Methods for Exploring Heterogeneity in Systematic Reviews of Randomized Controlled Trials

This thesis consisted of three major components: 1. A sample of randomized controlled trials of herbal medicines was collected and assessed with a recently developed extension of the CONSORT statement for herbal medicine trials. 2. A methodological review of proposed methods of assessing clinical heterogeneity in meta-analyses of randomized controlled trials, 3. The application of permutation based resampling in meta-regression of multiple covariates. An analysis of 406 RCTs of herbal medicine interventions revealed that these trials are regularly under reporting important aspects of the intervention. Next, the second project showed that there are many resources providing suggestions for investigating clinical heterogeneity in systematic reviews of controlled clinical trials and though there is minimal consensus some recommendations are common across sources. Finally, the third project found that permutation tests result in more conservative, larger, p-values potentially reducing the rate of false positive findings when exploring multiple covariates.