The use of the case-crossover design in studying illicit drug use.

Abstract

The case-crossover design was developed to study time-varying exposures that cause transient excess risk of acute health events. It is a variant of case-control and subject-as-own-control research designs, involving use of information about exposure history of each case to estimate the transient effect. This kind of self-control design can help to reduce sampling bias otherwise introduced in the selection of controls, as well as confounding bias that might be derived from enduring individual characteristics, especially personality traits and other long-standing inherited or acquired vulnerabilities. When the subject is used as his or her own control, these personal vulnerabilities are matched. In this paper we discuss strengths and weaknesses of the case-crossover design and suggest applications of the case-crossover design in epidemiologic studies on suspected hazards of illicit drug use, and in studies of drug use and co-occurring psychiatric disturbances. We conclude that the case-crossover design can play a useful role, but it discloses a need to secure fine-grained measurements in epidemiologic research on psychiatric comorbidity. As explained in the paper, we also believe the case-crossover method may be of use to criminologists who study the drugs-crime nexus, to services researchers and clinicians who seek to understand treatment entry and compliance behavior, and to etiologists interested in polydrug use.