Kim HM (HSR&D Center of Excellence, Ann Arbor VA), Zhou XA
(HSR&D Center of Excellence, VA Puget Sound Health Care System)
Although use of observational data may be more cost- and time-efficient, it suffers from selection bias where the intervention of interest is often provided for those who have an indication for the treatment, for example, more severely ill. The ultimate objective of this workshop is to introduce and discuss different methods in handling selection bias in using observational data where comparative inference is of interest. Statistical methods will include propensity stratification and weighing as well as instrumental variable analysis.
We will present statistical methods for analysis of observational data using VA administrative data. We will briefly describe available software for analysis of observational data and discuss how these methods are implemented. We will present case studies to illustrate how these methods are employed in practice. The strengths and limitations of potential approaches will also be described.
The workshop is intended for health services researchers and statisticians who are interested in understanding the issues in the use of VA administrative data to make comparative inferences and the methods available to address these issues.
Assumed Audience Familiarity with Topic:
The workshop is designed for the broad health services research audience with moderate levels of statistical understanding.