Challenges and New Approaches to Merging Longitudinal or Clustered Studies

• Activity Number: 154
• Type: Invited
• Date/Time: Monday, August 5, 2013 : 10:30 AM to 12:20 PM
• Sponsor: International Chinese Statistical Association
• Author(s): Lu Wang*+ and Fei Wang and Peter X.K. Song
• Companies: Univeristy of Michigan and Wayne State University and University of Michigan
• Keywords: Data merging ; Estimation efficiency ; Generalized estimating function ; Heterogeneity ; Meta analysis

Abstract

In this talk, we present new approaches to tackle important challenges arising from merging multiple longitudinal or clustered studies. We first develop a rigorous hypothesis testing procedure to assess the validity of data merging, and then propose a flexible joint estimation procedure for the merged data that can account for different within-subject correlations and follow-up schedules in different studies. Large sample properties for the proposed procedures are established. Comparing with meta analysis and generalized estimating equations, our test provides robust control of Type I error against misspecification of working correlation structures and heterogeneous dispersion parameters. The joint estimating procedure leads to an improvement in estimation efficiency on all regression coefficients after data merging is validated. We also propose an ordering adapted fused Lasso approach to automatically detect common parameters across studies when the number of studies or covariates are large.