Potential biases in case detection of alcohol involvement among adolescents in an emergency department

• Published in: Pediatric emergency care Vol. 18; no. 5; p. 350
• Main Authors: Colby, Suzanne M, Barnett, Nancy P, Eaton, Cheryl A, Spirito, Anthony, Woolard, Robert, Lewander, William, Rohsenow, Damaris J, Monti, Peter M
• Format: Journal Article
• Language: English
• Published: United States 01.10.2002
• Subjects: Adolescent; Alcoholic Intoxication - diagnosis; Alcoholic Intoxication - epidemiology; Decision Making; Female; Humans; Male; Medical Records - statistics & numerical data; New England - epidemiology; Practice Patterns, Physicians; Retrospective Studies; Selection Bias; Trauma Centers - statistics & numerical data; Wounds and Injuries - epidemiology; New England
• ISSN: 0749-5161; 1535-1815
• DOI: 10.1097/00006565-200210000-00005

To determine the factors associated with physician decisions to test for alcohol involvement in adolescents treated in an emergency department (ED) and to examine patient and event characteristics associated with being identified as alcohol positive, either by testing or by clinical examination. Medical chart reviews were conducted for all adolescent patients (n = 9,660; age range, 13-19 y) treated over a 1-year period in a Level I regional trauma center/ED. Among all 9,660 patients in the ED, 298 (3.1%) were identified as alcohol positive by test or clinical examination. Of the 9,660 patients, 464 (4.8%) were tested for alcohol, and 49% of these had alcohol-positive test results. Physicians were more likely to order alcohol tests when patients were male, older, injured, and treated during the overnight shift or on weekends. Testing was most common for suicide attempts, motor vehicle crashes, assaults, and intoxication. Patients being treated for an illness or for occupational or athletic injuries were rarely tested. A large proportion of alcohol-related treatment was for intoxicated, uninjured patients, whose profile was different (ie, younger, more often female, with higher blood alcohol concentrations, and admission distributed more evenly across ED shifts). Case detection rates for alcohol involvement may be biased and inflated when based on physician decisions to test for alcohol use. Rates based on comprehensive chart reviews and clinical examination may be better estimates but are also subject to methodologic limitations. Universal screening would yield the most accurate estimates of alcohol prevalence and would provide more accurate guidance to physicians regarding when to test for alcohol use. Universal screening as a clinical standard would help to identify more adolescents who might benefit from additional alcohol use intervention.