• We are sorry, but NCBI web applications do not support your browser and may not function properly. More information
Logo of bmjLink to Publisher's site
BMJ. Sep 4, 2004; 329(7465): 525–526.
PMCID: PMC516093

Stopping clinical trials early

Data monitoring committees may have important role
Adrian Grant, director

Halting high profile trials early has been in the news again this year.1 Legitimate questions have been raised about who decides that a trial should stop early, how they do it, and on what basis they reach such a decision.

Nearly all ongoing randomised clinical trials need some form of data monitoring.2-4 Differences between treatments may be larger than expected or unanticipated adverse effects may occur, and either of these may justify early termination of a trial. Although this has obvious bearing on current and potential participants in the trial, it also has potentially profound wider implications. In particular, the data from the trial may later constitute the evidence base for management of future patients. Hence the data should be sufficiently convincing to the wider clinical and patient community to determine future practice. Those taking a decision to stop a trial early therefore carry a heavy responsibility for ensuring that this would be in the best interests both of the individuals involved in the trial and of society.

Currently only a minority of trials have a formal group, often called a data monitoring committee, whose remit is to monitor a trial's data as they accumulate, although this practice is increasing and rightly is becoming standard practice for large trials.3,5 Usually the data monitoring committee consists of independent members who periodically consider confidential reports of interim analyses. Independence guards against the possibility that vested interests of the organisers, sponsors, or funders might influence the deliberations; the requirement for an independent data monitoring committee whose members have no competing interest is now the policy of most funders of large scale trials.3 Confidentiality protects against misinterpretation if the data were to be publicly available. Examples exist where the play of chance led to surprisingly large differences early in a trial, only to disappear or to be reversed over time.6

A data monitoring committee must show strong resolve when large unanticipated differences are inconsistent with existing evidence from outside the trial. The choice of membership can therefore be crucial. The selection of the chair is particularly important, and this person (and preferably other members too) should have had prior experience of serving on a data monitoring committee. Currently relatively few people have served on data monitoring committees, and this limits choice. In the United Kingdom a more formal system for training new members is being considered.3 The wider membership should be multidisciplinary, almost always including at least one person with appropriate clinical expertise and at least one statistician. Beyond this, the size and membership vary. Advantages exist for both large and small committees, and opinions differ about what is optimal.3 My own view is that four to six is appropriate for most trials. Extension beyond clinicians, statisticians, and trialists to include representatives of consumers (patients) and ethicists is currently uncommon and controversial.3 Lay membership should be encouraged as a way of involving non-professionals in these potentially difficult decisions but in the context of a multidisciplinary group.

In practice the data monitoring committee for most trials does no more than monitor progress and recommend that the trial continue as planned. Occasionally, interim analyses show unexpected differences in benefits or toxicity and the responsibility is then much heavier.3 A range of formal statistical approaches can be used as a basis for judging at what point such differences are so extreme as to be sufficiently unlikely to reflect the play of chance.2,3 However, statistical criteria for stopping should only be considered as guidelines. A data monitoring committee should also take into account other considerations, such as meta-analyses of data from comparable trials, other existing evidence external to the trial,7 and the nature of the condition and its alternative treatments.

In some data monitoring committees the analyses are blinded such that the members do not know to which treatment group they apply, the aim being to avoid biased interpretation. In practice, however, the data monitoring committee is making decisions about the balance of likely risks and benefits and needs to know the treatment to do this. Early stopping is sometimes suggested for futility—a decision that continuing a trial would not provide sufficiently useful information to warrant continuation. I believe that this is rarely justifiable, other than for poor recruitment, and the group responsible for running the trial (for example, the trial steering committee) can then take this decision.

The practical aspects of data monitoring and data monitoring committees have recently received increased attention.3-5,8 Ensuring that a monitoring committee's roles, responsibilities, and planned operations are explicitly agreed in advance with the trial's investigators, sponsors, and funders improves conduct and performance. This is perhaps best done through a formal charter, adapted for the circumstances of a particular trial, and agreed at an early meeting.3,9

This would cover aspects such as: who will attend meetings of the data monitoring committee; how often meetings will be held; whether they will include “open” sessions involving the investigators or funders, or both, or just closed sessions restricted to those with access to the interim analyses; the format of the interim reports presented to the data monitoring committee; whether the treatment groups will be labelled in the analyses or “blinded”; who will perform the interim analyses; what statistical criteria will be used; how decisions will be reached; and how and to whom the data monitoring committee will report.

Data monitoring committees act best in an advisory capacity, making recommendations to the group responsible for running the trial, possibly with recourse to an arbiter if this group disagrees with its recommendations. Deciding that a trial should be stopped early is never easy because of the range of issues and responsibilities that must be taken into consideration. Good research governance means that the organisers of trials need to give serious and early consideration to arrangements for monitoring accumulating data particularly the constitution, roles, responsibilities, and procedures of data monitoring committees.

Notes

Competing interests: AG was convener of a study group on issues in data monitoring and interim analysis of trials, funded by the UK NHS R&D Health Technology Assessment Programme (The DAMOCLES study).

References

1. Dyer O. Another HRT trial is stopped early. BMJ 2004;328: 305.
2. Pocock SJ. When to stop a clinical trial. BMJ 1992;305: 235-40. [PMC free article] [PubMed]
3. Grant AM, Sydes MR, McLeer SK, Clemens F, Altman DG, Babiker AB, et al. Issues in data monitoring and interim analysis of trials. Health Technol Assess (in press). [PubMed]
4. Sydes MR, Spiegelhalter DJ, Altman DG, Babiker AB, Parmar MKB, and the DAMOCLES Group. Systematic qualitative review of the literature on data monitoring committees for randomized controlled trials. Clin Trials 2004;1: 60-79. [PubMed]
5. Sydes MR, Altman DG, Babiker AB, Parmar MKB; Spiegelhalter DJ, and the DAMOCLES Group. The reported use of data monitoring committees in the main published reports of randomised controlled trials: a cross-sectional study. Clin Trials 2004;1: 48-59. [PubMed]
6. Doll R. The role of data monitoring committees. In: Duley L, Farrell B, eds. Clinical Trials. London: BMJ Publishing Group, 2001: 97-104.
7. Brocklehurst P, Elbourne D, Alfirovic A. The role of external evidence in monitoring clinical trials: reflections from a perinatal trial. BMJ 2000;320: 995-8. [PMC free article] [PubMed]
8. Ellenberg S. Fleming T, DeMets D. Data monitoring committees in clinical trials: a practical perspective. Chichester: Wiley, 2002.
9. The DAMOCLES Study Group. A proposed charter for clinical trial data monitoring committees: helping them to do their job well. Lancet (in press). [PubMed]

Articles from BMJ : British Medical Journal are provided here courtesy of BMJ Group
PubReader format: click here to try

Formats:

Related citations in PubMed

See reviews...See all...

Cited by other articles in PMC

See all...

Links

  • PubMed
    PubMed
    PubMed citations for these articles

Recent Activity

Your browsing activity is empty.

Activity recording is turned off.

Turn recording back on

See more...