The emergence of accreditation bodies will introduce new organizations with important roles to play in learning from lapses in the system to ensure continuous improvement, making it all the more important to spell out the roles and responsibilities of different parties when serious infractions come to light.
Large multicenter clinical trials now routinely include formal data safety and monitoring boards (DSMBs). DSMBs were initially established to assist research sponsors with analysis of their data, but their importance in assessing risk and monitoring safety has become apparent. Such boards are typically composed of researchers with expertise similar to that of the principal investigators, but they come from independent research institutions and are augmented by statisticians, bioethicists, and sometimes lawyers and consumers. The only personnel requirements for NIH DSMBs are that they include expert clinicians and experts in biometrics or statistics. These monitoring boards receive reports of study outcomes, including both intended effects and adverse events. They pool findings from multiple centers (findings which the individual centers often do not receive and to which only research sponsors would otherwise have access). DSMBs may stop a trial if it appears to be causing harm or if its study objective is met early. A DSMB can also become the locus for receiving reports of mishaps and complaints, as well as adverse events and research outcomes.
NIH has recently mandated that any NIH-sponsored clinical trial have a research monitoring plan and that the plan take into account the level of risk (NIH, 2000). The National Cancer Institute has mandated that any phase III trial (a large trial, typically conducted at many centers, intended to demonstrate the efficacy of an intervention) have a DSMB (NCI, 1999). The inclusion of such boards has been standard practice in most trials sponsored by private industry to test new drugs, devices, or biologics. The Good Clinical Practice portion of the International Conference on Harmonisation guidelines that govern clinical trials has an entire section (section 5.18) devoted to monitoring (International Conference on Harmonisation of Technical Requirements for Registration of Pharmaceuticals for Human Use, 1996, pp. 26–29). The connections between DSMBs and IRBs are not completely consistent, however. Although all DSMBs are accountable to research sponsors for the integrity of the data, their role in ensuring safety and in protecting research participants is less well articulated. They are not always clearly accountable to IRBs, and their responsibilities to research participants or groups representing the interests of research participants are sometimes not explicit.
The interaction between accreditation bodies and the organizations that they accredit can indicate new strategies for improving performance. Over the past three decades the constant lament of dozens of reports from a half dozen knowl-