|
Abstract:
|
The "bench-to-bedside" translational researcher is often a PhD trained in the basic sciences, a world steeped in the experimental method. Here, "like" individuals are randomized to study conditions set by the investigator and control of confounding is presumed by randomization. Discovery of effect modification is, at most, an exploration of biological synergism defined by study design. Appropriately, statistical training emphasizes analysis of variance, dose-response curve fitting, and the modeling of response surfaces, more generally. "Bedside" research, in contrast, often necessitates the use of observational study protocols and the measurement of dissimilar individuals. Statistical training of translational researchers might benefit from additional pedagogy in conditional likelihoods, epidemiologic study designs, control of confounding, and discovery of effect modification.
|
