The randomized, controlled trial (RCT), in which individuals are selected by chance to receive either an experimental intervention or a comparator, is considered the gold standard in clinical biomedical research.
Another methodology, which may be less well known, is the “cluster randomized trial.” The cluster trial randomizes at the group level. The group can vary: it could be a village, a hospital or specific units within a clinical setting. This type of research has become increasingly common in developing countries, where health authorities seek to implement procedures that are established in industrialized countries but have not yet been introduced in rural health centers and other low-resource settings.
This topic is receiving attention in the biomedical research literature, and I have just completed a month’s research exploring related ethical questions.
Implementation research using the cluster randomized design poses ethical questions that are not easily answered by the standard conception of ethics in RCTs. Consider the following example.
A study is designed to introduce an intervention to see whether it improves hygienic practices in birthing centers in a developing country. Six clinics in one province are randomized for researchers to look at outcome data. Three clinics will be provided with written materials for the staff that describe best practices in the use and cleaning of instruments and care for women during and after delivery. The clinics will be visited by an obstetrician and an obstetrical nurse, who will demonstrate the best practices. The other three clinics will not receive the intervention. Data will be collected from women’s records in both clinics after six months to compare rates of infection and other relevant outcomes. The health department wants to know whether it is worth spending money to introduce the intervention in all birthing centers in the province.
Who are the subjects in this cluster randomized trial? The healthcare workers in the birthing centers? The women who come to the center for delivery during this six-month period? Both the workers and the patients?
Is informed consent required of the workers, the women or both? If required, is it needed in both types of center, or only in the centers receiving the intervention? If some healthcare workers refuse to participate in the study, can the research yield valid results? Is it ethically acceptable to withhold information and training about proven infection-control procedures from healthcare workers in the control clusters?
The research is studying whether the intervention improves the hygienic practices of the healthcare workers—not as individuals, but as a group. The data are analyzed at the cluster level, not at the level of individual practitioners. Yet clearly, if there are any human subjects in this type of research, they are the healthcare workers. The women who come for delivery, on the other hand, are not research subjects, although their records will be used to look at outcome data in the intervention and nonintervention clusters. In today’s research environment, retrospective record review for research does not normally require consent from the patients whose records are reviewed. But what if the researchers take an additional step and seek to interview the women about the healthcare workers’ procedures? (Did they wash their hands? Did they wear gloves? Did they pick up used instruments that were lying around?) If the women are interviewed, then it seems their consent would be ethically required for that part of the research.
If the healthcare workers are considered research subjects, then their informed consent would be required and they could refuse to participate in the cluster randomized trial. But if not all workers in the intervention clinics participate, the research could not obtain valid results. Health authorities could simply introduce the new intervention in all health centers without a cluster randomized design, and it would then count as “quality improvement.” Healthcare workers would not have to give their consent; they would simply comply with the new required procedures.
These are only some of the ethical quandaries raised by the cluster randomized design methodology. Reasonable people may disagree with the answers I came up with in my study, but few would question the dilemmas posed by a type of research that departs from the typical RCT.