Emergency and Trauma Nursing

When judging the effects of treatment, health-care providers face the critical task of distinguishing truly causal relationships apart from mere (non-causal) associations. Unfortunately, the task of identifying ‘biasing pathways’ that alter the true effect of an exposure is often a daunting and frequently overlooked adventure. As a result, today’s health-care provider faces a bewilderment of seemingly contradictory reports, some of which appear in highly regarded medical journals. Epidemiologic studies are commonly constructed around three types of variables: Exposure, outcome and a ‘third’ variable that carries the potential to bias the exposure-outcome relationship. The recent merger of graphical probability theory with established methods for constructing causal diagrams has led to the creation of sophisticated yet highly intuitive tools for establishing causal inference. This seminar will provide an update on contemporary methods for identifying common threats to the validity of a clinical investigation. The speaker will argue that health care providers by virtue of their education and training are uniquely qualified to provide the expert knowledge requisite for establishing

causal inferences. Participants should be able to meet the following course objectives:

1. Define confounding from the traditional perspective and from the (modern) structural alternative.

2. Identify sources of confounding and selection bias given a directed acyclic graph that accurately conveys the

expert knowledge of the investigator.

3. Create strategies to adjust or remove biasing pathways responsible for creating confounding and selection bias.