- Monica Taljaard, scientist, assistant professor21,
- Charles Weijer, professor34,
- Jeremy M Grimshaw, senior scientist5, professor6,
- Martin P Eccles, professor of clinical effectiveness7
- the Ottawa Ethics of Cluster Randomised Trials Consensus Group
- 1Ottawa Hospital Research Institute, Clinical Epidemiology Program, Ottawa Hospital, 1053 Carling Avenue, ASB2-004, Ottawa, ON, K1Y 4E9, Canada
- 2Department of Epidemiology and Community Medicine, University of Ottawa, Ottawa, ON, Canada
- 3Rotman Institute of Philosophy, Department of Philosophy, Western University, London, ON, Canada
- 4Departments of Medicine and of Epidemiology and Biostatistics, University of Western Ontario, London, ON, Canada
- 5Ottawa Hospital Research Institute, Clinical Epidemiology Program, Centre for Practice-Changing Research, The Ottawa Hospital, Ottawa, ON, Canada
- 6Department of Medicine, University of Ottawa, Ottawa, ON, Canada
- 7Institute of Health and Society, Newcastle University, Newcastle upon Tyne, UK
- Correspondence to: M Taljaard
- Accepted 1 April 2013
An increasing number of studies in health system, public health, and knowledge translation research are using randomised controlled designs.1 Examples are studies of changes in healthcare policies across a region, mass media health promotion campaigns, and training of health professionals with the view to improving the quality of patient care. For these types of interventions, which are naturally administered at the group level, cluster randomisation may be the only feasible choice. In a cluster randomised trial, the unit of allocation is an intact social unit or group (for example, a medical practice, community, worksite, or school), whereas the unit of outcome measurement is individuals within the group (for example, a patient, citizen, employee, or student).2 3 Even if the intervention is to be administered at the individual level, such as vitamin supplementation or patient leaflets, cluster randomisation may nevertheless be selected for either scientific reasons, such as avoidance of experimental contamination, or practical reasons.
Cluster randomised trials raise specific methodological and ethical issues. Although there is no shortage of guidance for handling the methodological issues,4 5 6 little guidance exists for dealing with the ethical issues.7 8 9 10 Ethical issues arise because standard research ethics guidelines are not well attuned to the distinctive characteristics of cluster randomised trials. Most notably, cluster randomised trials may have differing units of allocation (for example, hospitals), intervention (for example, health professionals), and outcome measurement (for example, patients) within a single study. This complicates identification of those who ought to be considered as research participants and thus who are entitled to ethical and regulatory protections. Cluster level interventions and randomisation of clusters before identification of individual cluster members cause difficulties in seeking informed consent. Furthermore, cluster randomised trials often concern diverse social groups, communities, or organisations as the units of allocation, but the moral status of such groups, as well as the identification and authority of those who can morally speak on their behalf, are unclear. Because cluster randomised trials may have consequences for groups as well as for individuals, the assessments of benefits and harms are more challenging in such trials. In addition, protecting vulnerable participants in cluster randomised trials may require special consideration as their presence within clusters may go unnoticed, whereas risks to them may be amplified as a result of cluster membership.
The supplementary file presents a summary of the characteristics of cluster randomised trials and examples of the ethical challenges they raise. The absence of specific ethics guidelines for cluster randomised trials has led to considerable variability in the conduct and review of such trials11 and to serious shortcomings in their reporting.12 13
The Ottawa Statement14 aims to provide guidance for researchers and research ethics committees primarily, and for policy makers, journal editors, and potential study participants, on the ethical design and conduct of cluster randomised trials in health research. It supplements national and international ethics guidelines to take into account the unique characteristics of cluster randomised trials. The statement should be interpreted in light of the laws and regulations of the country or countries in which the cluster randomised trial is conducted and other applicable international standards.
The research team, comprising 15 investigators from Canada, the United Kingdom, and the United States (see supplementary file), was funded by the Canadian Institutes of Health Research in 2007. The goal of the project was to develop well grounded, international ethics guidelines for cluster randomised trials.15 It was designed as a mixed methods project consisting of parallel empirical and ethical streams culminating in a consensus process. A series of empirical studies was designed to identify ethical problems arising in cluster randomised trials, elicit the views and experiences of investigators conducting these types of trials, and study the ethical review process of cluster randomised trials.16 11 Based on the experience of the research team members and results of the empirical studies, the research team identified six ethical questions considered unique to cluster randomised trials.17 The team conducted an in-depth ethical analysis of the identified issues and published the results as a series of articles in an open access journal.18 19 20 21
In year 5 of the project, the research team convened a multidisciplinary expert panel consisting of six members of the team and 13 external members (ethicists, experienced cluster trial investigators, consumer representatives, research ethics committee members, policy makers, representatives from funding agencies, and journal editors). A consensus conference was held in Ottawa, Canada, during November 2011 (video available at www.rotman.uwo.ca/resources/video-audio/cluster-randomized-trials-international-consensus-conference/). A detailed description of the consensus process is provided elsewhere.12 A writing group produced a first draft of the guidelines, which the group circulated to the expert panel and then edited in response to their comments. In February 2012, the writing group posted the revised document on the project’s website for public comment. It then edited the document in response to the received comments. The final draft was approved by all members of the expert panel in June 2012 and the full length Ottawa Statement was published in November 2012.12
The Ottawa Statement sets out 15 recommendations, each with a detailed commentary. Recommendations address the justification of the cluster randomised trial design, research ethics review, the identification of research participants, informed consent, the role and authority of gatekeepers, the evaluation of benefits and harms, and the protection of vulnerable participants. The table⇓ presents the recommendations for each of these ethical issues in the form of a checklist that may be used by researchers when designing their studies and when applying to research ethics committees. Equally, research ethics committees may consider the recommendations as a tool to help them judge the attributes of cluster randomised trials submitted to them. We recommend that researchers include discussion of each of the main ethical issues in their study protocols. The table may then be appended to their ethics application to direct the attention of the research ethics committee to the guidance and how the ethical issues have been dealt with in the study protocol.
Here we outline key implications of the Ottawa Statement for researchers and research ethics committees.
Justifying the cluster randomised design
A cluster randomised trial is more complex to design and conduct and statistically inefficient than an individually randomised trial, and is vulnerable to multiple sources of biases. For these reasons, researchers should clearly justify their choice of cluster rather than individual randomisation (recommendation 1). Acceptable reasons include the evaluation of a cluster level intervention or group effects of an intervention; the need to avoid experimental contamination, reduce costs, enhance compliance, or secure cooperation of investigators; and administrative convenience. Researchers should not adopt this design in a veiled attempt to sidestep the requirements for informed consent.
Research ethics committee review
Research may be defined as a systematic investigation designed to produce generalisable knowledge.22 Most cluster randomised trials in health research, including those evaluating knowledge translation, quality improvement, and public health interventions, meet this definition.15 According to national and international research ethics guidelines, the trials must therefore be reviewed by a research ethics committee (recommendation 2). An expedited review may be appropriate if a cluster randomised trial poses a low risk to research participants and does not involve vulnerable participants.
Identifying research participants
Because cluster randomised trials may have differing units of allocation, intervention, and outcome measurement, determining which cluster members ought to be considered as research participants can be difficult. Yet appropriate identification of research participants is necessary for applying ethical and regulatory protections.17 Recommendation 3 provides a definition and criteria for identifying research participants. Any cluster member who is the recipient or the direct target of a study intervention (including the control condition), with whom researchers interact for study purposes or about whom identifiable private information is collected, is a research participant. Figure 1⇓ summarises the identification of research participants in cluster randomised trials.
The Ottawa Statement has three main implications for the identification of research participants. Firstly, researchers should clearly identify the research participants in the study protocol and other documents submitted for review by the research ethics committee. Secondly, in the case of a cluster level intervention, cluster members who are directly targeted by the intervention ought to be considered research participants. Thirdly, health professionals who are targeted by a knowledge translation intervention ought to be considered research participants; however, their patients are not research participants unless they are otherwise intervened upon or interacted with or their identifiable private information is collected for study purposes.
Obtaining informed consent
Informed consent procedures are complicated in cluster randomised trials for several reasons, including interventions administered at the cluster level, large cluster size, and randomisation of clusters before identification of individual research participants.18 Figure 2⇓ summarises the guidelines in the Ottawa Statement with respect to informed consent.
The Ottawa Statement has three main implications with respect to informed consent. Firstly, where possible, research participants (or their proxy decision makers) are required to provide informed consent, but not cluster members who are not research participants (recommendation 4). Secondly, as far as possible researchers should identify participants and seek their consent before cluster randomisation. If consent is not feasible before randomisation, it should be sought as soon as possible after cluster randomisation and before any study interventions or data collection procedures (recommendation 5). Researchers should be aware that post-randomisation consent can introduce selection biases and should consider adopting design strategies that minimise the risk of bias.23 Thirdly, because of the challenges in seeking informed consent in cluster randomised trials, waivers and alterations of consent procedures are especially important in this design. Waivers or alterations of consent can be justified only when the study would not otherwise be feasible and when the risks involved are minimal (recommendations 6 and 7). Minimal risk refers to the incremental risks associated with study participation and requires that such risks be consistent with the risks of daily life.24 A waiver of consent means that the research ethics committee removes the requirement to obtain informed consent, whereas an alteration of consent means that the committee permits changes to or removal of some of the standard elements of disclosure in the informed consent—for example, to maintain blinding in the case of a behavioural intervention.
Gatekeepers are individuals or bodies (such as a school principal or municipal council) who may be called on to protect the interests of organisations or communities that are the setting for a cluster randomised trial. Researchers have historically relied on “cluster gatekeepers” to perform a variety of roles, including providing permission for clusters to participate in a cluster randomised trial and providing consent on behalf of individuals in a cluster—for example, for a cluster level intervention.19
The Ottawa Statement has three main implications with respect to gatekeepers. Firstly, gatekeepers may not legitimately provide proxy consent on behalf of individuals in a cluster (recommendation 8). Legitimate proxy consent requires that the potential participants be incapable of making their own decisions and that the decision makers be well acquainted with the potential participants’ values and beliefs. Secondly, researchers should obtain permission from a gatekeeper to enrol a cluster or organisation in a cluster randomised trial when a trial poses important implications for group interests and a gatekeeper exists who has the legitimate authority to provide permission (recommendation 9). Thirdly, gatekeepers may help protect group interests in a cluster randomised trial by facilitating consultation between researchers and cluster members about the study (recommendation 10). Modes of consultation may include communication with communities through various media channels, open public forums, or meetings with opinion leaders.
Assessing benefits and harms
In individually randomised trials of clinical interventions, the analysis of benefits and harms usually focuses on the potential effects on individual patients, with the trial being justified on the basis of clinical equipoise. In a cluster randomised trial, however, study interventions may affect not only individuals but also clusters, organisations, and communities.16 Moreover, cluster randomised trials may investigate the effectiveness of public health, health systems, and knowledge translation interventions. The justification of a cluster randomised trial must therefore take interests of a broader range of stakeholders into account.
The Ottawa Statement has three main implications when assessing benefits and harms in cluster randomised trials. Firstly, researchers must ensure that study interventions are consistent with competent practice in the specialty of study relevant to the cluster randomised trial, such as medical practice, public health, or health policy (recommendation 11). Random assignment is justified when the relevant community of experts (for example, public health practitioners, health policy researchers) is uncertain as to the better performing practice. Secondly, researchers must adequately justify the choice of the control condition; generally, at a minimum the control condition is usual care within the study context (recommendation 12). Researchers and research ethics committees may consider whether the control arm should receive a minimal level of intervention or “augmented” care, but they should be aware that this may bias the estimate of the intervention effect towards the null and may reduce generalisability. Thirdly, the risks associated with data collection procedures must be minimised and stand in reasonable relation to the knowledge to be gained (recommendation 13).
Protecting vulnerable participants
Vulnerable research participants include children, incapable adults (that is, those unable to provide informed consent), those at undue risk of harm as a result of study participation, and those in subordinate positions within social or organisational structures. Cluster randomised trials are commonly conducted in organisations, and research participants may include professionals, service staff, or employees whose position within the cluster makes them less able to express a free choice about trial participation.
The Ottawa Statement has two main implications with respect to vulnerable participants in cluster randomised trials. Firstly, special care must be taken to identify and protect vulnerable participants as their presence may not be apparent within clusters (recommendation 14). Secondly, discussions about informed consent with research participants in cluster randomised trials involving organisations should limit the potential for coercive influence from employers or cluster leaders, and such discussions should include career related risks, including risks from detection of negligence or incompetence (recommendation 15).
The use of cluster randomisation in health research is increasing.1 The substantial methodological differences between cluster randomised trials and conventional randomised trials pose serious challenges to the current conceptual framework for research ethics. Researchers need direction on these ethical challenges to guide the optimal design and conduct of cluster randomised trials. Research ethics committees may be unfamiliar with this increasingly important study methodology and, in the absence of formal guidelines for cluster randomised trials, they may fail to consider all the relevant ethical issues generated by a study protocol, resulting in inadequate protection of participants. Variable interpretation of ethical requirements for cluster randomised trials may lead to problems initiating multijurisdictional trials and to unequal treatment of participants in different jurisdictions.
Until the publication of The Ottawa Statement on the ethical design and conduct of cluster randomised trials,12 there were no comprehensive ethics guidelines specific to this type of study design. The aim of the Ottawa Statement is to address this shortcoming by providing researchers and research ethics committees with recommendations for cluster randomised trials that meet international ethical standards in health research. Written by scientists and ethicists, the Ottawa Statement rests on a strong foundation of empirical study and detailed ethical analysis; it is the result of a transparent and robust consensus process. Applications of cluster randomised trials are found in a wide variety of sectors, including education, criminology, and welfare. Although the development of the Ottawa Statement has focused on applications in health research, it provides a starting point for future extension to these sectors.
Cluster randomised trials have many characteristics that complicate the application of standard research ethics guidelines; however, until recently there have been no comprehensive ethics guidelines specific to cluster randomised trials
The Ottawa Statement aims to provide researchers and research ethics committees with detailed guidance on the ethical design, conduct, and review of cluster randomised trials
The Ottawa Statement sets out 15 recommendations addressing the justification of the cluster randomised design, the need for ethics review, the identification of research participants, obtaining informed consent, the role of gatekeepers in protecting group interests, the assessment of benefits and harms, and the protection of vulnerable participants
Cite this as: BMJ 2013;346:f2838
The Ottawa Ethics of Cluster Randomised Trials Consensus Group includes: Fernando Althabe (Institute for Clinical Effectiveness and Health Policy, Buenos Aires, Argentina), Ariella Binik (Rotman Institute of Philosophy, London, Canada), Judith Belle Brown (Western University, London, Canada), Robert Boruch (University of Pennsylvania, Philadelphia, USA), Jamie C Brehaut (Ottawa Hospital Research Institute, Ottawa, Canada), Shazia Chaudhry (University of Ottawa, Ottawa, Canada), Allan Donner (Western University, London, Canada), Geneviève Dubois-Flynn (Canadian Institutes for Health Research Ethics Office, Ottawa, Canada), Martin P Eccles (Newcastle University, Newcastle upon Tyne, UK), Sarah Edwards (University College London, London, UK), Diana Elbourne (London School of Hygiene and Tropical Medicine, London, UK), Sandra Eldridge (Queen Mary University of London, London, UK), David Forster (Western IRB, Olympia, USA), Antonio Gallo (Rotman Institute of Philosophy, London, Canada), Jeremy M Grimshaw (Ottawa Hospital Research Institute, Ottawa, Canada), Catarina Kiefe, University of Massachusetts Medical School, Worcester, USA), Jonathan Kimmelman (McGill University, Montreal, Canada), Melody Lin (Office for Human Research Protections, Rockville, USA), Elizabeth Loder (Harvard Medical School, Boston, USA), Kathleen Lohr (RTI International, Research Triangle Park, USA), Andrew D McRae (University of Calgary, Calgary, Canada), Eileen S Naughton (Rhode Island House of Representatives, Providence, USA), Rex J Polson (Solihull Hospital, Solihull, UK), Raphael Saginur (Ottawa Hospital, Civic Campus, Ottawa, CanadaAbha Saxena (World Health Organization, Geneva, Switzerland), Julie Spence (St Michael’s Hospital, Toronto, Canada), Monica Taljaard (Ottawa Hospital Research Institute, Ottawa, Canada), Charles Weijer (Rotman Institute of Philosophy, London, Canada), Angela White (Rotman Institute of Philosophy, London, Canada), Gerald White (Health Council of Canada, Toronto, Canada), and Merrick Zwarenstein (Institute for Clinical Evaluative Studies, Toronto, Canada). Members of the Ottawa Ethics of Cluster Randomised Trials Consensus Group participated in this project as individuals and not as representatives of their employers. Authorship does not imply the approval of this document by the authors’ employers or other organisations with which they are affiliated.
Contributors: MT, CW, and JMG are co-principal investigators of the project. MPE is a co-investigator of the project and chaired the consensus process. CW led the ethical analysis. MT drafted this article and is the guarantor. All authors commented on drafts of this article and revised the article critically for important intellectual content. All authors approved the final version.
Funding: This study was funded by the Canadian Institutes of Health Research (operating grants MOP85066 and MOP89790). The funding agency had no role in the study design, collection, analysis or interpretation of data, writing of the manuscript, or decision to submit the manuscript for publication. JMG and CW both hold Canada research chairs.
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 which might have an interest in the submitted work in the previous three years; no other relationships or activities which could appear to have influenced the submitted work.
Ethical approval: This study was approved by the Ottawa Hospital research ethics board.
Provenance and peer review: Not commissioned; externally peer reviewed.