For trials of interventions that may entail the possibility of significant harm, as well as benefit, to participants, the trial sponsor should establish a DSMB—sometimes termed a committee (DSMC), a Data Monitoring Board (DMB) (or Committee (DMC)), or Independent Data Monitoring Committee (IDMC). The DSMB is independent of those conducting the trial and separate from the ethics review committee (ERC) to monitor the safety of the trial, while it is being conducted. Not all trials will require a DSMB, but listed in Box 7.1 are the types of trial for which WHO has recommended that it would be considered desirable to set up such a committee (World Health Organization, 2005).
A DSMB will usually be set up for trials which are double-blind, in which the investigators and sponsor do not know which intervention individual participants have received, but the DSMB will have access to the randomization code, which it can break during the course of the trial for specific reasons, including safety concerns or interim analyses (see Section 4.1.3). For trials where the intervention allocation is not blinded, the investigators and the sponsor can assess on a continuous basis if there is an excess of AE s in one of the intervention arms of a trial. Even so, it is usually a good idea to have a DSMB, as this committee may take the responsibility for advising the PI, steering committee, and sponsor on critical decisions, such as whether to stop a trial because of adverse effects or signs of failure of the intervention during the course of a trial. Although members of DSMBs are often not paid for their services, a budget will still be required to cover their meetings and any visits they may need to make to the trial site(s).
In this section, we outline the functions and responsibilities of a DSMB, the selection of members, the major issues with which it has to deal, and lines of reporting to those involved in the trial.
Box 7.1 WHO recommendations for the types of trial for which a DSMB is relevant
- Controlled studies with mortality and/or severe morbidity as a primary or secondary endpoint.
- ◆Randomized controlled studies focused on evaluating clinical efficacy and safety of a new intervention intended to reduce severe morbidity or mortality.
- ◆Early studies of a high-risk intervention (risk of non-preventable, potentially life-threatening, complications; or risk of common, preventable AEs of interest (especially adverse drug reactions)), whether or not randomized.
- ◆Studies in the early phases of a novel intervention, with very limited information on clinical safety or where prior information raises concern regarding potential serious adverse outcomes.
- ◆Studies where the design or expected data accrual are complex or where there may be ongoing questions with regard to the impact of accrued data on the study design and participants’ safety, particularly in studies of a long duration.
- ◆Studies where the data justify an early termination such as the case of an intervention intended to reduce severe morbidity or mortality, which might turn out to have adverse effects or lack of effect, resulting in increased morbidity or mortality.
- ◆Studies carried out in emergency situations.
- ◆Studies which involve vulnerable populations.
Reproduced with permission from the World Health Organization, Operational Guidelines for the Establishment and Functioning of Data and Safety Monitoring Boards, Copyright © World Health Organization on behalf of the Special Programme for Research and Training in Tropical Diseases 2005, available from <http://whqlibdoc.who.int/hq/2005/TDR_GEN_Guidelines_05.1_eng.pdf>. This box is not covered by the Creative Commons licence terms of this publication. For permission to reuse please contact the rights holder.
4.1 The functions of a Data and Safety Monitoring Board
The prime function of the DSMB of a trial is to safeguard the welfare of participants in the trial. A key aspect of this is to monitor the occurrence of adverse events (AEs) by trial arm and to recommend action to the investigators in the event of finding evidence of harm. In ‘blinded’ trials, the DSMB will be the only group to have access to the randomization codes during the conduct of the trial, so that, if necessary, they can ascertain which intervention an individual participant received. The DSMB may also be called upon to ‘break the code’ for a trial at pre-specified time points to make a recommendation as to whether or not a trial should be stopped prematurely because of ‘overwhelming efficacy’ or ‘futility’.
4.1.1 Monitoring the conduct of the trial
An important aspect of safeguarding the welfare of trial participants is to check that trial procedures are being followed, according to the protocol, and there are no significant deviations from the trial plan. If there is a trial steering committee, responsibility for monitoring trial conduct will lie principally with that committee, and the DSMB should receive reports from the steering committee—often the chair of the DSMB attends all or part of the steering committee’s meetings. However, if there is no steering committee, then the responsibility for monitoring trial conduct falls more heavily on the DSMB. Usually, this function is satisfied by the DSMB receiving detailed reports from the investigators on the progress and conduct of the trial at each DSMB meeting, but DSMB members may also make visits to the trial sites. Day-to-day monitoring of trial procedures, including data collection, is often provided by clinical trial monitors (see Chapter 16, Section 7.2) who should report regularly to the sponsor. If there are issues of concern, the sponsor has responsibility for reporting these to the DSMB, through the steering committee if there is one. Clearly, monitoring the conduct of a trial can be a major undertaking, and exactly what part the DSMB is expected to play in this should be detailed in the DSMB Charter (see Section 4.3).
4.1.2 Monitoring the safety of trial participants
Different kinds of AE should be reported to the DSMB on an ‘immediate’ or regular basis, the frequency depending on the seriousness of the AE.
Any serious AEs (SAEs) (see Chapter 13) that are judged by the investigators to be likely to have been due to the intervention (‘potentially intervention-related SAEs’) should be reported very quickly to the DSMB, within a few days of their occurrence or their first notification to the trial investigators. Similarly, any deaths among trial participants, whether or not judged related to the intervention, should be immediately reported to the DSMB. As much relevant detail as possible should be provided to the DSMB about the nature and circumstances of the death or SAE, along with a cumulative update of all such events. Deaths and potentially intervention-related SAEs would normally be reported to the sponsor at the same time as being reported to the DSMB. When such events are reported, the chair of the DSMB should communicate with members to ascertain if any members consider the events are sufficiently serious and linked to the intervention that they require further investigation or, in extreme cases, might require that the trial is paused or terminated. If major changes to the conduct of the trial are suggested, the committee would generally meet by telephone conference or have a face-to-face meeting.
For SAEs that are not considered by the investigators to be directly related to the intervention, the DSMB should be informed of these on a regular basis, possibly monthly, depending on the size of the trial. At regular meetings of the DSMB, the accumulated SAEs should be considered. They should be classified by the type of SAE (for example, by hospital admission diagnosis), when they occurred in relation to the application of the intervention, and whether they affected participants in the intervention or control group. Displaying data by intervention and control groups requires breaking the code, and this is usually done either by an independent statistician (i.e. not one of the investigators, but a statistician contracted by the sponsor to perform the analyses) or by a statistician on the DSMB. Ideally, the tabulations should be presented to the DSMB by trial arm, without specifying what each arm has received intervention or control, and they should consider whether they are concerned by the size of any relative excess of events in any of the trial arms. Only if the answer to this question is ‘yes’ should they ask which arm was which. This procedure is adopted to avoid unnecessarily exposing the DSMB to unblinded data, unless there is good reason for concern, and to avoid their being biased in their assessment of the distribution of SAEs by knowing which arm any excess is in.
For more minor AEs, such as a minor local reaction to a vaccine or mild nausea, the DSMB can be presented with analyses of these on an occasional basis, though usually this should happen at least once a year. Again, they should initially be presented without identification of what each arm has received. Data on AEs are usually presented for information, rather than action, though occasionally action might be considered if there is a substantial excess of AEs in the intervention arm.
4.1.3 Conducting interim analyses
In some trials, a plan is made to examine interim efficacy results before the trial’s expected end date. There are two main reasons for doing this. First, if the intervention proved much more effective than anticipated, then there might be grounds for terminating the trial early on the basis of ‘overwhelming efficacy’, such that it might not be considered ethical or necessary to continue with a control arm. Such analyses and their timing should be clearly specified in the trial protocol, as should the circumstances in which the results would lead to a recommendation to terminate the trial. In other words, the ‘stopping rule’ should be predefined. Second, the DSMB might recommend stopping a trial because of ‘futility’ if the interim results show a difference in study outcomes between the intervention group and the control group which is much less than expected if the intervention is effective and it is clear that, even if the trial is continued until its planned end, it is very unlikely to show an important difference in the rates of the primary outcome between the two groups. Again, such analyses should be planned in advance of starting the trial, and the stopping rules specified in the trial protocol. The DSMB has responsibility for conducting these analyses, because they require breaking the code of the trial; if the decision is to continue the trial, the investigators have not been compromised by knowing either the interim results or which participants were allocated to which intervention. In such circumstances, the DSMB should not be tempted to share the interim results with the investigators but should merely tell them to carry on as planned. Having an independent DSMB is very valuable if a decision needs to be made about stopping a trial early, because this usually has considerable logistic and funding implications and may not be popular with the investigators, staff (who may even need to be laid off early), or participants.
Another common reason for conducting an ‘interim’ analysis is if the incidence of the primary outcome in the trial is less than anticipated or if recruitment to the trial is slower than expected. In such circumstances, it may be clear that the funds for conducting the trial will be exhausted before the planned number of participants or outcome events has been achieved. The investigators may then wish to seek further support from the funding agency to complete the trial. That agency may well request an interim analysis to know if the results to date already show the one arm of the trial to be convincingly superior to the other(s) or, conversely, whether there is little difference between the results in the different arms of the trial and collecting further data is unlikely to produce a convincing result, so it would be futile to extend the trial.
4.1.4 Modification of trial procedures and other advice
During the course of a trial, there may be a need for the DSMB to recommend modifications to the study, because of considerations of patient safety such as eligibility criteria, dosages, treatment duration, and/or concomitant therapy. When there is a trial steering committee, the DSMB would normally propose these recommendations to that committee (as representing the sponsor).
In the absence of a trial steering committee, investigators may well turn to the DSMB for advice on other aspects of the conduct of the trial. The DSMB is a useful source of independent unbiased advice, especially as it will often include persons with substantial trial experience.
4.1.5 Reporting to the sponsor
After each meeting of the DSMB, minutes should be drawn up relating to confidential and non-confidential parts of the meeting. In the non-confidential parts of a meeting, the trial investigators or their representatives may be present, in order to update and inform the committee on the progress of the trial, any deviations from planned procedures, and any AEs among trial participants. Confidential (closed) parts of the meeting will be restricted to DSMB members and may involve looking at data from the trial on outcome measures or AEs, unblinded with respect to the intervention arms. The minutes of this part of the meeting should be kept securely and confidential to the DSMB members until the end of the trial, at which time they should be given to the sponsor.
At the end of each of its meetings, the DSMB should draw up a report to the sponsor. This is often short and along the lines of ‘We have reviewed the safety and other data from the trial and find no evidence of a safety concern that would lead us to suggest any change to trial procedures at this time’. However, if discussion of the trial data does cause the DSMB specific concerns and leads them to suggest specific changes to the conduct of the trial, these should be conveyed to the sponsor. The most extreme advice would be to halt the trial, but other advice might suggest, for example, changes to trial procedures, such as more frequent reporting of SAEs to the DSMB, more careful follow-up of a subset of participants, or changing diagnostic methods.
Sometimes, the investigators or sponsor will seek specific advice from the DSMB on aspects of trial procedures. For example, in some trials, the DSMB, or a subset of its members, may be asked to classify suspected cases of the disease of interest, according to levels of diagnostic certainty (without knowledge of which intervention they received)—though, depending on its composition in terms of expertise, this role might also be assumed by the steering committee or contracted to other independent experts (sometimes referred to as an ‘endpoint committee’).
It is important to note that the normal line of responsibility for reporting the deliberations of the DSMB is from the chair of the DSMB to the sponsor, and the responsibility for liaising with investigators and ethics committees lies with the sponsor. Sometimes, the sponsor will delegate this role to the trial steering committee, so that the DSMB reports directly to that committee. However, the DSMB should not report directly to the trial investigators, unless delegated to so do by the sponsor.
4.2 Composition and appointment of the Data and Safety Monitoring Board
The membership of a DSMB is usually decided by the trial sponsor, or the sponsor may delegate this task to the PIs or to the trial steering committee. Persons invited to join a DSMB are typically independent experts in the area of study of the trial, either working in the same field or in a related discipline, who have no personal or professional involvement with the intervention being tested, in that they will not profit either professionally or financially, according to the outcome of the trial. That is, the membership should be persons who are considered to be unbiased experts. Persons with strong views about the relative merits of the interventions under test would generally be considered unsuitable for DSMB membership. Those invited to join the committee should usually be familiar with, or have experience of, the conduct of RCTs. The chair of the committee should certainly have such experience and ideally should have experience of previous service on a DSMB.
The size of a DSMB will vary, according to the size and complexity of a trial and the likelihood that any of the trial interventions or procedures may cause significant harm to participants. The minimum size is three, and it is rare to have more than ten members, though the DSMB of large multicentre trials may approach that upper limit. The typical composition of a DSMB is outlined in Box 7.2.
In multicountry trials, it is common to have DSMB members drawn from at least some, if not all, of the countries included in the trial. Whether or not (lay) community members or advocates are included as members varies between trials. The inclusion of such members may help bring to the DSMB the perspectives of the population under study. Such members should not be participants in the trial, but the member could be someone with the disease or condition under study or a close relative of such an individual. For example, it has been common to include such persons in trials of HIV vaccines, but practice varies in trials of other interventions, and often lay members are not included.
For ‘high-profile’ trials of interventions, or with study procedures that might be controversial or have unusually high risks, the DSMB might include a medical ethicist knowledgeable in the design, conduct, and interpretation of clinical trials.
Anyone appointed to a DSMB must be prepared to respect the strict confidentiality of the discussions that take place within the committee and of the data that the committee may be given access to. They may also require training, for example, in the principles of GCP. Such training is now widely available either online (which is often free) or face to face.
Generally, members of a DSMB are not paid, but they may be recompensed for loss of earnings, travel, and other expenses incurred as a consequence of DSMB membership. However, in some industry-sponsored trials, members may be paid a fee for their participation.
Before an individual is appointed to a DSMB, it is important that they are given the opportunity to study the trial protocol, so that they fully understand the purposes of the trial and how it will be conducted. Often they will also be given the opportunity to suggest changes to the protocol—especially related to issues such as reporting of AEs and trial stopping rules. Once more, it is important to stress that their advice should go to the sponsor and the trial steering committee, and not directly to the investigators.
Box 7.2 Typical composition of a DSMB
- At least one clinician with knowledge of the disease(s) under study.
- A biostatistician knowledgeable about statistical methods for clinical trials and, if interim analyses are to be conducted, knowledgeable about the specific issues related to sequential analysis of trial data.
- At least one clinician or scientist familiar with the kinds of intervention(s) under test and their possible adverse effects.
- Others who bring special expertise to the committee relevant to the intervention or its application such as toxicologists, epidemiologists, and clinical pharmacologists.
It is possible for a single individual to cover more than one of these skill areas.
4.3 The Data and Safety Monitoring Board charter
In many trials, a specific ‘charter’ is drawn up by the sponsor that details exactly the terms of reference and responsibilities of DSMB members. The charter should take account of the particular needs of the trial and the questions it is addressing. It should detail the relationship between the DSMB and the sponsor, investigators, steering committee, ethics committees, and others with responsibilities in the study. It also gives details of how meetings will be organized, how often they will take place, how many members constitute a quorum, and how confidential and non-confidential minutes will be produced, distributed, and stored securely until the end of the trial. All members of the DSMB will be required to sign the charter at the time of their appointment. Guidelines are available on drawing up a DSMB charter (DAMOCLES Study Group, 2005).