Introduction. Research Report

Transcription

1 672036AOPXXX / Annals of PharmacotherapySen et al research-article2016 Research Report Evaluation of a Symptom-Triggered Benzodiazepine Protocol Utilizing SAS and CIWA-Ar Scoring for the Treatment of Alcohol Withdrawal Syndrome in the Critically Ill Annals of Pharmacotherapy 1 10 The Author(s) 2016 Reprints and permissions: sagepub.com/journalspermissions.nav DOI: / aop.sagepub.com Soumitra Sen, MD 1,2, Philip Grgurich, PharmD 1,3, Amanda Tulolo, PharmD 3,4, Andrew Smith-Freedman, BA 5, Yuxiu Lei, PhD 1, Anthony Gray, MD 1,2, and James Dargin, MD 1,2 Abstract Background: There are limited data on the efficacy of symptom-triggered therapy for alcohol withdrawal syndrome (AWS) in the intensive care unit (ICU). Objective: To evaluate the safety and efficacy of a symptom-triggered benzodiazepine protocol utilizing Riker Sedation Agitation Scale (SAS) scoring for the treatment of AWS in the ICU. Methods: We performed a before-and-after study in a medical ICU. A protocol incorporating SAS scoring and symptom-triggered benzodiazepine dosing was implemented in place of a protocol that utilized the Clinical Institute Withdrawal Assessment for Alcohol (CIWA-Ar) scale and fixed benzodiazepine dosing. Results: We enrolled 167 patients (135 in the preintervention and 32 in the postintervention group). The median duration of AWS was shorter in the postintervention (5, interquartile range [IQR] = 4-8 days) than in the preintervention group (8, IQR = 5-12 days; P < 0.01). Need for mechanical ventilation (31% vs 57%, P = 0.01), median ICU length of stay (LOS; 4, IQR = 2-7, vs 7, IQR = 4-11 days, P = 0.02), and hospital LOS (9, IQR = 6-13, vs 13, IQR = 9-18 days; P = 0.01) were less in the postintervention group. There was a reduction in mean total benzodiazepine exposure (74 ± 159 vs 450 ± 701 mg lorazepam; P < 0.01) in the postintervention group. Conclusion: A symptom-triggered benzodiazepine protocol utilizing SAS in critically ill patients is associated with a reduction in the duration of AWS treatment, benzodiazepine exposure, need for mechanical ventilation, and ICU and hospital LOS compared with a CIWA-Ar based protocol using fixed benzodiazepine dosing. Keywords critical care, benzodiazepines, alcohol abuse, alcohol withdrawal syndrome, alcohol withdrawal syndrome management Introduction Chronic alcoholism can complicate up to 40% of hospital admissions, and excess alcohol consumption contributes to 20% of all intensive care unit (ICU) admissions. 1 Furthermore, critically ill patients with alcohol withdrawal syndrome (AWS) have increased complications, including longer ICU and hospital length of stay (LOS); higher rates of pneumonia, urinary tract infections, sepsis, tracheostomy, and percutaneous endoscopic gastrostomy; and higher cost of care. 2 Historically, benzodiazepines have been the cornerstone of therapy for AWS. Treatment of AWS with benzodiazepines is associated with a significant reduction in seizures and delirium tremens as compared with other therapies. 3,4 In patients with severe AWS, fixed dosing of benzodiazepines, often with the use of benzodiazepine infusions, is common practice in many ICUs despite potential adverse effects, including excessive sedation, respiratory depression, delirium, and complications resulting from accumulation of propylene glycol. 5-9 In non-icu settings, symptom-triggered benzodiazepine therapy is associated with reduced benzodiazepine exposure, shorter hospital LOS, and shorter duration of therapy when compared with fixed dosing Few studies have 1 Lahey Hospital and Medical Center, Burlington, MA, USA 2 Tufts University School of Medicine, Boston, MA, USA 3 MCPHS University, Boston, MA, USA 4 Singapore General Hospital, Singapore, Singapore 5 Tufts University, Medford, MA, USA Corresponding Author: Soumitra Sen, Lahey Hospital and Medical Center Department of Pulmonary and Critical Care Medicine, Tufts University School of Medicine, 41 Mall Road, Burlington, MA , USA. soumitrasen.1983@gmail.com

2 2 Annals of Pharmacotherapy evaluated symptom-driven protocols for the treatment of severe AWS in critically ill patients, and these studies utilized benzodiazepine infusions or adjunctive barbiturate therapy. 7,13-15 Unfortunately, no guidelines exist for the management of AWS in the ICU, resulting in significant variation in practice patterns with regard to benzodiazepine dosing. 16 Further complicating matters, commonly used scales for assessing the severity of AWS can be difficult to apply in critical care settings. The most widely utilized assessment tool to administer symptom-triggered benzodiazepines in AWS is the Clinical Institute Withdrawal Assessment for Alcohol (CIWA-Ar) scale (Appendix A) The CIWA-Ar scale requires patients to have a clear history of recent alcohol use and intact ability to communicate effectively in response to questions associated with the assessment tool. Although widely used to assess AWS symptoms in critically ill patients, few studies have evaluated the use of the CIWA-Ar scale in AWS patients requiring intensive care. 17 Also, there are no studies addressing the challenges of using CIWA-Ar in mechanically ventilated patients. Our clinical experience has been that the CIWA-Ar is difficult to perform in many critically ill patients because of communication impairments relating to severity of illness, severe agitation, or decreased level of consciousness. For instance, it may be difficult or impossible to assess parts of the CIWA-Ar scale that require intact verbal communication, including the presence of tactile or auditory disturbances, severity of headache, and orientation. In fact, the CIWA-Ar scale cannot be used in up to 45% of hospitalized patients because patients often cannot communicate clearly. 18 These limitations may lead to inaccurate CIWA-Ar scoring, resulting in either overtreatment or undertreatment of AWS in ICU populations. In comparison, sedation scales are routinely used in the ICU to titrate sedative agents and recommended by the Society of Critical Care Medicine (SCCM) practice guidelines. 19,20 These scales are easy to utilize, familiar to ICU staff, and do not require patients to communicate effectively. One of the 2 sedation scales recommended by the SCCM practice guidelines is the Riker Sedation Agitation Scale (SAS) (Appendix B). The SAS ranges from 1 (unarousable) to 7 (dangerous agitation), with the target goal being 3 to 4 for most patients in the ICU. The SAS has been utilized to guide therapy in some studies of AWS in critically ill patients. 21,22 As a quality improvement initiative, we implemented a protocol for the treatment of AWS in the medical ICUs at our institution that utilized (1) more symptom-triggered benzodiazepine dosing in place of a protocol that included fixed dosing with benzodiazepine infusions and (2) SAS scoring for assessment of symptom severity when the CIWA-Ar score could not be accurately performed. We hypothesized that implementation of this protocol would result in shorter duration of AWS treatment and AWSrelated complications. Methods We performed a single-center, before-and-after study in 2 medical ICUs at a 350-bed tertiary care center from August 2011 to November The study was approved by the institutional review board at our hospital, and the need for written informed consent was waived. Patients in the preintervention group were treated according to a standardized order set using a combination of symptom-triggered and fixed dosing of benzodiazepines according to the CIWA-Ar scale (Figure 1). Patients could not receive greater than 10 mg of lorazepam per hour; if this limit was reached, a lorazepam infusion was initiated and titrated from 1 to 5 mg/h to target a CIWA-Ar score of less than 10 and SAS score of 3 to 4. In the postintervention group, patients were treated according to a symptom-triggered benzodiazepine dosing order set based on either the CIWA-Ar scale or SAS score (Figure 1). If the treating nurse deemed that the patient was unable to be assessed by the CIWA-Ar scale because of severe agitation or inability to communicate, the SAS score was used to assess the severity of alcohol withdrawal. In this group, patients could receive lower fixed benzodiazepine dosing with 0.5 to 2 mg of oral lorazepam every 6 hours for 24 hours if the CIWA-Ar score was less than 8 or the SAS score was 4. The symptom-triggered intravenous benzodiazepine dosing was different in the postintervention group because patients received 2 mg of intravenous lorazepam every 10 minutes for a CIWA-Ar score of 8 to 15 or SAS of 5, or 4-mg intravenous lorazepam every 10 minutes for CIWA-Ar scores greater than 15 or SAS greater than 5. There was no hourly limit on the amount of benzodiazepines that could be administered in this group. In addition, lorazepam infusions were not included in this order set but could be implemented separately at the discretion of the physician (Figure 1). In both groups, a midazolam or propofol infusion could be used in intubated patients at the discretion of the treating physician, but they were not included as part of either order set. Dexmedetomidine and other adjunctive medications such as haloperidol, atypical antipsychotics, and clonidine could be used in both intubated and nonintubated patients in either group at the discretion of the treating physician. We screened the electronic medical record using an automated reporting system for patients treated for alcohol withdrawal in 2 medical ICUs by searching for use of our hospital s alcohol withdrawal order sets. We then reviewed individual medical records to confirm documentation of AWS in 2 consecutive ICU daily progress notes and concomitant administration of benzodiazepines. We included all adults older than 18 years with documented treatment of AWS for a minimum of 48 hours. We excluded patients with head injury, severe baseline psychiatric illness (suidical ideation, bipolar disorder, or schizophrenia), non-alcohol-related

3 Sen et al 3 Figure 1. Comparison of preintervention and postintervention treatment protocols. Abbreviations: CIWA, Clinical Institute Withdrawal Assessment for Alcohol; IV, intravenous; SAS, Riker Sedation Agitation Scale. seizure disorder, barbiturate use during AWS treatment, and pregnancy. Preintervention data were collected retrospectively for consecutive patients treated during the 2-year period from August 2011 to August Postintervention data were collected for consecutive patients treated over the first 6 months after implementation of the new protocol, which was implemented in April Data collected included the following: demographics, comorbidities, Acute Physiology and Chronic Health Evaluation (APACHE) II score, duration of AWS treatment, duration of mechanical ventilation, hospital and ICU LOS, mortality, adjunctive medication use, and complications of AWS. The primary outcome was duration of AWS treatment. Treatment duration was measured from the first day a patient met inclusion criteria to the date of the last benzodiazepine administration or physician documentation of resolution of AWS (whichever came first). Secondary outcomes included the following: need for mechanical ventilation, duration of mechanical ventilation, use of neurological imaging (CT or MRI), seizure during AWS treatment, pneumonia, ICU LOS, hospital LOS, mortality, use of adjunctive medications for AWS treatment (quetiapine, haloperidol, propofol, dexmedetomidine, gabapentin, and clonidine), and overall benzodiazepine exposure. Benzodiazepine doses were measured in lorazepam equivalents (1 mg lorazepam = 10 mg diazepam; 1 mg lorazepam = 3 mg midazolam). 7,23 All statistical comparisons between preintervention and postintervention groups were unpaired, and all tests of significance were 2-tailed. Continuous variables were compared using the Student s t-test or Wilcoxon rank-sum test. The Fisher exact and χ 2 tests were used to compare categorical variables. For all analyses, a P value 0.05 was considered statistically significant. We performed univariate analysis to examine predictors of the duration of AWS. The influential variables with a P value 0.1 were included in a multivariable regression model. The statistical analysis was generated using Statistical Analysis Software, Version 9.2 for Windows (copyright ; SAS Institute Inc, Cary, NC).

4 4 Annals of Pharmacotherapy Figure 2. Flow diagram of patient selection: preintervention and postintervention groups. Abbreviations: AWS, alcohol withdrawal syndrome; ICD 9, International Classification of Diseases, Ninth Revision; ICU, intensive care unit. Results We included 167 patients treated for AWS in a medical ICU over the course of the study period (Figure 2). The mean age (53.9 ± 12.5 vs 53.4 ± 12.8 years, P = 0.83) and gender (76% vs 75% male, P = 1.0) were similar between the preintervention group and the postintervention group, respectively (Table 1). The median blood alcohol level was not statistically different between the preintervention and postintervention groups (50.0 mg/dl vs 97 mg/dl, P = 0.10). The preintervention group had a lower percentage of individuals with cirrhosis (14.8% vs 37.5%, P < 0.01) and congestive heart failure (2.2% vs 12.5%, P = 0.03), lower initial CIWA-Ar score (15.4 vs 23, P < 0.01), and higher APACHE II score (14.7 vs 11.3, P < 0.01) when compared with the postintervention group. In the postintervention group, SAS scoring was utilized to assess AWS in 50% (16/32) of participants. Primary and secondary outcomes are shown in Table 2. The median duration of AWS was significantly shorter in the postintervention group (5.0, IQR = days) when compared with the preintervention group (8.0, IQR = days; P < 0.01). Fewer individuals in the postintervention group required mechanical ventilation when compared with the preintervention group (31% vs 57%, P = 0.01). In the preintervention group, 3/77 (4%) individuals were already intubated prior to receiving treatment of AWS compared to 5/10 (50%) in the postintervention group. The median ICU LOS (4, IQR = 2-7, vs 7, IQR = 4-11 days; P = 0.02) and median hospital LOS (9, IQR = 6-13, vs 13, IQR = 9-18 days; P = 0.01) were shorter in the postintervention group compared with the preintervention group. ICU mortality was similar between groups (3.1% vs 2.2%, P = 0.56). Medications administrated for AWS for both treatment groups are shown in Table 3. There was a significant reduction in the percentage of patients who were treated with continuous benzodiazepine infusions in the postintervention group (9.4%) compared with the preintervention group (59%; P < 0.01). Three subjects in the postintervention group received benzodiazepine infusions under the specific direction of the medical ICU team for 2, 2, and 5 days respectively. The median duration of benzodiazepine infusion in the preintervention group was 5 days (IQR 3 7) compared to 2 days (IQR 2-2). In addition, the mean total benzodiazepine exposure was significantly reduced in the postintervention group compared with the preintervention group (74 ± 159 vs 450 ± 701 mg; P < 0.01). However, there was a significant increase in the use of dexmedetomidine in the postintervention group (37.5%) compared with the preintervention group (8.1%; P < 0.01). There was a significant reduction in the use of propofol in the postintervention group (15.6%) compared with the preintervention group (40.7%; P < 0.01). There was no significant difference in the use of antipsychotics (P = 0.40), antiepileptics (P = 0.16), or antihypertensives (P = 0.15) between the groups. A univariate analysis was performed to evaluate for variables associated with the primary outcome of duration of AWS treatment (Table 4). Univariate analysis demonstrated that treatment with the symptom-based protocol incorporating SAS scoring (P < 0.01) and a history of cirrhosis (P = 0.03) were associated with shorter durations of alcohol withdrawal treatment, whereas male gender (P = 0.04) and the use of propofol were associated with longer durations of AWS treatment. Multivariate analysis was performed using the influential variables, including the following: treatment with the symptom-based protocol, male gender, propofol, and cirrhosis. Multivariate analysis demonstrated that treatment with the symptom-based protocol incorporating SAS scoring and cirrhosis were associated with shorter duration of AWS therapy ( 2.0 days, P = 0.04, and 2.2 days, P = 0.02, respectively). Male gender and propofol use predicted greater duration of AWS treatment (+1.7 days, P = 0.04, and +3.0 days, P < 0.01, respectively).

5 Sen et al 5 Table 1. Baseline Patient Characteristics. Preintervention (n = 135) Postintervention (n = 32) P Value Age (years) a 53.9 ± ± Male (%) Depression b 33 (24.4) 7 (21.9) 0.82 Anxiety b 25 (18.5) 7(21.9) 0.63 Cirrhosis b 20 (14.8) 12 (37.5) 0.01 COPD b 18 (13.3) 3 (9.4) 0.77 CAD b 9 (6.7) 1 (3.1) 0.70 CHF b 3 (2.2) 4 (12.5) 0.03 Malignancy b 7 (5.2) 2 (5.6) 0.68 AST (IU/L) c 87.0 [ ] 67.0 [ ] 0.50 ALT (IU/L) c 47.0 [ ] 38.0 [ ] 0.13 Total bilirubin (mg/dl) c 1.5 [ ] 2.7 [ ] 0.05 INR c 1.0 [ ] 1.3 [ ] 0.09 Creatinine (mg/dl) c 0.9 [ ] 0.8 [ ] 0.09 Blood alcohol level (mg/dl) c 50.0 [ ] 97 [0-236] 0.10 (n = 81) (n = 18) APACHE II score a 14.7 ± ± 4.7 <0.01 Maximum initial CIWA score a,d 15.4 ± 7.8 (n = 100) 23.0 ± 10.0 (n = 18) <0.01 Maximum initial SAS score a,d 4.5 ± ± Abbreviations: ALT, alanine aminotransferase; APACHE, Acute Physiology and Chronic Health Evaluation; AST, aspartate aminotransferase; CAD, coronary artery disease; CHF, congestive heart failure; CIWA, Clinical Institute Withdrawal Assessment for Alcohol; COPD, chronic obstructive pulmonary disease; INR, international normalized ratio; SAS, Riker Sedation Agitation Scale. a Mean ± SD. b Number of patients (%). c Median [interquartile range]. d Maximal score refers to highest score within the first 24 hours of presentation. Table 2. Primary and Secondary Outcomes. Preintervention (n = 135) Postintervention (n = 32) P Value Duration of AWS treatment (days) a 8.0 [ ] 5.0 [ ] <0.01 Mechanical ventilation b 77 (57.0) 10 (31.3) 0.01 Mechanical ventilation days a 8 [4-10] 5 [2-9] 0.12 Required brain imaging b,c 43 (31.9) 15 (46.9) 0.15 Pneumonia b 51 (37.8) 8 (25.0) 0.22 ICU length of stay (days) a 7 [4-11] 4 [2-7] 0.02 Hospital length of stay (days) a 13 [9-18] 9 [6-13] 0.01 ICU mortality a 3 (2.2) 1 (3.1) 0.58 Seizures during AWS treatment 0 0 Abbreviations: AWS, alcohol withdrawal syndrome; ICU, intensive care unit. a Median [interquartile range]. b n (%). c Magnetic resonance imaging or computed tomography of the brain. Discussion AWS is a commonly encountered problem in the ICU. Critically ill patients experiencing AWS require resourceintense treatment, including frequent nursing evaluations for symptom-based benzodiazepine administration. 2 The optimal treatment of AWS in the ICU remains uncertain. 16 Previous studies have demonstrated improved outcomes with symptom-triggered therapy rather than fixed-dosing protocols in non critically ill patients. Although not rigorously studied in critically ill patients, one systematic review of alcohol withdrawal and delirium tremens in the critically ill concluded that early and aggressive titration of medications guided by symptoms is associated with improved outcomes. 17 However, assessing patients in the ICU using the traditional CIWA-Ar scale can be challenging. Indeed, portions of the CIWA-Ar scale assessment require patient cooperation and communication, and it can be difficult to calculate a complete and

6 6 Annals of Pharmacotherapy Table 3. Medication Administration. Preintervention (n = 135) Postintervention (n = 32) P Value Cumulative oral benzodiazepine use a 51 ± 85 mg 9 ± 14 mg <0.01 Cumulative bolus intravenous benzodiazepine use a 74 ± 93 mg 40 ± 56 mg 0.01 Cumulative continuous infusion benzodiazepine use a 375 ± 629 mg 195 ± 167 mg 0.57 Total benzodiazepine use (oral + intravenous + infusion) a 450 ± 701 mg 74 ± 159 mg <0.01 Use of benzodiazepine continuous infusion b 79 (58.5) 3 (9.4) <0.01 Antipsychotics c 95 (70.4) 20 (62.5) 0.40 Haloperidol c 82 (60.7) 11 (34.4) 0.01 Quetiapine c 60 (44.4) 5 (15.6) <0.01 Sedative c 57 (42.2) 17 (53.1) 0.32 Propofol c 55(40.7) 5 (15.6) 0.01 Dexmedetomidine c 11 (8.1) 12 (37.5) <0.01 Antihypertensives c 58 (43.0) 9 (28.1) 0.16 Clonidine c 40 (29.6) 3 (9.4) 0.02 Labetalol c 19 (14.1) 3 (9.4) 0.58 Antiepileptics c 48 (35.6) 7 (21.9) 0.15 Gabapentin c 26 (19.3) 6 (18.8) 1.00 a Mean dose in milligrams ± SD. b n (%). c Number of patients (%); all benzodiazepine amounts are expressed in lorazepam equivalents (1 mg lorazepam = 10 mg diazepam = 3 mg midozalam). Table 4. Univariate and Multivariate Analysis of Factors Affecting Duration of Alcohol Withdrawal Treatment. Univariate Analysis Multivariate Analysis Variable Duration of AWS Treatment P Value Duration of AWS Treatment P Value Treatment with SAS-based symptom-triggered protocol 3.2 days < days 0.04 Cirrhosis 2.2 days days 0.02 Male +1.9 days days 0.04 Propofol use +3.3 days < days <0.01 Anxiety/depression No effect 0.76 Dexmedetomidine use No effect 0.32 Age No effect 0.95 APACHE II No effect 0.42 Abbreviations: APACHE, Acute Physiology and Chronic Health Evaluation; AWS, alcohol withdrawal syndrome; SAS, Riker Sedation-Agitation Score. accurate score in some ICU patients because of the severity of illness (eg, need for mechanical ventilation), agitation, delirium, or decreased level of consciousness. Alternatively, commonly used ICU sedation scales, such as SAS, are easily performed in the vast majority of ICU patients regardless of cooperativeness, agitation, or severity of illness and may be used to assess patients who cannot be scored with the CIWA-Ar scale. Thus, we performed a before-and-after study investigating the effectiveness of a symptom-triggered benzodiazepine protocol incorporating SAS scoring for selected patients compared with a CIWA-Ar based protocol that relied on fixed-dosing with benzodiazepine infusions for severe cases. We observed a significant reduction in duration of AWS treatment and benzodiazepine dose in ICU patients receiving symptom-based treatment with benzodiazepines utilizing both SAS and CIWA-Ar scoring as compared with patients managed exclusively with the CIWA-Ar scale for symptom assessment and more fixed dosing of benzodiazepines. Despite a reduction in benzodiazepine exposure and duration of treatment, we did not observe an increase in complications from inadequate treatment of AWS, such as seizures. Furthermore, patients treated with symptom-triggered benzodiazepine dosing using SAS scoring had a lower likelihood of needing mechanical ventilation and had reduced ICU and hospital LOS. We also did observe an absolute reduction in the use of antipsychotic and antihypertensive medications during the course of AWS treatment, but this difference was not statistically significant. Our results are similar to those published by Duby et al 14 who performed a pre-post intervention study in a mixed medical-surgical population of ICU patients with AWS. The

7 Sen et al 7 preintervention patients were treated according to physician preference, and the postintervention patients were treated with a protocol that used escalating doses of benzodiazepines or phenobarbital titrated to light sedation based on the Richmond Agitation Sedation Scale (RASS) score. The authors reported a decreased ICU LOS, reduced need for mechanical ventilation, increased ventilator-free days, and reduced need for continuous sedation in the postintervention group. Our study is unique in that it incorporates the SAS scale to assess and dose benzodiazepine therapy without using barbiturate adjunctive therapy. Because the SAS scale is easy to utilize and does not depend on patient communication, it may be used to effectively and efficiently assess severity of AWS symptoms in place of the CIWA-Ar scale for critically ill patients who may have limited ability to communicate. Although no study has compared use of the RASS and SAS scales to direct AWS treatment, both scales are recommended for routine assessment in critically ill patients. 20 By primarily using escalating doses of benzodiazepines to manage symptoms of AWS, our protocol avoids potential concerns associated with adjunctive barbiturates, including respiratory depression, the possible need for intubation, altered mental status, and somnolence in patients who may have already received benzodiazepines. It also reduces the risk of prolonged and heightened effects in patients with significant hepatic impairment. Our results add to the evidence that symptom-triggered therapy without benzodiazepine infusions for critically ill patients suffering from AWS is beneficial. In a recent, single-center observational study, the use of continuous benzodiazepine infusions for severe alcohol withdrawal was associated with a relatively low intubation rate of 20%. 5 Although the authors concluded that treatment with highdose, continuous sedative infusions was not associated with excess morbidity and mortality, the severity of illness was relatively low in this study, and there was no comparison group. We have observed that use of benzodiazepine infusions in severe AWS is often associated with inadequate initial control of AWS symptoms, prolonged duration of AWS treatment and mechanical ventilation, and a high incidence of delirium several days into therapy. By treating AWS symptoms with benzodiazepine bolus therapy, we demonstrated a reduction in intubation rates and a shorter duration of therapy. We suspect that the use of dexmedetomidine in the postintervention group also contributed to a reduction in benzodiazepine exposure, which has been reported in other studies. 6,22,24-26 There are a few limitations to this study. First, this was a single-center study performed at a tertiary care center and may not be generalizable to other centers. The study evaluated treatment of medical ICU patients and may not be generalizable to other critically ill populations. The before-and-after design may have led to unmeasured differences between the preintervention and postintervention groups. The postintervention group had more individuals with cirrhosis, congestive heart failure, greater initial total bilirubin levels, and higher initial CIWA scores, suggesting that this group may have had more complications from chronic alcoholism compared with the preintervention group and paradoxically required lower doses of benzodiazepines to treat AWS because of worse hepatic function and drug metabolism. However, the postintervention group did have lower APACHE II scores, suggesting that this patient cohort may have been less critically ill. Although this difference is statistically significant, the APACHE II scores between the 2 groups are unlikely to be clinically significant, and the 2 groups would have similar predicted mortality based on the mean scores observed. We did attempt to control for these differences in our multivariate analysis. In addition, our order set involved changes in benzodiazepine dosing and the use of SAS scoring. Therefore, we cannot determine the contribution these individual changes made toward the improved outcomes we observed. Another limitation of our study is that the primary outcome of duration of AWS treatment was based on timing of the last benzodiazepine dose or physician documentation of resolution of AWS. Physician documentation of resolution of AWS may be subjective, and accuracy of this end point hinges on good physician documentation practices. There could also have been temporal changes in practice patterns for AWS during our study period. For example, in 2013, the SCCM published updated guidelines for the management of pain, agitation, and delirium, which discouraged benzodiazepines for agitation unrelated to AWS. 20 Although this may partly account for the significant reduction of benzodiazepine infusions and increased use of dexmedetomidine we report in our study, our new AWS protocol discouraged benzodiazepine infusions. The use of dexmedetomidine in the postintervention group could have contributed to the decreased benzodiazepine exposure we observed. This is similar to the findings of other studies that reported that dexmedetomidine use in ICU patients with AWS has been associated with reduced benzodiazepine requirements. 6,22,24-26 However, in our univariate analysis, dexmedetomidine use was not an independent predictor of duration of alcohol withdrawal treatment. Conclusion A symptom-triggered benzodiazepine protocol utilizing SAS and CIWA-Ar scoring for treatment of AWS in critically ill patients is associated with a reduction in the duration of AWS treatment, benzodiazepines exposure, need for mechanical ventilation, and ICU and hospital LOS when compared with a CIWA-Ar based protocol using fixed dosing with benzodiazepine infusions. These findings suggest that a randomized controlled trial should be conducted to provide more definitive data to help direct the care of AWS patients in the ICU.

8 8 Annals of Pharmacotherapy Appendix A Clinical Institute Withdrawal Assessment for Alcohol (CIWA-Ar)27 Nausea and vomiting Ask, Do you feel sick to your stomach? Have you vomited? 0 = no nausea and no vomiting 1 = mild nausea with no vomiting = intermittent nausea with dry heaves = constant nausea, frequent dry heaves and vomiting Tremor Arms extended and fingers spread apart. 0 = no tremor 1 = not visible, but can be felt fingertip to fingertip = moderate, with patient s arms extended = severe, even with arms extended Paroxysmal sweats 0 = no sweat visible 1 = barely perceptible sweating, palms moist = beads of sweat obvious on forehead = drenching sweats Anxiety Ask, Do you feel nervous? 0 = no anxiety, at ease 1 = mildly anxious = moderately anxious, or guarded, so anxiety is inferred = equivalent to acute panic states as seen in severe delirium or acute schizophrenic reaction Agitation 0 = normal activity 1 = somewhat more than normal activity = moderately fidgety and restless = paces back and forth during most of the interview, or constantly thrashes about Tactile disturbances Ask, Have you any itching, pins and needles sensations, burning sensations, numbness or do you feel bugs crawling on or under your skin? 0 = None 1. very mild itching, pins and needles, burning and numbness 2. mild itching, pins and needles, burning and numbness 3. moderate itching, pins and needles, burning and numbness 4. moderately severe hallucinations 5. severe hallucinations 6. extremely severe hallucinations 7. continuous hallucinations Auditory disturbances Ask, Are you more aware of sounds around you? Are they harsh? Do they frighten you? Are you hearing anything that is disturbing to you? Are you hearing things you know are not there? 0 = not present 1. very mild harshness or ability to frighten 2. mild harshness or ability to frighten 3. moderate harshness or ability to frighten 4. moderately severe hallucinations 5. severe hallucinations 6. extremely severe hallucinations 7. continuous hallucinations Visual disturbances Ask, Does the light appear to be too bright? Is its color different? Does it hurt your eyes? Are you seeing anything that is disturbing to you? Are you seeing things you know are not there? 0 = not present 1. very mild sensitivity 2. mild sensitivity 3. moderate sensitivity 4. moderately severe hallucinations 5. severe hallucinations 6. extremely severe hallucinations 7. continuous hallucinations Headache, fullness in head Ask, Does your head feel different? Does it feel as if there is a band around your head? Do not rate for dizziness or lightheadedness. Otherwise, rate severity. 0 = not present 1. very mild 2. mild 3. moderate 4. moderately severe 5. severe 6. very severe 7. extremely severe Orientation and clouding of sensorium Ask, What day is this? Where are you? Who am I? 0 = oriented and can do serial additions 1. cannot do serial additions or is uncertain about date 2. disoriented for date by no more than 2 calendar days 3. disoriented for date by more than 2 calendar days 4. disoriented for place and/or person Total CIWA Score (maximum possible score: 67)

9 Sen et al 9 Appendix B Sedation Agitation Scale (SAS)28 7 Dangerous agitation Pulling at endotracheal tube, trying to remove catheters, climbing over bed rail, striking at staff, thrashing side-toside 6 Very agitated Does not calm, despite frequent verbal reminding of limits; requires physical restraints, biting ET tube 5 Agitated Anxious or mildly agitated, attempting to sit up, calms down to verbal instructions 4 Calm and Calm, awakens easily, follows commands cooperative 3 Sedated Difficult to arouse, awakens to verbal stimuli or gentle shaking but drifts off again, follows simple commands 2 Very sedated Arouses to physical stimuli but does not communicate or follow commands, may move spontaneously 1 Unarousable Minimal or no response to noxious stimuli, does not communicate or follow commands Declaration of Conflicting Interests The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article. Funding The authors received no financial support for the research, authorship, and/or publication of this article. References 1. Uusaro A, Parviainen I, Tenhunen JJ, Ruokonen E. The proportion of intensive care unit admissions related to alcohol use: a prospective cohort study. Acta Anaesthesiol Scand. 2005;49: Bard MR, Goettler CE, Toschlog EA, et al. Alcohol withdrawal syndrome: turning minor injuries into a major problem. J Trauma. 2006;61: Mayo-Smith MF, Beecher LH, Fischer TL, et al. Management of alcohol withdrawal delirium: an evidence-based practice guideline. Arch Intern Med. 2004;164: Amato L, Minozzi S, Davoli M. Efficacy and safety of pharmacological interventions for the treatment of the alcohol withdrawal syndrome. Cochrane Database Syst Rev. 2011;(6):CD Stewart R, Perez R, Musial B, Lukens C, Adjepong YA, Manthous CA. Outcomes of patients with alcohol withdrawal syndrome treated with high-dose sedatives and deferred intubation. Ann Am Thorac Soc. 2016;13: Crispo AL, Daley MJ, Pepin JL, Harford PH, Brown CV. Comparison of clinical outcomes in nonintubated patients with severe alcohol withdrawal syndrome treated with continuous-infusion sedatives: dexmedetomidine versus benzodiazepines. Pharmacotherapy. 2014;34: Decarolis DD, Rice KL, Ho L, Willenbring ML, Cassaro S. Symptom-driven lorazepam protocol for the treatment of severe alcohol withdrawal delirium in the intensive care unit. Pharmacotherapy. 2007;27: Horinek EL, Kiser TH, Fish DN, MacLaren R. Propylene glycol accumulation in critically ill patients receiving continuous intravenous lorazepam infusions. Ann Pharmacother. 2009;43: Weaver MF, Hoffman HK, Johnson RE, Mauck K. Alcohol withdrawal pharmacotherapy for inpatients with medical comorbidity. J Addict Dis. 2006;25: 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: Daeppen JB, Gache P, Landry U, et al. Symptom-triggered versus fixed-schedule doses of benzodiazepine for alcohol withdrawal: a randomized treatment trial. Arch Intern Med. 2002;162: Jaeger TM, Lohr RH, Pankratz VS. Symptom-triggered therapy for alcohol withdrawal syndrome in medical inpatients. Mayo Clin Proc. 2001;76: Spies CD, Otter HE, Huske B, et al. Alcohol withdrawal severity is decreased by symptom-oriented adjusted bolus therapy in the ICU. Intensive Care Med. 2003;29: Duby JJ, Berry AJ, Ghayyem P, Wilson MD, Cacanour CS. Alcohol withdrawal syndrome in critically ill patients: protocolized versus nonprotocolized management. J Trauma Acute Care Surg. 2014;77: 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: Wong A, Benedict NJ, Kane-Gill SL. Multicenter evaluation of pharmacologic management and outcomes associated with severe resistant alcohol withdrawal. J Crit Care. 2015;30: Awissi DK, Lebrun G, Coursin DB, Riker RR, Skrobik Y. Alcohol withdrawal and delirium tremens in the critically ill: a systemic review and commentary. Intensive Care Med. 2013;39: Hecksel KA, Bostwick JM, Jaeger TM, Cha SS. Inappropriate use of symptom-triggered therapy for alcohol withdrawal in the general hospital. Mayo Clin Proc. 2008;83: Riker RR, Picard JT, Fraser GL. Prospective evaluation of the sedation-agitation scale in adult ICU patients. Crit Care Med. 1999;27: Barr J, Fraser GL, Puntillo 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: Weinberg JA, Magnotti LJ, Fischer PE, et al. Comparison of intravenous ethanol versus diazepam for alcohol withdrawal prophylaxis in the trauma ICU: results of a randomized trial. J Trauma. 2008;64: 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:

10 10 Annals of Pharmacotherapy 23. Reeves JG, Glass P, Lubarsky DA, et al. Intravenous anesthetics. In: Miller RD, Eriksson LI, Fleischer LA, et al, eds. Miller s Anesthesia. 7th ed. Philadelphia, PA: Churchill Livingstone; 2010: Bielka K, Kuchyn I, Glumcher F. Addition of dexmedetomidine to benzodiazepines for patients with alcohol withdrawal syndrome in the intensive care unit: a randomized controlled study. Ann Intensive Care. 2015;5: Ludtke KA, Stanley KS, Yount NL, Gerkin RD. Retrospective review of critically ill patients experiencing alcohol withdrawal: dexmedetomidine versus propofol and/or lorazepam continuous infusions. Hosp Pharm. 2015;50: Linn DD, Loeser KC. Dexmedetomidine for alcohol withdrawal syndrome. Ann Pharmacother. 2015;49: Sullivan JT, Sykora K, Schneiderman J, Naranjo CA, Sellers EM. Assessment of alcohol withdrawal: the revised Clinical Institute Withdrawal Assessment for Alcohol Scale (CIWA-Ar). Br J Addict. 1989;84: Riker RR, Picard JT, Fraser GL. Prospective evaluation of the Sedation-Agitation Scale for adult critically ill patients. Crit Care Med. 1999;27: