# Development of the Alcohol Use Disorders Identification Test-Consumption (AUDIT-C) Score as a Predictor of Alcohol Withdrawal Syndrome in Trauma Patients at a Community Level 1 Trauma Center

## TL;DR

This study shows that the AUDIT-C score can effectively predict alcohol withdrawal syndrome in trauma patients, helping guide treatment decisions.

## Contribution

The study introduces a validated AUDIT-C threshold for predicting AWS in trauma patients with moderate alcohol use.

## Key findings

- AUDIT-C scores ≥ 5 predicted AWS with 90% sensitivity and 96.2% specificity.
- An AUDIT-C threshold of ≥ 5 was identified using logistic regression and ROC analysis.
- The study population included 662 trauma patients, mostly geriatric and female.

## Abstract

Background and objective

The Clinical Institute Withdrawal Assessment for Alcohol-Revised (CIWA-Ar) is an assessment tool that guides symptom-triggered therapy (STT) in alcohol withdrawal syndrome (AWS) patients. Institutionally, CIWA-Ar is used for STT when patients admit to daily alcohol use or arrive intoxicated. Given the lack of validated screening tools for predicting AWS, we hypothesized that CIWA-Ar and STT were used inefficiently, causing poor resource stewardship and overtreatment. Our current protocol is to complete the Alcohol Use Disorders Identification Test-Consumption (AUDIT-C), an evidence-based screening tool for hazardous alcohol use. Given this protocol and the absence of verified screening tools for alcohol withdrawal prediction, we aimed to analyze AUDIT-C efficacy in predicting AWS to guide STT.

Methods

A retrospective review was performed of admission AUDIT-C responses between January 1, 2018, and December 31, 2018. Given the vague documentation of AWS diagnosis, an alcohol withdrawal syndrome score (AWS Score) was created based on current literature and was statistically confirmed. Per our criteria, AWS was defined as an AWS score of ≥ 3 in patients with moderate alcohol use.

Results

The study population included 662 trauma patients, predominantly geriatric (age ≥ 65 years, 68%) and female (60%). In the setting of moderate alcohol use, AUDIT-C was a statistically significant predictor for AWS (logistic regression model, χ2(1) = 172.371, p < 0.0005), with a 90.0% sensitivity, 96.2% specificity, positive predictive value of 52.9%, and negative predictive value of 99.5%. To provide clinicians a guide for more objective utilization of alcohol withdrawal protocols, an AUDIT-C threshold of ≥ 5 was identified using binary logistic regression and receiver-operating characteristic curve (ROC) analyses.

Conclusion

We noted AUDIT-C scores of ≥ 5 at the time of admission in hospitalized trauma patients with moderate alcohol use to predict AWS. Given these findings, we propose that AUDIT-C scores may be reliable guides for implementing alcohol withdrawal protocols for the treatment of this patient population.