"There aren't enough small numbers to meet the many demands made of them."
Guy, Richard K. (October 1988)
Meta-analysis, or analysis of analysis, is a statistical method for contrasting and combining results from different studies. This is often particularly important when individual studies and their conclusions are based on small numbers. Larger numbers, obtained by combining studies, reduces chance findings and permits analyses of important subgroups. Meta-analysis has the advantage of using all previously published information, but the ‘‘meta-analyst’’ is stuck with analysis decisions made by the original authors, which often differ across studies.
In contrast, pooled analysis uses raw data from previous studies, and thus, allows the “pooled-analyst” to apply identical analyses to all included studies. However, pooled analysis is more labor intensive and requires access to the original data from the studies. Pooled analysis is considered the gold standard for synthesizing results from multiple studies. It allows for comparison across different studies and metrics, free of artifacts introduced by analytic differences, and allows for derivation of statistically more stable results. The choices of cut points, reference groups, metrics, etc., in a pooled analysis may differ from the choices made in the original studies and may result in changes in the study-specific effect estimates. However, results from both meta- and pooled analyses, are prone to the same biases that might have been operating in the original studies.
Use of meta- and pooled analyses in medicine has skyrocketed to overcome limitations of small sample sizes in clinical research. In this talk, I will present how such analyses are conducted, including some medical insights gained from these approaches.