Back to Annual Meeting
Back to Annual Meeting
|
Back to Annual Meeting
|
|
Back to Annual Meeting
|
APHA Scientific Session and Event Listing |
3029.0: Monday, November 06, 2006: 8:30 AM-10:00 AM | |||
Oral | |||
| |||
| The three most widely discussed biases in epidemiologic research are those due to confounding, selection effects, and measurement error. Methods for minimizing bias from these sources—e.g., stratification and modeling for confounding—are taught in basic methods courses. However, traditional methods may not be appropriate in many common applications. In this session, the speakers will describe the causal origins of bias due to confounding, selection, or measurement error and will present emerging methods that use causal knowledge to assess and control bias. The first speaker will give an introduction to directed acyclic graphs (DAGs), which will be used by all three speakers to illustrate the causal structure underlying their analyses. In addition, she will present an application of instrumental variable techniques to the control of confounding. The second speaker will discuss measurement error from a structural standpoint, and will likewise show how instrumental variables can be used to control bias. The third speaker will discuss how two-phase designs (e.g., a case-control study nested within an enumerated cohort), which were developed to efficiently collect high-quality exposure and covariate data, can be subject to selection/response bias if the two-phase nature of the data is ignored. By using a two-phase analysis, the estimates will be less biased and more precise compared with those from conventional analyses of the second-phase data alone. | |||
| Learning Objectives: At the conclusion of the session, the participant (learner) in this session will be able to: 1. Describe the causal origins of bias due to confounding, selection effects, and measurement error using directed acyclic graphs. 2. Discuss three non-traditional analytic approaches to the control of these biases. | |||
| Katherine J. Hoggatt, PhD | |||
| Tim Lash, ScD | |||
| Introductory Remarks | |||
| Using Instrumental Variable Estimators to Control Confounding M. Maria Glymour, ScD | |||
| A Structural Classification of Measurement Bias Miguel Hernan, DrPH | |||
| Controlling for Selection/Response Bias in Two-Phase Studies Katherine J. Hoggatt, PhD | |||
| Discussion | |||
| Questions and wrap-up | |||
| See individual abstracts for presenting author's disclosure statement and author's information. | |||
| Organized by: | Epidemiology | ||
| CE Credits: | CME, Health Education (CHES), Nursing | ||
The 134th Annual Meeting & Exposition (November 4-8, 2006) of APHA