|Thursday, February 23|
|PS1 Poster Session 1 and Opening Mixer||
Thu, Feb 23, 5:30 PM - 7:00 PM
Conference Center AB
Good Statistical Practices: An Example of Meta-Analysis of Odds Ratios (303446)*Bei-Hung Chang, University of Massachusetts Medical School
David C Hoaglin, University of Massachusetts Medical School
Keywords: Meta analysis, odds ratio, statistical software
Statistics is a fast growing field. Many new statistical methods have been developed and available in commonly used software. However, many users are not aware of these methods and continue to use conventional methods that are not reliable. We use meta-analysis as an example to demonstrate a good statistical practice by applying an alternative method that is more reliable than conventional approaches. We focus on the case when odds ratio is used for intervention effect in randomized clinical trials. The conventional methods of estimating an overall odds ratio involve weighted averages of the individual trials’ estimates of the logarithm of the odds ratio. This approach has several shortcomings, arising from assumptions and approximations, that render the results unreliable. A well-developed alternative approach avoids the approximations by working directly with the numbers of subjects and events in the arms of the individual trials. We use an example of 19 randomized trials and demonstrate the use of SAS, Stata, and R software for the analysis. The results from the two approaches differ in ways that can affect inferences. The use of the alternative method is recommended.