Effect of early and focused benzodiazepine therapy on length of stay in severe alcohol withdrawal syndrome

Abstract

Objective: Current evidence supports symptom-triggered therapy for alcohol withdrawal syndrome (AWS). Early, escalating therapy with benzodiazepines (BZD) appears to decrease ICU length of stay (LOS); however, the effect on hospital LOS remains unknown. The hypothesis of this study is that focused BZD treatment in the first 24 h will decrease hospital LOS.
Design: Pre–post cohort study.
Setting: Academic medical center.
Patients: This study included patients with severe AWS. The pre-intervention cohort (PRE) was admitted between January and November 2015. The post-intervention cohort (POST) was admitted between April 2016 and March 2017. Severe AWS was defined as patients requiring diazepam doses of >30 mg. Focused treatment was defined as >50% of total diazepam usage within the first 24 h of recognition of AWS.
Intervention: In the PRE group, patients received symptom-triggered, escalating doses of diazepam and phenobarbital based on their Richmond Agitation-Sedation Scale (RASS). In the POST group, patients received a revised, time-limited course of therapy: escalating doses of BZD and phenobarbital were given during a 24-h loading phase, and all therapy was discontinued after a 72-h tapering phase. The SHOT scale was used as an adjunct to RASS to assess non-agitation symptoms of AWS and guide additional diazepam doses.
Measurements and main results: The primary outcome was hospital LOS; secondary outcomes included ICU LOS, BZD use, and ventilator-free days. Five hundred thirty-two patients were treated using the AWS protocol; 113 experienced severe AWS. The PRE (n = 75) and POST (n = 38) groups were evenly matched in age, sex, history of AWS, and severity of illness. There was a substantial difference in POST patients who received focused treatment (51.3% vs. 73.7%, p = .03). The POST group had a significant decrease in hospital LOS (14.0 vs. 9.8 days, p = .03) and ICU LOS (7.4 vs. 4.4 days, p = .03).
Conclusion: Early, focused management of severe AWS was associated with a decrease in ICU and hospital LOS.

Full text

Full Terms & Conditions of access and use can be found at
https://www.tandfonline.com/action/journalInformation?journalCode=ictx20
Clinical Toxicology
ISSN: 1556-3650 (Print) 1556-9519 (Online) Journal homepage: https://www.tandfonline.com/loi/ictx20
Effect of early and focused benzodiazepine
therapy on length of stay in severe alcohol
withdrawal syndrome
Jin A. Lee, Jeremiah J. Duby & Christine S. Cocanour
To cite this article: Jin A. Lee, Jeremiah J. Duby & Christine S. Cocanour (2019): Effect of early
and focused benzodiazepine therapy on length of stay in severe alcohol withdrawal syndrome,
Clinical Toxicology, DOI: 10.1080/15563650.2018.1542701
To link to this article: https://doi.org/10.1080/15563650.2018.1542701
View supplementary material
Published online: 07 Feb 2019.
Submit your article to this journal
Article views: 179
View Crossmark data

CLINICAL RESEARCH
Effect of early and focused benzodiazepine therapy on length of stay in severe
alcohol withdrawal syndrome
Jin A. Lee
a
, Jeremiah J. Duby
a
and Christine S. Cocanour
b
a
Department of Pharmacy Services, University of California, Davis Medical Center, Sacramento, CA, USA;
b
Department of Surgery, University
of California, Davis Medical Center, Sacramento, CA, USA
ABSTRACT
Objective: Current evidence supports symptom-triggered therapy for alcohol withdrawal syndrome
(AWS). Early, escalating therapy with benzodiazepines (BZD) appears to decrease ICU length of stay
(LOS); however, the effect on hospital LOS remains unknown. The hypothesis of this study is that
focused BZD treatment in the first 24 h will decrease hospital LOS.
Design: Pre–post cohort study.
Setting: Academic medical center.
Patients: This study included patients with severe AWS. The pre-intervention cohort (PRE) was admit-
ted between January and November 2015. The post-intervention cohort (POST) was admitted between
April 2016 and March 2017. Severe AWS was defined as patients requiring diazepam doses of >30 mg.
Focused treatment was defined as >50% of total diazepam usage within the first 24h of recognition
of AWS.
Intervention: In the PRE group, patients received symptom-triggered, escalating doses of diazepam
and phenobarbital based on their Richmond Agitation-Sedation Scale (RASS). In the POST group,
patients received a revised, time-limited course of therapy: escalating doses of BZD and phenobarbital
were given during a 24-h loading phase, and all therapy was discontinued after a 72-h tapering phase.
The SHOT scale was used as an adjunct to RASS to assess non-agitation symptoms of AWS and guide
additional diazepam doses.
Measurements and main results: The primary outcome was hospital LOS; secondary outcomes
included ICU LOS, BZD use, and ventilator-free days. Five hundred thirty-two patients were treated
using the AWS protocol; 113 experienced severe AWS. The PRE (n¼75) and POST (n¼38) groups
were evenly matched in age, sex, history of AWS, and severity of illness. There was a substantial differ-
ence in POST patients who received focused treatment (51.3% vs. 73.7%, p¼.03). The POST group
had a significant decrease in hospital LOS (14.0 vs. 9.8 days, p¼.03) and ICU LOS (7.4 vs. 4.4
days, p¼.03).
Conclusion: Early, focused management of severe AWS was associated with a decrease in ICU and
hospital LOS.
ARTICLE HISTORY
Received 22 June 2018
Revised 27 September 2018
Accepted 22 October 2018
Published online 6 February
2019
KEYWORDS
Alcohol withdrawal; alcohol;
critical care;
benzodiazepine;
phenobarbital; delirium
Introduction
An estimated 14.5 million people in the United States report
having alcohol use disorder [1]. Alcohol-associated events are
responsible for >500,000 emergency department visits annu-
ally. Up to 33% of all patients admitted to the intensive care
unit (ICU) are at risk for developing symptoms of alcohol
withdrawal syndrome (AWS) [2]. Patients who progress to
delirium tremens (DT), a severe complication of severe AWS,
are estimated to have a 5–15% mortality rate [3]. Alcoholism
and AWS are associated with increased hospital length of
stay (LOS), nosocomial infection, and duration of mechanical
ventilation [4,5]. Despite significant mortality and morbidity,
a standard of care does not exist.
AWS is primarily due to an imbalance of the inhibitory
neurotransmitter, gamma-aminobutyric acid (GABA), and the
excitatory amino acid, glutamate. Due to the depressant
effects of alcohol, chronic alcoholism leads to down-regula-
tion of the GABA receptor with a compensatory up-regula-
tion of the N-methyl-D-aspartate (NMDA) receptor, the
binding site for glutamate. In AWS, the depressant effects of
alcohol are acutely removed from the system, and the excita-
tory glutamate floods the central nervous system (CNS). The
primary goal of AWS treatment is to prevent progression to
DT by correcting this imbalance. Benzodiazepines (BZDs) are
widely regarded as the cornerstone of therapy due to their
GABA
A
-potentiating effects. More recently, phenobarbital has
gained traction for treatment of AWS by addressing potential
BZD resistance through its dual mechanism of GABA
A
-
potentiation and glutamate inhibition [6].
Current evidence indicates that a symptom-triggered
treatment strategy utilizing BZDs decreases ICU LOS [7,8]. It
CONTACT Jin A. Lee xjlee@ucdavis.edu Department of Pharmacy Services, University of California, Davis –Medical Center, 2315 Stockton Boulevard,
Sacramento, CA 95817, USA
Supplementary data for this article can be accessed here
ß2019 Informa UK Limited, trading as Taylor & Francis Group
CLINICAL TOXICOLOGY
https://doi.org/10.1080/15563650.2018.1542701

has been suggested that early and aggressive titration of
medication guided by symptoms may improve AWS treat-
ment outcomes by avoiding progression to DT [5]; however,
the effects of such a strategy are unknown. The hypothesis
of this study is that focused BZD treatment in the first 24 h
after recognition of AWS will decrease overall hospital LOS.
Methods
This pre–post cohort study was conducted at a large, tertiary,
academic medical center in adult patients (18 years of age or
older) diagnosed with severe AWS admitted between
January 2015 and March 2017. Severe AWS was defined as
patients who received at least one 30-mg dose of diazepam.
This inclusion criterion, though somewhat arbitrary, is similar
to previous reports that utilized a BZD dose cutoff for ICU
admission [7]. Patients were excluded for gross protocol
deviation, i.e., incorrect BZD dose escalation.
The pre-intervention cohort (PRE, January 2015 to
November 2015) was treated based on an algorithm utilizing
escalating bolus doses of diazepam and phenobarbital based
on RASS, which our institution previously developed for the
management of severe AWS and was shown to decrease ICU
LOS, time of mechanical ventilation, and BZD requirements
[8]. The post-intervention cohort (POST, April 2016 to March
2017) was treated based on a revised algorithm emphasizing
a time-limited loading phase. Escalating doses of diazepam
and phenobarbital were administered based on RASS >2 for
the first 24 h, then de-escalated to a tapering phase for the
following 72 h (Supplementary Figure 1). BZD escalation in
the loading phase was simplified in the treatment algorithm
with an emphasis on identifying and repeating the minimum
effective dose. To address the non-agitation symptoms of
withdrawal, the SHOT scale was utilized (Table 1). The SHOT
scale (sweating, hallucinations, orientation, and tremor) was
originally devised to assess pretreatment severity of alcohol
withdrawal in the emergency department [9]. A SHOT score
of greater than 3 triggered administration of a 10-mg dose
of diazepam.
The primary objective of this study was to assess the
effect of a revised AWS protocol emphasizing focused BZD
use on hospital LOS. Focused BZD use was defined as having
received 50% of a patient’s total diazepam use within the
first 24 h of onset of severe AWS. Secondary endpoints
included ICU LOS, total BZD use in diazepam equivalents,
total phenobarbital requirements, ventilator-free days, and
use of flumazenil.
Demographic information, laboratory data, ventilator days,
hospital and ICU LOS, and medication administration infor-
mation were gathered through the electronic medical record.
Descriptive statistics were utilized to analyze demographic
data. Comparison of baseline characteristics and outcomes
were assessed using Student’st-test, Chi-square test, or
Mann–Whitney Utest as appropriate. Kaplan–Meier curves
were utilized to determine the effectiveness of treatment in
terms of LOS. The study was approved by the Institutional
Review Board at University of California, Davis and the
requirement for informed consent was waived.
Results
Of 532 patient admissions utilizing the AWS protocol, a total
of 113 progressed to severe AWS (Figure 1). In the PRE group
(n= 75), 30% of the patients met the definition of severe
AWS. In the POST group (n= 38), only 13% of the patients
met the definition of severe AWS
The PRE and POST groups were evenly matched in age,
sex, history of alcohol withdrawal, and Sequential Organ
Failure Assessment (SOFA) scores (Table 2). The majority of
patients in each group were admitted to the medical ICU,
and 15–20% were admitted to the surgical ICU. The only
statistically significant difference noted in the baseline
Table 1. SHOT scale.
Symptom Points
Sweating 0 –No sweating visible
1–Palms moderately moist
2–Beads of sweat visible on forehead
Hallucinations 0 –No hallucinations
1–Tactile hallucinations only
2–Visual and/or auditory hallucinations
Orientation 0 –Oriented
1–Disoriented for date by 1 month or more
2–Disoriented to place or person
Tremor 0 –No tremor
1–Minimally visible tremor
2–Mild
3–Moderate
4–Severe
Table 2. Baseline characteristics of PRE and POST groups.
Characteristic PRE (n¼75) POST (n¼38) pvalue
Age, mean ± SD 50.8 ± 11.2 49.1 ± 9.4 .48
Males, n(%) 67 (89.3) 30 (79.0) .14
History of AWS, n(%) 54 (72.0) 26 (68.4) .69
BAL on admit, mg/dL 161.8 95.9 .04
SOFA score on admit 4.0 ± 2.6 3.5 ± 2.9 .45
Service, n(%)
Medical ICU 49 (65.3) 25 (65.8) .96
Surgical ICU 17 (22.7) 6 (15.8) .39
Other 9 (12.0) 7 (18.4) .36
AWS: alcohol withdrawal syndrome; BAL: blood alcohol level; SOFA: sequential
organ failure assessment.
Figure 1. Flow diagram of patients included in study analysis.
2 J. A. LEE ET AL.

characteristics was a higher blood alcohol level (BAL) noted
upon admission in the PRE group compared to the POST
group (161.8 vs. 95.9 mg/dL, p= .04).
Both PRE and POST cohorts had similar time to first BZD
dose, and there was no statistically significant difference in
total diazepam use, days of BZD exposure, or total phenobar-
bital use (Table 3). However, there was a statistically signifi-
cant difference noted in the percent of patients in each
cohort who received focused treatment, defined as patients
who received 50% of their total diazepam usage within the
first 24 h of onset of AWS (51.3% vs. 73.7%, p= .03).
Table 4 shows primary and secondary outcomes. The pri-
mary outcome of hospital LOS was significantly shorter in
the POST group (14.0 vs. 9.8 days, p= .03). A similar
decrease was noted in ICU LOS (7.4 vs. 4.4 days, p= .03).
Rates of intubation, intubation due to AWS, duration of ven-
tilation, and ventilator-free days were not significantly differ-
ent between groups. There was no difference in the use of
flumazenil between the two groups.
Discussion
While there have been previous studies in AWS management
to guide selection of medications and certain treatment
strategies (i.e., bolus dosing of BZDs and barbiturates in a
symptom-triggered manner), these studies did not apply the
approach of focused therapy to the first 24 h after starting
treatment. In this study, focused therapy was implemented
by limiting higher loading doses (i.e., diazepam 10–50 mg
every 15 min) to the first 24 h after detection of severe AWS.
The tapering phase consisted of lower doses (i.e., diazepam
10–20 mg every 30 min) until 96 h after detection of severe
AWS, after which all orders self-discontinued. This approach
utilizes the natural timeline of alcohol withdrawal to prevent
progression to DT, while avoiding prolonged use of BZDs to
limit development of delirium. To our knowledge, this study
is the first to assess the potential impact of a 24-h focused
treatment strategy of BZDs for the treatment of severe AWS.
Our findings appear to confirm the recommendations of
recent reviews supporting early and aggressive management
of AWS [3].
The other novel element of this study was the use of the
SHOT scale as a symptom assessment tool to measure non-
agitation symptoms of alcohol withdrawal. Traditionally, the
CIWA-Ar scale has been utilized to assess symptoms of alco-
hol withdrawal [10]. However, this scale tends to be time-
consuming and difficult to assess in an actively withdrawing
ICU patient on a frequent (e.g., every 15 min) basis. Gray
et al. selected specific elements of the CIWA-Ar to create the
SHOT scale that relied less on self-report (i.e., anxiety,
headache, nausea) and agitation, which can be measured
with the RASS scale [9]. The SHOT scale was validated as an
alternative to CIWA-Ar, such that a positive CIWA-Ar score of
greater than or equal to 10 strongly correlated with a SHOT
score greater than or equal to 2 [9].
In this study, the SHOT scale was utilized in a similar man-
ner to differentiate severe from mild–moderate alcohol with-
drawal. Additionally, it was also found to be an effective tool
to manage the non-agitation symptoms of AWS during the
loading phase, which was a gap identified in the original
symptom-triggered protocol that solely relied on RASS. SHOT
also filled the gap identified during the tapering phase when
agitation typically subsides rapidly during AWS treatment
but non-agitation symptoms (i.e., sweating, hallucinations,
orientation, tremors) may still be present. Treating these
symptoms may prevent the re-escalation of AWS symptoms.
Although all BZDs bind to GABA
A
- to increase the fre-
quency of GABA chloride channel opening and effectively
potentiate the inhibitory effects of GABA, diazepam was
selected as the backbone BZD due to its favorable pharma-
cokinetic profile. Diazepam exhibits a rapid time to peak
(1–3 min) in contrast to lorazepam (30 min), which facilitates
up-titration during the loading phase while avoiding dose-
stacking. Furthermore, its context-sensitive half-time allows
for rapid elimination with smaller initial doses, but the dur-
ation of action rapidly increases with subsequent accumula-
tion of drug, leading to a smooth taper and decreased risk
of iatrogenic BZD withdrawal. Diazepam is also more lipo-
philic than lorazepam; therefore, although the half-life of its
metabolites is reportedly longer than lorazepam, the clinic-
ally effective duration of action is much shorter due to its
high volume of distribution, and the active drug is able to
rapidly cross into and out of the blood brain barrier. Relying
on a single BZD can become problematic in times of supply
issues, but given the dynamic nature of drug shortages,
these issues must be addressed on a case-by-case basis.
Phenobarbital has recently emerged as a useful adjunct to
BZD therapy for the treatment of AWS due to its inhibition
of the excitatory glutamate neurotransmitter to the N-
methyl-D-aspartate (NMDA) receptor [7,11]. The largest pro-
spective study on phenobarbital for treatment of AWS to
date studied its effects in a single, 10 mg/kg loading dose
[6]. In contrast, this study utilized phenobarbital after the
patient escalated to Step 3 of the loading dose
(Supplementary Figure 1), which provides a 30-mg dose of
diazepam for acute agitation (RASS greater than 1). The
study investigators wanted to ensure that patients would not
Table 3. Treatment of PRE and POST groups.
Treatment PRE (n¼75) POST (n¼38) pvalue
Time to first BZD, hours, mean ± SD 7.7 ± 0.6 7.0 ± 0.5 .83
Total diazepam
a
, mg, mean ± SD 949.6 ± 980.4 812.2 ± 566.6 .36
Focused treatment, % 51.3 73.7 .03
BZD exposure, days, mean ± SD 6.0 ± 6.8 5.2 ± 3.5 .38
Phenobarbital, mg, mean ± SD 251.7 ± 371.8 301.8 ± 310.7 .46
a
Lorazepam equivalents: 190.0 mg (PRE) versus 162.8 mg (POST).
Table 4. Primary and secondary outcomes.
Outcome PRE (n¼75) POST (n¼38) pvalue
HLOS, days, mean ± SD 14.0 ± 12.6 9.8 ± 7.3 .03
ILOS, days, mean ± SD 7.4 ± 10.2 4.4 ± 4.6 .03
Mechanical ventilation
Intubated, n(%) 27 (36.0) 8 (21.1) .11
Intubated due to AWS, n(%) 13 (17.3) 2 (5.3) .08
Duration of ventilation, days,
mean ± SD
2.0 ± 5.7 1.1 ± 2.8 .27
Ventilator-free days, median (IQR) 28 (26.3–28.0) 28 (27.7–28.0) .31
Flumazenil, n(%) 2 (2.7) 0 (0) .31
HLOS: hospital length of stay; ILOS: ICU length of stay; AWS: AWS
CLINICAL TOXICOLOGY 3

be oversedated with a long-acting barbiturate upon admis-
sion in case the primary reason for acute agitation and
altered mental status was not AWS (e.g., hypoxia, sepsis,
metabolic disorders, brain injuries, overdose, psychosis, pain,
etc.). By placing phenobarbital later in the algorithm, the ini-
tial smaller doses of diazepam (10–20 mg) were effectively
used as a test dose to assess whether the reason for altered
mental status was due to GABA imbalance.
The two cohorts were evenly matched in regards to age,
sex, history of alcohol withdrawal, SOFA score, and admission
to medical versus surgical ICUs. There was a statistically sig-
nificant difference noted in BAL on admission, but this was
not noted to be of clinical significance. A minority of the 513
patients for whom severe AWS was suspected progressed to
the point of needing high doses of BZDs in both cohorts.
Notably, the proportion of patients who progressed to severe
AWS was decreased in the POST group (30% vs. 13%).
There are a number of limitations associated with a sin-
gle-center, retrospective study. The management of both the
PRE and POST groups was based on an institution-specific
protocol. There was a difference noted between the groups
in BAL. However, the extended time interval between patient
admission and first dose of BZD likely mitigated any effect of
this difference on ordering treatment for AWS. Although a
prospective randomized trial would be ideal, the pre–post
study design is a practical approach to measure the impact
of a revised protocol in otherwise similar cohorts.
A strength of this approach is its ability to personalize
treatment to each patient with AWS based on their individ-
ual degree of GABA imbalance by seeking the minimal
effective dose in the loading phase. However, increasing indi-
vidualization in an order set naturally lends itself to a more
complicated treatment strategy. Although every attempt was
made to simplify the algorithm, extensive education was
required for all nurses who cared for patients with
severe AWS.
Conclusion
The findings of this study suggest that focused BZD use in
the first 24 h of AWS treatment may decrease hospital LOS.
Further, the SHOT scale can be safely used to triage and
treat non-agitation symptoms of AWS. Future prospective
studies are needed to confirm these findings and determine
an optimal treatment strategy for AWS.
Disclosure statement
The authors have no conflict of interest to declare. This was an investi-
gator-initiated study with no funding source. This study was approved
by the Institutional Review Board at the University of California, Davis.
References
[1] Center for Behavioral Health Statistics and Quality. Behavioral
health trends in the United States: Results from the 2017
National Survey on Drug Use and Health (HHS Publication No.
SMA 18-5068, NSDUH Series H-53). Rockville, MD; 2018.
[2] Gold JA, Nelson LS. Ethanol withdrawal. In: Weitz M, Nalieri C,
editors. Goldfrank’s toxicologic emergencies. 10th ed. New York:
McGraw-Hill; 2015.
[3] Trevisan LA, Boutros N, Petrakis IL, et al. Complications of alcohol
withdrawal: pathophysiological insights. Alcohol Health Res
World. 1998;22:61–66.
[4] de Wit M, Jones DG, Sessler CN, et al. Alcohol-use disorders in
the critically ill patient. Chest. 2010;138:994–1003.
[5] Awissi DK, Lebrun G, Coursin DB, et al. Alcohol withdrawal and
delirium tremens in the critically ill: a systematic review and com-
mentary. Intensive Care Med. 2013;39:16–30.
[6] Rosenson J, Clements C, Simon B, et al. Phenobarbital for acute
alcohol withdrawal: a prospective randomized double-blind pla-
cebo-controlled study. J Emerg Med. 2013;44:592–598.
[7] Gold JA, Rimal B, Nolan A, et al. A strategy of escalating doses of
benzodiazepines and phenobarbital administration reduces the
need for mechanical ventilation in delirium tremens. Crit Care
Med. 2007;35:724–730.
[8] Duby JJ, Berry AJ, Ghayyem P, et al. Alcohol withdrawal syn-
drome in critically ill patients: protocolized versus nonprotocol-
ized management. J Trauma Acute Care Surg. 2014;77:938–943.
[9] Gray S, Borgundvaag B, Sirvastava A, et al. Feasibility and reliabil-
ity of the SHOT: a short scale for measuring pretreatment severity
of alcohol withdrawal in the emergency department. Acad Emerg
Med. 2010;17:1048–1054.
[10] Sullivan JT, Sykora K, Schneiderman J, et al. Assessment of alco-
hol withdrawal; The revised clinical institute withdrawal assess-
ment for alcohol scale (CIWA-Ar). Br J Addict. 1989;84:1353–1357.
[11] Dixit D, Endicott J, Burry L, et al. Management of acute alcohol
withdrawal syndrome in critically ill patients. Pharmacotherapy.
2016;36:797–822.
4 J. A. LEE ET AL.