Page 20

219126_NCS_Currents_March_2_eMag

JOURNAL WATCH Journal Watch By Chad Miller, MD and Susanne Muehlschlegel, MD, MPH Protocol-Driven Treatment of Alcohol This study suggests that a protocolized, symptom-triggered Withdrawal Syndrome in Critically Ill approach to AWS leads to shorter ICU stays and less overall Patients sedation. While many practitioners may have anticipated the Duby JJ, Berry AJ, Ghayyem P, et al. J Trauma findings of this study, the benefits of this approach have not Acute Care Surg 2014;77:938-943 previously been demonstrated in critically ill patients with This single-center, pre- and post-intervention various primary diagnoses. The studied AWS protocol varies from trial sought to determine the value of others (like CIWA) in that it is designed to be applicable to the protocolized, symptom-triggered, dose- assessment of a critically ill or intubated patient. Chad Miller, MD escalation management of Alcohol The findings of this study are limited by our understanding of Withdrawal Syndrome (AWS) upon length how the PRE group was managed. Furthermore, it is concerning of stay in the ICU. Subjects included adult that, despite half the enrollment time, the number of POST group patients admitted to the ICU of a tertiary subjects was greater than the PRE group. This raises concerns academic medical center with AWS, regardless about selection bias that may be further substantiated by the of concomitant medical illness. Patients disparity of SOFA scores between the two groups. It is possible enrolled in the two-year period prior to that, as the institution promoted its AWS order set, it was applied protocol implementation (PRE) were to a greater variety of patients, including those who were less managed utilizing strategies of the treating acutely ill and destined for more rapid recovery. However, the physicians’ preference. benefit of treatment in the POST group was maintained after Susanne logistic regression analysis.Though management details were not Muehlschlegel, specifically characterized, the institutional Simple BRAIN Score to Predict Hematoma Growth in ICH MD, MPH standard included use of continuous infusion Wang X, Arima H, Al-Shahi Salman R, et al. Stroke 2015;46:376-381 or scheduled dosing of benzodiazepines. In the third year of Simple clinical risk scores are useful for risk stratification for both enrollment (POST), a Richmond Agitation Sedation Score (RASS clinical care and research. The authors of this study analyzed goal 0 to -2)-triggered dose-escalation protocol was constructed non-contrast head CTs from the pilot and main phases of the which utilized diazepam and phenobarbital as primary and INTERACT 1 and 2 trials to develop and validate a simple clinical secondary sedative options. The treatment team was guided risk score to predict hematoma volume growth (≥6 ml) within by an electronic order set, which specified sedative dosing after 24 hours after ICH onset. Patients from the CT sub-studies of patient assessment conducted at 15-30 minute intervals. Patients each trial were identified, all of whom underwent repeat CT at 24 with severe brain injury (Glasgow Coma Scale <8) were excluded hours after randomization. The INTERACT 2 cohort was used for from enrollment. derivation (n=964) and the INTERACT 1 cohort (n=346) was used The primary study aim was assessment of ICU length of stay for validation. (LOS). Sequential Organ Failure ICH volumes were calculated centrally – using mostly volumetric Assessment (SOFA) scores and other key medical demographics software with a small proportion using the ABC/2 method when were collected. Differences between cohorts (PRE and POST) a digital CT was not available – and blind to clinical data. Kappa were included in a logistic regression analysis to control for their statistics for head CT reads were 0.97 and 0.96 for INTERACT 1 impact upon primary and secondary outcome measures. 135 total and 2, respectively. Clinically significant growth was defined as subjects were enrolled (60 PRE, 75 POST), approximately half an absolute growth of ≥6 ml from baseline. Using multivariable of whom were cared for in a Medical ICU and 30% in a Surgical logistic regression and backward elimination of non-significant Trauma ICU. covariates, predictors remained in the model for p<0.05. PRE patients were older (55.7 versus 50.7 years, p=0.03) and Regression coefficients were then used to generate point scores. had higher SOFA scores (6.1 versus 3.9, p=0.0004). Both groups In the derivation cohort, 181/964 (18.8%) and in the validation had similar mean blood alcohol levels and rates of seizures and cohort 43/346 (12.4%) developed significant ICH growth. psychosis. A non-significant trend toward greater rates of prior Variables forced into the model included age, sex, diabetes alcohol withdrawal (40% versus 30.6%, p=0.28) and delirium mellitus, deep location of ICH, and randomization to intensive tremens (10% versus 4%, p=0.19) was seen in the PRE group. The BP-lowering treatment. Significant co-variables remaining in the PRE group had greater ICU LOS (9.6 versus 5.2 days, p=0.0004), model included baseline ICH volume, recurrent ICH, warfarin use days on ventilator (5.6 versus 1.3 days, p<0.0001), intubation due at onset, intraventricular hemorrhage (IVH), and number of hours to AWS (22% versus 5%, p<0.001), and duration of sedation (10.8 to baseline CT from onset. versus 3.5 days, p<0.01). The BRAIN score prediction algorithm score was developed. Cumulative lorazepam-equivalent benzodiazepine use was three The median probability of ICH growth was 16.4% (IQR 7.7%- times greater in the PRE group compared to the POST group 27.2%), ranging from 3.4% (score 0) to 85.8% (max score of (median 33 mg versus 23 mg), whereas phenobarbital use was 24), with a good discriminative ability (C-statistic 0.73). In twice as great in the POST group. Patients admitted for AWS were the independent validation cohort, the discriminative ability more likely to be intubated compared to those with AWS primarily remained the same (C-statistic 0.73) and had good calibration admitted to the ICU for other reasons (p=0.004). (Hosmer-Lemeshow Chi-Square statistic 4.25, p=0.82). Here, the median probability of ICH growth was 9.3% (IQR 5.1%-16.6%), with excellent linear correlation between predicted and observed probabilities of ICH growth. 20 Continued on page 21


219126_NCS_Currents_March_2_eMag
To see the actual publication please follow the link above