In Reply: We agree with Dr Kong and colleagues that the evidence used by regulators should represent best available science. In our view, however, researchers are not presenting the best available science. For the highest-risk medical devices, only 14% of studies were blinded, 27% randomized, and only half used controls (31% of which were retrospectively selected). In addition, 88% of primary end points in the studies were surrogate outcomes, and large numbers of enrolled patients were frequently excluded from safety and efficacy analyses.
Redberg RF, Dhruva SS, Bero LA. Evidence Strength in FDA Premarket Approval of Cardiovascular Devices—Reply. JAMA. 2010;303(16):1599-1601. doi:10.1001/jama.2010.512