The Ethics of Deciding Something Isn't Research

Research ethics committees are increasingly being asked to demonstrate the quality of their work:

Are reviews consistent? Are decisions proportionate? Are the right ethical issues being identified? Are committees appropriately constituted and trained?

These are important questions. But there is another decision in the research ethics system that receives much less scrutiny:

The decision that ethics review isn't required at all.

Quality improvement and audit activities don't automatically require research ethics review. Nor should they. Health services and other organisations need to be able to examine their own performance, identify problems and improve practice without turning every improvement activity into a research project.

But deciding that an activity isn't research is itself a consequential decision. It determines which governance pathway the activity follows and, potentially, which ethical protections and forms of independent scrutiny apply.

In other words, there is ethics involved in deciding that something isn't research.

A recent reminder from Norway

A case currently unfolding in Norway provides a striking example.

Researchers at Oslo University Hospital used health information from more than 1,000 children and young people in work they maintained was quality assurance. Results were subsequently published in peer-reviewed journals, including analyses of demographics, mental health and treatment trajectories.

Norwegian ethics bodies ultimately reached a different conclusion. The National Research Ethics Committee for Medicine and Health Sciences considered whether the activity was genuinely directed at improving patient care in a local setting and whether practice was being assessed against established standards. It concluded that the project was instead intended to generate new and generalisable knowledge and upheld an earlier determination that it constituted health research requiring prior approval.

The consequences have been significant. A joint institutional committee subsequently found five researchers had committed scientific misconduct through gross negligence in relation to two of the studies. The researchers are appealing that misconduct finding. One paper has already been retracted and another has been reported as due for retraction.

It should be noted that the particulars of this case belong within the Norwegian regulatory system, and the misconduct findings remain contested by the researchers.

But the broader governance question travels remarkably well to our context:

How do we make sure the decision to classify something as quality assurance or audit was the right one?

The gateway matters

We have written previously about the difficult boundary between research, quality improvement and audit. There is no single characteristic that always resolves it.

Publication doesn't automatically make something research. Using existing data doesn't automatically make something quality assurance. The absence of an intervention doesn't settle the question either.

Purpose, methodology, use of data, intended outputs and the context of the activity all matter. In other words, determination should be case-by-case.

Projects can also evolve. Something that genuinely begins as an internal quality-improvement activity may later develop different questions, analytical approaches or intended uses.

This is why classification isn't merely an administrative label: It is a gateway decision.

On one side, research may undergo independent ethics review, including scrutiny of consent, privacy, risk, participant interests and the justification for accessing and using data.

On the other, a legitimate quality activity may appropriately proceed through organisational governance processes without research ethics review.

Self-classification has limits

Researchers and service teams are often well placed to understand the purpose of their work. However, a robust system shouldn't depend entirely on people correctly classifying their own activity.

This isn't necessarily because anyone is trying to avoid ethics review. The boundary can genuinely be difficult. Researchers may be deeply embedded in the service they are evaluating. Definitions can be interpreted differently. Projects change over time. And teams may understandably see an activity through the lens in which it began.

That is precisely why some form of independent or sufficiently separate assessment can be valuable. The question shouldn't simply be:

“Do the researchers say this is quality improvement?”

It should be:

“On what basis has that determination been made?”

Quality assurance needs quality assurance

This is where I think we need to take the conversation further.

Organisations may have processes for determining whether an activity is research, audit, evaluation or quality improvement. But how often do we test whether those processes are actually working?

Australia provides a useful comparison, although its framework is slightly different. The NHMRC's National Statement provides for exemption from ethics review for some activities that are still research, rather than simply determining that an investigative activity isn't research.

Importantly, those exemption decisions don't disappear from oversight. Institutions are expected to have processes for making and recording them, creating an opportunity to monitor whether the gateway is working as intended.

That is an important principle: being outside ethics review doesn't have to mean being outside oversight.

A robust governance system might document classification decisions and their rationale, provide an escalation pathway for uncertain cases, and require reassessment where the purpose or methodology of a project materially changes. It could also periodically audit its own determinations.

For example, an organisation could review a sample of projects classified as quality improvement or audit and ask:

• Were the criteria applied appropriately?

• Were similar activities classified consistently?

• Did any projects subsequently change in purpose or scope?

• Did the eventual use or dissemination of the findings remain consistent with the original determination?

• Were there projects that, with hindsight, should have been referred for ethics review?

The purpose would not be to catch people out. It would be to understand whether the gateway itself is functioning as intended.

However, the problem becomes if there is no pathway to do this (like an application form or locality approval), how do we actually know what’s going on in the institution?

This shouldn't mean sending everything to an ethics committee

There is an obvious danger here. If the response to uncertainty is simply to require ethics review for every audit, evaluation and quality-improvement activity, we haven't improved the system.

We have avoided making the difficult distinction.

That would create unnecessary workload for researchers, governance teams and already stretched ethics committees and secretariats. It could also slow legitimate improvement activity without providing meaningful additional protection.

The objective should be better classification, not more classification as research.

Clear criteria, documented reasoning, access to advice and proportionate oversight are more useful than a system in which everything is sent to an ethics committee “just in case”.

The decision not to review still deserves scrutiny

Research ethics quality assurance understandably focuses on the work that ethics committees actually see. But that gives us only part of the picture.

A high-performing ethics committee cannot protect participants in research that never reaches it because the activity has been incorrectly classified outside its scope. Nor can good standards compensate for a gateway that is applied inconsistently.

If we are serious about strengthening the quality of research ethics systems, quality assurance therefore needs to begin before an application reaches the committee.

We should be able to have confidence not only that research requiring ethics review is reviewed well, but that activities which genuinely do not require that review are being identified on a sound and defensible basis.

The decision not to seek ethics review is still an ethical governance decision.

And consequential decisions deserve quality assurance too.

Additional Reading:

Retraction Watch — Researchers appeal misconduct finding over studies of trans youth (2026) - https://retractionwatch.com/2026/09/09/norway-national-center-gender-incongruence-appeal-misconduct-finding-trans-youth/