SEARCH

SEARCH BY CITATION

Aims. The factor structure of DSM-IV substance disorder criteria is examined among alcohol, cannabis, cocaine and opiate users to determine the dimensionality of abuse and dependence criteria within each of these drug classes and whether a common construct can be generalized across drug classes. Design. 12-month criterion prevalence was assessed as part of the World Health Organization's Study on the Reliability and Validity of the Alcohol and Drug Use Disorder Instruments in various settings at eight sites around the world using the Schedules for Clinical Assessment in Neuropsychiatry (SCAN). A majority of respondents were recruited from non-treatment settings. In addition to exploratory factor analysis, confirmatory factor analysis was used to analyse factor structures using weighted least square methods and tetrachoric correlation matrices. Multi-sample analysis techniques were used to model differences between drug-classes. Findings. In the full data analyses identified a single factor solution for each user population and across user populations. However, analyses of data from users reporting low to moderate symptomatology identified a two-dimensional construct among alcohol, cannabis and opiate users consisting of a major "dependence" factor and a lesser "abuse" factor. In addition, results showed that neither the abuse criterion "(A2) use in physical hazardous situations" or the dependence criterion "(D7) use despite knowledge of psychological/physical problems" were central to the latent construct in any of the user populations, except for D7 among alcohol users. Conclusions. The multi-dimensional results found among users with low to moderate symptomatology indicate that: (1) previous results from relatively homogeneous populations may have been biased towards lesser order solutions, and that (2) the DSM-IV substance disorder criteria describe at least two distinct phenomena, supporting the current DSM-IV organization of substance disorder criteria. Further work needs to evaluate whether prevalent symptoms are present in random or predictable combinations, whether combinations reflecting a specific hierarchy of severity can be identified, and whether incident symptoms are accumulated in a predictable pattern, within specific user populations and across user populations.