In a commentary published in the December 2, 2009 issue of the Journal of the American Medical Association, Charles H. Hennekens, M.D., the first Sir Richard Doll Research Professor in the Charles E. Schmidt College of Biomedical Science at Florida Atlantic University, said, "To detect the most likely small to moderate benefits or harm of drugs requires randomized trials of sufficient size and duration. Overreliance on small trials, meta-analyses or subgroup analyses can lead to misleading conclusions."
In the commentary, Hennekens notes that 35% of the time, meta-analyses did not accurately predict outcomes of the large randomized trials. "If the totality of evidence is incomplete, it is appropriate to remain uncertain," said Hennekens.
Hennekens also emphasizes in the commentary that small trials, meta-analyses of small trials and subgroup analyses are useful mainly to formulate a hypothesis, but not to test it. To make the most reliable clinical decisions for individual patients and policy decisions for the health of the general public, a sufficient totality of evidence should generally include some large scale randomized phase 3 trials with high adherence and enough clinical endpoints to distinguish reliably between no effect and moderate benefit or, on occasion, harm.
Professor Nancy Ip: Pioneering New Paths in Neurodegenerative Therapy