Alcohol withdrawal syndrome in ICU patients: Clinical features, management, and outcome predictors.

BACKGROUND: Alcohol withdrawal syndrome (AWS) is a common condition in hospitalized patients, yet its epidemiology in the ICU remains poorly characterized.
METHODS: Retrospective cohort of patients admitted to the Nantes University Hospital ICU between January 1, 2017, and December 31, 2019, and coded for AWS using ICD-10 criteria. The objective of the study was to identify factors associated with complicated hospital stay defined as ICU length of stay ≥7 days or hospital mortality.
RESULTS: Among 5,641 patients admitted to the ICU during the study period, 246 (4.4%) were coded as having AWS. Among them, 42 had exclusion criteria and 204 were included in the study. The three main reasons for ICU admission were sepsis (29.9%), altered consciousness (29.4%), and seizures (24%). At ICU admission, median Cushman's score was 6 [4-9] and median SOFA score was 3 [2-6]. Delirium tremens occurred in half the patients, seizures in one fifth and pneumonia in one third. Overall, 48% of patients developed complicated hospital stay, of whom 92.8% stayed in the ICU for ≥7 days, 36.7% received MV for ≥7 days, and 16.3% died during hospital stay. By multivariable analysis, two factors were associated with complicated hospital stay: a higher number of organ dysfunctions at ICU admission was associated with a higher risk of complicated hospital stay (OR, 1.18; 95CI, 1.05-1.32, P = 0.005), whereas ICU admission for seizures was associated with a lower risk of complicated hospital stay (OR, 0.14; 95%CI, 0.026-0.80; P = 0.026).
CONCLUSIONS: AWS in ICU patients chiefly affects young adults and is often associated with additional factors such as sepsis, trauma, or surgery. Half the patients experienced an extended ICU stay or death during the hospital stay. The likelihood of developing complicated hospital stay relied on the reason for ICU admission and the number of organ dysfunctions at ICU admission.

Introduction

Alcohol is the most commonly used psychoactive substance in adults and a major cause of hospitalization, morbidity, and mortality worldwide. In 2016, 32.5% of the world’s population were current drinkers and 2.8 million deaths were attributed to alcohol use [1]. In France, an estimated 5 million people have alcohol-use disorder [2], and alcohol is a leading risk factor for premature death and disability [3]. One of the adverse consequences of chronic alcohol use is alcohol withdrawal syndrome (AWS). AWS may lead to acute neurotoxicity due to an extensive release of glutamate neurotransmitters and a massive opening of post-synaptic calcium channels which induces neuronal apoptosis [4]. Among heavy alcohol users, approximately 50% experience some degree of withdrawal symptoms when their consumption is reduced or stopped [5-8] and about 10% have withdrawal seizures [5-7]. Moreover, AWS can progress to delirium tremens, a state characterized by severe confusion and hallucinations associated with severe autonomic hyperactivity [5-7]. The most severe forms of AWS may require ICU admission, and a study conducted in Finland found that 20% of ICU admissions were related to alcohol use [9]. Several studies have attempted to identify risk factors for developing AWS and delirium tremens [6,10-18], while others focused on the therapeutic strategy [17-24]. However, the epidemiology of AWS in ICU patients is poorly known, its optimal management remains chiefly empirical, and its outcome is largely unstudied. A better understanding of the epidemiology, treatment, and outcome in ICU patients with AWS may help guide clinical practice and research.

We therefore conducted an epidemiological study in a French university-affiliated ICU, by using the International Classification of Diseases 10th Revision (ICD-10) coding system to identify patients with AWS. We aimed to test the hypothesis that AWS in ICU patients could result in extended ICU stay or death, and to identify factors associated with such outcomes.

Methods

This retrospective study was approved by the ethics committee of the French Intensive Care Society (CE SRLF 21–10) on February 01, 2021 with a waiver for informed consent. The study is reported in compliance with the STROBE recommendations [25].

Study design, setting, and population

We identified consecutive adults (≥18 years of age) admitted to the ICU of the Nantes University Hospital between January 1, 2017, and December 31, 2019, and registered in the electronic hospital database with any of the codes for AWS in the ICD-10 (F10.3, F10.30, F10.31, F10.4, F10.40, F10.41, F10.03, F10.05, F10.06). For patients who had multiple admissions during the study period, only the first admission was considered. In our institution, coding is done at the time of ICU discharge by the physician in charge, using the patient’s formal discharge summary. Each medical file was reviewed by AV and EC to confirm the diagnosis of AWS based on the Diagnostic and Statistical Manual of Mental Disorders, Fifth Edition (DSM-5) [26]. All four major criteria had to be present in the electronic medical record of each patient to diagnose AWS. For major criterion B, 2 or more of the 8 symptoms had to be present. Data were extracted from the doctors and nurses notes (S1 Fig). Delirium tremens was defined as a patient with AWS who developed a state of confusion, acute agitation and hallucinations recorded in the medical file during the ICU stay. Exclusion criteria were absence of signs and/or symptoms of AWS recorded in the medical file, withdrawal syndrome of unclear origin (absence of chronic alcoholism or withdrawal of a substance other than alcohol), and preventive treatment for AWS without subsequent AWS.

Data collection

Data were extracted from the electronic medical records of the ICU (CERNER Millenium®, Nantes, France). We obtained data for baseline patient characteristics, including demographics, comorbidities, chronic medications, and habits of alcohol consumption. Habits of alcohol consumption and alcohol history are part of the standard intake procedure in our hospital. Data were obtained from patients’ interview. When the patient’s clinical condition made the interview impossible, data were obtained either from the next of kin or from the patient at the time of discharge. The onset of AWS was the date when AWS was first recorded in the medical file. For each patient, AWS recovery was assessed by reading the daily notes of nurses and doctors from the EMR. The date of resolution was either the date of resolution recorded in the medical file or the last date of recording of AWS with no further signs or symptoms of AWS recorded for at least 48 hours. If neither of these two conditions was met, the episode of AWS was classified as persistent. When patients had underlying dementia or other neurocognitive disorders, a worsening of the clinical state during hospitalization had to be mentioned in the patients’ EMR to classify a patient with persistent confusion. AWS severity was assessed using Cushman’s score [27]. The following complications of AWS during the ICU stay were recorded: delirium tremens, seizures, status epilepticus, rhabdomyolysis, acute kidney injury, and pneumonia. The drugs administered intravenously or orally during the AWS episode were extracted from the electronic prescription database of the hospital. The life-sustaining therapies used during the ICU stay (high-flow oxygen, non-invasive ventilation, mechanical ventilation (MV), vasopressors, and/or renal replacement therapy) were extracted from the electronic medical records. Vital status and destination at hospital discharge (home, psychiatry ward, rehabilitation center, or discharge against medical advice) were also recorded.

Objectives

The primary objective of the study was to identify factors associated with complicated hospital stay. Complicated hospital stay was defined as ICU length of stay ≥7 days or death before hospital discharge. In the absence of these criteria, patients were considered to have uncomplicated hospital stay.

The secondary objectives were to describe the clinical features, treatments, and outcomes of ICU patients with AWS.

Statistical analysis

Continuous variables are described as median and interquartile range [IQR] and compared using Wilcoxon’s test. Categorical variables are described as counts (percent) and compared using the exact Fisher’s test. The occurrence of complicated hospital stay (versus uncomplicated hospital stay) was analyzed as a binary variable. Logistic regression analyses were performed to identify variables associated with complicated hospital stay, with estimated odds ratios (ORs) and their 95% confidence intervals (95%CIs). For the multivariable model, we preselected candidate variables which plausibly fit with complicated hospital stay based on knowledge from the literature (SOFA, comorbidities, and mortality) and our assumptions (chronic use of BZD or antipsychotics, history of AWS, reason for ICU admission, and extended stay in the ICU). We carefully checked to avoid collinearity between variables and we applied the rule of selecting a maximum of 1 variable per 8 events (total of 12 variables in our study). All tests were two-sided, and P values lower than 5% were considered to indicate significant associations. Statistical tests were conducted using the R statistics program, version 3.5.0 (R Foundation for Statistical Computing, Vienna, Austria; www.R-project.org/).

Results

Study population

Among 5,641 patients admitted to the ICU during the study period, 246 (4.4%) were coded as having AWS. A detailed analysis of the medical files showed that 42 patients had exclusion criteria. The remaining 204 patients were included in the study (Fig 1). Table 1 reports their main features. Median daily alcohol intake was 129 (72–216) grams (missing data, n = 51). Patients were admitted from the emergency department (67.2%), wards (19%), or pre-hospital emergency medical service (14%). Computed tomography (CT) of the brain was performed in 77 (37.8%) patients, of whom 22 (28.6%) had the following abnormal findings: subarachnoid bleeding, n = 6; subdural hematoma, n = 6; stroke, n = 4; intracranial hematoma, n = 2; extradural hematoma, n = 2; chronic hydrocephalus, n = 1; and cortical atrophy compatible with Wernicke encephalopathy, n = 1.

Fig 1

Study flowchart.

ICD-10: International Classification of Diseases 10th Revision.

Table 1

Baseline characteristics of the 204 study participants.

Variable	All patients (n = 204)	Complicated hospital stay (n = 98)	Uncomplicated hospital stay (n = 106)	P value
Demographics
Age, median [IQR], years	53 [46–60]	54.5 [48–61]	50 [44–58]	0.099
Male sex, n (%)	172 (84.3)	83 (84.7)	89 (84.0)	0.88
Charlson’s index, median [IQR]	1 [0–3]	2 [0.25–4]	1 [0–3]	0.019
Alcohol withdrawal history
History of AWS, n (%)	42 (20.6)	23 (23.5)	19 (18.0)	0.387
History of DT, n (%)	10 (4.9)	7 (7.1)	3 (2.8)	0.200
History of withdrawal seizures, n (%)	25 (12.3)	15 (15.3)	10 (9.4)	0.201
Psychiatric history
Substance use disorder other than alcohol, n (%)	30 (14.7)	12 (12.2)	18 (17.0)	0.429
Any psychiatric disorder, n (%)a	71 (34.8)	33 (33.7)	38 (35.9)	0.598
Mood disorders, n (%)	7 (3.4)	4 (4.1)	3 (2.8)	0.712
Anxiety disorders, n (%)	64 (31.4)	29 (29.6)	35 (33)	0.651
Chronic medications
Benzodiazepines, n (%)	71 (34.8)	27 (27.6)	44 (41.5)	0.036
Antipsychotic drugs, n (%)	26 (12.8)	6 (6.1)	21 (18.9)	0.0063
Time from hospital admission to ICU admission, median [IQR], days	0 [0–1]	0 [0–1]	0 [0–1]	0.168
ICU admission from the ED, n (%)	137 (67.2)	63 (64.3)	74 (69.8)	0.401
Cushman’s score at ED admission, median [IQR]	7 [4–9]	7 [4–9]	7 [4–9]	0.827
Reason for ICU admission, n (%)				0.00073
Sepsis	61 (29.9)	39 (39.8)	22 (20.8)
Altered consciousness	60 (29.4)	22 (22.5)	38 (35.9)
Seizures	24 (11.7)	4 (4.1)	20 (18.9)
Trauma	20 (9.8)	9 (9.2)	11 (10.4)
Surgery	12 (5.9)	6 (6.1)	6 (5.7)
AKI	12 (5.9)	7 (7.2)	5 (4.7)
Otherb	15 (7.4)	11 (11.2)	4 (3.8)
Clinical variables and measures at ICU admission
HR, median [IQR], bpm	104 [88–120]	109 [98–123]	99 [85–117]	0.038
SBP, median [IQR], mmHg	125 [106–147]	121 [100–140]	126 [112–150]	0.038
Glasgow Coma Scale score, median [IQR]	14 [12–15]	14 [11–15]	14 [13–15]	0.84
RR, median [IQR]	22 [18–26]	22.5 [19.3–27]	20 [17–25]	0.015
Cushman score	6 [4–9]	6 [4–9]	7 [4–9]	0.196
SOFA	3 [2–6]	5 [3–8]	3 [1–5]	<0.0001
SAPS II	24 [16–34]	27 [17–37]	20 [13–31]	0.0067

AKI: Acute kidney injury; AWS: Alcohol withdrawal syndrome; BPM: Beats per minute; DT: Delirium tremens; ED: Emergency department; HR: Heart rate; ICU: Intensive care unit; IQR: Interquartile range; RR: Respiratory rate; SAPS II: Simplified Acute Physiology Score, version II; SBP: Systolic blood pressure; SOFA: Sequential Organ Failure Assessment.

aAny psychiatric disorder, n (%): Including 6 patients with underlying dementia.

bOther: Cardiac or respiratory arrest; upper gastrointestinal hemorrhage; acute pancreatitis; ketoacidosis; mesenteric ischemia.

Clinical features of AWS and ICU management

AWS was typically diagnosed 1 [1-2] day after ICU admission and lasted 5 [3-8] days. Table 2 provides details about the treatments used and complications observed. All patients were treated with a combination of vitamins B1 and B6 and intravenous hydration during the first 24 hours. Benzodiazepines were given to 99% of patients. Diazepam and oxazepam were often given intermittently. Continuous midazolam or propofol were prescribed to 27.9% and 12.3% of the patients, respectively. Among patients with pneumonia, the most commonly recovered micro-organisms were Streptococcus pneumoniae, methicillin-sensitive Staphylococcus aureus, and Haemophilus influenzae. In the patients who required MV, the time interval between the diagnosis of AWS and endotracheal intubation ranged from -1 to +2 days after the diagnosis of AWS. The most common reasons for intubation were coma (75.6%) and acute respiratory failure (24.4%).

Table 2

Clinical features of AWS and pharmacological management.

Variable	All patients (n = 204)	Complicated hospital stay (n = 98)	Uncomplicated hospital stay (n = 106)	P value
Clinical features of AWS
Time from ICU admission to AWS onset, days, median [IQR]	1 [1–2]	1 [1–3]	1 [1–2]	0.6083
Worst Cushman score during the ICU stay, median [IQR]	11 [8–14]	12 [9–15]	11 [8–13]	0.099
IV fluids during the first 24 h, L median [IQR]	2 [1.5–3]	2 [1.5–3]	2 [1.5–3]	0.484
B1 and B6 vitamin therapy, n (%)	204 (100)	98 (100)	106 (100)	1.00
Drugs administered at the time of AWS
Benzodiazepines
Diazepam (IV/PO), n (%)	170 (83.3)	82 (83.7)	88 (83)	0.9003
Oxazepam (PO), n (%)	102 (50)	52 (53.1)	50 (47.2)	0.4005
Midazolam (continuous IV), n (%)	57 (27.9)	39 (39.8)	18 (17)	0.0028
Length of treatment with BZD, days, median [IQR]	4 [3–7]	7 [4–9]	3 [2–4]	<0.0001
Antipsychotics
Haloperidol (IV)	26 (12.3)	13 (13.3)	13 (12.3)	0.83
Cyamemazine (PO)	37 (18.1)	25 (25.5)	12 (11.3)	0.0085
Loxapine (IM)	6 (2.9)	5 (5.1)	1 (0.9)	0.1056
Other drug
Propofol (continuous IV)	25 (12.3)	22 (22.5)	3 (2.8)	<0.0001
AWS-related diagnoses, n (%)
Delirium tremens	108 (52.9)	57 (58.2)	51 (48.1)	0.15
Seizures	39 (19.1)	16 (16.3)	23 (21.7)	0.329
Status epilepticus	18 (8.8)	5 (5.1)	13 (12.3)	0.071
Pneumonia	66 (32.4)	47 (48)	19 (17.9)	<0.0001
AKI	60 (29.4)	31 (31.6)	29 (27.4)	0.503
Rhabdomyolysis	12 (5.9)	7 (7.1)	5 (4.7)	0.461
Duration of AWS, days, median [IQR]	5 [3–8]	8 [6.25–13]	4 [3–5]	<0.0001
Life-sustaining therapies
High flow oxygen, n (%)	9 (4.4)	5 (5.1)	4 (3.8)	0.644
Non-invasive ventilation, n (%)	16 (7.8)	13 (13.3)	3 (2.8)	0.0056
MV, n (%)	86 (42.2)	64 (65.3)	22 (20.8)	<0.0001
MV duration, days, median [IQR]	5.5 [2–10]	7.50 [4–12.3]	2 [1.25–2]	<0.0001
Vasopressors, n (%)	33 (16.2)	27 (27.6)	6 (5.7)	<0.0001
Renal replacement therapy, n (%)	2 (1)	2 (2)	0 (0)	0.2295

AKI: Acute kidney injury; AWS: Alcohol withdrawal syndrome; BZD: Benzodiazepine; ICU: Intensive care unit; IM: Intramuscular; IQR: Interquartile range; IV: Intravenous; MV: Mechanical ventilation; PO: Per os.

Outcomes

During the study period, the occurrence of complicated hospital stay in ICU patients was 48% in patients with AWS and 31% in patients without AWS (p<0.001, S1 Table). Among AWS patients who developed complicated hospital stay, 92.8% stayed in the ICU for ≥7 days, 36.7% required MV for ≥7 days, and 16.3% died during the hospital stay (Table 3). The duration of AWS in patients with complicated hospital stay was twice that in patients with uncomplicated hospital stay (8 [6-13] versus 4 [3-5] days). Patients with complicated hospital stay were twice as likely to have persistent confusion at ICU discharge and had more than twice the hospital length of stay, compared to patients with uncomplicated hospital stay. Finally, the destination at hospital discharge differed between the two groups (Table 3).

Table 3

Outcomes.

Variable	All patients (n = 204)	Complicated hospital stay (n = 98)	Uncomplicated hospital stay (n = 106)	P value
AWS outcome
Persistent confusion at ICU discharge, n (%)	51 (25)	34 (34.7)	17 (16.0)	0.0003
Persistent agitation at ICU discharge, n (%)	12 (5.9)	8 (8.2)	4 (3.8)	0.120
Length of stay
ICU, days, median [IQR]	6 [4–10.3]	11 [8–15.8]	4 [3–5]	<0.0001
Hospital, days, median [IQR]	13 [7–26.3]	23 [13.5–34]	9 [5–14]	0.0085
Vital status
ICU mortality, n (%)	11 (5.4)	11 (11.2)	0 (0)	0.00039
Hospital mortality, n (%)	16 (7.8)	16 (16.3)	0 (0)	<0.0001
Destination at hospital discharge a
Home, n (%)	118 (57.8)	51 (52.0)	67 (63.2)	0.959
Follow-up care and rehabilitation unit, n (%)	29 (14.2)	20 (20.4)	9 (8.5)	0.0022
Psychiatric ward, n (%)	16 (7.8)	3 (3.1)	13 (12.7)	0.037
Left against medical advice, n (%)	7 (3.4)	1 (1.0)	6 (5.7)	0.113
Addictology unit, n (%)	3 (1.47)	0 (0)	3 (2.8)	0.126
Alive on day 28, n (%)	189 (92.7)	83.9 (85.6)	106 (100)	0.00027

AWS: Alcohol withdrawal syndrome; IQR: Interquartile range; ICU: Intensive care unit; LOS: Length of stay.

a: Missing data n = 31 (15.2%).

Factors associated with complicated hospital stay

By univariate analysis, comorbidities, organ dysfunctions at ICU admission, tachycardia, low blood pressure, high respiratory rate, and sepsis were associated with an increased risk of developing complicated hospital stay. In contrast, chronic use of benzodiazepines or neuroleptic drugs was more common in patients with uncomplicated hospital stay. By multivariable analysis, only two factors were independently associated with developing complicated hospital stay: a higher number of organ dysfunctions at ICU admission was associated with a higher risk of complicated hospital stay, while ICU admission for seizures was associated with a lower risk of complicated hospital stay (Table 4). In another multivariable model which included all patients admitted to the ICU during the study period, AWS was by itself a factor associated with a higher risk of complicated hospital stay (S2 Table).

Table 4

Logistic regression analyses for factors associated with complicated hospital stay.

Factors	Univariate analysis		Multivariable analysisa
	OR (95%CI)	P value	OR (95%CI)	P value
Demographics
Age (per year)	1.02 (0.99–1.05)	0.10
Male sex	0.95 (0.44–2.02)	0.87
Charlson’s index	1.15 (1.02–1.31)	0.02	1.09 (0.95–1.23)	0.24
Alcohol withdrawal history
History of AWS	1.39 (0.71–2.73)	0.33	1.95 (0.87–4.57)	0.11
Chronic medications
Benzodiazepines	0.54 (0.30–0.97)	0.04	0.58 (0.29–1.15)	0.12
Antipsychotic drugs	0.28 (0.11–0.73)	0.009	0.35 (0.11–0.98)	0.056
Cushman’s score at ED admission	1.01 (0.92–1.10)	0.826
Reason for ICU admission
Altered consciousness	1		1
Sepsis	3.06 (1.45–6.45)	0.031	2.02 (0.91–4.58)	0.088
Seizures	0.35 (0.10–1.15)	0.081	0.22 (0.05–0.071)	0.019
Trauma	1.41 (0.50–3.97)	0.51	1.47 (0.48–4.32)	0.48
Surgery	1.73 (0.49–6.06)	0.39	0.93 (0.23–3.74)	0.92
AKI	2.42 (0.68–8.61)	0.17	1.50 (0.37–6.31)	0.57
Otherb	4.75 (1.34–16.86)	0.015	2.77 (0.74–12.0)	0.14
Clinical variables and measures at ICU admission
SOFA at ICU admission	1.23 (1.12–1.36)	0.00004	1.18 (1.06–1.33)	0.005
HR	1.01 (1.00–1.02)	0.041
SBP	0.99 (0.98–0.99)	0.041
RR	1.05 (1.01–1.09)	0.018

AKI: Acute kidney injury; AWS: Alcohol withdrawal syndrome; CI: Confidence interval; ED: Emergency department; HR: Heart rate; ICU: Intensive care unit; OR: Odds ratio; RR: Respiratory rate; SBP: Systolic blood pressure; SOFA: Sepsis-related Organ Failure Assessment.

aPreselected candidate variables included in the multivariable model were: Charlson’s index, history of AWS, chronic use of benzodiazepines, chronic use of antipsychotic drugs, reason for ICU admission, and SOFA at ICU admission.

bOther: Cardiac or respiratory arrest; upper gastrointestinal hemorrhage; acute pancreatitis; ketoacidosis; mesenteric ischemia.

Discussion

Key findings

We used the ICD-10 coding system and DSM-5 criteria to identify ICU patients who developed AWS. We found that patients with AWS accounted for approximately 4% of all ICU admissions and that half of them developed complicated hospital stay despite having low severity scores at ICU admission. Furthermore, patients with complicated hospital stay had more co-morbidities, were more likely to be admitted for sepsis, displayed higher SOFA scores at ICU admission, and were more likely to require follow-up care or rehabilitation at hospital discharge compared to patients with uncomplicated AWS. Finally, the likelihood of developing complicated hospital stay was lower in patients with seizures and higher in patients with a higher number of organ failures at ICU admission. Neither Cushman’s score nor the occurrence of delirium tremens was associated with the risk of complicated hospital stay.

Comparison with previous studies

The epidemiology of AWS in ICU patients is difficult to ascertain. The available studies were conducted in specific populations (emergency departments, addiction units, psychiatry wards, trauma centers, or medical wards). They found highly variable incidences ranging from 0.3% to 52% [8,24,28,29]. In a recent review, the incidence of AWS in the ICU patients ranged from <1% in unselected patients to 60% in highly selected alcohol-dependent patients [24]. Studies differed in the tools they used to diagnose and assess AWS, making comparisons difficult. We used both ICD-10 codes and DSM-5 criteria to identify AWS in an unselected population admitted to a university-affiliated ICU. According to these criteria, approximately 4% of ICU patients had AWS.

Male sex, older age, heavier drinking, past history of AWS or withdrawal seizures, greater severity of AWS at hospital admission, concurrent substance use disorder, and mental health conditions have been reported to be associated with a higher risk of developing severe AWS or delirium tremens [6,8,24]. We found that 85% of patients were males and heavy drinkers and that one-third had underlying psychiatric disorders, whereas only a fifth had a history of AWS. Approximately half the patients had sepsis, trauma, or surgery identified as a precipitating factor for AWS, in keeping with previous studies [24,30].

The optimal management of AWS has yet to be determined. Benzodiazepines are considered the cornerstone of therapy despite the lack of a high level of evidence [31], with symptom-triggered bolus administration being the recommended modality [24,32]. In addition, short-acting antipsychotics or alpha2-agonists are often required to treat agitation and autonomic hyperactivity [19,21,33]. In a study conducted in three US hospitals, as many as 16 different medications and 74 combinations of medications were used to treat AWS [21]. In our study, intermittent administration of diazepam or oxazepam was the first-line treatment in nearly all the patients, a continuous infusion of midazolam or propofol was added in nearly 30% of patients, and antipsychotics were used in one fourth of patients. However, our data cannot allow conclusions about the effectiveness of specific treatments on patient outcomes, highlighting the need for further trials. Interestingly, a recent study reported that the implementation of a hospital-wide protocol for the management of AWS resulted in significant improvements in quality of care, decreased the need for ICU admission and the rate of intubation, reduced hospital length of stay, and was cost-savings [34].

The assessment of AWS severity is important to identify patients at high risk for adverse outcomes and to adjust the pharmacological interventions accordingly. However, the definition and assessment of severe AWS has varied across studies [35]. Most studies used scales to grade clinical symptoms, such as the revised Clinical Institute Withdrawal Assessment for Alcohol scale (CIWA-Ar) [5,35,36]. However, this scale was not developed and has not been validated for ICU patients. Moreover, the CIWA-Ar excludes mechanically ventilated patients and has limited accuracy for predicting severe AWS and its complications [35]. We used Cushman’s scale [27] to evaluate the severity of AWS. Unlike the CIWA-Ar, Cushman’s scale can be used in uncooperative ICU patients. In our study, Cushman’s scores obtained at several time points were not associated with the need for prolonged MV, an extended ICU stay, or mortality.

The severity of AWS is often defined as the occurrence of seizures and delirium tremens. However, whether seizures or delirium tremens are associated with clinically relevant adverse outcomes such as a prolonged ICU stay or mortality is unclear [8,24]. In our study, delirium tremens was diagnosed in half of our patients but was not associated with prolonged MV, an extended ICU stay, or mortality. In contrast, we found ICU admission for seizures to be associated with uncomplicated hospital stay. Seizures can result in substantial complications, including status epilepticus or aspiration pneumonia [8]. However, withdrawal seizures can be efficiently treated with benzodiazepines with a potential rapid improvement of clinical status compared to other reasons for ICU admission such as sepsis, acute pancreatitis or acute kidney injury which are more complex to treat. AWS can result in significant morbidity, including aspiration pneumonia, acute kidney injury, and arrhythmia [37]. In historical studies, mortality rates of up to 15% were observed [38,39]. However, in a recent study conducted among trauma patients, AWS-associated mortality was 7% [28]. Our experience was similar, with an overall ICU mortality of 5.4%.

Data on the long-term outcomes of ICU patients with AWS are limited. In a Spanish study, 72% of ICU patients with delirium tremens were readmitted multiple times to the emergency department within the next 2 years [40]. Although AWS is by definition an acute syndrome, a quarter of our patients had persistent confusion at ICU discharge, more than 40% were unable to return home at hospital discharge, and 7% died within 28 days of ICU admission.

Study implications

The findings from our study imply that AWS in ICU patients is often triggered by a precipitating factor, with sepsis being the most commonly reported. Therefore, patients with AWS should be routinely screened for sepsis to identify those who require early investigations and treatment. Moreover, our findings imply that, although Cushman’s score may help clinicians to titrate the treatment of AWS, it is unable to identify patients at risk for an extended ICU stay, or hospital mortality. In contrast, early detection of organ dysfunctions identifies a population at high risk for adverse outcomes. Thus, the prompt identification, regular re-assessment, and early treatment of organ dysfunctions may improve patient outcomes. Finally, the high frequency of persistent confusion at ICU discharge, high proportion of patients who could not be discharged home, and significant mortality within 28 days after ICU admission support the view that closer surveillance of this vulnerable population may be justified.

Strengths and limitations

This study has a number of strengths. First, we used both the ICD-10 coding system and a detailed review of each medical file for DSM-5 criteria to identify patients with AWS. This minimized potential bias related to the retrospective selection of the study patients. Second, we evaluated and identified risk factors for clinically relevant endpoints (ICU stay ≥7 days and in-hospital mortality). Thus, we provide new data for identifying patients at risk for poor outcomes. Third, we obtained detailed information on the outcome after ICU discharge, an area rarely explored in previous ICU studies.

Our study also has several limitations. First, the retrospective design implies information bias with a possibility of missing data. For example, we had no information on other potential causes of health disparities, such as income, health care coverage or country of birth, which may have influenced patients’ outcomes. Second, the study was conducted in a single institution, where the case mix may have significantly influenced our findings. Nonetheless, we conducted this study in an unselected ICU population in a large university-affiliated center, and our results should therefore apply to similar settings in high-income countries. Third, although ICD-10 discharge coding combined with DSM-5 criteria has strong reliability for diagnosing AWS, sensitivity may be limited [41]. We therefore may have underestimated the true incidence of AWS and studied a particular cohort of patients with more easily diagnosed and, perhaps, more severe and prolonged AWS. However, there is no consensus on the best method for identifying AWS. Finally, the lack of a standardized protocol for managing AWS resulted in substantial variability in the drugs used, their dosages, and their combinations. This prevented us from evaluating how treatments may have affected patient outcomes. However, there is no agreement on the optimal pharmacological treatment of AWS. Current recommendations rely mostly on expert opinion with a low level of evidence [31,42-48].

Conclusion

In conclusion, ICU patients in this sample drawn from a single hospital in France were predominantly male (84%) with a median age of 53 (IQR 46–60) and were commonly admitted with additional diagnoses including sepsis, trauma, or following elective or urgent surgery. Despite having low severity scores at ICU admission, half the patients experienced an extended ICU stay, or death during hospital stay. The likelihood of developing complicated hospital stay was lower in patients with seizures and higher in those with multiple organ dysfunctions at ICU admission. A previous history of AWS, Cushman’s score, and delirium tremens were not associated with outcomes. These findings suggest that early identification of organ dysfunctions and prompt recognition and treatment of sepsis may improve patient outcomes. Additional trials are needed to determine the optimal therapeutic strategy for decreasing the morbidity and mortality of AWS. Finally, the high frequency of persistent confusion at ICU discharge underlines the need for studies focusing on the long-term outcomes of AWS.

Diagnostic criteria for alcohol withdrawal syndrome according to the Diagnostic and Statistical Manual of Mental Disorders (Fifth Edition) (DSM-5).

(DOCX)

Click here for additional data file.

Comparison of ICU patients with and without AWS during the study period.

(DOCX)

Click here for additional data file.

Logistic regression analyses for factors associated with complicated hospital stay among the 5,641 patients admitted to the ICU during the study period.

(DOCX)

Click here for additional data file.

(CSV)

Click here for additional data file.

2 Nov 2021

PONE-D-21-30656Alcohol Withdrawal Syndrome in ICU Patients: Clinical Features, Management, and Outcome PredictorsPLOS ONE

Dear Dr. Canet,

Thank you for submitting your manuscript to PLOS ONE. After careful consideration, we feel that it has merit but does not fully meet PLOS ONE’s publication criteria as it currently stands. Therefore, we invite you to submit a revised version of the manuscript that addresses the points raised during the review process. Please submit your revised manuscript by Dec 17 2021 11:59PM. If you will need more time than this to complete your revisions, please reply to this message or contact the journal office at plosone@plos.org. When you're ready to submit your revision, log on to https://www.editorialmanager.com/pone/ and select the 'Submissions Needing Revision' folder to locate your manuscript file.

Please include the following items when submitting your revised manuscript:

A rebuttal letter that responds to each point raised by the academic editor and reviewer(s). You should upload this letter as a separate file labeled 'Response to Reviewers'.

A marked-up copy of your manuscript that highlights changes made to the original version. You should upload this as a separate file labeled 'Revised Manuscript with Track Changes'.

An unmarked version of your revised paper without tracked changes. You should upload this as a separate file labeled 'Manuscript'.

If you would like to make changes to your financial disclosure, please include your updated statement in your cover letter. Guidelines for resubmitting your figure files are available below the reviewer comments at the end of this letter.

If applicable, we recommend that you deposit your laboratory protocols in protocols.io to enhance the reproducibility of your results. Protocols.io assigns your protocol its own identifier (DOI) so that it can be cited independently in the future. For instructions see: https://journals.plos.org/plosone/s/submission-guidelines#loc-laboratory-protocols. Additionally, PLOS ONE offers an option for publishing peer-reviewed Lab Protocol articles, which describe protocols hosted on protocols.io. Read more information on sharing protocols at https://plos.org/protocols?utm_medium=editorial-email&utm_source=authorletters&utm_campaign=protocols.

We look forward to receiving your revised manuscript.

Kind regards,

Aleksandar R. Zivkovic

Academic Editor

PLOS ONE

Journal Requirements:

When submitting your revision, we need you to address these additional requirements.

1. Please ensure that your manuscript meets PLOS ONE's style requirements, including those for file naming. The PLOS ONE style templates can be found at

https://journals.plos.org/plosone/s/file?id=wjVg/PLOSOne_formatting_sample_main_body.pdf and

https://journals.plos.org/plosone/s/file?id=ba62/PLOSOne_formatting_sample_title_authors_affiliations.pdf.

2. Please provide additional details regarding participant consent. In the Methods section, please ensure that you have specified (1) whether consent was informed and (2) what type you obtained (for instance, written or verbal). If your study included minors, state whether you obtained consent from parents or guardians. If the need for consent was waived by the ethics committee, please include this information.

3. Thank you for stating the following in the Competing Interests section:

[EC received fees for lectures and conference talks and had travel and accommodation expenses to attend scientific meetings covered by Gilead, Baxter, and Sanofi-Genzyme.

The authors declare that they have no competing interests.]

Please confirm that this does not alter your adherence to all PLOS ONE policies on sharing data and materials, by including the following statement: "This does not alter our adherence to  PLOS ONE policies on sharing data and materials.” (as detailed online in our guide for authors http://journals.plos.org/plosone/s/competing-interests).  If there are restrictions on sharing of data and/or materials, please state these. Please note that we cannot proceed with consideration of your article until this information has been declared.

Please include your updated Competing Interests statement in your cover letter; we will change the online submission form on your behalf.

4. We note that you have stated that you will provide repository information for your data at acceptance. Should your manuscript be accepted for publication, we will hold it until you provide the relevant accession numbers or DOIs necessary to access your data. If you wish to make changes to your Data Availability statement, please describe these changes in your cover letter and we will update your Data Availability statement to reflect the information you provide.

[Note: HTML markup is below. Please do not edit.]

Reviewers' comments:

Reviewer #1: 1) Table 1: Why is patient demographics is limited to age + gender? Health disparities may influence outcomes. Income, access to care, native vs. foreign-born, primary language may factors influencing time of presentation

2) Table 1: complicated v. uncomplicated appear to be 2 different patient populations based on chronic medication use (BZD, antipsychotics). Cessation of BZD can result in seizures. Antipsychotics may lower seizure threshold. Both may result in apparent alcohol withdrawal seizures.

3) Table 2: What was time from hospital admission to ICU admission? Medications received prior to ICU? Treatment prior to ICU arrival may influence ICU outcomes (see #2). For instance, oversedation in ED or hospital wards may lead to aspiration, PNA, sepsis, etc. See Melkonian et al (2019)

4) Table 3: how was persistent confusion at ICU discharge assessed? CAMS-ICU? Clinical impression? How much was delirium vs. dementia? What was baseline confusional status?

5) Discussion: based on multivariable analysis, authors argue that sicker patients (MOD) have complicated stays. It would be interesting to learn how this compares to patients without AWS.

6) Abstract: based on multivariable analysis, authors also argue that seizures are protective. I find this statement illogical since seizures are, by definition, harmful.

7) Discussion: the presence of DTs or alcohol withdrawal seizures may suggest opportunities to standardize treatment of AWS in the ED and general wards, which may hopefully reduce incidence of ICU admission.

Reviewer #2: PONE-D-21-30656 — Alcohol Withdrawal Syndrome in ICU Patients: Clinical Features, Management, and Outcome Predictors

The authors perform a retrospective cohort study of ICU patients with AWS, using manual chart review for data extraction. The main objective was to describe a wide array of factors associated with ICU stay ≥ 7 days and/or in-hospital mortality – a combined outcome the authors labeled “complicated hospital stay.” This is mainly a descriptive study, including descriptions of many patient-level factors stratified by the primary outcome (i.e., complicated hospital stay): demographical characteristics, baseline diagnoses, chronic medications, reason for ICU admission, clinical features associated with acute illness (e.g., SOFA scores), AWS therapies, AWS-related clinical scores (i.e., Cushman) and diagnoses (e.g., seizures), duration of AWS, life-sustaining ICU therapies (e.g., mechanical ventilation), persistent confusion at ICU discharge, length of stay, mortality, and disposition upon hospital discharge. Logistic regression analyses were also performed to further evaluate the association between certain patient factors and the primary composite outcome, but it is not entirely clear how/why the much smaller list of independent variables were selected and included in the model.

The authors should be commended for investigating a grossly understudied yet common ICU condition, offering insights regarding the basic epidemiology of AWS in the ICU and “real world” treatment approaches and hospital course. Although the study is interesting, it lacks focus and does not seem to be driven by a central hypothesis or research question. As a result, the reader gets lost. Many conclusions are stated throughout the discussion that cannot be drawn from this study. Given these broad issues, I am offering general feedback with some specific examples below. Overall, I think the study design and resulting manuscript needs significant restructuring.

Examples of issues that need to be addressed:

Lack of transparency re. methods – the data for this study was mainly obtained via manual chart review. This could be a strength if rigorously approached but was incompletely described. As it stands, the methods section leaves many questions unaddressed. For example, were “habits of alcohol consumption” obtained via patient interview? Is the alcohol history part of the standard hospital intake procedure? What was the frequency of missing data (many ICU patients are too sick to provide history)? Did authors AV and EC manually extract ALL data for the study or just confirm the diagnosis of AWS? How were DSM-5 criteria for AWS applied to information recorded in electronic medical records that was not necessarily designed for assessing these criteria? For example, how was “increased hand tremor” (included in the DSM-5 criteria) assessed via the electronic medical records?

Statistical analysis section is confusing – “Quantitative” and “qualitative” variables are referred to in the first sentence of this section—do the authors mean continuous and categorical variables? It seems a statistical approach (“significant” variables in univariate analyses) versus a hypothesis driven approach was used to design the multivariable logistic regression model. Presumably the authors are using this multivariable model to address confounding, but the underlying hypothesis regarding how these variables relate is unclear. An excellent reference for thinking about study design and presentation of results is: Lederer et al. Annals Am Thorac Soc 2019;16(1):22-28.

“Overreaching” conclusions – In the final paragraph (and similarly stated in the abstract), “AWS in ICU patients chiefly affects young patients with few comorbidities and is often triggered by a precipitating factor such as sepsis, trauma, or surgery.” A more accurate statement from my perspective might be: “ICU patients in this sample drawn from a single hospital in France were predominantly male (84%) with a median age of 53 (IQR 46-60) and were commonly admitted with additional comorbidities including sepsis, trauma, or following (elective?) surgery.” We do not know that AWS was “triggered” by comorbid conditions like sepsis. AWS is more likely triggered by heavy alcohol use that is ALSO possibly associated with these other conditions (based on data from other studies).

Implicit comparisons to a larger ICU sample, not included in the study – The authors seem to make comparisons to a broader ICU cohort. For example, “Despite having low severity scores at ICU admission”… or “the high frequency of persistent confusion” – these statements imply comparisons but the comparison group (implicitly, ICU patients at large) is not defined for the reader. The authors’ tendency to make such comparisons illustrates perhaps the fundamental design flaw of the study. Descriptions of ICU patients with AWS are provided, but without context. The reader is left wondering, how does this compare to “average” ICU patients at the study hospital? The association identified between organ dysfunction and the combined outcome of ICU stay ≥ 7 days and/or in-hospital mortality is not surprising in ICU patients. Whether or not AWS modifies this relationship would be the interesting question; for example, testing the hypothesis that the known association between organ dysfunction and ICU length of stay and/or in-hospital mortality is more pronounced in patients with AWS compared to patients without AWS.

[NOTE: If reviewer comments were submitted as an attachment file, they will be attached to this email and accessible via the submission site. Please log into your account, locate the manuscript record, and check for the action link "View Attachments". If this link does not appear, there are no attachment files.]

While revising your submission, please upload your figure files to the Preflight Analysis and Conversion Engine (PACE) digital diagnostic tool, https://pacev2.apexcovantage.com/. PACE helps ensure that figures meet PLOS requirements. To use PACE, you must first register as a user. Registration is free. Then, login and navigate to the UPLOAD tab, where you will find detailed instructions on how to use the tool. If you encounter any issues or have any questions when using PACE, please email PLOS at figures@plos.org. Please note that Supporting Information files do not need this step.

24 Nov 2021

PONE-D-21-30656

Alcohol Withdrawal Syndrome in ICU Patients: Clinical Features, Management, and Outcome Predictors

PLOS ONE

Dear Dr. Canet,

Thank you for submitting your manuscript to PLOS ONE. After careful consideration, we feel that it has merit but does not fully meet PLOS ONE’s publication criteria as it currently stands. Therefore, we invite you to submit a revised version of the manuscript that addresses the points raised during the review process.

Please submit your revised manuscript by Dec 17 2021 11:59PM. If you will need more time than this to complete your revisions, please reply to this message or contact the journal office at plosone@plos.org. When you're ready to submit your revision, log on to https://www.editorialmanager.com/pone/ and select the 'Submissions Needing Revision' folder to locate your manuscript file.

Author’s reply: Thank you. We are grateful to the editor for giving us the opportunity to improve our manuscript and submit a revised version. All comments have been taken into account in the revised version of the manuscript.

Please include the following items when submitting your revised manuscript:

• A rebuttal letter that responds to each point raised by the academic editor and reviewer(s). You should upload this letter as a separate file labeled 'Response to Reviewers'.

• A marked-up copy of your manuscript that highlights changes made to the original version. You should upload this as a separate file labeled 'Revised Manuscript with Track Changes'.

• An unmarked version of your revised paper without tracked changes. You should upload this as a separate file labeled 'Manuscript'.

If you would like to make changes to your financial disclosure, please include your updated statement in your cover letter. Guidelines for resubmitting your figure files are available below the reviewer comments at the end of this letter.

If applicable, we recommend that you deposit your laboratory protocols in protocols.io to enhance the reproducibility of your results. Protocols.io assigns your protocol its own identifier (DOI) so that it can be cited independently in the future. For instructions see: https://journals.plos.org/plosone/s/submission-guidelines#loc-laboratory-protocols. Additionally, PLOS ONE offers an option for publishing peer-reviewed Lab Protocol articles, which describe protocols hosted on protocols.io. Read more information on sharing protocols at https://plos.org/protocols?utm_medium=editorial-email&utm_source=authorletters&utm_campaign=protocols.

We look forward to receiving your revised manuscript.

Kind regards,

Aleksandar R. Zivkovic

Academic Editor

PLOS ONE

Journal Requirements:

When submitting your revision, we need you to address these additional requirements.

1. Please ensure that your manuscript meets PLOS ONE's style requirements, including those for file naming. The PLOS ONE style templates can be found at

https://journals.plos.org/plosone/s/file?id=wjVg/PLOSOne_formatting_sample_main_body.pdf and

https://journals.plos.org/plosone/s/file?id=ba62/PLOSOne_formatting_sample_title_authors_affiliations.pdf.

Author’s reply: The manuscript has been revised and now fulfilled all PLOS ONE’s style requirements.

2. Please provide additional details regarding participant consent. In the Methods section, please ensure that you have specified (1) whether consent was informed and (2) what type you obtained (for instance, written or verbal). If your study included minors, state whether you obtained consent from parents or guardians. If the need for consent was waived by the ethics committee, please include this information.

Author’s reply: The requested information is now detailed in the revised version of the manuscript.

3. Thank you for stating the following in the Competing Interests section:

[EC received fees for lectures and conference talks and had travel and accommodation expenses to attend scientific meetings covered by Gilead, Baxter, and Sanofi-Genzyme.

The authors declare that they have no competing interests.]

Please confirm that this does not alter your adherence to all PLOS ONE policies on sharing data and materials, by including the following statement: "This does not alter our adherence to PLOS ONE policies on sharing data and materials.” (as detailed online in our guide for authors http://journals.plos.org/plosone/s/competing-interests). If there are restrictions on sharing of data and/or materials, please state these. Please note that we cannot proceed with consideration of your article until this information has been declared.

Please include your updated Competing Interests statement in your cover letter; we will change the online submission form on your behalf.

Author’s reply: The requested information has been added in the revised version of the manuscript.

4. We note that you have stated that you will provide repository information for your data at acceptance. Should your manuscript be accepted for publication, we will hold it until you provide the relevant accession numbers or DOIs necessary to access your data. If you wish to make changes to your Data Availability statement, please describe these changes in your cover letter and we will update your Data Availability statement to reflect the information you provide.

Author’s reply: We made changes to our Data Availability statement. This information has been added to the cover letter.

[Note: HTML markup is below. Please do not edit.]

Reviewers' comments:

Reviewer #1:

Author’s reply: We would like to thank the reviewer for making constructive comments that have helped us to clarify and improve our manuscript.

1) Table 1: Why is patient demographics is limited to age + gender? Health disparities may influence outcomes. Income, access to care, native vs. foreign-born, primary language may factors influencing time of presentation

Author’s reply: The reviewer raises an important point. We agree that other social and demographic factors may influence patients’ outcome. We report age, gender and the burden of comorbidities measured by the Charlson’s index, but the risk of uncaptured confounding factors is significant. Unfortunately, in our institution, the number of demographic factors routinely recorded in the electronic medical health records is limited. Data related to income, place of birth and primary language were not available. Access to care in public hospitals is unrestricted in France, even for patients not covered by the statutory French health insurance (which account for a very limited number of patients). Although we agree that such patients may experience issues related to follow-up after hospital discharge and access to chronic treatments (key determinants of long-term outcome), our purpose was to focus on the ICU setting and the short-term (day-28 after ICU).

This limitation has been clearly acknowledged in the revised version of the manuscript (discussion section, limitations) as follow:

“Our study also has several limitations. First, the retrospective design implies information bias with a possibility of missing important data. For example, we had no information on other potential causes of health disparities, such as income, health care coverage or country of birth, which may have influenced patients’ outcomes.”

2) Table 1: complicated v. uncomplicated appear to be 2 different patient populations based on chronic medication use (BZD, antipsychotics). Cessation of BZD can result in seizures. Antipsychotics may lower seizure threshold. Both may result in apparent alcohol withdrawal seizures.

Author’s reply: We agree that these 2 populations share similarities (age, AWS and psychiatric history) but also have significant differences (chronic medications, reason for ICU admission). Although we agree that BZD withdrawal syndrome and the use of antipsychotics increase the risk of seizure, in our study patients who had a complicated hospital stay were four times less likely to be admitted to the ICU for seizures than patients who had an uncomplicated hospital stay (4.1% versus 18.9%, table 1). Therefore, we hypothesize that patients who had such chronic medications did not experience significant BZD deficiency or antipsychotics overdose. Thus, we believe that AWS was not “overdiagnosed” in patients with a complicated hospital stay. Moreover, as AWS is a difficult clinical diagnosis without gold standard, we carefully double checked the plausibility of AWS diagnosis using the combination of ICD10 criteria and DSM-5 criteria with a thorough analysis of each electronic medical record by 2 investigators (AV and EC). Patients who had isolated seizures without the other DSM-5 criteria were not classified as having AWS. Using such methodology, 42 patients were excluded from the analysis because the diagnosis of AWS was unclear (Figure 1, flowchart).

This important point has been clarified in the revised version of the manuscript (methods section) as follow:

“Each medical file was reviewed by AV and EC to confirm the diagnosis of AWS based on the Diagnostic and Statistical Manual of Mental Disorders, Fifth Edition (DSM-5). All four major criteria had to be present in the electronic medical record of each patient to diagnose AWS. For major criterion B, 2 or more of the 8 symptoms had to be present. Data were extracted from the doctors and nurses notes (Supplementary appendix, Figure 1)”.

3) Table 2: What was time from hospital admission to ICU admission? Medications received prior to ICU? Treatment prior to ICU arrival may influence ICU outcomes (see #2). For instance, oversedation in ED or hospital wards may lead to aspiration, PNA, sepsis, etc. See Melkonian et al (2019)

Author’s reply: Time from hospital admission to ICU admission was 0[0-1] day without significant difference between the complicated group (0[0-1] day) and the uncomplicated group (0[0-1] day) (p=0.168). Of note, more than two-thirds of the patients were admitted to the ICU directly from the ED. In the subgroup of patients who were admitted to the ICU from the medical or surgical wards, median time from hospital to ICU admission was 0[0-2.5] day, without significant difference between the complicated and uncomplicated groups (1[0-3] vs 0[0-1.25], p=0.143). Unfortunately, medications received prior to ICU were not available. Taken all together, our results suggest that time from hospital to ICU admission had no influence on patients’ outcome in our study. However, our results are undoubtedly influenced by the single-centre design of our study, and thus by the policy of ICU referral and admission in our hospital. Even if we do not have predefined criteria for MET/RRT activation, our policy strongly encourages early assessment of all unstable patients by the intensivist and early ICU admission when ED or ward staff are worried about a patient. Therefore, all AWS patients not responding to a first line of treatment (intermittent administration of BZD) are admitted to the ICU and treatment escalation is not administered outside the ICU.

As suggested by the reviewer, we have added the information in the revised version of the manuscript (Table 1) and the suggested reference in the discussion section.

Page 15 lines 297-300

Variable

All patients

(n = 204)

Complicated hospital stay

(n = 98)

Uncomplicated hospital stay

(n = 106)

P value

Demographics

Age, median [IQR], years 53 [46-60] 54.5 [48-61] 50 [44-58] 0.099

Male sex, n (%) 172 (84.3) 83 (84.7) 89 (84.0) 0.88

Charlson’s index, median [IQR] 1 [0-3] 2 [0.25-4] 1 [0-3] 0.019

Alcohol withdrawal history

History of AWS, n (%) 42 (20.6) 23 (23.5) 19 (18.0) 0.387

History of DT, n (%) 10 (4.9) 7 (7.1) 3 (2.8) 0.200

History of withdrawal seizures, n (%) 25 (12.3) 15 (15.3) 10 (9.4) 0.201

Psychiatric history

Substance use disorder other than alcohol, n (%) 30 (14.7) 12 (12.2) 18 (17.0) 0.429

Any psychiatric disorder, n (%) 71 (34.8) 33 (33.7) 38 (35.9) 0.598

Mood disorders, n (%) 7 (3.4) 4 (4.1) 3 (2.8) 0.712

Anxiety disorders, n (%) 64 (31.4) 29 (29.6) 35 (33) 0.651

Chronic medications

Benzodiazepines, n (%) 71 (34.8) 27 (27.6) 44 (41.5) 0.036

Antipsychotic drugs, n (%) 26 (12.8) 6 (6.1) 21 (18.9) 0.0063

Time from hospital admission to ICU admission, median [IQR], days 0 [0-1] 0 [0-1] 0 [0-1] 0.168

ICU admission from the ED, n (%) 137 (67.2) 63 (64.3) 74 (69.8) 0.401

Cushman’s score at ED admission, median [IQR] 7 [4-9] 7 [4-9] 7 [4-9] 0.827

Reason for ICU admission, n (%) 0.00073

Sepsis 61 (29.9) 39 (39.8) 22 (20.8)

Altered consciousness 60 (29.4) 22 (22.5) 38 (35.9)

Seizures 24 (11.7) 4 (4.1) 20 (18.9)

Trauma 20 (9.8) 9 (9.2) 11 (10.4)

Surgery 12 (5.9) 6 (6.1) 6 (5.7)

AKI 12 (5.9) 7 (7.2) 5 (4.7)

Other* 15 (7.4) 11 (11.2) 4 (3.8)

Clinical variables and measures at ICU admission

HR, median [IQR], bpm 104 [88-120] 109 [98-123] 99 [85-117] 0.038

SBP, median [IQR], mmHg 125 [106-147] 121 [100-140] 126 [112-150] 0.038

Glasgow Coma Scale score, median [IQR] 14 [12-15] 14 [11-15] 14 [13-15]

0.84

RR, median [IQR] 22 [18-26] 22.5 [19.3-27] 20 [17-25] 0.015

Cushman score 6 [4-9] 6 [4-9] 7 [4-9] 0.196

SOFA 3 [2-6] 5 [3-8] 3 [1-5] <0.0001

SAPS II 24 [16-34] 27 [17-37] 20 [13-31] 0.0067

4) Table 3: how was persistent confusion at ICU discharge assessed? CAMS-ICU? Clinical impression? How much was delirium vs. dementia? What was baseline confusional status?

Author’s reply: We agree that this point needs to be clarified. For each patient, we used the daily notes of nurses and doctors in the electronic medical record to retrospectively assess the date of AWS resolution or the status of persistent confusion. The date of AWS resolution was the date either where the episode of AWS was stated to have resolved in the EMR, or the last date where AWS was mentioned if it was followed by no further signs or symptoms of AWS for a period of at least 48 hours and no confusion was reported at the time of discharge. If none of these 2 conditions were reported, the episode of AWS was considered as persistent. We agree that this is a non-validated method. However, we provide information on confusion persistence after AWS in the ICU, an area rarely explored.

Six patients (2.9%) had dementia mentioned in their past medical history. For such patients a worsening of the clinical state during hospitalization and at the time of discharge had to be mentioned in the patients’ EMR to classify the patient in the “persistent confusion” category.

As suggested by the reviewer, we have added the information in the revised version of the manuscript (Methods section and Table 1) as follow:

“For each patient, AWS recovery was assessed by reading the daily notes of nurses and doctors from the EMR. The date of resolution was either the date of resolution recorded in the medical file or the last date of recording of AWS with no further signs or symptoms of AWS recorded for at least 48 hours. If neither of these two conditions was met, the episode of AWS was classified as persistent. When patients had underlying dementia or other neurocognitive disorders, a worsening of the clinical state during hospitalization had to be mentioned in the patients’ EMR to classify a patient with persistent confusion”

Table 1. Baseline characteristics of the 204 study participants

Variable

All patients

(n = 204)

Complicated hospital stay

(n = 98)

Uncomplicated hospital stay

(n = 106)

P value

Demographics

Age, median [IQR], years 53 [46-60] 54.5 [48-61] 50 [44-58] 0.099

Male sex, n (%) 172 (84.3) 83 (84.7) 89 (84.0) 0.88

Charlson’s index, median [IQR] 1 [0-3] 2 [0.25-4] 1 [0-3] 0.019

Alcohol withdrawal history

History of AWS, n (%) 42 (20.6) 23 (23.5) 19 (18.0) 0.387

History of DT, n (%) 10 (4.9) 7 (7.1) 3 (2.8) 0.200

History of withdrawal seizures, n (%) 25 (12.3) 15 (15.3) 10 (9.4) 0.201

Psychiatric history

Substance use disorder other than alcohol, n (%) 30 (14.7) 12 (12.2) 18 (17.0) 0.429

Any psychiatric disorder, n (%)* 71 (34.8) 33 (33.7) 38 (35.9) 0.598

Mood disorders, n (%) 7 (3.4) 4 (4.1) 3 (2.8) 0.712

Anxiety disorders, n (%) 64 (31.4) 29 (29.6) 35 (33) 0.651

Chronic medications

Benzodiazepines, n (%) 71 (34.8) 27 (27.6) 44 (41.5) 0.036

Antipsychotic drugs, n (%) 26 (12.8) 6 (6.1) 21 (18.9) 0.0063

ICU admission from the ED, n (%) 137 (67.2) 63 (64.3) 74 (69.8) 0.401

Cushman’s score at ED admission, median [IQR] 7 [4-9] 7 [4-9] 7 [4-9] 0.827

Reason for ICU admission, n (%) 0.00073

Sepsis 61 (29.9) 39 (39.8) 22 (20.8)

Altered consciousness 60 (29.4) 22 (22.5) 38 (35.9)

Seizures 24 (11.7) 4 (4.1) 20 (18.9)

Trauma 20 (9.8) 9 (9.2) 11 (10.4)

Surgery 12 (5.9) 6 (6.1) 6 (5.7)

AKI 12 (5.9) 7 (7.2) 5 (4.7)

Other** 15 (7.4) 11 (11.2) 4 (3.8)

Clinical variables and measures at ICU admission

HR, median [IQR], bpm 104 [88-120] 109 [98-123] 99 [85-117] 0.038

SBP, median [IQR], mmHg 125 [106-147] 121 [100-140] 126 [112-150] 0.038

Glasgow Coma Scale score, median [IQR] 14 [12-15] 14 [11-15] 14 [13-15]

0.84

RR, median [IQR] 22 [18-26] 22.5 [19.3-27] 20 [17-25] 0.015

Cushman score 6 [4-9] 6 [4-9] 7 [4-9] 0.196

SOFA 3 [2-6] 5 [3-8] 3 [1-5] <0.0001

SAPS II 24 [16-34] 27 [17-37] 20 [13-31] 0.0067

*Any psychiatric disorder, n (%): including 6 patients with underlying dementia

** Other: cardiac or respiratory arrest; upper gastrointestinal hemorrhage; acute pancreatitis; ketoacidosis; mesenteric ischemia

5) Discussion: based on multivariable analysis, authors argue that sicker patients (MOD) have complicated stays. It would be interesting to learn how this compares to patients without AWS.

Author’s reply: We agree that some of our findings in AWS patients (influence of multiple organ dysfunctions on patients’ outcome) may apply to many other diseases or conditions in the ICU setting (Sepsis, ARDS, trauma, pancreatitis,…). As suggested by the reviewer, we collected the available data of patients without AWS admitted to the ICU during the study period. Patients with AWS were younger and had lower severity scores at ICU admission than patients without AWS. The incidence of complicated hospital stay was 48% in patients with AWS and 31% in patients without AWS (p<0.001). Interestingly, the incidence of complicated hospital stay was explained by a higher incidence of extended stay in the ICU while the mortality was lower. In a multivariable model which included age, SAPSII and AWS, AWS had the highest aOR for complicated hospital stay.

As suggested by the reviewer, this information has been added to the result section of the revised version of the manuscript (page 11 lines 210-211, page 12 line 237, and page 13 lines 238-239). Both tables have been added to the supplementary appendix.

SA Table 1. Comparison of ICU patients with and without AWS during the study period

Variable

Patients

with AWS

(n = 204)

Patients without AWS

(n =5437)

P value

Demographics

Age, median [IQR], years 53 [46-60] 60 [45-70] 0.002

SAPS II, median [IQR], years 24 [16-34] 35 [24-52] 0.001

Outcome

ICU LOS, median [IQR], days 6 [4-10.3] 3 [2-6] <0.001

ICU LOS≥7days, n (%) 90 (44) 1095 (20.1) <0.001

Hospital mortality, n (%) 16 (7.8) 786 (14.5) 0.008

Complicated hospital stay

ICU LOS ≥7 days or hospital death, n (%) 98 (48) 1685 (31) <0.001

AWS: alcohol withdrawal syndrome; ICU: intensive care unit; IQR: interquartile range; LOS: length of stay; SAPS II: Simplified Acute Physiology Score, version II

SA Table 2: Logistic regression analyses for factors associated with complicated hospital stay among the 5,641 patients admitted to the ICU during the study period.

Factors

Multivariable analysis

OR (95%CI) P value

Age (per year) 0.99 (0.99-1.00) 0.368

SAPS II (per point) 1.06 (1.05-1.06) <0.001

Alcohol withdrawal syndrome 3.53 (2.60-4.81) <0.001

SAPS II: Simplified Acute Physiology Score, version II

Candidate predictors were: Age, SAPS II, and alcohol withdrawal syndrome.

6) Abstract: based on multivariable analysis, authors also argue that seizures are protective. I find this statement illogical since seizures are, by definition, harmful.

Author’s reply: We agree with the reviewer that we need to rephrase the conclusion, which is confusing and could be misinterpreted and misunderstood.

As suggested by the reviewer, we modified the abstract’s conclusion as follow:

“The likelihood of developing complicated hospital stay relied on the reason for ICU admission and the number of organ dysfunctions at ICU admission.”

7) Discussion: the presence of DTs or alcohol withdrawal seizures may suggest opportunities to standardize treatment of AWS in the ED and general wards, which may hopefully reduce incidence of ICU admission.

Author’s reply: We agree with the reviewer that preventing clinical deterioration of AWS by improving its early identification and standardizing its treatment can improve quality of care and patient safety, as reported by Melkonian et al.

As suggested by the reviewer, this point has been added to the revised version of the manuscript (discussion section – comparison with previous studies – optimal management - page 15 lines 297-300) as follow:

“Interestingly, a recent study reported that the implementation of a hospital-wide protocol for the management of AWS resulted in significant improvements in quality of care, decreased the need for ICU admission and the rate of intubation, reduced hospital length of stay, and was cost-savings (34).”

(34) Melkonian A, Patel R, Magh A, Ferm S, Hwang C. Assessment of a Hospital-Wide CIWA-Ar Protocol for Management of Alcohol Withdrawal Syndrome. Mayo Clin Proc Innov Qual Outcomes. 2019 Aug 23;3(3):344-349. doi: 10.1016/j.mayocpiqo.2019.06.005. eCollection 2019 Sep.

Reviewer #2: PONE-D-21-30656 — Alcohol Withdrawal Syndrome in ICU Patients: Clinical Features, Management, and Outcome Predictors

The authors perform a retrospective cohort study of ICU patients with AWS, using manual chart review for data extraction. The main objective was to describe a wide array of factors associated with ICU stay ≥ 7 days and/or in-hospital mortality – a combined outcome the authors labeled “complicated hospital stay.” This is mainly a descriptive study, including descriptions of many patient-level factors stratified by the primary outcome (i.e., complicated hospital stay): demographical characteristics, baseline diagnoses, chronic medications, reason for ICU admission, clinical features associated with acute illness (e.g., SOFA scores), AWS therapies, AWS-related clinical scores (i.e., Cushman) and diagnoses (e.g., seizures), duration of AWS, life-sustaining ICU therapies (e.g., mechanical ventilation), persistent confusion at ICU discharge, length of stay, mortality, and disposition upon hospital discharge. Logistic regression analyses were also performed to further evaluate the association between certain patient factors and the primary composite outcome, but it is not entirely clear how/why the much smaller list of independent variables were selected and included in the model.

The authors should be commended for investigating a grossly understudied yet common ICU condition, offering insights regarding the basic epidemiology of AWS in the ICU and “real world” treatment approaches and hospital course. Although the study is interesting, it lacks focus and does not seem to be driven by a central hypothesis or research question. As a result, the reader gets lost. Many conclusions are stated throughout the discussion that cannot be drawn from this study. Given these broad issues, I am offering general feedback with some specific examples below. Overall, I think the study design and resulting manuscript needs significant restructuring.

Author’s reply: We thank the reviewer for all the comments and for giving us the opportunity to improve our manuscript and to submit a revised version. We intended to conduct an epidemiological study of AWS in the ICU setting, to describe its clinical features, course, treatment and outcome in the ICU. We hypothesized that a proportion of ICU patients with AWS would experience a complicated hospital stay (defined by hospital death or extended stay in the ICU) and we aimed to identify factors associated with such outcomes.

As suggested by the reviewer, the research purpose has been clarified in the revised version of the manuscript and the methods section has been thoroughly revised.

Examples of issues that need to be addressed:

Lack of transparency re. methods – the data for this study was mainly obtained via manual chart review. This could be a strength if rigorously approached but was incompletely described. As it stands, the methods section leaves many questions unaddressed. For example, were “habits of alcohol consumption” obtained via patient interview? Is the alcohol history part of the standard hospital intake procedure? What was the frequency of missing data (many ICU patients are too sick to provide history)? Did authors AV and EC manually extract ALL data for the study or just confirm the diagnosis of AWS? How were DSM-5 criteria for AWS applied to information recorded in electronic medical records that was not necessarily designed for assessing these criteria? For example, how was “increased hand tremor” (included in the DSM-5 criteria) assessed via the electronic medical records?

Author’s reply: We agree with the reviewer that this point needs to be clarified. Yes, all data were manually extracted from the EMRs for each patient by 2 investigators (AV and EC) to minimize biais and improve accuracy.

Habits of alcohol consumption and alcohol history are part of the standard intake procedure in our hospital. Data were obtained from patients’ interview. When the patient’s clinical condition made the interview impossible, data were obtained either from the next of kin or from the patient at the time of discharge (when he recovered from the acute illness). However, data on the daily alcohol intake was missing in 51 patients (25%). This information has been added in the revised version of the manuscript (methods and results sections).

We agree with the reviewer that AWS is a difficult clinical diagnosis with no gold standard criteria. We carefully double checked the plausibility of AWS diagnosis using the combination of ICD10 criteria and DSM-5 criteria. For each patient, a thorough analysis of the electronic medical record (doctors and nurses notes) was done by 2 investigators (AV and EC). All four major criteria had to be present in the electronic medical record of each patient to diagnose AWS. For major criterion B, 2 or more of the 8 symptoms had to be present. Using such methodology, 42 patients were excluded from the analysis because the diagnosis of AWS was unclear (Figure 1, flowchart).

This important point has been clarified in the revised version of the manuscript (methods section) as follow:

“Each medical file was reviewed by AV and EC to confirm the diagnosis of AWS based on the Diagnostic and Statistical Manual of Mental Disorders, Fifth Edition (DSM-5). All four major criteria had to be present in the electronic medical records of each patient to diagnose AWS. For major criterion B, 2 or more of the 8 symptoms had to be present. Data were extracted from the doctors and nurses notes (S1 Fig.)”.

Statistical analysis section is confusing – “Quantitative” and “qualitative” variables are referred to in the first sentence of this section—do the authors mean continuous and categorical variables? It seems a statistical approach (“significant” variables in univariate analyses) versus a hypothesis driven approach was used to design the multivariable logistic regression model. Presumably the authors are using this multivariable model to address confounding, but the underlying hypothesis regarding how these variables relate is unclear.

Author’s reply: We agree with the reviewer that our statistical section is unclear and needs to be clarified. As suggested by the reviewer, “quantitative” and “qualitative” have been replaced by “continuous” and “categorical” variables in the revised version of the manuscript.

Most of the literature on AWS focused on identifying predictors of delirium tremens (DT) or seizures. However, whether DT or seizures are associated with other relevant patients-centered outcomes (mortality, extended ICU stay) remains unclear. Therefore, we purposefully chose to identify factors associated with hospital death or extended ICU stay (combined outcome analyzed as a binary variable), an area almost unstudied in ICU patients with AWS. Our aim was to identify frontline variables (available at the time of ICU admission) associated with such outcomes to help clinicians for early identification of patients at risk of clinical deterioration who may benefit the most from close monitoring and therapeutic interventions.

Therefore, we purposefully preselected variables which plausibly fit these outcomes based on knowledge from the literature (SOFA, comorbidities, and mortality) and assumptions (chronic use of BZD or antipsychotics, history of AWS, reason for ICU admission, and extended stay in the ICU). We carefully checked to avoid collinearity (for example: SOFA and heart rate, blood pressure, and respiratory rate) and we applied the rule of selecting a maximum of 1 variable per 8 events (total of 12 variables in our study). All variables included in the model are displayed and thus, we believe our assumptions are transparent and explicit.

As suggested by the reviewer, we have revised the statistical methods section as follow:

“Continuous variables are described as median and interquartile range [IQR] and compared using Wilcoxon’s test. Categorical variables are described as counts (percent) and compared using the exact Fisher’s test. The occurrence of complicated hospital stay (versus uncomplicated hospital stay) was analyzed as a binary variable. Logistic regression analyses were performed to identify variables associated with complicated hospital stay, with estimated odds ratios (ORs) and their 95% confidence intervals (95%CIs). For the multivariable model, we preselected candidate variables which plausibly fit with complicated hospital stay based on knowledge from the literature (SOFA, comorbidities, and mortality) and our assumptions (chronic use of BZD or antipsychotics, history of AWS, reason for ICU admission, and extended stay in the ICU). We carefully checked to avoid collinearity between variables and we applied the rule of selecting a maximum of 1 variable per 8 events (total of 12 variables in our study). All tests were two-sided, and P values lower than 5% were considered to indicate significant associations. Statistical tests were conducted using the R statistics program, version 3.5.0 (R Foundation for Statistical Computing, Vienna, Austria; www.R-project.org/).”

In addition, preselected candidate variables included in the multivariable model are now clearly stated in the footnote of the revised version of table 4.

An excellent reference for thinking about study design and presentation of results is: Lederer et al. Annals Am Thorac Soc 2019;16(1):22-28.

“Overreaching” conclusions – In the final paragraph (and similarly stated in the abstract), “AWS in ICU patients chiefly affects young patients with few comorbidities and is often triggered by a precipitating factor such as sepsis, trauma, or surgery.” A more accurate statement from my perspective might be: “ICU patients in this sample drawn from a single hospital in France were predominantly male (84%) with a median age of 53 (IQR 46-60) and were commonly admitted with additional comorbidities including sepsis, trauma, or following (elective?) surgery.” We do not know that AWS was “triggered” by comorbid conditions like sepsis. AWS is more likely triggered by heavy alcohol use that is ALSO possibly associated with these other conditions (based on data from other studies).

Author’s reply: We agree that our conclusions may go beyond our results and should be rephrased and tempered. We fully agree that no causal effect can be drawn from our study. However we suggest swapping “additional comorbidities” by “additional diagnoses” to avoid confusion between comorbidities (diabetes, hypertension,…) and acute conditions (sepsis, surgery,…).

As suggested by the reviewer, we have modified the abstract conclusion and the final paragraph of the manuscript as follow:

Abstract conclusion

“AWS in ICU patients chiefly affects young adults and is often associated with additional factors such as sepsis, trauma, or surgery. Half the patients experienced an extended ICU stay or death during the hospital stay. The likelihood of developing complicated hospital stay relied on the reason for ICU admission and the number of organ dysfunctions at ICU admission.”

Manuscript conclusion

“ICU patients in this sample drawn from a single hospital in France were predominantly male (84%) with a median age of 53 (IQR 46-60) and were commonly admitted with additional diagnoses including sepsis, trauma, or following elective or urgent surgery.”

Implicit comparisons to a larger ICU sample, not included in the study – The authors seem to make comparisons to a broader ICU cohort. For example, “Despite having low severity scores at ICU admission”… or “the high frequency of persistent confusion” – these statements imply comparisons but the comparison group (implicitly, ICU patients at large) is not defined for the reader. The authors’ tendency to make such comparisons illustrates perhaps the fundamental design flaw of the study. Descriptions of ICU patients with AWS are provided, but without context. The reader is left wondering, how does this compare to “average” ICU patients at the study hospital? The association identified between organ dysfunction and the combined outcome of ICU stay ≥ 7 days and/or in-hospital mortality is not surprising in ICU patients. Whether or not AWS modifies this relationship would be the interesting question; for example, testing the hypothesis that the known association between organ dysfunction and ICU length of stay and/or in-hospital mortality is more pronounced in patients with AWS compared to patients without AWS.

Author’s reply: We agree that some of our findings in AWS patients (influence of multiple organ dysfunctions on patients’ outcome) may apply to many other diseases or conditions in the ICU setting (Sepsis, ARDS, trauma, pancreatitis,…). As suggested by the reviewer, we collected the available data of patients without AWS admitted to the ICU during the study period. Patients with AWS were younger and had lower severity scores at ICU admission than patients without AWS. The incidence of complicated hospital stay was 48% in patients with AWS and 31% in patients without AWS (p<0.001). Interestingly, the incidence of complicated hospital stay was explained by a higher incidence of extended stay in the ICU while the mortality was lower. In a multivariable model which included age, SAPSII and AWS, AWS had the highest aOR for complicated hospital stay.

As suggested by the reviewer, this information has been added to the result section of the revised version of the manuscript (page 11 lines 210-211, page 12 line 237, and page 13 lines 238-239). Both tables have been added to the supplementary appendix.

SA Table 1. Comparison of ICU patients with and without AWS during the study period

Variable

Patients

with AWS

(n = 204)

Patients without AWS

(n =5437)

P value

Demographics

Age, median [IQR], years 53 [46-60] 60 [45-70] 0.002

SAPS II, median [IQR], years 24 [16-34] 35 [24-52] 0.001

Outcome

ICU LOS, median [IQR], days 6 [4-10.3] 3 [2-6] <0.001

ICU LOS≥7days, n (%) 90 (44) 1095 (20.1) <0.001

Hospital mortality, n (%) 16 (7.8) 786 (14.5) 0.008

Complicated hospital stay

ICU LOS ≥7 days or hospital death, n (%) 98 (48) 1685 (31) <0.001

AWS: alcohol withdrawal syndrome; ICU: intensive care unit; IQR: interquartile range; LOS: length of stay; SAPS II: Simplified Acute Physiology Score, version II

SA Table 2: Logistic regression analyses for factors associated with complicated hospital stay among the 5,641 patients admitted to the ICU during the study period.

Factors

Multivariable analysis

OR (95%CI) P value

Age (per year) 0.99 (0.99-1.00) 0.368

SAPS II (per point) 1.06 (1.05-1.06) <0.001

Alcohol withdrawal syndrome 3.53 (2.60-4.81) <0.001

SAPS II: Simplified Acute Physiology Score, version II

Candidate predictors were: Age, SAPS II, and alcohol withdrawal syndrome.

Submitted filename: Response to reviewers AWS ICU.docx

Click here for additional data file.

2 Dec 2021

Alcohol Withdrawal Syndrome in ICU Patients: Clinical Features, Management, and Outcome Predictors

PONE-D-21-30656R1

Dear Dr. Canet,

We’re pleased to inform you that your manuscript has been judged scientifically suitable for publication and will be formally accepted for publication once it meets all outstanding technical requirements.

Within one week, you’ll receive an e-mail detailing the required amendments. When these have been addressed, you’ll receive a formal acceptance letter and your manuscript will be scheduled for publication.

An invoice for payment will follow shortly after the formal acceptance. To ensure an efficient process, please log into Editorial Manager at http://www.editorialmanager.com/pone/, click the 'Update My Information' link at the top of the page, and double check that your user information is up-to-date. If you have any billing related questions, please contact our Author Billing department directly at authorbilling@plos.org.

If your institution or institutions have a press office, please notify them about your upcoming paper to help maximize its impact. If they’ll be preparing press materials, please inform our press team as soon as possible -- no later than 48 hours after receiving the formal acceptance. Your manuscript will remain under strict press embargo until 2 pm Eastern Time on the date of publication. For more information, please contact onepress@plos.org.

Kind regards,

Aleksandar R. Zivkovic

Academic Editor

PLOS ONE

9 Dec 2021

PONE-D-21-30656R1

Alcohol withdrawal syndrome in ICU patients: clinical features, management, and outcome predictors

Dear Dr. Canet:

I'm pleased to inform you that your manuscript has been deemed suitable for publication in PLOS ONE. Congratulations! Your manuscript is now with our production department.

If your institution or institutions have a press office, please let them know about your upcoming paper now to help maximize its impact. If they'll be preparing press materials, please inform our press team within the next 48 hours. Your manuscript will remain under strict press embargo until 2 pm Eastern Time on the date of publication. For more information please contact onepress@plos.org.

If we can help with anything else, please email us at plosone@plos.org.

Thank you for submitting your work to PLOS ONE and supporting open access.

Kind regards,

PLOS ONE Editorial Office Staff

on behalf of

Dr. Aleksandar R. Zivkovic

Academic Editor

PLOS ONE