Regulators rallied stakeholders to discuss best practices for the conduct and assessment of meta-analyses.

Credit: Garry Gay/Alamy

The lowdown: Meta-analyses offer a unique opportunity to combine and analyse data from multiple studies and draw inferences from populations of interest. But despite their widespread use, the US Food and Drug Administration (FDA) says that there is a lack of consensus on the best statistical practices and design of these studies, which undermines their utility. The FDA therefore held a workshop in November to establish a robust and transparent approach to meta-analyses. A resulting guidance document “will provide a consistent framework for how meta-analyses should be designed, analyzed, reported, and interpreted in the context of product safety regulation,” said the agency.

Key topics covered at the workshop included bias and multiplicity in meta-analyses, the hierarchy of evidence, the need for prospective planning, the quality of constituent studies and the statistical persuasiveness of generated evidence. The agency also presented its perspective in a white paper published before the workshop. Participants at the workshop discussed the possibility of a “refuse-to-file” action for submitted meta-analyses that do not meet adequate FDA standards.

On the same day as the workshop, the FDA retracted some of the restrictions on GlaxoSmithKline's antidiabetic drug rosiglitazone. A meta-analysis published in 2007 suggested that the peroxisome proliferator-activated receptor-γ modulator increases the risk of cardiovascular events, triggering a high-profile investigation into rosiglitazone. The agency subsequently restricted access to the drug, but has revised its position in light of a re-review of data from a clinical trial known as RECORD.