evaluable

When a trial reports a result it does not always report it for everyone who joined. Some participants withdraw, some do not complete the treatment as designed, and some have measurements that could not be taken or could not be used. Those who remain, whose data can be assessed against the outcome being measured, are often described as evaluable.

The word is informal, and it is worth knowing that regulators do not define it on its own. What the international guideline on statistical principles for clinical trials defines is the per protocol set: the participants who complied with the protocol closely enough that their data are likely to show the treatment's effects. The guideline notes that this set is sometimes described as the valid cases, the efficacy sample, or the evaluable subjects sample.

This is worth noticing in a headline. A result given as sixteen of twenty may be describing the evaluable group rather than everyone enrolled. The contrasting approach is the intention-to-treat principle, under which the primary analysis includes all randomised participants whatever happened next. Neither is dishonest and both are standard, but they answer different questions, and the evaluable figure is generally the more flattering of the two.