Outcomes of Patients with Alcohol Withdrawal Syndrome Treated with High-Dose Sedatives and Deferred Intubation

Outcomes of Patients with Alcohol Withdrawal Syndrome Treated with High-Dose Sedatives and Deferred Intubation Robert Stewart 1, Ricardo Perez 2, Bogdan Musial 2, Carrie Lukens 3, Yaw Amoateng Adjepong 4, and Constantine A. Manthous 2 * 1 Department of Internal Medicine, Texas A&M University, College Station, Texas; 2 Hospital of Central Connecticut, New Britain, Connecticut; 3 Department of Internal Medicine, University of Connecticut, Farmington, Connecticut; and 4 Department of Internal Medicine, Yale University, New Haven, Connecticut Abstract Rationale: High doses of sedating drugs are often used to manage critically ill patients with alcohol withdrawal syndrome. Objectives: To describe outcomes and risks for pneumonia and endotracheal intubation in patients with alcohol withdrawal syndrome treated with high-dose intravenous sedatives and deferred endotracheal intubation. Methods: Observational cohort study of consecutive patients treated in the intensive care unit (ICU) of a university-affiliated, community hospital for alcohol withdrawal syndrome, where patients were not routinely intubated to receive high-dose or continuously infused sedating medications. Measurements and Main Results: We studied 188 patients hospitalized with alcohol withdrawal syndrome from 2008 through 2012 at one medical center. The mean age (SD) of the subjects was 50.8 6 9.0 years and their mean ICU admission APACHE (Acute Physiology and Chronic Health Evaluation) II score was 6.2 6 3.4. Thirty subjects (16%) developed pneumonia, and 38 (20.2%) required intubation. All of the 188 patients received lorazepam (median total dose, 42.5 mg), and 170 of 188 received midazolam, all but 2 by continuous intravenous infusion (median total dose, 527 mg; all administered in ICU); 19 received propofol (median total dose, 6,000 mg); and 19 received dexmedetomidine (median total dose, 1,075 mg). Intubated patients received substantially more benzodiazepine (median total dose, 761 mg of lorazepam equivalent vs. 229 mg for subjects in the nonintubated group; P, 0.0001). Endotracheal intubation was associated with pneumonia and higher acuity of illness (APACHE II score,.10). Intubated patients had a longer duration of hospital stay (median, 15 d vs. 6 d; P < 0.0001). One patient did not survive hospitalization. Conclusions: In this single-center, observational study, where endotracheal intubation was deferred until aspiration or cardiopulmonary decompensation, treatment of alcohol withdrawal syndrome with high-dose, continuously infused sedating medications was not associated with excess morbidity or mortality. Keywords: alcohol withdrawal syndrome; delirium tremens; endotracheal intubation; sedation (Received in original form July 19, 2015; accepted in final form December 9, 2015 ) *Present affiliation: Lawrence and Memorial Hospital, New London, CT. Author Contributions: R.S., B.M., C.A.M.: study design; R.S., R.P., B.M., C.A.M.: data acquisition; C.L., Y.A.A., C.A.M.; data analysis; all authors: manuscript preparation and approval. Correspondence and requests for reprints should be addressed to Constantine A. Manthous, M.D., 6 Hemingway Road, Niantic, CT 06357. E-mail: constantinemanthous@gmail.com Ann Am Thorac Soc Vol 13, No 2, pp 248 252, Feb 2016 Copyright 2016 by the American Thoracic Society DOI: 10.1513/AnnalsATS.201507-448BC Internet address: www.atsjournals.org Alcohol withdrawal syndrome is a common cause of hospital admission and use of intensive care (1). Rigorous prospective randomized studies of optimal management of critically ill patients with alcohol withdrawal have not been undertaken. Nonetheless, administration of sedating drugs to mitigate the catecholamine storm and agitation that characterize this syndrome is considered the standard of care (1 6). These medications have a relatively narrow therapeutic window and, even at low doses, may cause cardiorespiratory instability. Because airway incompetence, that is, loss of airway reflexes that protect the lower airways from upper airway secretions, may predispose to oropharyngeal aspiration, pneumonia or lung injury may complicate both under- and oversedation of patients. When to secure the airways of these patients is an unanswered clinical question, and few studies lend insight on the epidemiology of intensive care unit 248 AnnalsATS Volume 13 Number 2 February 2016

(ICU)-requiring alcohol withdrawal syndrome. Here we report clinical outcomes of patients treated at a single universityaffiliated medical center where the approach was to monitor ICU patients with alcohol withdrawal by cardiac monitoring, oximetry, and capnography, and to defer endotracheal intubation until clinically apparent aspiration or cardiopulmonary decompensation. A secondary goal was to explore relationships between sedative medication doses and pneumonia, endotracheal intubation, and prolonged hospitalization. Methods An administrative database was maintained at the Hospital of Central Connecticut (New Britain, CT) to identify all inpatients who generated ICU bed charges. Data for patients admitted to the ICU between 2008 and 2012 with a diagnosis of alcohol withdrawal syndrome (based on ICD9 codes) were extracted from the database. Demographic information (age, sex, and APACHE [Acute Physiology and Chronic Health Evaluation] score), physiological information (heart rate, blood pressure, gas exchange, and ventilator parameters when present), cumulative doses of administered sedating drugs, and administrative/ outcomes data (diagnosis of pneumonia more than 48 h after admission, endotracheal intubation, hospital length of stay, and mortality) were captured electronically and by review of all medical records for the hospital admission. The hospital investigational review board approved this project and granted an exemption from further review in accordance with U.S. federal regulation HHS46.101. Patient Care Protocols The protocol on hospital wards, during the study period 2008 2012, was to administer to patients undergoing alcohol withdrawal continuous infusions of lowdose lorazepam (up to 1.2 mg/h) with intermittent boluses of 1 2 mg of lorazepam titrated to a previously standardized clinical score (Clinical Institute Withdrawal Assessment for Alcohol score, 6). Patients who were agitated or who became critically ill despite this approach were assessed and transferred to the ICU by resident physicians who served on the rapid response team. The alcohol withdrawal syndrome practice in the hospital s ICUs was to administer parenteral sedating drugs (most frequently continuously infused midazolam up to 20 mg/h, but also haloperidol, propofol, and dexmedetomidine). There was no expectation that patients receiving continuously infused sedating drugs would be intubated. Intubation was performed only for patients whose continuous pulse oximetry or nasal capnography demonstrated worsening gas exchange suggestive of aspiration or oversedation, or for clinically apparent macroaspiration. No explicit criteria mandated intubation, so clinicians determined ad hoc the degree of gas exchange failure or apparent aspiration that warranted intubation. Although attending physicians were present most days between 8 A.M. and 5 P.M., bedside decisions were made largely by ICU residents and advanced practice registered nurses in concert with bedside nurses, occasionally but not always with the input of a fellow or attending physician. Recording of Sedating Drug Total and Average Daily Doses For purposes of this analysis, the total (hospital) benzodiazepine dose was calculated using 2 mg of midazolam as Table 1. Characteristics of study cohort Variable equivalent to 1 mg of lorazepam (7). Thus the total benzodiazepine dose, in milligrams of lorazepam equivalent, was recorded as the sum of the total lorazepam dose plus half the total midazolam dose (total mg of lorazepam dose equivalent = total mg of lorazepam 1 0.5 3 total mg of midazolam) (7). The exact daily dose of sedating drugs for each patient was not available in the database. To estimate average daily benzodiazepine use, the total benzodiazepine dose was divided by the length of stay. A maximum length of stay of 20 days was used for this estimation. (Thus, for the 12 patients with length of stay of 21 d or more, the average daily benzodiazepine = total benzodiazepine dose/20, presuming that a majority of sedative was likely administered in the first 20 d.) Statistical Analysis Features and outcomes of patients who were intubated were compared with those for patients who were not intubated by nonpaired Student t test, analysis of variance, and Fisher exact test where appropriate. Logistic regression analyses, which included plausible physiological effectors and variables associated with outcomes in univariate analyses, were conducted to adjust for confounding. Multivariate analyses were conducted with Findings Total no. 188 Age, yr Mean 50.8 6 9.0 Median (25th, 75th percentiles) 51 (46, 56) Sex Men 174 (92.6%) Women 14 (7.4%) Initial admit location Ward 143 (76.1%) ICU 45 (23.9%) Acuity of illness (APACHE II) Mean 6.2 6 3.4 Median (25th, 75th percentile) 6 (3.5, 8.5) APACHE II > 10 32 (17.3%) Seizures 24 (12.8%) Pneumonias 30 (16.0%) Length of hospital stay, d Mean 9.6 6 11.7 Median (25th, 75th percentiles) 7 (5, 10) Maximum LOS 133 LOS > 21 d 12 (6.4%) Definition of abbreviations: APACHE = Acute Physiology and Chronic Health Evaluation; ICU = intensive care unit; LOS = length of stay. Stewart, Perez, Musial, et al.: ICU Alcohol Withdrawal Syndrome 249

Table 2. Comparison of outcomes for patients who were or were not intubated Variable Intubated Nonintubated P Value Total, no. (%) 38 (100%) 150 (100%) Age, yr Median (25th, 75th percentiles) 52 (45, 57) 50.5 (46.0, 56.0) 0.6 Mean 51.4 6 10.6 50.6 6 8.6 0.6 Sex, no. (%) Men 35 (92.1%) 139 (92.7%) 0.6 Women 3 (7.9%) 11 (7.3%) Admit location, no. (%) ICU 13 (34.2%) 32 (21.3%) 0.08 Ward 25 (65.8%) 118 (78.7%) APACHE II score Mean 7.4 6 4.6 5.9 6 3.0 0.01 >10, no. (%) 13 (35.1%) 19 (12.8%) 0.003 Medical history, no. (%) Seizure history 14 (36.8%) 60 (40.0%) 0.4 DT history 6 (15.8%) 30 (20.0%) 0.4 Cirrhosis 4 (10.5%) 4 (2.7%) 0.05 CHF 5 (13.2%) 2 (1.3%) 0.004 CAD 2 (5.3%) 9 (6.0%) 0.6 Diabetes mellitus 1 (2.6%) 9 (6.0%) 0.4 Medications used, median (25th, 75th percentiles) Total midazolam dose, mg 1,435 (440, 2,860) 331 (104, 834),0.0001 Total lorazepam dose, mg 43.2 (21.0, 84.5) 42.5 (27.5, 69.5) 0.9 Dexmedetomidine (n = 19), mg 1,448 (400, 3,300) 686 (600, 1,075) 0.2 Propofol (n = 19), mg 6,000 (2,000, 12,000) N/A Haloperidol (n = 120), mg 15 (10, 32) 10 (5, 20) 0.09 Total benzodiazepines used, mg of lorazepam equivalents Median (25th, 75th percentiles) 761 (310, 1,444) 229 (103, 485),0.0001 Average daily benzodiazepines used, mg of lorazepam equivalent Median (25th, 75th percentiles) 64.9 (28.6, 101.5) 41.7 (19.8, 62.2) 0.01 Outcome, no. (%) Death 1 (2.6%) 0 (0%) N/A Pneumonia 21 (55.3%) 9 (6.0%),0.0001 Seizures 8 (21.1%) 16 (10.7%) 0.08 LOS, d Median (25th, 75th percentiles) 14.7 (9.8, 18.7) 6.0 (4.6, 8.7),0.0001 Mean 18.5 6 20.7 7.4 6 6.8,0.0001 LOS > 21 d, no. (%) 8 (21.1%) 4 (2.7%) 0.003 Definition of abbreviations: APACHE = Acute Physiology and Chronic Health Evaluation; CAD = coronary artery disease; CHF = congestive heart failure; DT = delirium tremens; ICU = intensive care unit; LOS = length of stay; N/A = not applicable. pneumonia, endotracheal intubation, and prolonged length of stay (.8 d)as dependent variables/outcomes of interest. Results We studied 188 episodes of alcohol withdrawal syndrome, of mostly (174, 92.6%) male patients, with a mean age (SD) of 50.8 6 9.0 years and mean ICU admission APACHE II score of 6.2 6 3.4 (Table 1). One hundred and forty-three patients (76.1%) were initially admitted to hospital wards, and the remaining patients were admitted directly to the ICU. Twenty-four patients had at least one seizure. Thirty patients (16%) developed pneumonia more than 48 hours after hospital admission. Overall, 38 patients (20.2%) required intubation at some point during their hospital stay. All of the 188 patients received lorazepam (median total dose, 42.5 mg); 170 of 188 received midazolam, all but 2 by continuous infusion (median total dose, 527 mg); 19 received propofol (median total dose, 6,000 mg); and 19 received dexmedetomidine (median total dose, 1,075 mg). All patients receiving propofol and 12 of the 19 patients receiving dexmedetomidine were intubated. Haloperidol was used in 23 (19.2%) of intubated and 97 (80.8%) of nonintubated patients. The median duration of hospital stay was 7 days (interquartile range, 5 10 d; mean, 9.5 6 11.7 d). Twelve patients (6.4%) were hospitalized for 21 days or more, with a maximum length of stay of 133 days. One patient did not survive hospital admission. Table 2 contrasts the characteristics of the 38 patients who underwent endotracheal intubation with the 150 patients who were not intubated. There was no statistically significant difference in age between intubated and nonintubated patients. Intubated patients had higher acuity of illness at the time of admission, with 35.1% having an APACHE II score of at least 10, compared with only 12.8% in nonintubated patients (P = 0.003). Of intubated patients, 34.2% were admitted directly to the ICU compared with 21.3% of nonintubated patients (P = 0.08). Intubated 250 AnnalsATS Volume 13 Number 2 February 2016

Table 3. Multivariate logistic regression assessing factors associated with endotracheal intubation Variable Adjusted Odds Ratio 95% CI P Value Age > 60 yr (yes/no) 2.16 0.56, 8.4 0.3 APACHE II > 10 (yes/no) 5.26 1.79, 15.5 0.003 Average daily benzodiazepines > 50 mg 2.11 0.76, 5.86 0.15 Pneumonia (yes/no) 23.54 7.97, 69.46,0.0001 Seizures (yes/no) 3.36 0.97, 11.59 0.055 Sex (male/female) 0.96 0.19, 4.89 0.96 Definition of abbreviations: APACHE = Acute Physiology and Chronic Health Evaluation; CI = confidence interval. patients received substantially more benzodiazepine (median total benzodiazepine dose, 761 mg [lorazepam equivalent] vs. 229 mg in the nonintubated group; P, 0.0001). Adjusting for length of stay, the median daily benzodiazepine dose was 64.9 mg (lorazepam equivalent) for the intubated patients compared with 41.7 mg for nonintubated patients (P =0.01). In multivariate analyses, endotracheal intubation was associated with pneumonia and acuity of illness (APACHE II score,.10) (see Table 3). Higher doses (.50 mg/d) of benzodiazepine were not associated with a greater rate of endotracheal intubation after adjustment for acuity of illness, pneumonia, seizures, and sedative drug dose (P = 0.15; Table 3). Intubated patients had a longer duration of stay compared with nonintubated patients (median, 15 vs.6 d; P < 0.0001). Only 2.7% of nonintubated patients had a length of stay of 21 or more days compared with 21.1% of the intubated patients (P < 0.0001). Discussion This was retrospective cohort study of patients treated for alcohol withdrawal with high-dose, continuously infused sedating drugs in an ICU where endotracheal intubation was performed only for clinically manifest aspiration or cardiopulmonary decompensation. The observed incidence of pneumonia was 16% (estimated incidence of 16 cases per 1,000 d of hospitalization); 20.2% of the patients were intubated; and the mortality rate was 0.5% (1 of 188). To our knowledge, this is the largest and, arguably, first study to report outcomes of patients treated with high-dose intravenous sedatives essentially conscious sedation with systematically deferred intubation. Alcohol is a leading cause of morbidity and mortality worldwide. The aggregate cost to society is estimated at up to 1% of gross domestic product (8). Acute alcohol-related disease is a relatively common cause of hospitalization and ICU admission (1, 9). Although the role of benzodiazepines in treating alcohol withdrawal was well established by the late 1960s (10), there remains a relative paucity of high-quality research to define optimal therapy for critically ill patients with this syndrome. Awissi and colleagues performed a comprehensive review of studies of ICU management of alcohol withdrawal (1). Many of these studies had small sample sizes, disparate entry criteria and definitions,anduseda panoply of management regimens. Accordingly, optimal evidence-based management cannot be surmised from the literature. One widely cited study included only 36 critically ill patients with alcohol withdrawal treated before versus after protocol-guided, symptom-titrated benzodiazepine administration. The total doses of medication used (lorazepam equivalents, 1,677 vs. 1,044 mg, before vs. after the protocol), rates of endotracheal intubation (5%), and mortality (1 of 36) were similar to the results of our study. Indications for intubation, however, were not reported. A study of critically ill trauma patients who developed alcohol withdrawal syndrome during management (11) showed similarly low mortality rates (3%). Another study (12), which is a reasonable comparator for our study, required preemptive intubation if withdrawing patients required continuous infusions of sedatives. Of 54 patients studied, 26 (51%) required intubation and 28% developed pneumonia. Accordingly, the rate of pneumonia noted in our study with deferred intubation (16%) is not excessive. Without prospective randomized studies that control for the clinical variables that contribute to outcomes (pneumonia, intubation, hospital stay, and mortality), no definitive conclusions can be drawn about optimal management. We can only describe that the methods used in this study yielded mortality, and rates of pneumonia and endotracheal intubation, that are not excessive compared with previous studies (1). Our study has several notable limitations. The retrospective design did not permit us to standardize the dosing of sedating drugs, quantify the daily benzodiazepine dose, detect the exact time of onset of pneumonia (or even the accuracy of that diagnosis), or standardize criteria for intubation. Protocols and practices were not strictly monitored for adherence. Almost certainly, management decision to transfer to ICU and then if and when to intubate varied substantially, albeit within the overlying principle of deferring intubation as described previously. The study site experiences a sizable number of alcohol withdrawal cases (roughly 4% of ICU admissions) and more than half were admitted to wards before being admitted to the ICU. It is possible that these patients were undertreated with the inpatient ward protocol (i.e., lorazepam at 1.2 mg/h with 1- to 2-mg boluses every 4 h as needed), potentiating a greater number of patients who progressed to more severe illness. In addition, our study cannot shed light on whether high-dose sedation is a cause of pneumonia and endotracheal intubation or simply marks patients who experience more severe withdrawal. Because airway incompetence, observed aspiration, and hypoxemia were informal criteria for and preceded intubation, it is not surprising that pneumonia occurred more often in intubated patients. Because other hospitals may have different protocols both for inpatient ward and ICU management, generalizing our findings may be impossible. Nonetheless, our study adds to those cited previously suggesting that high doses of sedatives can be given without mandatory Stewart, Perez, Musial, et al.: ICU Alcohol Withdrawal Syndrome 251

intubation, provided patients are closely monitored. Whether this practice is safer and more effective than preemptive intubation for such patients remains an open question. In conclusion, our study ranks among the largest reported series of ICU patients treated for alcohol withdrawal to date, and describes outcomes associated with a specific approach to management using high-dose midazolam up to 20 mg/h and reserving endotracheal intubation only for patients who clinically fail this regimen. Our study complements the review (1) highlighting how little is known about optimal management of this relatively common illness. Even less is known about why patients require such huge doses of sedatives (e.g., resistance to vs. rapid metabolism of medications). Although cohort studies such as ours may suggest hypotheses and imply general safety, well-constructed, prospective, randomized, carefully controlled clinical studies are required to examine the relative effectiveness of various regimens. n Author disclosures are available with the text of this article at www.atsjournals.org. References 1 Awissi DK, Lebrun G, Coursin DB, Riker RR, Skrobik Y. Alcohol withdrawal and delirium tremens in the critically ill: a systematic review and commentary. Intensive Care Med 2013;39:16 30. 2 Amato L, Minozzi S, Vecchi S, Davoli M. Benzodiazepines for alcohol withdrawal. Cochrane Database Syst Rev [serial on the Internet]. 2010 [accessed January 2016];17:CD005063. Available from: http://summaries.cochrane.org/ 3 Daeppen JB, Gache P, Landry U, Sekera E, Schweizer V, Gloor S, Yersin B. Symptom-triggered vs fixed-schedule doses of benzodiazepine for alcohol withdrawal: a randomized treatment trial. Arch Intern Med 2002;162:1117 1121. 4 Saitz R, Mayo-Smith MF, Roberts MS, Redmond HA, Bernard DR, Calkins DR. Individualized treatment for alcohol withdrawal: a randomized double-blind controlled trial. JAMA 1994;272:519 523. 5 Mayo-Smith MF; American Society of Addiction Medicine Working Group on Pharmacological Management of Alcohol Withdrawal. Pharmacological management of alcohol withdrawal: a metaanalysis and evidence-based practice guideline. JAMA 1997;278: 144 151. 6 DeCarolis DD, Rice KL, Ho L, Willenbring ML, Cassaro S. Symptomdriven lorazepam protocol for treatment of severe alcohol withdrawal delirium in the intensive care unit. Pharmacotherapy 2007;27: 510 518. 7 Barr J, Zomorodi K, Bertaccini EJ, Shafer SL, Geller E. A double-blind, randomized comparison of i.v. lorazepam versus midazolam for sedation of ICU patients via a pharmacologic model. Anesthesiology 2001;95:286 298. 8 Rehm J, Mathers C, Popova S, Thavorncharoensap M, Teerawattananon Y, Patra J. Global burden of disease and injury and economic cost attributable to alcohol use and alcohol-use disorders. Lancet 2009;373:2223 2233. 9 Sarff M, Gold JA. Alcohol withdrawal syndromes in the intensive care unit. Crit Care Med 2010;38(9, Suppl):S494 S501. 10 Kaim SC, Klett CJ, Rothfeld B. Treatment of the acute alcohol withdrawal state: a comparison of four drugs. Am J Psychiatry 1969; 125:1640 1646. 11 Duby JJ, Berry AJ, Ghayyem P, Wilson MD, Cocanour CS. Alcohol withdrawal syndrome in critically ill patients: protocolized versus nonprotocolized management. J Trauma Acute Care Surg 2014;77: 938 943. 12 Gold JA, Rimal B, Nolan A, Nelson LS. 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. 252 AnnalsATS Volume 13 Number 2 February 2016