Trial Registration and Results Reporting: Closing the Gap Between Studies Run and Studies Seen

When only the trials with favorable results get published, the evidence base is distorted before anyone reads it. Prospective registration requires a trial to be listed publicly, with its planned outcomes, before the first participant enrolls, and results reporting is a separate duty. Together they let a reader check a published paper against the study that was actually planned.

The problem: outcomes that quietly disappear

Imagine ten trials of the same treatment. Six find nothing and are never published; four find a benefit and appear in journals. A reader who sees only the four gets a badly distorted picture, and so does any meta-analysis built on the published record. This is publication bias, and its quieter cousin is a single trial that measures several outcomes but reports only the ones that turned out well.

Trial registration and results reporting exist to close this gap between the studies that were run and the studies the world gets to see.

Prospective registration, before the first participant

The core rule is timing. A trial should be registered in a public registry at or before the enrollment of its first participant, with its planned outcomes stated in advance. Editors following the major recommendations require this prospective registration as a condition of later publication.

The reason registration must come first is subtle but decisive. If a study can declare its main outcome after seeing the data, it can always find something that looks positive and call that the plan. Registering the intended outcomes in advance locks down what the study set out to measure, so that later reporting can be checked against it.

The WHO dataset and the registry network

Registration only works if entries are consistent and public. The World Health Organization coordinates this through a platform that connects registries around the world and defines a common minimum dataset, a set of items every registered trial should carry, covering its design, sponsor, and outcomes.

Acceptable registries are publicly accessible at no cost, searchable, and validated. A trial registered in any registry within this network, which includes national registries and the large United States registry that feeds into it, can be recognized across journals. The network turns scattered national systems into something a reader anywhere can search.

Results reporting is a separate duty

Registering a trial is only half the obligation. The results also need to be reported, both posted back to the registry and, ideally, published. A registered trial that enrolls participants and then never reports anything still leaves the evidence base incomplete, and it means the participants' contribution never informs anyone.

This is why registration and reporting are treated as distinct duties. A study can be perfectly registered and still go dark. Efforts to track reporting rates exist precisely because the second duty is easier to neglect than the first.

Selective outcome reporting, and how registration exposes it

The most useful thing a registry entry gives a reader is a before picture. Because the planned outcomes are on record from the start, anyone can compare them to what the final paper actually reported. When a pre-specified primary outcome is missing from the publication, or a different outcome has been promoted to primary, that switch is now visible.

This does not by itself prove wrongdoing; plans can change for legitimate reasons that a good paper explains. But it moves the change into the open, where it can be questioned, rather than leaving the reader unaware that the reported outcome was not the one the trial was designed around.

How a reader uses a registry entry

You can do this yourself. Take the registration number a paper cites, look it up in the registry, and check three things: whether the trial was registered before it started enrolling, whether the primary outcome in the registry matches the one in the paper, and whether results have been posted.

None of these steps requires statistical training, and together they catch some of the most common ways a published account can drift from the study that was actually planned. A registry entry is one of the few tools that lets an ordinary reader hold a paper to its own original design.