Risk Factors for Delirium Tremens Development Jeffrey A. Ferguson, MD, MPH, Christopher J. Suelzer, MD, George J. Eckert, MS, Xiao-Hua Zhou, PhD, Robert S. Dittus, MD, MPH OBJECTIVE: To identify clinical characteristics associated with inpatient development of delirium tremens so that fu- ture treatment efforts can focus on patients most likely to benefit from aggressive therapy. DESIGN: Retrospective cohort study among patients dis- charged with diagnoses related to alcohol abuse. SETTING: University-affiliated inner-city hospital. PATIENTS~PARTICIPANTS: Two hundred consecutive patients discharged between June 1991 and August 1992 who under- went evaluation and treatment for alcohol withdrawal or detoxiflcation. MEASUREMENTS AND MAIN RESULTS: Mean age was 41.9 years, 85% were male, 57% were white and 84% were unmar- ried. Forty-eight (24%) of the patients developed delirium tremens during hospitalization. Bivariate analysis indicated that those who developed delirium tremens were more likely to be African-American, unemployed, and homeless, and were more likely to have gone more days since their last drink, and to have concurrent acute medical illness, high admission blood urea nitrogen level and respiratory rate, and low admis- sion albumin level and systolic blood pressure. In multiple lo- gistic regression analyses, patients who developed delirium tremens were more likely to have gone more days since their last drink (odds ratio [OR) 1.3; 95% confidence interval [CI] 1.09, 1.61) and to have concurrent acute medical illness (OR 5.1; 95% CI 2.07, 12.55). These risk factors were combined for assessment of their ability to predict the occurrence of delirium tremens. If no factors were present, 9% developed delirium tremens; if one factor was present, 25% developed delirium tremens; and if two factors were present, 54% devel- oped delirium tremens. CONCLUSIONS: Inpatient development of delirium tremens was common among patients treated for alcohol detoxifica- tion or withdrawal and correlated with several readily avail- able clinical variables. KEY WORDS: Delirium tremens; hospitalized patients; detox- iflcation; alcohol-related complications. J GEN INTERN MED 1996;11:410-414. Received from the Divisions of General Internal Medicine (JAF, CJS, RSD) and Biostatistics (GJE, XHZ), Department of Medi- cine, Indiana University School of Medicine; Regenstrief lnstitute for Health Care (JAF, GE, XHZ, RSD); and Richard L. Roude- bush Veterans Affairs Medical Center (RSD); Indianapolis, Ind. The conclusions presented herein do not necessarily repre- sent those of the.federal government. Abstract presented at the Society of General Internal Medicine annual meeting, Washing- ton, DC, April 29, 1993. Address correspondence and reprint requests to Dr. Fergu- son: Richard L. Roudebush VAMC, D4, 1481 W. 10th St., India- napolis, IN 46202. 410 A lcohol abuse and dependence are prevalent health problems responsible for substantial morbidity, mor- tality, and health services utilization with approximately 44 billion dollars spent annually in the United States to treat alcohol-related complications.l Once alcohol depen- dence is recognized, formal detoxification and abstinence can be successful in preventing or attenuating further complications. However, detoxification carries some risk, because alcohol cessation can initiate alcohol withdrawal, ranging from mild and asymptomatic to severe and life- threatening. The severest form, defrium tremens, is esti- mated to occur in approximately 5% to 20% of all patients undergoing detoxification, and of those who experience delirium tremens approximately 5% suffer fatal complica- tions. 2-5 Prospectively identifying patients with increased risk of developing severe alcohol withdrawal would be ex- tremely useful to clinicians. Valid multivariable models would allow clinicians to tailor therapeutic decisions based on the risk of delirium tremens development. How- ever, existing literature suggesting associations between patient factors and risk of delirium tremens either lack empirical data or are limited by inadequate sample size or lack of multivariable analyses, a-8 In addition, most stud- ies have focused on detoxification unit settings where pa- tients with concurrent acute medical illnesses are not treated. 5-7 Therefore, physicians who assess and treat pa- tients undergoing alcohol withdrawal in acute care set- tings do not have available sufficient evidence to apply risk stratifcation criteria to their treatment strategies. In this study, we estimated the incidence rate of delir- ium tremens among a cohort of patients admitted to an acute care hospital for evaluation or treatment of alcohol withdrawal or detoxification. We also identified factors as- sociated with delirium tremens development as a first step in helping clinicians develop targeted interventions to decrease the risk of such events. METHODS Setting This retrospective cohort study was conducted among patients admitted to the internal medicine service at Wishard Hospital, a university-affiliated 380-bed inner- city hospital. Inpatient care is managed by approximately 110 housestaff physicians with supervision by general in- ternal medicine and internal medicine subspecialty fac- ulty. Approximately 480 patients are admitted to the in- ternal medicine service per month. The initial evaluation and treatment of alcohol withdrawal and detoxification are determined by housestaff physicians without the use