DSMBs are formal independent committees generally used for more complex, possibly higher-risk, multisite research. DSMBs generally include an experienced statistician, medical experts from the setting being studied, and an expert in the ethical and regulatory aspects of the research. In pragmatic clinical trials, study designs, such as cluster randomization and stepped-wedge designs, are often used. The DSMB statistician should have experience with the challenges these designs can present. In addition to clinicians, in pragmatic clinical trials, medical experts may include bioinformaticians and healthcare systems experts who understand relevant institutional operations. In addition, it may be important to include a patient representative (Flynn et al 2013).

Given the focus on patient-centered outcomes in [pragmatic clinical trials], the role of a patient representative [on data monitoring committees] may be particularly important. — Ellenberg et al 2015

DSMBs have formal charters that detail their logistics, responsibilities, and operations. Before the start of the study, the DSMB must review and approve the protocol and, if there is one, the consent form. The DSMB and the study team must agree on how often information about the study (such as interim results and adverse events) will be reported to the DSMB and what results or events would trigger consideration of early termination of the study. The Ethics and Regulatory Core Working Group of the NIH Pragmatic Trials Collaboratory developed a template charter for data monitoring committees in pragmatic trials.

There are several factors to consider in determining when a DSMB is the preferred approach to data and safety monitoring. In general, a DSMB is appropriate if participant safety requires regular interim review of safety and efficacy data. Use of a DSMB has also been encouraged when a study may have a substantial public health impact and ongoing monitoring of the validity and reliability of the data would be beneficial.

In addition, research sponsors may require the use of a DSMB as a condition of a grant award. For example, NIH policy requires most NIH-supported randomized clinical trials to have a DSMB if they are a multicenter trial and pose any material risk to participants. Likewise, the Patient-Centered Outcomes Research Institute (PCORI) mandates that a DSMB be appointed in any trial it sponsors if required by the IRB or a regulatory agency or if it is determined to be appropriate "after considering factors such as potential risks; target study subject population, nature, and size; and the research project's scope and complexity."

Although the Food and Drug Administration (FDA) is rarely involved in pragmatic clinical trials, the following modified FDA guidance provides a helpful list of criteria for when to consider a DSMB:

• Is the trial a large, multicenter study of long duration?
• Is the study endpoint such that a finding in the interim analysis might ethically require termination of the study before its planned completion?
• Are there a priorireasons for a particular safety concern (for example, a particularly invasive treatment?)
• Is there prior information suggesting the potential for serious toxicity due to the study treatment?
• Is the study being performed in potentially scientifically complex or vulnerable populations (such as children, pregnant women, very elderly patients, terminally ill patients, those with diminished mental capacity)?
• Is the study being performed in a population at elevated risk of death or other serious outcomes, even when the study objective addresses a less serious endpoint?

If a study has one or more of these characteristics, the FDA recommends that sponsors consider the use of a DSMB to further protect study participants.

Investigators conducting pragmatic clinical trials should work with the study sponsor to implement an appropriate data monitoring plan and determine whether a DSMB will be used. The NIH Pragmatic Trials Collaboratory has included brief descriptions of the data monitoring plans for all of its NIH Collaboratory Trials as part of the publicly available ethics and regulatory documentation for these trials. The data monitoring details are excerpted in the table below. Not all of the NIH Collaboratory Trials use an independent DSMB, but all have a data monitoring plan that was discussed with the NIH sponsor and determined to be in compliance with relevant policies.

Table. Approaches to Data Monitoring in the NIH Pragmatic Trials Collaboratory NIH Collaboratory Trials

Source: NIH Collaboratory Trial ethics and regulatory documentation. For more information about the NIH Collaboratory Trials, including study populations and primary outcomes, see the NIH Collaboratory Trials table in the What Is a Pragmatic Clinical Trial? chapter of the Living Textbook.

In the sections that follow, we review special considerations for data monitoring in pragmatic trials, including monitoring protocol adherence when information on "real-world" use is desired, issues associated with use of electronic health record (EHR) data such as data quality and timeliness, complexities of monitoring adverse events in pragmatic trials, and whether pragmatic trials should ever be stopped early. These considerations may be useful with DSMBs and other types of data and safety monitoring approaches.