Predicting delirium tremens in patients with alcohol withdrawal seizures

April 13, 2015, 9:16 pm

delirium_tremens_label_2★★☆☆☆

Clinical predictors for delirium tremens in patients with alcohol withdrawal seizures. Kim DW et al. Am J Emerg Med 2015 Feb 23 [Epub ahead of print]

Abstract

Being able to predict which patients with alcohol withdrawal seizures will go on to develop delirium tremens (DTs) may lead to improved clinical outcomes and decreased morbidity and mortality.  The goal of this retrospective Korean study was to identify clinical and laboratory findings in emergency department (ED) patients with seizures attributed to alcohol withdrawal and would predict progression to delirium tremens.

ED patients presenting to 4 tertiary referral centers with seizures over a 22 month period were identified retrospectively. Patients with seizure etiology other than alcohol withdrawal were excluded. Eligible patients were observed for a minimum of 48 hours. Diagnosis of DTs was made according to the definition in the Diagnostic and Statistical Manual of Mental Disorders, Fourth Edition (DSM-IV.)

The authors identified 97 eligible patients with alcohol withdrawal seizures. Thirty-four (35%) of these went on to be diagnosed with delirium tremens. High blood homocysteine levels and low platelet count were independent risk factors for progression to delirium tremens.

Unfortunately, these results will, in my opinion, turn out to be less than clinically useful. For one thing, homocysteine levels are not readily available at most institutions, and are unlikely to be so in the foreseeable future. Secondly, making multiple comparisons (the authors analyze 24 clinical and laboratory factors) means that some apparent distinguishing characteristics will just appear by chance. Finally, even given these limitations, the factors the authors identify are that good, despite using cutoffs that were not predetermined. A platelet count of 137,000/μL was only 73% sensitive, and a homocysteine level of 14.5 micromoles per liter was only 81% sensitive. These population numbers are not good enough to be helpful in the individual patient.
Related posts:

Surely the NEJM could do a better review of delirium tremens

Comments are closed.