Meta-analysis

Abstract

Meta-analysis is a process to estimate a summary statistic from all available studies. The statistic may be a difference in survival rate, hazard ratio, or others. To calculate a summary statistic, a method using variance by Taylor series, a Mantel-Haenszel method, Peto method, or DerSimonian-Laird method are used. Although they all give a close estimation; the DerSimonian- Laird method is most commonly applied. This method is based on a random effect model with the assumption that an estimate from each study distributes normally around a mean derived from whole studies.