOH, phenobarbital, propofol,
anticonvulsants, beta-blockers, opioids, thiamine, physo-
stigmine, or dexmedetomidine for AWS therapy in ICUs
are based on case report or expert opinion evidence and
see these agents mostly as second-line therapy or adjuncts
for specic symptoms or comorbidities. (Al-Sanouri et al.,
2005; Darrouj et al., 2008; Dittmar, 1991, 1994; Engel-
hardt, 1996; Eyer et al., 2011; Hayner et al., 2009; van
Klei et al., 2000; Krumholz, 1996; McCowan and Marik,
2000; Muhl, 2006; Powell, 1999; Remennik et al., 2005;
Sar and Gold, 2010; Spies and Rommelspacher, 1999;
Vagts and Noldge-Schomburg, 2002; de Wit et al., 2007;
Zielmann et al., 2003).
Limitation
One important limitation of this review might consist
in the overall thin evidence base: some of the available
controlled trials had small patient samples (DeCarolis et al.,
2007; Elsing et al., 2009; Heil et al., 1990; Lansford
et al., 2008; Mertes et al., 1996) and inadequate (Mertes
et al., 1996) or too numerous outcome items to allow strong
and concrete AWS prevention or therapy recommendations.
Furthermore, SRs are always potentially subject to publica-
tion bias, as we do not know if studies with negative or unfa-
vorable results have remained unpublished. The authors
neutrality might be aected by the fact that 3 of the analyzed
studies as well as some of other cited publications originate
from their own department. On the other hand, it can be seen
as strength that the authors used the PRISMA criteria and
only highlighted recommendations based on evidence of
higher level.
CONCLUSIONS
BZO can be regarded as the standard of care for AWS pre-
vention and treatment. Symptom-triggered bolus administra-
tion is more benecial than continuous infusion for AWS
therapy. Delirium monitoring with a validated score is obli-
gatory. Dierences in dosage and optimal adjunct substances
are still to be found in further studies.
UNGUR ET AL.
Clomethiazol should not be used in critically ill patients
due to its risk of tracheobronchitis and pneumonia through
higher bronchial secretion.
EtOH is eective in AWS prevention but not recom-
mended for AWS therapy.
GHB was found to be eective for AWS therapy but is not
recommended as rst-choice agent because of safety con-
cerns, insucient potency against hallucinations, and legal
concerns. It is also not available in all countries.
Safety concerns are seen in clonidine (cardio circulatory
side eects) and haloperidol (QT prolongation), but both
drugs are well established and have a proven ecacy as
adjunctive substances for specic AWS autonomic (cloni-
dine) or psychotic (haloperidol) symptom control. As
clonidine is not available in all countries, a more potent
alpha2-agonist, dexmedetomidine might also be used (LoE
4; Tang et al., 2011).
For all other possibly helpful agents, the riskbenet
ratios compared to the substances above remain to be eval-
uated in controlled trials. Useful further study designs
could be placebo-controlled AWS prevention trials to evalu-
ate a general benet of screening and prevention of AWS
risk. To optimize AWS therapy, a next step could be to
compare longer with shorter acting BZO in double-blinded
trials or to evaluate other adjuncts in placebo-controlled
trials.
DECLARATION OF INTEREST
Three analyzed trials have been performed by the authors
department. The authors report no further conicts of inter-
est. The authors alone are responsible for content and writ-
ing of this article.
ACKNOWLEDGMENTS
The authors thank Dr. Yoanna Skrobik for an interesting,
fruitful discussion. This systematic review has been funded
by institutional resources of the Department of Anesthesiol-
ogy and Operative Intensive Care Medicine at the Charite
Universitatsmedizin Berlin.
REFERENCES
Al-Sanouri I, Dikin M, Soubani AO (2005) Critical care aspects of alcohol
abuse. South Med J 98:372381.
Bard MR, Goettler CE, Toschlog EA, Sagraves SG, Schenarts PJ, Newell
MA, Fugate M, Rotondo MF (2006) Alcohol withdrawal syndrome:
turning minor injuries into a major problem. J Trauma 61:14411445; dis-
cussion 14451446.
Batel P, Lariviere P (2000) Late-onset alcohol withdrawal syndrome. Ann
Med Interne (Paris) 151:B27B29.
Croissant B, Mann K (2000) Alcohol withdrawal syndrome and its treat-
ment. Ther Umsch 57:257260.
Darrouj J, Puri N, Prince E, Lomonaco A, Spevetz A, Gerber DR (2008)
Dexmedetomidine infusion as adjunctive therapy to benzodiazepines for
acute alcohol withdrawal. Ann Pharmacother 42:17031705.
REVIEW: ALCOHOL WITHDRAWAL ON ICU
DeCarolis DD, Rice KL, Ho L, Willenbring ML, Cassaro S (2007) Symp-
tom-driven lorazepam protocol for treatment of severe alcohol withdrawal
delirium in the intensive care unit. Pharmacotherapy 27:510518.
Dissanaike S, Halldorsson A, Frezza EE, Griswold J (2006) An ethanol proto-
col to prevent alcohol withdrawal syndrome. J Am Coll Surg 203:186191.
Dittmar G (1991) Alcohol deliriumpathogenesis and therapy. Med Klin
(Munich) 86:607612.
Dittmar G (1994) Alcoholic deliriuma potentially life-threatening condi-
tion. Fortschr Med 112:274276.
Eggers V, Tio J, Neumann T, Pragst F, Muller C, Schmidt LG, Kox WJ,
Spies CD (2002) Blood alcohol concentration for monitoring ethanol
treatment to prevent alcohol withdrawal in the intensive care unit. Inten-
sive Care Med 28:14751482.
Elsing C, Stremmel W, Grenda U, Herrmann T (2009) Gamma-hydroxybu-
tyric acid versus clomethiazole for the treatment of alcohol withdrawal
syndrome in a medical intensive care unit: an open, single-center random-
ized study. Am J Drug Alcohol Abuse 35:189192.
Engelhardt W (1996) Alcohol administration for prevention of withdrawal
delirium in alcohol dependent surgical intensive care patients: pro. Anas-
thesiol Intensivmed Notfallmed Schmerzther 31:183184.
Eyer F, Schreckenberg M, Hecht D, Adorjan K, Schuster T, Felgenhauer N,
Pfab R, Strubel T, Zilker T (2011) Carbamazepine and valproate as
adjuncts in the treatment of alcohol withdrawal syndrome: a retrospective
cohort study. Alcohol Alcohol 46:177184.
Feuerlein W, Ringer C, Kufner H, Antons K (1979) Diagnosis of alcoholism:
the Munich Alcoholism Test (MALT). Curr Alcohol 7:137147.
Gold JA, Rimal B, Nolan A, Nelson LS (2007) A strategy of escalating
doses of benzodiazepines and phenobarbital administration reduces the
need for mechanical ventilation in delirium tremens. Crit Care Med
35:724730.
Hayner CE, Wuestefeld NL, Bolton PJ (2009) Phenobarbital treatment in a
patient with resistant alcohol withdrawal syndrome. Pharmacotherapy
29:875878.
Heil T, Martens D, Eyrich K (1990) Alcohol withdrawal syndrome in the
postoperative phasetherapy or prevention? Langenbecks Arch Chir Sup-
pl II Verh Dtsch Ges Chir 1990:11371140.
Hoey LL, Nahum A, Vance-Bryan K (1994) A prospective evaluation of
benzodiazepine guidelines in the management of patients hospitalized for
alcohol withdrawal. Pharmacotherapy 14:579585.
Horstmann E, Conrad E, Daweke H (1989) Severe course of delirium tre-
mens. Results of treatment and late prognosis. Med Klin (Munich) 84:
569573.
Huber FT, Bartels H, Siewert JR (1990) Treatment of postoperative alcohol
withdrawal syndrome after esophageal resection. Langenbecks Arch Chir
Suppl II Verh Dtsch Ges Chir 1990:11411143.
van Klei WA, Havenaar JM, Klijn FA, van Dijk A (2000) Ethanol for
treatment of delirium in alcohol dependent patients on intensive care
units in the Netherlands: ecacy not proven. Ned Tijdschr Geneeskd
144:710713.
Kleinschmidt S, Mertzlut F (1995) Gamma-hydroxybutyric acidsigni- Spies CD, Otter HE, Huske B, Sinha P, Neumann T, Rettig J, Lenzenhuber
cance for anesthesia and intensive care medicine? Anasthesiol Intensivmed
Notfallmed Schmerzther 30:393402.
Kork F, Neumann T, Spies C (2010) Perioperative management of patients
with alcohol, tobacco and drug dependency. Curr Opin Anaesthesiol
23:384390.
Krumholz W (1996) Alcohol as therapeutic drug in alcohol dependent inten-
sive care patients: contra. Anasthesiol Intensivmed Notfallmed Schmerz-
ther 31:181182.
Lansford CD, Guerriero CH, Kocan MJ, Turley R, Groves MW, Bahl V,
Abrahamse P, Bradford CR, Chepeha DB, Moyer J, Prince ME, Wolf
GT, Aebersold ML, Teknos TN (2008) Improved outcomes in patients
with head and neck cancer using a standardized care protocol for post-
operative alcohol withdrawal. Arch Otolaryngol Head Neck Surg 134:
865872.
Lenzenhuber E, Muller C, Rommelspacher H, Spies C (1999) Gamma-
hydroxybutyrate for treatment of alcohol withdrawal syndrome in inten-
685
sive care patients. A comparison between with two symptom-oriented
therapeutic concepts. Anaesthesist 48:8996.
McCowan C, Marik P (2000) Refractory delirium tremens treated with prop-
ofol: a case series. Crit Care Med 28:17811784.
Mertes N, Goeters C, Kuhmann M, Zander JF (1996) Postoperative alpha
2-adrenergic stimulation attenuates protein catabolism. Anesth Analg
82:258263.
Meyer S, Kleinschmidt S, Gottschling S, Gortner L, Strittmatter M (2005)
Gamma-hydroxy butyric acid: neurotransmitter, sedative and party drug.
Wien Med Wochenschr 155:315322.
Muhl E (2006) Delirium and intensive care unit syndrome. Chirurg 77:
463471; quiz 472.
Naber M, Franz W, Overbeck W (1991) Characteristics of alcohol with-
drawal delirium in surgical patients and recommendations for treatment.
Chirurg 62:133137.
Petignat PA (2005) The management of the alcohol withdrawal syndrome in
the intensive care unit. Rev Med Suisse 1:29052911.
Phillips S, Haycock C, Boyle D (2006) Development of an alcohol with-
drawal protocol: CNS collaborative exemplar. Clin Nurse Spec 20:190
198; quiz 199200.
Powell AH (1999) Alcohol withdrawal in critical care. Dimens Crit Care
Nurs 18:2428.
Remennik AG, Nepomnyashchikh LM, Remennik VI (2005) Progressive
toxemia with acetaldehyde in a reactive form of alcohol withdrawal
syndrome. Bull Exp Biol Med 139:732734.
Sander M, Neumann T, von Dossow V, Schonfeld H, Lau A, Eggers V, Spies
C (2006) Alcohol use disorder: risks in anesthesia and intensive care medi-
cine. Internist (Berl) 47:332, 334336, 338, passim.
Sar M, Gold JA (2010) Alcohol withdrawal syndromes in the intensive care
unit. Crit Care Med 38:S494S501.
Skrobik Y (2007) Protocols, practice, and patientsthe case of alcohol with-
drawal. Crit Care Med 35:955.
Skrobik Y (2009) Broadening our perspectives on ICU delirium risk factors.
Crit Care 13:160.
Spies CD, Dubisz N, Funk W, Blum S, Muller C, Rommelspacher H,
Brummer G, Specht M, Hannemann L, Striebel HW, Schaartzik W
(1995) Prophylaxis of alcohol withdrawal syndrome in alcohol-dependent
patients admitted to the intensive care unit after tumour resection. Br J
Anaesth 75:734739.
Spies CD, Dubisz N, Neumann T, Blum S, Muller C, Rommelspacher H,
Brummer G, Specht M, Sanft C, Hannemann L, Striebel HW, Schaartzik
W (1996a) Therapy of alcohol withdrawal syndrome in intensive care unit
patients following trauma: results of a prospective, randomized trial. Crit
Care Med 24:414422.
Spies CD, Neuner B, Neumann T, Blum S, Muller C, Rommelspacher H,
Rieger A, Sanft C, Specht M, Hannemann L, Striebel HW, Schaartzik
W (1996b) Intercurrent complications in chronic alcoholic men admitted
to the intensive care unit following trauma. Intensive Care Med 22:
286293.
E, Kox WJ, Sellers EM (2003) Alcohol withdrawal severity is decreased by
symptom-orientated adjusted bolus therapy in the ICU. Intensive Care
Med 29:22302238.
Spies CD, Rommelspacher H (1999) Alcohol withdrawal in the surgical
patient: prevention and treatment. Anesth Analg 88:946954.
Tang JF, Chen PL, Tang EJ, May TA, Stiver SI (2011) Dexmedetomidine
controls agitation and facilitates reliable, serial neurological examinations
in a non-intubated patient with traumatic brain injury. Neurocrit Care
15:175181.
Tiecks FP, Einhaupl KM (1994) Treatment alternatives of alcohol with-
drawal delirium. Nervenarzt 65:213219.
Vagts DA, Iber T, Noldge-Schomburg GF (2003) Alcohola perioperative
problem of anaesthesia and intensive care medicine. Anasthesiol Intensiv-
med Notfallmed Schmerzther 38:747761.
Vagts DA, Noldge-Schomburg GF (2002) Managing anesthesia in the
alcoholic patient. Anaesthesiol Reanim 27:160167.
686 UNGUR ET AL.

Verstraete L, Joosten E, Milisen K (2008) Opinions of physicians and nurses Zielmann S, Petrow H, Walther P, Henze T (2003) Intensive care of delirium
regarding the prevention, diagnosis and management of delirium. Tijdschr syndromes. Anaesthesiol Reanim 28:812.
Gerontol Geriatr 39:2634.
Weinberg JA, Magnotti LJ, Fischer PE, Edwards NM, Schroeppel T, Fabian
TC, Croce MA (2008) Comparison of intravenous ethanol versus diaze- SUPPORTING INFORMATION
pam for alcohol withdrawal prophylaxis in the trauma ICU: results of a
randomized trial. J Trauma 64:99104. Additional Supporting Information may be found in the
de Wit M, Wan SY, Gill S, Jenvey WI, Best AM, Tomlinson J, Weaver MF
(2007) Prevalence and impact of alcohol and other drug use disorders on
online version of this article:
sedation and mechanical ventilation: a retrospective study. BMC Anesthe- Data S1. Addendum.
siol 7:3.