Data sources on drug safety evaluation: a review of recent published meta-analyses

Meta-analysis is a quantitative approach to summarize the findings from several studies and has been applied with increasing frequency to clinical trials. Because of their sample size and duration limitations, experimental studies (ESs) could not be able to detect late or rare adverse events (AEs), which may be identified in well-designed observational studies (OSs). This study aims to identify and analyze meta-analyses from both ES and OS where safety was found to be an outcome measure.

The meta-analyses inclusion criteria was established as at least one AE as primary outcome. Safety outcomes were considered as the increase in the risk for an AE after a pharmacological intervention. A MEDLINE search for meta-analyses published in the New England Journal of Medicine, The Lancet, Journal of American Medical Association, British Medical Journal, Annals of Internal Medicine, PLoS Medicine, Annual Review of Medicine, and Archives of Internal Medicine, between October 2005 and September 2010, was carried out.

Sixty meta-analyses met the inclusion criteria. Of these, 53 included only ES, 4 included both ES and OS, and 2 included only OS. Of the 6 meta-analyses that included OS, 4 included cohort and case-control studies, and 2 included cohort, case-control, and cross-sectional studies. One meta-analysis did not report the type of studies included.

Experimental studies were found to be the main source of meta-analyses on drug safety. The role of meta-analyses in pharmacovigilance is a matter of ongoing debate, and efforts are being made to develop guidelines on the use of meta-analysis in drug safety assessments, to better combine evidence about harms.