The case-time-control design is an extension of the case-crossover design capable of handling time trends in the exposure of the general population. Time-invariant confounders are controlled for by the design itself. The idea is to compare the exposure status of a person in one or several reference periods during which no event occurred with the exposure status of the same person in the index period where the event occurred. By comparing case-crossover results in cases to case-crossover results in controls, the exposure-outcome association can be estimated by conditional logistic regression. We review the mathematical assumptions underlying the case-time-control design and examine sensitivity to deviations from the assumed independence of within-individual exposure history. Results from simulating various scenarios suggest that the design is quite robust to deviations from this model assumption. In addition, we show that changes in exposure probability over time can be modeled in a flexible way using splines.
From the aDepartment of Biostatistics, University of Copenhagen, Copenhagen, Denmark; and bDepartment of Cardiology, Copenhagen University Hospital, Gentofte, Denmark.
The authors report no conflicts of interest.
The TRYG foundation (Denmark) supported this study.
Supplemental digital content is available through direct URL citations in the HTML and PDF versions of this article (www.epidem.com). This content is not peer-reviewed or copy-edited; it is the sole responsibility of the author.
Correspondence: Aksel Karl Georg Jensen, Department of Biostatistics, University of Copenhagen, Oster Farimagsgade 5, Box 2099, DK-1014 Copenhagen, Denmark. E-mail: email@example.com.