Title :  Relative efficiency of using summary and individual data in meta-analysis

Abstract: Meta-analysis is a statistical methodology for combining information from diverse sources to reach a more reliable and efficient conclusion. It can be performed by either synthesizing study-level summary statistics or modeling individual participant data (IPD), if available. The latter is traditionally viewed as the “gold standard” approach. However, it remains not fully understood whether the use of IPD indeed gains additional efficiency over summary statistics. In this talk, we discuss the relative efficiency of the two methods under a general likelihood inference setting. We show theoretically that there is no gain of efficiency asymptotically by analyzing IPD, provided that the random-effects follow the Gaussian distribution and maximum likelihood estimation is used to obtain summary statistics. Our findings are confirmed by simulation studies and a real data analysis of beta-blocker treatment effect for myocardial infarction. This is a joint work among Professors Dungang Liu, Xiaoyi Min and Heping Zhang.