20 May Flawed meta-analyses can cause confusion

Well-conducted meta-analyses can inform public health policy. They make information accessible for decision-makers, in a summary format that provides a critical overview of the evidence. Poorly performed meta-analyses can mislead decision-makers when, for example, they exclude some studies on purpose, introduce bias, alter original data, fail to control for confounders, and use methods that cannot be replicated by other scientists. Although labour-intensive, it is important for scientists to review the detailed methods used by researchers who present meta-analyses that produce counterintuitive or contrary results and to assess the merit of their conclusions. We set out to do this for a meta-analysis that concluded that male circumcision increases the risk of various common sexually transmitted infections. Our in-depth analysis of the methods results in a clear rejection of the findings and conclusions of the original meta-analysis on multiple grounds. An extensive and robust body of scientific literature underpins the widely acknowledged and accepted conclusion that male circumcision reduces the risk of many sexually transmitted infections including herpes simplex-2, syphilis, chancroid, oncogenic HPV (the cause of cervical cancer in women and penile cancer in men), and HIV infection.

Click here to read the full article