March 22, 2021 guest blog
 | Ariella Binik McMaster University, Canada |
 | Karla Hemming University of Birmingham, UK |
 | Charles Weijer Western University, Canada |
Is the stepped wedge cluster randomised design ethically advantageous? Does it resolve ethical concerns about depriving participants of access to the study interventions? Should it be used when the study intervention seems particularly likely to be effective? Each of these questions has been raised in connection with the stepped wedge trial, but they have received only limited attention in the ethics literature and guidelines. In what follows, we address key ethical issues associated with the stepped wedge design, including the choice of the design, the permissibility of delaying access to the intervention, research ethics committee review, and informed consent.
Background
Cluster randomised trials randomly allocate groups rather than individuals to study arms. They often evaluate study interventions targeted at communities, health systems, or health care professionals, including public health, health service delivery, or educational interventions. In a stepped-wedge cluster randomised trial, clusters randomly and sequentially cross from control to the study intervention, until (usually) all clusters have been observed under the intervention condition.
Ethical guidelines were developed with individual randomisation in mind and, as a result, they are challenging to apply to trials randomising groups. For example, ethical guidelines require obtaining informed consent from each prospective participant, but these guidelines were developed with an eye to interventions on individuals, and so might not be an obvious fit for cluster trials evaluating interventions that target entire health systems or communities. There can thus be a tension between ethical guidelines and cluster randomised trials. Moreover, while some ethical questions in the stepped wedge design are common to all cluster randomised trials, the nature of the staggered roll-out under the stepped-wedge design can pose further ethical issues. Two international ethics guidelines are specific to cluster randomised trials. The Ottawa Statement on the Ethical Design and Conduct of Cluster Randomised Trials offers detailed guidance for researchers and research ethics committees. In addition, the 2016 revision of the Council for International Organizations of Medical Sciences (CIOMS) International Ethical Guidelines for Health-related Research involving Humans includes a section specific to cluster trials (section 21). At the moment, these are the only ethical guidelines addressing cluster randomised trials and neither focuses on the stepped wedge design. This is a limitation given that unique features of the stepped wedge merit ethical consideration.
What are ethical considerations for stepped wedge trials?
There are at least four ethical issues that merit careful consideration when planning or conducting a stepped wedge trial: (1) the choice of the stepped wedge design; (2) the permissibility of delaying intervention; (3) research ethics committee review; and (4) informed consent.
1. The choice of the stepped wedge design
Cluster randomised trials are less efficient statistically and more susceptible to bias than comparable individually randomised trials. Because of these complexities, the use of a cluster randomised design requires justification. The stepped wedge cluster randomised trial is at even more at risk of bias than parallel arm cluster trials. This is largely because the design is confounded with time (the data observed under the control condition is at a systematically earlier calendar time than that under the intervention condition). Consequently, a clear rationale should be provided for the use of a stepped wedge design. Indeed, it is likely that the stepped-wedge design is being chosen when other study designs might have been preferable.
It is frequently thought that ethics is a good rationale for selecting the stepped wedge trial. Indeed, a systematic review of 37 stepped wedge trials published between 2010 and 2014 found that 16 cited ethical reasons in the choice of design. In these trials, “authors described a lack of equipoise for the intervention based on positive pilot study results or prior literature, and felt it would be unethical to deny the intervention to some groups.” On this view, the balance of evidence has already been tipped in favour of the study intervention and a stepped wedge trial is ethically preferable because it ensures that all clusters will receive the intervention by the end of the study. This view is mistaken. New programs in health systems and public health are commonly introduced with insufficient evidence about their effectiveness and the implementation of these programs should be accompanied by research evaluation. It follows that the implementation of a policy should not necessarily be taken to suggest it will be effective and, in the absence of sufficient evidence of effectiveness, there is no ethical reason to prefer a design in which all clusters receive the intervention. Moreover, the principle of equipoise (discussed below) should be used to determine when there is an ethical obligation to evaluate an intervention. In the presence of equipoise, there are good rationales for selecting a stepped wedge trial, including facilitating cluster recruitment, logistical constraints (e.g., the rollout of a scarce resource), or increased statistical power when clusters are limited. But ethics is not a good rationale.
2. The permissibility of delaying interventions
In the situation where a novel health policy or public health intervention is thought to be effective (and so equipoise is believed not to hold), then depriving some groups of the study intervention for some or all of the duration of the trial will be problematic. In conventional cluster randomised trials, it would thus not be ethically permissible to withhold the study intervention from those in the control group. The stepped wedge trial has been portrayed as an alternative that both permits randomised evaluation and mitigates these ethical concerns. Since all clusters receive the study intervention by the end of the trial, the stepped wedge trial delays rather than deprives access to the intervention and delay seems preferable to deprivation. Yet, the ethical permissibility of delaying the study intervention is an issue from which stepped wedge trials are not immune
This is therefore not a persuasive ethical defense. If a cluster (and the people within it) have a moral claim to receive an intervention, then the intervention must be provided. The intuition that “delaying is better than depriving” has a certain common-sense appeal. But if the intervention is one to which people have a right, then both delaying and depriving violate our ethical obligation to research participants. In such cases, ethical considerations likely preclude the use of any randomized design. What’s more, whilst all clusters ultimately receive the intervention, this is not necessarily true of all people within clusters. It is ethical to randomise when both intervention and control are in equipoise. Equipoise means that there is “uncertainty about the comparative benefits and harms of the intervention in the experimental arm versus the control arm, according to a community of experts.” Given the range of interventions evaluated in stepped wedge trials, the relevant expert community should be understood broadly, and may include public health officials, health systems managers, or medical experts. If at the outset of a stepped wedge trial, the evidence in favour of the study intervention is insufficient to disrupt equipoise, then there is no ethical obligation to provide control clusters with it. At a minimum, people within control clusters should receive “usual care,” and they must not be “deprived of effective care or programs to which they would have access, were there no trial.”
3. Research ethics committee review
Research involving human participants generally requires review and approval by a research ethics committee. Research refers to “a systematic investigation designed to produce generalizable knowledge.” A recent systematic review found that 75% of stepped wedge trials reported research ethics committee review and 50% reported trial registration, suggesting that some may not view these trials as research. Particularly contentious may be stepped wedge trials evaluating an intervention within the context of routine health system rollout. In such cases, the stepped wedge trial meets the definition of research with human participants and research ethics committee review is required. Even when the intervention is not designed by or under the control of researchers (e.g., a government programme), the use of a stepped wedge design “signals that researchers and government are working together to plan the roll-out of a programme so that it may be evaluated. As a result, equipoise issues are triggered that must be addressed by the [research ethics committee].”
4. Informed consent
Informed consent is generally required from participants in health research. Informed consent upholds the ethical principle of respect for persons and the value of individual autonomy. But there are exceptional conditions under which a waiver of consent may be permitted. A research ethics committee may approve a waiver of consent if: “the research would not be feasible or practicable to carry out without the waiver…; the research has important social value; and the research poses no more than minimal risks to participants.” Waivers of consent are common in health research involving review of medical records and psychology studies in which revealing the hypothesis would undermine the study’s validity.
The Ottawa Statement sets out a new application for waiver of consent in cluster randomised trials evaluating cluster level interventions (i.e., cluster-cluster trials). Cluster level interventions are defined as interventions, such as community-wide public health messaging, that are indivisible at the level of the individual. In such cases, it is difficult or impossible for cluster members to avoid the study intervention. As a result, refusal of consent for the study intervention would be meaningless. Provided that the study holds the prospect of social value and the intervention poses only minimal risk to participants, a research ethics committee may approve a waiver of consent for the cluster level intervention. Where a waiver of consent for the study intervention is approved, “consent for data collection procedures may nonetheless be required.”
Despite broad ethical requirements to obtain consent, a recent systematic review found that only 16 of 32 stepped wedge trials reported obtaining informed consent from research participants. “In the subgroup of 20 trials with research participants at cluster level, only 4 (20%) reported informed consent from such participants; in 26 trials with individual-level research participants, only 15 (58%) reported their informed consent.” Stepped wedge trials often evaluate cluster level interventions, such as public health or health systems interventions, and consequently, they may be eligible for a waiver of consent provided that risks to participants are minimal. It is important, however, not to conflate cluster and individual level interventions. The rollout of a drug or vaccine across clusters may be framed as a policy or government program; but, as they are divisible at the level of the individual, they are correctly viewed as individual level interventions. Stepped wedge trials of drugs or vaccines exclude participants who have a medical contraindication to the intervention. Informed refusal of study participation should similarly be regarded as an exclusion. As stated above, a waiver of consent for the study intervention does not preclude requiring consent for data collection in a trial.
Concluding thoughts
To return to the questions with which we began, there are good reasons to select a stepped wedge trial in various instances, but ethics is not a persuasive rationale. The stepped wedge design is not preferable ethically when the study intervention appears likely to be effective. The ethics of randomizing clusters to intervention and control conditions should be assessed by appealing to equipoise. In the absence of equipoise, and when clusters (and the people within them) have a right to the intervention, both delaying and depriving access violate our ethical obligation to research participants. However, provided that the evidence in favour of the study intervention is insufficient to disrupt equipoise, there is no ethical obligation to provide all clusters with the study intervention. Stepped wedge trials, including those evaluating the rollout of an intervention across a health system, must be reviewed by a research ethics committee. Finally, informed consent is an important ethical requirement in stepped wedge trials, but a waiver of consent may be approved by a research ethics committee for a cluster level intervention, provided risks to participants are minimal.
 | Go back to the beginning: What is a stepped wedge trial? |