Keeping patients safe while avoiding bias in randomised trials

BMJ 2012; 344 doi: https://doi.org/10.1136/bmj.e3297 (Published 14 May 2012) Cite this as: BMJ 2012;344:e3297
  1. Nick Freemantle, professor of clinical epidemiology and biostatistics1,
  2. Frans Van de Werf, professor and chairman2
  1. 1PRIMENT Clinical Trials Unit, Department of Primary Care and Population Health, UCL Medical School, London NW3 2PF, UK
  2. 2Department of Cardiovascular Medicine, University Hospitals Leuven, Belgium
  1. nicholas.freemantle{at}ucl.ac.uk

Safety monitoring in non-pharmaceutical trials should be stronger

Investigators in randomised clinical trials have a duty to prevent undue hazards for their study patients. To protect patients in trials an independent committee—usually called the data and safety monitoring board, data monitoring and ethics committee, or independent data monitoring committee—is formed to monitor the results as they unfold. To avoid bias, monitoring should be conducted in confidence and the results not communicated to study staff, investigators, or patients. Unfortunately, this does not always happen in device trials and investigator led trials. Further guidance is needed to bring all trials of human subjects up to adequate standards for safety monitoring.

Trials sponsored or run by drug companies are required to follow good clinical practice as described by the International Conference on Harmonisation of Technical Requirements for Registration of Pharmaceuticals for Human Use (ICH E6).1 In addition, ICH E9 and ICH E6 make specific recommendations on how data safety and monitoring boards should monitor safety during trials to ensure that patients are not placed at undue risk while the internal validity of the trial is maintained.1 2 The guiding ethical principle of research on humans is that a study must have the potential to advance knowledge that is beneficial to health and wellbeing while the interests of the individual participating in the research are held above those of society.3 It may be in the interests of society not to modify a trial so that the trial’s scientific objectives can be achieved, but if this materially disadvantages the participants then not making these safety changes would be unethical.

Thus, a data safety and monitoring board will make recommendations to the trial steering committee and may recommend that a trial’s design is stopped or modified either to avoid harm to patients or when there is convincing evidence of benefit. Alternatively, the board may simply state that the trial can continue as planned. Even if the monitoring board noticed that some aspect of the conduct of a trial (such as the choice of primary outcome) militated against the trial achieving its scientific goal, the board should not advise a change because this would introduce bias.4

Decisions about modifications to a trial require knowledge of the ongoing results and thus unblinded access to the data. A data monitoring board should use a prespecified plan for its analyses, including rules to guide future decisions on whether to advise stopping or continuing the trial, and some kind of “alpha spending” plan to account for taking multiple looks at the data and to avoid a type 1 error (concluding that there is a concern when there is not). ICH E9 provides unusually strongly worded guidance: “All staff involved in the conduct of the trial should remain blind to the results of such analyses, because of the possibility that their attitudes to the trial will be modified and cause changes in the characteristics of patients to be recruited or biases in treatment comparisons.”2 Drug trials often go to considerable lengths to ensure confidentiality, with analyses conducted by an independent statistician working for a different organisation who alone has access to the dataset, including the randomised groups.

Although trials of drugs conducted by industry are bound by the ICH guidance (even when not conducted with the aim of seeking marketing authorisation), trials of medical devices are not subject to the same level of regulatory attention.5 6 Trials conducted by universities and other non-commercial bodies are also not required to follow best practice in the conduct of their data safety and monitoring boards.

It is difficult to know exactly how trials are conducted because we rely on information that trialists choose or are able to disclose. Trials funded by commercial organisations are typically protected by confidentiality agreements, and clinical trials units may choose not to share their standard operating procedures, making it difficult to ascertain current practice in academic trials. Nevertheless, our experience indicates that we cannot always rely on trialists using best practice if this is not required by regulation. Indeed our personal experience of data safety and monitoring boards for trials not governed by ICH guidance is that trialists often take a much less rigorous view of research methods than that required for drugs. Thus, interim analyses may be conducted and the results shared with the study team, and study statisticians (who are aware of the ongoing results of a trial) commonly conduct analyses for the data monitoring board while in a position to influence the planning and continued conduct of the trial.

Introduction of the ICH guidance strengthened patient safety in industry sponsored drug trials. Requiring all trials (not just those conducted by drug companies) to follow the same basic standards would protect patients and trials. Device trials and investigator led trials should learn from the ICH guidance, be subject to a higher level of regulation, and implement best practice to ensure patient safety while avoiding bias.


Cite this as: BMJ 2012;344:e3297


  • Competing interests: All authors have completed the ICMJE uniform disclosure form at www.icmje.org/coi_disclosure.pdf (available on request from the corresponding author) and declare: no support from any organisation for the submitted work; no financial relationships with any organisations that might have an interest in the submitted work in the previous three years; no other relationships or activities that could appear to have influenced the submitted work.

  • Provenance and peer review: Not commissioned; externally peer reviewed.


View Abstract

Sign in

Log in through your institution