Inclusion and Exclusion Criteria

Inclusion criteria state the characteristics a person must have to enter a study; exclusion criteria state what disqualifies them despite meeting the inclusions. Together they define the enrolled population and therefore the boundary of the conclusion. They are written into the protocol before recruitment for two reasons: to select a group in whom a treatment effect can plausibly be detected, and to keep out people for whom the risk of participation is judged unacceptable.

Their effect on interpretation is easy to underestimate. The STEP obesity programme illustrates it, since STEP 1 studied adults with obesity and excluded type 2 diabetes while STEP 2 was conducted specifically in people with type 2 diabetes and produced smaller mean weight reduction, so quoting one figure for both populations misstates the drug.

For a reader the criteria are the first place to check whether a result applies to the patient in front of them. A trial reporting a large effect in participants aged 18 to 65 with no significant comorbidity, screened for adherence and free of interacting drugs, has demonstrated something real about that group and nothing directly about an 80-year-old taking nine medicines. Tight criteria raise the chance of detecting a true effect by reducing noise, and narrow external validity by the same amount.

Restrictive criteria are often invisible in the abstract, so the effect size travels into practice without the population attached to it. The less obvious problem is a screening cascade that selects for something the criteria never mention, such as tolerating a run-in period, attending frequent visits, or being willing to inject, all of which enrich the sample for likely responders. When a peptide is marketed on trial numbers, ask how many people were screened for every participant randomised.