COMPARISON OF SEDATION FOR ALCOHOL WITHDRAWAL Crispo et al 911

Transcription

1 Comparison of Clinical Outcomes in Nonintubated Patients with Severe Alcohol Withdrawal Syndrome Treated with Continuous-Infusion Sedatives: Dexmedetomidine versus Benzodiazepines Angela L. Crispo, 1 Mitchell J. Daley, 1 Jodie L. Pepin, 2 Paul H. Harford, 3 and Carlos V.R. Brown 4 1 Department of Pharmacy Services, University Medical Center Brackenridge, Austin, Texas; 2 Department of Pharmacy Services, Seton Medical Center Williamson, Round Rock, Texas; 3 Medical Intensive Care Unit, University Medical Center Brackenridge, Austin, Texas; 4 Trauma Services University Medical Center Brackenridge, Austin, Texas STUDY OBJECTIVE To compare efficacy and safety outcomes in nonintubated patients with severe alcohol withdrawal syndrome (AWS) who required a continuous infusion of a benzodiazepine or dexmedetomidine in addition to standard medical therapy for AWS. DESIGN Retrospective cohort study. SETTING Two hospitals within the same network that used different treatment strategies for AWS. PATIENTS A total of 61 nonintubated adults who received a continuous infusion of either a benzodiazepine (BZD) (lorazepam or midazolam; 33 patients) or dexmedetomidine (DEX) (28 patients) for severe AWS between April 1, 2011, and October 31, 2012, as well as standard medical therapy for AWS. MEASUREMENTS AND MAIN RESULTS The primary outcome was a composite end point including rates of respiratory distress requiring endotracheal intubation or occurrence of alcohol withdrawal seizures. No significant differences in the composite end point were noted between the BZD and DEX groups (9.1% and 7.1%, respectively, p>0.99) or its individual components of respiratory distress (9.1% and 7.1%, respectively, p>0.99) or alcohol withdrawal seizures (0% and 3.6%, respectively, p=0.46). The DEX group received a lower median total dose of lorazepam equivalents after initiation of the study drug (median [interquartile range] 105 [ ] mg in the BZD group vs 3.5 [0 12] mg in the DEX group), but this did not translate into a reduced requirement for endotracheal intubation or decreased length of stay. DEX was associated with more adverse drug events including hypotension and bradycardia. Of concern, DEX may impair the ability to assess symptoms appropriately and administer BZDs in a symptom-triggered fashion. Although the total cost of hospitalization was similar between groups, DEX was associated with a higher study drug cost per patient. CONCLUSION DEX demonstrated a BZD-sparing effect in the treatment of AWS; however, this surrogate end point should be interpreted with caution. Although this study cannot disprove the possibility of a protective effect of DEX in preventing the requirement for endotracheal intubation in patients with AWS, an increased rate of adverse drug events and increased study drug costs were observed. If DEX is used in clinical practice, it should only be used as adjunctive therapy with BZDs that have a proven benefit in AWS. KEY WORDS dexmedetomidine, benzodiazepines, lorazepam, midazolam, alcohol withdrawal syndrome, critical care. (Pharmacotherapy 2014;34(9): ) doi: /phar.1448

2 COMPARISON OF SEDATION FOR ALCOHOL WITHDRAWAL Crispo et al 911 Alcohol is the most frequently abused drug in the world. Alcohol use disorder affected ~18 million Americans (7.3% of the population) in Up to 40% of hospitalized patients have a history of alcohol use disorder, and an estimated 18% of these patients are predicted to develop alcohol withdrawal syndrome (AWS) while hospitalized. 1, 2 The most common symptoms of AWS are autonomic instability including agitation, tremor, hypertension, and tachycardia, occurring within several hours and up to 5 days after cessation or reduction of alcohol use. 2 Of patients who experience AWS, 5 20% progress to severe symptoms such as seizures or delirium tremens that significantly increase the morbidity and mortality associated with AWS. 1, 3 The primary initial pharmacologic therapy for AWS involves c-aminobutyric acid (GABA) receptor agonism, most often achieved with the intermittent administration of benzodiazepines (BZDs), to replace the neurodepressant activity of alcohol. 4 For the treatment of alcohol withdrawal, BZDs reduced the rate of recurrent seizures by 21% compared with placebo and reduced the risk of mortality compared with neuroleptic agents (relative risk 6.6, 95% confidence interval [CI] ). 4, 5 However, patients with considerable alcohol tolerance can demonstrate cross-tolerance to BZDs, which has led to the use of high doses, oversedation, and, potentially, respiratory distress requiring mechanical ventilation. 1 In a single-center evaluation of a diazepam and phenobarbital dose-escalation strategy for the management of refractory alcohol withdrawal, 22 47% of patients required endotracheal intubation for respiratory distress. 6 In a prospective analysis of severe alcohol withdrawal, 45.8% of patients were intubated at study entry. 7 BZD doses necessary to control AWS symptoms vary greatly and have been reported to be 100-fold higher in some patients. 8, 9 When standard protocols using intermittent BZD doses are no longer effective, alternative strategies are often attempted including the initiation of a continuous infusion of a BZD or use of alternative GABA agents (e.g., phenobarbital or propofol). 4 Presented as a poster at the Society of Critical Care Medicine 43rd Critical Care Congress, San Francisco, California, January 10, *Address for correspondence: Mitchell J. Daley, Department of Pharmacy Services, University Medical Center Brackenridge, 601 E. 15th Street, Austin, TX 78701; mjdaley@seton.org. Ó 2014 Pharmacotherapy Publications, Inc. Regardless, all of these pharmacologic agents may interfere with respiratory drive, exposing a patient to a higher risk of endotracheal intubation for mechanical ventilation. Dexmedetomidine (DEX) is an attractive adjunctive therapy in severe AWS due to its ability to reduce autonomic hyperactivity without compromising respiratory drive. 10 Published case reports and retrospective cohort studies constitute small populations of patients that support a potential role for DEX in the management of AWS as adjunctive therapy with BZDs, largely based on surrogate end points such as a decreased requirement for BZDs. 2, 6, Only recently was a randomized placebo-controlled trial published that investigated the effect of various doses of DEX on BZD requirement in the treatment of AWS; however, approximately half of the study cohort required endotracheal intubation prior to receipt of the study drug. 7 Many clinicians have used DEX off label in the treatment of AWS to prevent the requirement for endotracheal intubation, yet DEX is not without risks in the treatment of AWS. 14, 16 Mainly, DEX lacks the GABA activity necessary to prevent and treat delirium tremens or seizures, requiring that it only be used as adjunct therapy. 10 Therefore, this study evaluated the efficacy and safety of DEX in patients without endotracheal intubation. In addition to standard medical therapy, we specifically evaluated patients who required a continuous infusion of DEX compared with those who required a continuous infusion of a BZD for the treatment of severe AWS. Methods Study Design, Setting, and Patient Population In this retrospective multicenter cohort study, we evaluated clinical outcomes in patients treated for severe alcohol withdrawal at two hospitals within the same network in which differing treatment strategies for AWS are used among the institutions. The lack of clear superiority of one treatment strategy over another and the lack of a network protocol for AWS enables clinicians to individualize treatment at their discretion based on the patient s specific needs and drug safety profiles. The two hospitals were a 181-bed community-based hospital that has routinely used DEX as adjunct support for the management of severe AWS and a 249-bed urban teaching hospital where BZD doses are escalated to a continuous infusion when required for severe

3 912 PHARMACOTHERAPY Volume 34, Number 9, 2014 symptoms. The Seton Healthcare Family institutional review board approved the study protocol prior to evaluation of patient data. Study patients were identified by cross-matching a list of primary and secondary International Classification of Diseases, Ninth Revision diagnosis codes for AWS with a drug utilization report that identified patients who received a continuous infusion of a BZD (midazolam or lorazepam) or DEX. Review of the medical record and discharge summary was used to validate appropriateness of study enrollment. All patients admitted between April 1, 2011, and October 31, 2012, were included if they were between 18 and 89 years of age and had received a continuous infusion of the study drug for the treatment of AWS. Patients were excluded if they were intubated before or concurrent to the initiation of continuous infusion of the study drug, had a history of seizure disorder, had incomplete medical records, or had a previous AWS admission in the previous 30 days. End Points The primary outcome was a composite end point of rate of respiratory distress requiring endotracheal intubation or occurrence of alcohol withdrawal seizures. The composite end point was intended to represent a balanced nonbiased comparison representative of the risks associated with the administration of each study drug when used in patients for AWS. Considering that respiratory distress requiring endotracheal intubation was expected to occur more frequently than seizures, we hypothesized that adjunct DEX would reduce rates of the composite end point. 5, 6 Secondary outcomes included the following: individual components of the composite end point, Richmond Agitation Sedation Scale (RASS) score as a surrogate for efficacy, time to discontinuation of the continuous infusion, total doses of BZDs and adjunctive therapies before and during administration of study drug, length of hospital stay, all-cause in-hospital mortality, associated costs (hospital and study drug) and occurrence of adverse events including bradycardia and hypotension. Definitions A BZD infusion consisted of either lorazepam or midazolam because BZD drug shortages occurred during the study period. All BZD doses, including continuous infusions and intermittent administrations, were recorded in lorazepam equivalents. One milligram of lorazepam (one lorazepam equivalent) was considered equivalent to 2 mg midazolam, 5 mg diazepam, and 10 mg 17, 18 chlordiazepoxide. Adjunctive therapy consisted of haloperidol, olanzapine, phenobarbital, or clonidine. Discontinuation of the study drug was defined as cessation of the continuous infusion for greater than 24 hours. Composite end point events were identified through several documentation sources within the electronic medical record including clinical event reporting, oxygenation details, and discharge notes. The Clinical Institute Withdrawal Assessment for Alcohol Scale, Revised (CIWA-Ar) is used in practice, but titration of medications to target a CIWA-Ar is based on physician orders, and documentation is inconsistent. Therefore, RASS was included to assess ongoing agitation and as a surrogate for efficacy. RASS score was assessed at 4, 8, 12, and 24 hours after initiation of the study drug. Bradycardia and hypotension were defined as a heart rate less than 50 beats/minute and a systolic blood pressure less than 90 mm Hg, respectively. The DEX network protocol does not allow for a bolus given the risk of hemodynamic compromise, and the maximum dose used is 1.5 lg/kg/hour. Statistical Analysis The study was designed with 80% power to detect a 20% difference in the proportion of patients who experienced the composite end point based on an estimated 30% event rate (respiratory distress leading to endotracheal intubation or seizure) in the BZD continuousinfusion group. 6 To achieve power, a total of 72 patients were needed in each group. However, all patients were included who met study enrollment within the defined time frame, which was limited based on physician prescribing patterns and medical record documentation. The primary outcome and other nominal data were analyzed by using the Fisher exact test or a v 2 test when appropriate. A univariate logistic regression analysis was used to evaluate the association of DEX and the composite end point. Ordinal data, including RASS score and Charlson Comorbidity Index, were analyzed by using a Mann-Whitney U test. Continuous data were analyzed by using a Shapiro-Wilk test to determine parametric assumptions, and a Student t test or Mann-Whitney U (rank sum) test as appropriate was applied to compare means and medians.

4 COMPARISON OF SEDATION FOR ALCOHOL WITHDRAWAL Crispo et al 913 Results A total of 97 patients were evaluated for study inclusion; of these patients, 36 were excluded due to following reasons: endotracheal intubation prior to the start of the continuous infusion (20 patients), history of non-aws seizures (4 patients), prior admission for AWS within 30 of the previous days (1 patient), and never received the study drug (11 patients). Of the remaining 61 patients, 33 patients were included in the BZD group and 28 in the DEX group. Baseline demographic and clinical characteristics were similar between the two groups, with significant differences noted in age and race/ethnicity (Table 1). Prior to administration of the study drug, the BZD received a significantly higher median (interquartile range [IQR]) cumulative dose of BZDs in lorazepam equivalents compared with the DEX (BZD 17 [10 24] mg vs DEX 8 [4 11] mg, p<0.01), and a higher proportion of patients received adjunctive olanzapine (BZD 54.5% vs DEX 10.7%, p<0.01) at higher median (IQR) cumulative doses (BZD 5 [0 10] mg vs 0 [0] mg, p<0.01). A larger proportion of patients who received DEX were managed in the intensive care unit (BZD 66.7% vs DEX 100%, p<0.01), although this was confounded by lack of an intermediate care unit at the study site where DEX was routinely used. No significant difference between groups was observed in the use of intermittent haloperidol. Median RASS scores at baseline were similar between groups (BZD 3 [1 3] vs DEX 2 [1 2], p=0.11). The median (IQR) infusion rate of lorazepam equivalents was 2.1 ( ) mg/hour with a mean SD duration of hours, whereas the DEX median (IQR) infusion rate was 0.54 ( ) lg/kg/hour with mean SD duration of hours (Table 2). The total dose Table 1. Baseline Demographic and Clinical Characteristics of the Study Patients Characteristic Benzodiazepine group (n=33) Dexmedetomidine group (n=28) p value Age, yrs Male sex 30 (90.9) 22 (78.6) 0.28 Race/ethnicity White 18 (54.5) 26 (92.9) < 0.01 Hispanic 13 (39.4) 1 (3.6) < 0.01 African-American 2 (6.1) 1 (3.6) > 0.99 Admitting diagnosis Primary alcohol withdrawal 22 (67) 15 (54) 0.3 Primary medical diagnosis 8 (24) 9 (32) 0.49 Primary trauma diagnosis 3 (9) 4 (14) 0.53 Charlson Comorbidity Index 2 (1 2) 2.5 (1.75 4) 0.43 History of cirrhosis 8 (24.2) 3 (10.7) 0.2 Reported alcoholic drinks/day 12 (6 18) 12 (6 17) 0.24 History of drug abuse 7 (21.2) 1 (3.6) 0.06 Positive urine drug screen 3 (9.1) 3 (10.7) 0.68 Prothrombin time, sec International normalized ratio Alanine aminotransferase level, U/L 53 (35 82) 49 (27 100) 0.11 Total bilirubin level, mg/dl 0.9 ( ) 1.4 ( ) 0.37 Serum creatinine level, mg/dl RASS score 3 (1 3) 2 (1 2) 0.11 Hospital location during treatment Intensive care unit 22 (66.7) 28 (100) < 0.01 Intermediate care unit 11 (33.3) 0 (0) < 0.01 Cumulative dose of benzodiazepines in lorazepam equivalents, mg Mean SD < 0.01 Median (IQR) 17 (10 24) 8 (4 11) < 0.01 Received adjunctive olanzapine 18 (54.5) 3 (10.7) < 0.01 Cumulative olanzapine dose, mg Mean SD < 0.01 Median (IQR) 5 (0 10) 0 (0) < 0.01 Received adjunctive haloperidol 11 (33.3) 11 (39.3) 0.58 Cumulative haloperidol dose, mg Mean SD Median (IQR) 0 (0 10) 0 (0 10) 0.52 RASS = Richmond Agitation Sedation Scale. Data are mean SD, no. (%) of patients, or median (interquartile range [IQR]).

5 914 PHARMACOTHERAPY Volume 34, Number 9, 2014 Table 2. Study Drug and Adjunctive Therapy Requirements Study drug or adjunctive therapy Benzodiazepine group (n=33) Dexmedetomidine group (n=28) Study drug Mean SD Duration, hrs Total dose mg lg NA Infusion rate mg/hr lg/kg/hr NA Median (IQR) Duration, hrs 55.5 ( ) 45.9 ( ) 0.43 Total dose 105 ( ) mg ( ) lg NA Infusion rate 2.1 ( ) mg/hr 0.54 ( ) lg/kg/hr NA No. (%) of patients who received intermittent benzodiazepine 13 (39.4) 19 (67.9) 0.03 Cumulative lorazepam equivalent dosing during continuous infusion Mean SD mg mg 17 < 0.01 Median (IQR) 105 ( ) 3.5 mg (0 12) < 0.01 No. (%) of patients who received adjunctive olanzapine 14 (42.4) 2 (7.1) < 0.01 Cumulative olanzapine dose, mg Mean SD < 0.01 Median (IQR) 0 (0 20) 0 (0) < 0.01 No. (%) of patients who received adjunctive haloperidol 11 (33.3) 8 (28.6) 0.69 Cumulative haloperidol dose, mg Mean SD Median (IQR) 0 (0 20) 0 (0 5) 0.27 IQR = interquartile range; NA = not applicable. p value of lorazepam equivalents administered while receiving the study drug, including intermittent administration of lorazepam with DEX, was significantly higher for patients who received a continuous infusion of a BZD (BZD 105 [ ] mg vs DEX 3.5 [0 12] mg, p<0.01). Nine patients who were treated with DEX did not receive an additional BZD dose after initiation of the study drug (BZD 100% vs DEX 68%, p<0.01). All patients who received intermittent BZDs or antipsychotics (olanzapine or haloperidol) were facilitated by a mix of single-order entry (physician facilitated) and as-needed orders (physician and nurse facilitated). No significant differences were noted between the BZD and DEX in the occurrence of the composite outcome (BZD 9.1% vs DEX 7.1%, p>0.99) or its individual components of respiratory distress requiring endotracheal intubation (BZD 9.1% vs DEX 7.1%, p>0.99) or alcohol withdrawal seizures (BZD 0% vs DEX 3.6%, p=0.46; Table 3). One patient in the DEX experienced both a seizure and respiratory distress and was counted only once within the composite end point. In the univariate logistic regression analysis, the use of DEX was not predictive of the composite event (odds ratio 0.76, 95% CI ). Median RASS scores after initiation of study drug were similar between groups, although DEX appeared to achieve a deeper sedation (BZD 0.5 [ 2 to 0] vs DEX 2 [ 4 to 1], p=0.15). All patients achieved an RASS score less than +1 within 24 hours of study drug initiation, with a similar median time to RASS Table 3. Efficacy and Safety End Points End point Benzodiazepine group (n=33) Dexmedetomidine group (n=28) Composite efficacy end point 3 (9.1) 2 (7.1) > 0.99 Respiratory distress 3 (9.1) 2 (7.1) > 0.99 Seizure 0 1 (3.6) 0.46 RASS score <+1within 24 hrs 33 (100) 28 (100) > 0.99 Change in RASS score after initiation of study 0.5 ( 2 to0) 2 ( 4 to 1) 0.15 drug Time, hrs, until RASS score <+1 4 (4 12) 4 (4 4) 0.11 Bradycardia 5 (15.2) 13 (46.4) < 0.01 Hypotension 4 (12.1) 12 (42.9) < 0.01 RASS = Richmond Agitation Sedation Scale. Data are no. (%) of patients or median (interquartile range). p value

6 COMPARISON OF SEDATION FOR ALCOHOL WITHDRAWAL Crispo et al 915 score less than +1 (BZD 4 [4 12] hrs vs DEX 4 [4 4] hrs, p=0.11). Patients who developed respiratory distress requiring endotracheal intubation had a similar duration of mechanical ventilation (BZD days vs DEX days, p=0.83). Bradycardia (BZD 15.2% vs DEX 46.4, p<0.01) and hypotension (BZD 12.1% vs DEX 42.9%, p<0.01) were significantly higher in the DEX compared with the BZD group. The portion of patients who developed hemodynamic adverse events received concomitant antihypertensives (BZD 37.5% vs DEX 10%), which were discontinued in all cases. Interventions after identification of adverse events included discontinuation of the study drug (BZD 0% vs DEX 35%, p=0.08), administration of vasopressor agents (BZD 0% vs DEX 20%, p=0.17), and fluid administration (BZD 100% vs DEX 60%, p=0.06). Total cost of hospitalization, including all relevant charges during the course of hospitalization, was similar between groups (mean SD BZD $11, vs DEX $17, , p=0.11; Table 4). DEX was associated with a higher study drug cost/patient (mean SD BZD $ vs DEX $ , p<0.01). A significantly lower proportion of patients who received DEX required olanzapine after initiation of the study drug (BZD 42.4% vs DEX 7.1%, p<0.01). There was no significant difference in length of hospital stay between groups (BZD vs DEX days, p=0.88). No patient in either group received clonidine or phenobarbital during the study period. Discussion BZDs are the mainstay of therapy for AWS due to their ability to prevent seizures, treat delirium tremens, and reduce mortality compared with neuroleptic agents. 4 Escalated doses may pose a risk for respiratory distress, and therefore off-label use of DEX has become common in nonintubated patients despite limited supporting data. 10, 11 One study described 10 cases of AWS in which DEX and BZDs blunted autonomic hyperactivity by reducing heart rate by 10.5% and mean arterial pressure by 3% in mild to moderate AWS. 8 Similarly, another study evaluated 20 patients with AWS and found a 60% decrease in BZD use and ~50% reduction of haloperidol doses with the addition of DEX. 3 More recently, a study reported the first randomized controlled trial that demonstrated adjunct DEX reduced the 24-hour lorazepam requirement by 54% compared with placebo but found no significant differences in the total doses of lorazepam received at 7 days. 7 Similar to previous publications, our investigation found that patients who received DEX required lower BZD and neuroleptic doses to manage AWS; however, this surrogate end point should be interpreted with caution to justify offlabel use of DEX given the clear benefit of BZDs in the management of severe AWS. Current guidelines recommend against using adjunctive therapies with no GABA activity alone due to their inability to prevent seizures and/or delirium tremens. 4, 19 Non-BZD alternatives, such as DEX, when used as adjunct therapy, have the capability to attenuate symptoms of withdrawal that might otherwise necessitate the use of symptom-triggered BZDs. For example, one study found that DEX improved hemodynamic parameters in patients with AWS, reducing heart rate by 17 beats/minute and mean arterial pressure by 14 mm Hg. 15 Although the BZD-sparing effects and improvement in hemodynamics observed with DEX are often interpreted as beneficial, in our study, a third of patients who received DEX did not receive concomitant BZDs following initiation of the study drug, increasing the risk of progressing AWS and seizures. One patient, who was not treated with BZDs for 6 days after the initiation of DEX, eventually developed an alcohol withdrawal seizure leading to terminal and irreversible encephalopathy. This raises major concerns for the safe use of DEX as adjunctive therapy in the setting of severe AWS, which may be unrecognized in the current literature due to limited sample sizes. If DEX is used within clinical practice, BZDs Table 4. Length of Stay, Mortality, and Cost Outcomes Outcome Benzodiazepine group (n=33) Dexmedetomidine group (n=28) p value Hospital length of stay, days No. (%) of deaths 0 1 (3.6) 0.46 Cost/hospitalization, $ 11, , Study drug cost/patient, $ < 0.01 Data are mean SD unless otherwise specified.

7 916 PHARMACOTHERAPY Volume 34, Number 9, 2014 should be routinely administered on a scheduled basis to decrease risk of alcohol withdrawal seizures. In addition, although adjunctive DEX has been shown to have a BZD-sparing effect, it is unclear whether this translates into decreased rates of respiratory distress leading to endotracheal intubation in the treatment of severe AWS. To our knowledge, our study is the first comparative study evaluating exclusively nonintubated patients at the time of study drug initiation, and we found similar rates for the requirement for endotracheal intubation between the BZD and DEX groups. However, a significant potential for a type II error exists, and the lack of a protective effect of DEX cannot be proven. Given that the overall event rate of respiratory distress requiring endotracheal intubation and alcohol withdrawal seizure was lower than expected, a total of 5842 patients would be needed to achieve adequate power to detect differences between groups based on the observed results. Patients who developed either respiratory distress leading to endotracheal intubation or alcohol withdrawal seizure were more likely to have a history of drug abuse (50% vs 10.9%, p=0.09), without significant differences in median total lorazepam equivalent doses received (53 [IQR ] mg vs (66 [IQR ] mg, p=0.82). Comparatively, in a single-center study evaluating DEX as adjunctive therapy versus usual care, 11 of 24 patients enrolled required intubation prior to initiation of the study drug, making it difficult to draw conclusions as to whether the lorazepam-sparing effects of DEX decreased rates of respiratory distress leading to endotracheal intubation. 7 However, of the patients who received BZDs alone after randomization, no additional patients developed respiratory distress despite receiving a mean total lorazepam dose of 77 mg within the first 24 hours. One study also found that the implementation of an institutional guideline that focused on escalating doses of diazepam and phenobarbital given in a symptom-triggered fashion led to higher total doses of diazepam administered compared with administering intermittent boluses of diazepam before guideline implementation (562 mg post-guideline vs 248 mg pre-guideline, p<0.01), yet reduced need for mechanical ventilation (22% post-guideline vs 47% pre-guideline, p<0.01). 6 In all cases, it is difficult to control for subjective variables that may influence the decision favoring endotracheal intubation. Escalating doses of BZDs or use of sedatives or hypnotics as a continuous infusion in nonintubated patients is an example that confounds extrapolation of the available literature. The respiratory depressant effects of BZDs are variable and influenced by individual characteristics such as diminished GABA receptor function observed in patients with chronic alcohol abuse. 20 Given that the data supporting adjunctive DEX to prevent endotracheal intubation in patients with AWS is best categorized as suppositional, off-label drug use within clinical practice should either be confined to research protocols or prospectively monitored with consideration for written 21, 22 informed consent. Our study failed to find a significant difference between the BZD and DEX groups in RASS score, which served as surrogate marker for efficacy in the absence of alcohol withdrawal objective assessments such as CIWA-Ar. Other potential efficacy markers, such as duration of treatment, were also not significantly different between the groups. DEX was associated with significantly more hypotension and bradycardia, which affected approximately half of the study cohort. Similarly, DEX had no effect on hospital length of stay compared with BZDs, similar to the study cited earlier. 7 The commonality of off-label use of DEX in patients with AWS may be due to extrapolation of the literature regarding sedation for mechanically ventilated patients Pharmacoeconomic studies have supported DEX as a primary sedative compared with BZDs due to a lower total hospital cost despite higher acquisition costs. 26 Yet the findings of these studies should be interpreted with caution when applying them in the setting of severe alcohol withdrawal due to different physiologic and therapeutic goals. In our investigation, total hospital costs were not statistically significantly different between the BZD and DEX groups, but study drug costs favored BZDs without an observed benefit for indirect costs, such as length of stay. Total intensive care unit charges appear to have had the most influence on disparity in costs between groups, considering the lack of an intermediate care unit at the institution where DEX was routinely used. Several limitations of this study are inherent to a retrospective design. Although to our knowledge this study represents the largest cohort of patients to receive DEX for AWS, the sample size is limited to exclude the probability of a type II error. Therefore, we cannot disprove the potential protective effect of DEX on preventing respiratory distress leading to endotracheal

8 COMPARISON OF SEDATION FOR ALCOHOL WITHDRAWAL Crispo et al 917 intubation. Additional enrollment was limited by the study time frame, which was defined based on physician prescribing practices and consistent documentation. Small variations seen in the baseline demographics of the study patients may be attributed to the unique patient populations of the individual hospitals. The CIWA-Ar was not consistently documented to assess efficacy, and the RASS is not a validated assessment scale for treatment of AWS, which increases the potential for bias in the use of DEX for AWS. Practice variations may have existed due to the lack of a standardized protocol within the network at the time of the investigation. Finally, data collection relied on consistency of documentation in the medical record. Conclusion DEX demonstrated a BZD-sparing effect in the treatment of AWS; however, this surrogate end point should be interpreted with caution. Although our study cannot disprove the possibility of a protective effect of DEX in preventing the requirement for endotracheal intubation in patients with AWS, an increased rate of adverse drug events and increased study drug costs were observed. Of concern, DEX may impair the ability to assess symptoms appropriately and administer BZDs, which have a proven benefit in AWS. For these reasons, off-label use of DEX in an attempt to prevent endotracheal intubation in patients with AWS should be done with caution, and DEX should only be used as adjunctive therapy with BZDs. References 1. de Wit M, Jones DG, Sessler CN, Zilberberg MD, Weaver MF. Alcohol-use disorders in the critically ill patient. Chest 2010;138: American Psychiatric Association. Substance-Related Disorders. Diagnostic and Statistical Manual of Mental Disorders, 4th ed. text revision. Arlington, VA: American Psychiatric Publishing, Inc., Rayner SG, Weinert CR, Peng H, Jepsen S, Broccard AF. Dexmedetomidine as adjunct treatment for severe alcohol withdrawal in the ICU. Ann Intensive Care 2012;2: Mayo-Smith MF, Beecher LH, Fischer TL, et al. Management of alcohol withdrawal delirium. An evidence-based practice guideline. Arch Intern Med 2004;164: D Onofrio G, Rathlev NK, Ulrich AS, Fish SS, Freeland ES. Lorazepam for the prevention of recurrent seizures related to alcohol. N Engl J Med 1999;340: 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: Mueller SW, Preslaski CR, Kiser TH, et al. A randomized, double-blind, placebo-controlled dose range study of dexmedetomidine as adjunctive therapy for alcohol withdrawal. Crit Care Med 2014;42: DeMuro JP, Botros DG, Wirkowski E, Hanna AF. Use of dexmedetomidine for the treatment of alcohol withdrawal syndrome in critically ill patients: a retrospective case series. J Anesth 2012;26: Spies CD, Rommelspacher H. Alcohol withdrawal in the surgical patient: prevention and treatment. Anesth Analg 1999;88: Muzyk AJ, Fowler JA, Norwood DK, Chilipko A. Role of alpha2-agonists in the treatment of acute alcohol withdrawal. Ann Pharmacother 2011;45: Bhana N, Goa KL, McClellan KJ. Dexmedetomidine. Drugs 2000;59: Yu SB. Dexmedetomidine sedation in ICU. Korean J Anesthesiol 2012;62: Iirola T, Aantaa R, Laitio R, et al. Pharmacokinetics of prolonged infusion of high-dose dexmedetomidine in critically ill patients. Crit Care 2011;15:R MacLaren R, Krisl JC, Cochrane RE, Mueller SW. A casebased approach to the practical application of dexmedetomidine in critically ill adults. Pharmacotherapy 2013;33: Frazee EN, Personett HA, Leung JG, et al. Influence of dexmedetomidine therapy on the management of severe alcohol withdrawal syndrome in critically ill patients. J Crit Care 2014;29: Muzyk AJ, Kerns S, Brudney S, Gagliardi JP. Dexmedetomidine for the treatment of alcohol withdrawal syndrome: rationale and current status of research. CNS Drugs 2013;27: 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: Barr J, Zomorodi K, Bertaccini EJ, Shafer SL, Geller E. A double-blind, randomized comparison of IV lorazepam versus midazolam for sedation of ICU patients via a pharmacologic model. Anesthesiology 2001;95: Chouinard G. Issues in the clinical use of benzodiazepines: potency, withdrawal, and rebound. J Clin Psychiatry 2004;65: Sarff M, Gold JA. Alcohol withdrawal syndromes in the intensive care unit. Crit Care Med 2010;38:S Largent EA, Miller FG, Pearson SD. Going off-label without venturing off-course: evidence and ethical off-label prescribing. Arch Intern Med 2009;169: Freedman B. Equipoise and the ethics of clinical research. N Engl J Med 1987;317: Riker RR, Shehabi Y, Bokesch PM, et al. Dexmedetomidine vs. midazolam for sedation of critically ill patients. JAMA 2009;301: Pandharipande PP, Pun BT, Herr DL, et al. Effect of sedation with dexmedetomidine vs lorazepam on acute brain dysfunction in mechanically ventilated patients. The MENDS randomized controlled trial. JAMA 2007;298: Barr J, Fraser GL, Runtillo K, et al. Clinical practice guidelines for the management of pain, agitation and delirium in adult patients in the intensive care unit. Crit Care Med 2013;41: Dasta JF, Kane-Gill SL, Pencina M, et al. A cost-minimization analysis of dexmedetomidine compared with midazolam for long-term sedation in the intensive care unit. Crit Care Med 2010;38: