Medical trials frequently measure something other than what patients actually want changed. The substitute is called a surrogate endpoint, and understanding when it works clarifies a recurring source of confusion.
The substitution buys time and size
Outcomes such as death, heart attack or fracture take years to accumulate in sufficient numbers, requiring long trials with many participants.
A surrogate is a measurable marker expected to track that outcome: a laboratory value, a scan finding, a physiological measurement.
Because markers change within months, trials using them are shorter and smaller, which is the entire reason they are used.
Correlation is not the property required
A valid surrogate must lie on the causal pathway between the treatment and the outcome, so that changing it necessarily changes the outcome.
Many markers are merely associated with an outcome because both are driven by an underlying process, without the marker causing anything.
Moving such a marker directly leaves the underlying process untouched, and the outcome does not follow.
A drug can have effects beyond the marker
Even where a marker is genuinely causal, a treatment acts on more than one thing.
A drug can improve the surrogate through one mechanism while causing harm through another, and the net effect on the outcome then depends on the balance.
This is the pattern behind several historical cases where a treatment improved a measured value while outcomes worsened, which is how the limits of surrogates were established.
Validation is specific to context
A surrogate validated for one drug class does not automatically transfer to another, because the second class may act through a different mechanism.
It may not transfer between populations either, since the relationship between marker and outcome can differ by age or by disease stage.
Formal validation requires showing that the treatment effect on the surrogate reliably predicts the treatment effect on the outcome across multiple trials, which is a demanding standard.
How to read a result built on a surrogate
A trial reporting an improved marker has established that the drug does something measurable. It has not established benefit unless the surrogate is validated for that context.
Confirmatory studies with hard outcomes are the mechanism for closing that gap, and they sometimes overturn earlier expectations.
This is one reason headlines about a treatment lowering a number should be read as an intermediate finding rather than a conclusion about health.