Case-crossover and case-time-control designs as alternatives in pharmacoepidemiologic research

Area of Science:

• Pharmacoepidemiology
• Biostatistics
• Drug Safety Research

Background:

• Standard cohort and case-control designs are suitable for chronic exposure effects but susceptible to confounding by indication.
• Automated databases have increased but cohort studies remain time-consuming and expensive.
• Case-control studies face challenges in control selection, especially for rare exposures.

Purpose of the Study:

• To present a structured decision table for selecting appropriate pharmacoepidemiologic research designs.
• To compare the strengths and weaknesses of cohort, case-control, case-crossover, and case-time-control studies.
• To guide researchers in choosing designs that minimize confounding by indication and address transient exposure effects.

Main Methods:

• Comparison of standard cohort and case-control designs with case-crossover and case-time-control designs.
• Analysis of susceptibility to confounding by indication and suitability for intermittent versus chronic exposures.
• Evaluation of designs for time-critical decisions and handling of rare exposures.

Main Results:

• Case-crossover and case-time-control designs are less susceptible to confounding by indication.
• Case-crossover studies effectively examine transient exposures and acute events, using the case as their own control.
• Case-time-control designs refine case-crossover by incorporating control group data to adjust for temporal prescribing changes.

Conclusions:

• Case-crossover and case-time-control designs are preferred for separating acute from chronic effects of transient exposures.
• These designs are particularly valuable when confounding by indication is a significant issue in pharmacoepidemiologic research.
• The choice of design depends on the nature of the exposure (chronic vs. intermittent) and the research question regarding effect timing.