Abstract by Nolan Cole
Improving Clinical Trials Through Meta-Analysis: Estimating Heterogeneity in Meta-Analysis for Binary Outcomes.
Meta-analysis is a statistical procedure that combines data from multiple studies. This is particularly useful in clinical research when multiple studies produce conflicting results or when reaching a definitive conclusion on the effectiveness of a medication. Determining the extent to which these studies differ from one another due to differences in treatment administration and patient populations is key to identifying treatment efficacy. Estimation of the differences the studies demonstrate is known as heterogeneity. Estimating the heterogeneity test statistic tau for meta-analysis is key to ensuring the validity of clinical research. We investigate the ability of available heterogeneity methods for dichotomous data by analyzing data from multiple clinical trials; we have found that specific methods of estimating heterogeneity outperform others while a select few consistently underperform. Understanding which method performs best will allow clinical researchers to better estimate heterogeneity, thereby improving their ability to find new treatments.