Course Level: Beginner
Format Requirements: Didactic lectures and interactive discussion of theory, potential confounders and limitations, and statistical methods, using case study examples from published medical literature.
Background: Meta-analysis is a formal, systematic method to synthesize the results of independent studies, considering and integrating the combined weight of evidence to determine the effect of an intervention. Meta-analysis is being used increasingly in the medical and health sciences to inform and guide practice and policy, in areas as disparate as estimating the effectiveness of mammography in detection of breast cancer and the consistency of gene-disease association studies. A Google Scholar search on meta-analysis identified 589,000 hits in medicine, 293,000 in health policy, and 102,000 in genetics. The information explosion in almost every field coupled with the movement towards evidence-based decision making and cost-effective analysis has catalyzed development of more rigorous procedures to synthesize the results of independent studies.
Description and Objectives:
- Understand the potential value of and theory underlying the conduct of meta-analysis of independent studies
- Understand conditions under which meta-analyses can be performed and common factors that limit or confound the meta-analysis conduct and interpretation.
- Learn and understand a range of statistical methods for analyzing and interpreting meta-analysis studies.
- Introduce the concept of network meta-analysis
This workshop will provide an historical perspective of meta-analysis, and discuss methodological issues such as various types of bias and heterogeneity on the conduct and interpretation of meta-analyses. There will be extensive discussion of the appropriateness and use of statistical methods for combining data across studies, including nonparametric and parametric models; effect sizes for proportions, fixed versus random effects, regression and ANOVA models; multivariate models for proportions and standardized mean differences, treatment of zero cells, models with missing data, and special methods and issues in genetic applications.