Using Meta-analysis to Answer Clinical Questions—Reply

In Reply: We agree with Dr Laheij that exploration of heterogeneity sources may provide valuable information, especially for controversial medical topics for which therapeutic or preventive effects of different magnitude are obtained by different study designs.¹ We also agree that empirical meta-analytic evaluations can be applied in other types of studies as well.

Dr Theodoropoulos and colleagues reduce meta-analysis to a method that simply lumps information and biases. We do not share this view. Meta-analysis has the potential to systematically examine the strengths and weaknesses of the accumulated evidence, explore heterogeneity between studies (including those of different designs), and identify potential biases. As a form of original research, meta-analysis does not compete with or substitute for other original work. It functions at a different level than single original studies and has been widely accepted as the highest level in the hierarchy of evidence.^2- 3 We welcome criticism of meta-analysis, which may help refine further its methods. However, when a meta-analysis shows that no more research is needed or that there are overt biases that should be appropriately corrected, then it is equivalent to misconduct to perform further clinical studies or to conduct new studies without correcting known biases.⁴