Analytical validity, clinical validity and clinical utility

In one sentence

Analytical validity asks whether a test measures correctly, clinical validity asks whether its result relates to a clinical condition, and clinical utility asks whether using it helps patients.

The intuition

Imagine a bathroom scale that always reads your weight correctly. That is useful, but it cannot tell you which medicine to take. A sound measurement, a meaningful medical interpretation and a helpful action are three different achievements.

This analogy has limits. Cancer tests may measure a molecule, produce a risk estimate, or classify a tumor. There may be no single perfect reference. Still, the three questions help you keep the evidence in order.

How it works

Analytical validity concerns the measurement itself. Does a test find the signal it claims to find, with acceptable error? Studies examine accuracy, repeatability, interference and the smallest signal the method can reliably detect. Validation depends on the specimen and method. Performance in carefully prepared material may not transfer to damaged clinical samples.

Clinical validity concerns the link between a result and a defined clinical condition or outcome. Does a positive result identify disease, or does a score relate to later recurrence, in the relevant population? The setting matters: a test studied before treatment may behave differently after surgery. A useful association must be distinguished from a treatment-selection claim; see prognostic versus predictive biomarkers.

Clinical utility concerns using the information. Compare a specified test-guided strategy with a meaningful alternative. Does the strategy improve health, avoid harm, or improve a decision patients value? A test can predict risk accurately while the best response to that risk remains unknown. Utility includes downsides such as unnecessary treatment and missed disease.

These are dimensions of evidence, rather than prizes awarded once for every future use.

Question	Evidence to inspect	Tempting overread
Does it measure correctly?	Reference samples, repeat runs, detection limits, failed samples	A reliable measurement must predict benefit
Does the result mean something clinically?	A defined population, outcome and follow-up; independent validation	Association means the proposed action works
Does acting on it help?	Outcomes of the stated decision strategy, including harms	A treatment changed, therefore care improved

Why it matters in cancer

A laboratory can report an alteration correctly without showing that a named drug helps people with that alteration. Likewise, a recurrence-risk model can separate higher- and lower-risk groups without showing that adding a particular treatment improves their outcomes.

Worked example

In a fictional study, a blood test repeatedly detects a tumor-associated signal. A separate cohort shows that people with that signal recur more often. Neither study tests whether starting Drug A after a positive result helps. Measurement and risk association have support; the Drug A strategy remains a research question.

Common confusions

Validation is use-specific. A study in one disease, specimen or treatment setting does not establish every proposed use.
Changing a decision is not itself a health benefit. The changed decision may help, harm, or do neither.
Regulatory status is a separate question. Laboratory and product labels describe oversight or intended use; they do not replace the evidence for a claim.

Sources and scope

Source check: October 9, 2026. The scale and blood-test examples are fictional. Expert and learner review remain pending.

MedlinePlus Genetics: how to judge a valid and useful genetic test — the three evidence questions.
National Human Genome Research Institute: regulation of genetic tests — definitions and the distinction between laboratory quality and clinical evidence.
FDA–NIH BEST: predictive biomarker — a treatment-effect claim differs from a general outcome association.