A Rasch Model Analysis of DSM-IV Alcohol Abuse and Dependence Items in the National Epidemiological Survey on Alcohol and Related Conditions


  • The National Epidemiological Survey on Alcohol and Related Conditions was conducted and funded by the National Institute on Alcohol Abuse and Alcoholism, with supplemental support from the National Institute on Drug Abuse.

Reprint requests: Christopher W. Kahler, PhD, Center for Alcohol and Addiction Studies, Brown University, Box G-BH, Providence, RI 02912; Fax: 401-444-1850; E-mail: Christopher_Kahler@brown.edu


Background: Questions persist about the diagnoses of alcohol abuse and alcohol dependence as defined by Diagnostic and Statistical Manual of Mental Disorder, Fourth Edition (DSM-IV). These questions center around the performance of individual alcohol use disorder (AUD) criteria and the nature of items used to assess them. In this study, we conduct in-depth analyses of interview items used to assess alcohol abuse and dependence.

Method: We conducted Rasch model analyses of responses to 33 items assessing alcohol abuse and dependence criteria in the National Epidemiological Survey on Alcohol and Related Conditions, a nationally representative sample of 34,550 participants. We also examined whether items performed differentially by gender, race/ethnicity, age, education, and time frame of assessment (i.e., past year vs lifetime).

Results: A single latent dimension explained the vast majority of shared variance among the items. Abuse symptoms were not of consistently lower severity than dependence symptoms, and some were consistently prone to differential functioning across subgroups. Items related to tolerance also functioned differentially across subgroups especially in the comparison of past-year symptoms between young adults (ages 18–24) and older adults. Items differed widely in their empirically estimated severity, even when they were intended to assess the same DSM-IV criterion.

Conclusions: The use of Rasch model analyses demonstrated the essential unidimensionality of AUD criteria and the lack of hierarchical ordering between abuse and dependence criteria, as well as potential sources of bias in the abuse diagnosis and the tolerance criterion. Results highlight that the prevalence and even the meaning of AUD criteria can differ substantially depending on the wording of items used to assess them. Implications for assessment of specific DSM criteria and for development of DSM-V are discussed.