To the Editor. —
The article by Chaisson et al1 suggests that intravenous (IV) cocaine use is associated with human immunodeficiency virus infection. However, their data analysis contains methodological flaws and their conclusions make unjustified population generalizations.The multivariate model depicted in Table 1 of the article by Chaisson et al contains two interaction variables, "daily cocaine use, blacks and Hispanics," and "injection >3 times daily, blacks, before methadone treatment," yet includes only one of the four corresponding main effects (black race).The etiologic and mathematical interpretation of nonhierarchical models becomes complex.2,3 In the presence of a significant interaction, the meaning of the main terms is affected by how they are coded; hence, it is unwise to drop them from the model solely on the basis of tests of statistical significance. Hypotheses based on models that include significant interactions without their corresponding main effects are viewed by some
French JF. Intravenous Cocaine and HIV Infection. JAMA. 1989;262(11):1471. doi:10.1001/jama.1989.03430110061022
Customize your JAMA Network experience by selecting one or more topics from the list below.
Create a personal account or sign in to: