A trial’s enrollment number changed. What actually happened?

Why are most enrollment changes not really changes?

Because the registry stores one enrollment field with a label, and the label matters more than the number. While a trial runs, the field holds an estimated target. When it wraps up, the sponsor replaces it with the actual count. That replacement looks like a change, and in our archive it accounts for 67,955 of 82,987 enrollment changes: 82%.

So four out of five times an enrollment number moves, nothing about the plan changed at all. The trial finished counting. Any monitoring approach that treats every enrollment edit as a signal will bury the twenty percent that are, which is why the first question about any enrollment change is the label: did an estimate become an actual, or did a forecast change to a different forecast?

When a sponsor really revises a target, which way does it go?

Up, twice as often as down. Of the 10,501 genuine revisions in our archive, where a still-forecast target changed to a different forecast, 67.1% raised the target and 32.9% cut it.

That is the opposite of the folk assumption, and it makes sense on reflection: sponsors expand trials that are working, add arms and sites and populations, more readily than they publicise retreat. It also sharpens the rare event. A genuine, labelled cut to a still-recruiting target is a one-in-three minority of a category that is itself only 18% of enrollment changes. When one lands on a trial you compete with, it has earned your attention.

Do trials actually hit their enrollment targets?

A number circulates in this industry that 80% of trials miss their enrollment targets. Our data cannot confirm it, and the honest measurement is more interesting. Comparing each trial’s actual final enrollment against the last target its sponsor stated, across 67,955 finalisations:

Two things to hold at once. Against the final stated target, hitting enrollment is close to a coin flip and the median trial lands at exactly 100%. And yet more than a fifth of trials finalise with fewer than half the participants their own record promised. The important caveat cuts against us as much as anyone: sponsors revise targets while trials run, so this measures delivery against the last stated plan, not against the original ambition. A trial that cut its target from 500 to 200 and then enrolled 210 counts here as exceeding it.

How should you read a single enrollment change on a watched trial?

In this order. Check the label first: an estimate becoming an actual is a completion event, and the interesting question becomes how the final count compares with the target, not that it changed. If it is a genuine revision, the direction is the signal: upward revisions are the common case and usually read as expansion; a labelled cut is the rare event worth escalating. Then look at what moved with it: an enrollment cut arriving in the same update as a completion date slip is a co-ordinated edit, and co-ordinated edits are how real schedule resets look in registry data.