Outcome Switching: Why the Registered Protocol Is the Receipt

Outcome switching is when a trial reports a different primary outcome than the one it registered in advance, letting a post hoc finding pose as a planned result. The registered protocol is a timestamped record of intent, so comparing it against the published paper is the simplest integrity check a reader can run. Studies find switching is common, and that significant outcomes are reported more fully than null ones.

What outcome switching means

When a trial is registered, its investigators declare in advance which outcome is the primary one the study is built to answer. Outcome switching is when the published paper reports a different primary outcome: the original is quietly dropped, demoted to a footnote, or replaced by something that turned out to look better.

Because the switch happens after the results are known, it lets a chance finding masquerade as a planned discovery. The reader who sees only the paper has no way to know the headline was not the original question.

Why the registered protocol is the receipt

A public registry entry, filed before results exist, is a timestamped record of intent. It is the receipt you check the final bill against. Without it, you cannot tell whether an impressive primary outcome was the plan all along or was selected from among many because it happened to reach significance.

This is why prospective registration became a condition of publication at major journals and a legal requirement for many trials. The registry turns a private intention into a public commitment that anyone can later verify.

What the evidence shows

Comparisons of protocols with publications have repeatedly found switching. One study of trials with available protocols found that a majority had at least one primary outcome changed, introduced, or omitted between protocol and publication.

The same work found that statistically significant outcomes were substantially more likely to be fully reported than nonsignificant ones. That selective visibility is exactly how outcome reporting bias tilts the published record toward apparent benefit.

The COMPare project

One team took this to the front line, checking every trial in five leading journals against its registered outcomes and writing a correction letter for each discrepancy. They found trials reporting many outcomes that were never prespecified while dropping some that were, and that journals often resisted publishing the corrections.

The exercise showed two things at once: how common switching is, and how weak the enforcement can be even at journals that formally endorse reporting standards. Endorsing a standard and applying it turned out to be different things.

How switching poisons meta-analysis

The damage does not stop at one paper. When outcomes are reported selectively based on their results, systematic reviews that pool those papers inherit the bias without knowing it.

Analyses of Cochrane reviews found that a meaningful share contained at least one trial with high suspicion of outcome reporting bias, and that adjusting for it reduced or in some cases erased the pooled effect. A biased brick makes a biased wall, so the integrity of the synthesis depends on the integrity of the reporting beneath it.

How to check a trial yourself

Find the registry entry or protocol, note its date, and compare its prespecified primary outcome with the one headlined in the paper. Check whether the registration predates enrollment, since a record filed after data collection began offers little protection.

Watch for a swarm of secondary outcomes with one starred winner, which is a common signature of a search for significance. None of this requires statistics. It requires only the willingness to read the paper against its own earlier commitment.