Finding a successful approach
At Statistical Revelations we stay on top of industry trends and developments. We have experience in meta-analysis, using both fixed effects and random effects models. Meta-analysis is an expanding area of medical statistical endeavor, where the results from several independent studies are statistically combined to produce an estimate of effect. This is essentially a non-biased form of literature review.
Meta-analysis is used in several areas of clinical development, the most commonly known being re-imbursement. Meta-analysis can also be a useful tool early in development to determine the state of current knowledge in a particular disease area and gain an understanding of anticipated benefits from treatment.
Meta-analysis can be performed by non-statisticians in many cases, however, not all areas of research lend themselves readily to the “ideal” randomized, blinded parallel group design which is frequently stipulated to be a requirement for meta-analysis. At Statistical Revelations we have experience in performing meta-analysis where the “non-ideal” study designs can also be incorporated to ensure that every possible study is used in the final modeling. By incorporating a large number of studies, methods such as meta-regression are possible and can prove to be useful.