Modern Epidemiological Study Designs

Although the prototypical non-experimental study designs to measure relative incidence conditional on putative risk factors (cohort, case-control) have been foundational to modern epidemiology, a growing recognition of their inherent limitations have impelled methodologists to develop a new generation of designs to surmount them. The premise underlying some of these is the recognition that validity can be advantaged by comparing individuals to themselves under a theoretical duality of exposures and hence a duality of potential outcomes. This has in turn led to the genesis of observational case-only studies, some of which are introduced in this chapter, including case-crossover studies, case-time-control studies, self-controlled case series studies, and case-specular studies. In addition, case-control studies have advanced far beyond their original model of comparing cases to non-cases, and the most bountiful advances have come through an improved understanding of how control selection not only impacts interpretation of effect measures, but also how it can be used both for enhanced validity and precision. This applies to case-cohort studies and nested case-control studies, which can be used to estimate both relative measures of risk and incidence rates. In addition, sampling cases and non-cases within covariate strata and then applying adjustments for stratum-specific weights represent another ongoing area of development, as does incorporation of information from the non-sampled members of the cohort. These and other studies will take on growing prominence as greater challenges in measuring exposure effects on health, such as in environmental epidemiology and pharmacoepidemiology, are in need of solutions.