Methodological Challenges in Observational Research: A Pharmacoepidemiological Perspective

Pharmacoepidemiology is the science of studying the effects of utilizing pharmaceutical and biological products in the population, and in most cases this science is conducted in observational epidemiological designs, including retrospective database analysis. Observational pharmacoepidemiology is associated with a myriad of methodological challenges that affect study conclusions and related causal inferences. However, if these challenges are addressed and effectively dealt with, observational studies can have important and impactful clinical, regulatory, and public health outcomes. This article examines common challenges in retrospective database analysis and serves as an introductory text to important methodological concepts in research involving medication use, including confounding by indication, time-dependent confounding, informative censoring, depletion of susceptibles, and immortal time bias.