What a Cluster Trial Is, and Why Consent Differs

Part of Consent and Vulnerable Populations: Cluster Trials and Simplified Consent.

About 6 minutes. Free, in your language, no account needed.

Why this matters

In most trials the unit of allocation is a person: this participant receives A, that one receives B.

In a cluster trial the unit is a group. A clinic, a ward, a school, a village, a general practice. Every person in the cluster receives whatever that cluster was allocated, and no individual can be allocated differently.

This breaks the ordinary consent model in a specific way. A person can consent or refuse to participate in data collection. They frequently cannot refuse the intervention, because it is being delivered to the place they live, work or receive care.

Why cluster designs are used

They are not a convenience. There are questions that can only be answered this way.

The intervention is inherently group-level. A change to how a clinic triages patients, a water supply intervention, a school curriculum. There is no way to deliver it to some individuals in the building and not others.

Contamination. Where an individually allocated intervention would spread to the control group, allocating by cluster prevents it. Training a clinician in a new approach for some of their patients and not others does not work.

The effect operates at group level. Herd effects in vaccination, changes in a community's behavior. The individual-level question is not the question.

What individuals can and cannot refuse

They usually cannot refuse the intervention, where it is delivered to the setting. A person in a village whose water supply is treated cannot opt out of the water.

They can usually refuse data collection. Answering a questionnaire, giving a sample, having their records reviewed.

They can sometimes refuse an individually delivered component. Where the intervention is delivered person by person within an allocated cluster, such as a vaccine offered to individuals in allocated villages, individual refusal remains available and must be.

The distinction between those situations determines what the consent process can honestly offer, and a process that implies a choice which does not exist is misleading regardless of its wording.

The gatekeeper

Cluster trials involve someone who agrees on behalf of the cluster: a clinic director, a head teacher, a village leader, a practice partner.

This person is a gatekeeper, and the term matters because it names what they are and are not.

They can agree to the cluster's participation in the research: that the site takes part, that the intervention is delivered there, that data may be collected subject to individual consent.

They cannot consent on behalf of the individuals in it. A head teacher agreeing that a school participates has not consented for any child, and a village leader agreeing has not consented for any resident.

Treating the gatekeeper's agreement as a substitute for individual consent is the characteristic cluster trial error, and it is easy to make because the gatekeeper's agreement is genuinely necessary and genuinely authoritative for what it covers.

What the frameworks say

Regulation (EU) No 536/2014 provides for simplified means of obtaining informed consent in the context of cluster trials, subject to conditions including that the trial's methodology requires groups of subjects rather than individuals to be allocated, that the trial is low-intervention, that the medicinal products are used in accordance with their marketing authorization, and that no procedures additional to standard treatment apply. The member state concerned must permit it, and the subject must not object.

Note the conditions. Simplified consent is available for a defined and relatively low-risk class of trial, not for cluster trials generally. A cluster trial of an unauthorized product, or one adding procedures beyond standard care, is not in this category.

In practice

A trial allocates general practices to one of two approaches for managing a condition, using medicines already authorized for it, with no additional procedures.

The design team's initial plan is that practice partners agree and no individual consent is sought, since patients are receiving authorized medicines within their normal care.

The ethics review draws the distinction. The practices may agree to participate, which is the gatekeeper's proper role. The patients' data is being used for research, and that requires their knowledge and the ability to object.

The revised design: practices consent as clusters; every patient is informed in writing that their practice is participating and that their routine data will be used, with a clear and easy route to opt out; and individual consent is sought for the additional questionnaire that some patients complete.

Three levels, each matched to what is actually being asked. The gatekeeper agreement covers the practice, the notification and opt-out covers the data, and full consent covers the only element that goes beyond routine care.

What this looks like when it goes wrong

A cluster trial allocates villages to an intervention. Village leaders agree, and the intervention is delivered.

Data is collected from residents through household surveys, on the basis that the village has consented.

Some residents are surveyed who did not know the village had agreed to anything, and some who did know did not understand that answering the survey was research.

The village leaders' agreement was proper for what it covered. It was treated as covering individual data collection, which it never could.

The finding is that a large number of people provided personal information for research without consenting to it, and that the trial's design confused an authority to permit research in a place with an authority to consent on behalf of the people in it.

Key takeaways