State of the art and science associated with the use of statistical models used in the meta-analysis of clinical data.

This meeting will be held at the MBC Offices, located on the ninth floor of One Cambridge Center, Cambridge, MA 02142.

Schedule

8:00-8:20 Networking and Breakfast

8:20-9:30 Presentation and Q&A

9:30-10:00 Networking

TOPIC

State of the art and science associated with the use of statistical models used in the meta-analysis of clinical data.

ABSTRACT

Meta-analysis, the quantitative analysis of data collected in a systematic review of the literature, is a key part of evidence based medicine. It consists of a set of statistical techniques for combining information from the published results of clinical studies to summarize treatment efficacy or diagnostic accuracy. Although meta-analysis has traditionally provided a single summary estimate of effect, many reviews suggest substantial heterogeneity of results which an appropriate summary of the information should not ignore. The power of meta-analysis lies then not just in enabling the use of more data to increase precision of a single number summary, but also in informing how efficacy or accuracy varies across different groups of individuals. In turn, this promotes targeting the use of drugs and devices to those for whom they are most effective. This talk will discuss a number of statistical techniques used in meta-analysis, starting with an overview of the general idea of weighted averages and proceeding to an overview of techniques for appropriately incorporating sources of heterogeneity and variation that characterize data from systematic reviews.

SPEAKER

Dr. Schmid is Director of the Biostatistics Research Center in the Institute for Clinical Research and Health Policy Studies at Tufts-NEMC and Professor of Medicine at Tufts University School of Medicine and of Clinical Research in the Sackler School of Graduate Biomedical Sciences.

Biography for Dr. Schmid

Dr. Schmid is Director of the Biostatistics Research Center in the Institute for Clinical Research and Health Policy Studies at Tufts-NEMC and Professor of Medicine at Tufts University School of Medicine and of Clinical Research in the Sackler School of Graduate Biomedical Sciences. He directs the biostatistics/epidemiology concentration in Clinical Research. He received his PhD in statistics from Harvard in 1991 and his BA in Mathematics from Haverford College in 1983. Major research interests include development and application of Bayesian models, statistical methods and computational tools for meta-analysis of diagnostic tests and clinical efficacy, methods for combining and analyzing multiple databases and methods for handling missing time-dependent data in longitudinal studies. Examples of recent work include the use of hierarchical Bayesian models to explain heterogeneity in meta-analyses of efficacy and diagnostic test studies, meta-analysis of community-based N-of-1 trials, hierarchical models for saltatory growth, Bayesian approaches to sample size calculations, development of prediction equations for the glomerular filtration rate, clinical and genetic studies of chronic Lyme disease patients, and missing data methods in predictive models. Dr. Schmid serves on several advisory boards, grant review panels and data safety monitoring committees and consults extensively on clinical research with industry, academia and government.