Model for quantifying the drug involvement of medical students

We have applied a two-parameter normal item-response model to analyze data on "drugs ever used" by freshman medical students, then used the result to predict their subsequent drug use patterns over the first 18 months of medical school. Our results provide strong evidence for the existence of a single latent dimension of "drug involvement," and they demonstrate that a student's initial position on this dimension predicts subsequent drug use well. The model applies to male and female students equally well, suggesting the dimension is not sex specific. Students' drug use is cumulative in the sense that an individual tends to add new drugs to all previous drugs, rather than "graduating" from a previous to a new drug, as he/she becomes more "drug involved." Greater "drug involvement" is also associated with more frequent use of most drug categories.