Narrative
Alcohol use disorder remains a highly prevalent issue that frequently results in emergency department (ED) visits and hospitalizations. Severe alcohol withdrawal syndrome (SAWS) in hospitalized patients has significant morbidity and mortality, warranting a need to identify patients at high risk of developing alcohol withdrawal on admission. Identifying patients at high risk of SAWS could improve patient outcomes. The systematic review discussed here evaluates the elements of clinical history, signs and symptoms, or clinical assessment tools to aid in identifying patients at risk of developing SAWS.
In this evidence-based summary, we review the systematic review by Wood et al. “Will This Hospitalized Patient Develop Severe Alcohol Withdrawal Syndrome?: The Rational Clinical Examination Systematic Review”. Wood et al. conducted a meta-analysis on a total of 14 studies (one case control study, five prospective cohort studies, and eight retrospective cohort studies) that assessed the value of medical history, signs and symptoms at presentation, and relevant laboratory findings in predicting SAWS in hospitalized patients. SAWS was either explicitly defined in individual studies, or was extrapolated in the systematic review by the development of delirium tremens or withdrawal seizures if SAWS was not an independent outcome. Some trials also assessed the clinical utility of scoring tools in predicting SAWS. The trials in aggregate enrolled 71,295 admitted patients with 1355 cases developing SAWS while admitted. Patients were admitted for several different indications, including alcohol detoxification or withdrawal, seizures, and trauma.
A previous history of delirium tremens had a likelihood ratio for a positive test (+LR) of 2.9 [95% CI, 1.7–5.2] in predicting SAWS. Additionally, having had 3 or more seizures as a part of the current presentation had an LR of 2.8 [95% CI, 1.4–5.9], and 1–2 seizures had an LR of 1.6 [95% CI, 1.4–2.2] for progressing to delirium tremens. Analysis of the patient's sex and race did not show an increased risk of SAWS.
Of the signs and symptoms studied, an elevated systolic blood pressure (≥ 140 mmHg) had an LR+ of 1.7 [95% CI, 1.3–2.3] for predicting SAWS. However, a normal blood pressure did not indicate a lower risk of developing severe alcohol withdrawal. Of note, tachycardia did not have adequate predictive power due to its small effect size. No single sign or symptom was identified as a useful predictor of SAWS.
The systematic review also evaluated the value of laboratory findings. Patients presenting for trauma with a Blood Alcohol Concentration (BAC) of > 200 mg/dL had an LR+ of 3.5 [95% CI, 3.0–4.0]; blood urea nitrogen of > 26 mg/dL had an LR+ of 1.6 [95% CI, 0.67–3.88]; and thrombocytopenia (< 150 × 103/μL) had an LR+ of 2.2 [95% CI, 1.4–3.4]
, ,
Rather than using individual items to predict the risk of developing SAWS, multiple scoring tools have been studied that combine multiple items to improve test characteristics. One example of such a scoring tool is the Prediction of Alcohol Withdrawal Severity Scale (PAWSS), see Table 1. A score of ≥ 4 suggests a high risk of SAWS with an LR of 174 [95% CI, 42–696]. A score of < 4 on PAWSS had an LR for a negative test (−LR) of 0.07 [95% CI, 0.02–0.26]. Across all studies, PAWSS had a sensitivity of 93% and a specificity of 99% for predicting SAWS.
Table 1. Prediction of Alcohol Withdrawal Severity Scale (PAWSS) questions.
Another scoring tool evaluated by the systematic review was the simplified Luebeck Alcohol Withdrawal Scale (LARS-10). LARS-10 was designed to predict the likelihood of SAWS in patients with a history of “alcohol dependence”. LARS-10 contains 10 questions; see Table 2.
Table 2. List of questions in LARS-10 and point values.
Scoring ≥ 9 had an LR+ of 12 [95% CI, 5.8–27] of severe withdrawal, with a sensitivity of 95% and a specificity of 93%. Scoring < 9 had a LR− of 0.05 [95% CI, 0.02–0.37].
Caveats
The 14 studies included in this meta-analysis were case–control and cohort studies. By design, cohort studies inherently limit the ability to control exposures, increasing the potential for bias resulting from confounding variables. Furthermore, as most studies were conducted among hospitalized patients, selection bias is likely if the reader wishes to apply scoring tools on all patients presenting to the ED with alcohol use disorder. Hospitalized patients tended to be more medically ill due to alcohol use. Different studies included in the systematic review used varying definitions of SAWS, making it difficult to standardize outcomes across the meta-analysis. The studies included a mix of inpatient populations, which may not be fully comparable in terms of baseline risk, comorbidities, or treatment settings. Some of the included studies used non-standardized tools or non-validated measures to identify or predict SAWS (e.g., CIWA-Ar, DSM criteria, or local clinical assessments). This inconsistency can affect the reliability of pooled diagnostic accuracy measures. Although the review identified several potential predictors (e.g., history of withdrawal seizures, prior delirium tremens, high blood alcohol level), data on some predictors were limited or of low quality, reducing the ability to draw strong conclusions for certain variables. Most of the studies were conducted in academic or tertiary care centers, which may limit the generalizability of the findings to community hospitals or non-hospitalized populations. The systematic review noted a moderate to high risk of bias in some studies due to retrospective designs, incomplete outcome data, or lack of blinding. A common problem with clinical scoring tools such as PAWSS is incorporation bias. PAWSS includes “evidence of increased autonomic activity (e.g., HR > 120 bpm, tremor, sweating, agitation, nausea),” which is a symptom of alcohol withdrawal, potentially biasing the score.
Variability in treatment protocols, such as the use of benzodiazepines or symptom-triggered therapy, may have influenced the progression to SAWS, confounding the relationship between predictors and outcomes. While some predictors were statistically significant, none had high enough predictive accuracy alone to be clinically decisive. The authors caution against over-reliance on any single factor.
Relevance to Emergency Medicine
The ED often serves as the first point of contact for patients in metabolic crisis, providing a unique opportunity to screen for alcohol use disorders systematically. PAWSS can be effectively applied in emergency medicine because many patients at risk for severe alcohol withdrawal are first encountered in the ED, where early identification is critical when the decision to admit is coupled with a determination of level of care. A high index of suspicion in the ED allows the admitting team to be alerted promptly and initiate preventative measures, potentially averting adverse events. Early risk stratification using PAWSS in this setting can guide resource allocation, such as ICU admission or need for closer monitoring, and help prioritize care in a high-risk environment.
In summary, patients with a history of withdrawal seizures or delirium tremens are at increased risk for severe alcohol withdrawal, but no single predictor should be used alone in determining the level of care and initiation of preventative measures. Clinical decisions should rely on a combination of factors. The PAWSS clinical decision tool provides a framework for consolidating pertinent historical risk factors and objective clinical data, allowing reliable risk stratification for patients at risk of severe alcohol withdrawal.
The original manuscript was published in
Academic Emergency Medicine as part of the partnership between TheNNT.com and AEM.
Author
Stephanie Cung, MD; Sierra F. Williams, DO, MS; Noah Berland, MD
Supervising Editors: Shahriar Zehtabchi, MD
Published/Updated
October 14, 2025
What are Likelihood Ratios?
LR, pretest probability and posttest (or posterior) probability are daunting terms that describe simple concepts that we all intuitively understand.
Let's start with pretest probability: that's just a fancy term for my initial impression, before we perform whatever test it is that we're going to use.
For example, a patient with prior stents comes in sweating and clutching his chest in agony, I have a pretty high suspicion that he's having an MI – let's say, 60%. That is my pretest probability.
He immediately gets an ECG (known here as the "test") showing an obvious STEMI.
Now, I know there are some STEMI mimics, so I'm not quite 100%, but based on my experience I'm 99.5% sure that he's having an MI right now. This is my posttest probability - the new impression I have that the patient has the disease after we did our test.
And likelihood ration? That's just the name for the statistical tool that converted the pretest probability to the posttest probability - it's just a mathematical description of the strength of that test.
Using an online calculator, that means the LR+ that got me from 60% to 99.5% is 145, which is about as high an LR you can get (and the actual LR for an emergency physician who thinks an ECG shows an obvious STEMI).
(Thank you to Seth Trueger, MD for this explanation!)
