• Cannabis;
  • cocaine;
  • dependence;
  • DSM-IV;
  • hallucinogens;
  • item–response theory;
  • sedatives;
  • stimulants;
  • opioids;
  • substance abuse


Aims  This paper explored, in a population-based sample of males, the factorial structure of criteria for substance abuse and dependence, and compared qualitatively the performance of these criteria across drug categories using item–response theory (IRT).

Design  Marginal maximum likelihood was used to explore the factor structure of criteria within drug classes, and a two-parameter IRT model was used to determine how the difficulty and discrimination of individual criteria differ across drug classes.

Participants  A total of 4234 males born from 1940 to 1974 from the population-based Virginia Twin Registry were approached to participate.

Measurements  DSM-IV drug use, abuse and dependence criteria for cannabis, sedatives, stimulants, cocaine and opiates.

Findings  For each drug class, the pattern of endorsement of individual criteria for abuse and dependence, conditioned on initiation and use, could be best explained by a single factor. There were large differences in individual item performance across substances in terms of item difficulty and discrimination. Cocaine users were more likely to have encountered legal, social, physical and psychological consequences.

Conclusions  The DSM-IV abuse and dependence criteria, within each drug class, are not distinct but best described in terms of a single underlying continuum of risk. Because individual criteria performed very differently across substances in IRT analyses, the assumption that these items are measuring equivalent levels of severity or liability with the same discrimination across different substances is unsustainable. Compared to other drugs, cocaine usage is associated with more detrimental effects and negative consequences, whereas the effects of cannabis and hallucinogens appear to be less harmful. Implications for other drug classes are discussed.