Why Screening Tests Need A Cut-Off Point

Screening tests produce a result described as normal or abnormal. Underneath almost every one is a continuous measurement and a chosen line drawn across it.

Biology does not come in two groups

Most things measured in screening, such as pressure, concentration or density, vary continuously across a population without a natural gap between healthy and unhealthy.

People with a condition and people without it overlap on almost any single measure. The distributions sit on top of each other rather than side by side.

A test must nonetheless produce an actionable answer, so a threshold is imposed on a scale that has no inherent step in it.

Moving the line trades one error for another

Setting the threshold lower catches more true cases, because fewer affected people fall below it. It also captures more unaffected people.

Setting it higher reduces false alarms and unnecessary follow-up, at the cost of missing cases that sit just under the line.

There is no setting that eliminates both errors while the distributions overlap. The choice is about which error is more tolerable, not about accuracy.

Consequences on each side decide the setting

Where a missed case is dangerous and rapidly progressive, thresholds are set to catch nearly everything and accept a high false alarm rate.

Where the follow-up test is invasive, risky or expensive, the threshold is set more conservatively because the harm of a false alarm is substantial.

This is why two tests of similar statistical quality can be pitched very differently. The mathematics is the same; the cost of each mistake is not.

How common the condition is changes everything

The chance that an abnormal result reflects real disease depends heavily on how common the condition is in the group being tested.

In a rare condition, even a very accurate test produces more false alarms than true findings, simply because there are so many more unaffected people being tested.

This is the main reason screening programmes are offered to defined groups rather than everyone, and why age ranges and risk criteria appear in invitations.

What an abnormal result actually says

An abnormal screening result is a statement that further testing is warranted, not a diagnosis. Screening sorts a population; diagnosis examines an individual.

Thresholds also change over time as evidence accumulates, which is why a value considered borderline in one period may be handled differently later.

Interpreting any individual result requires the context of a person's history and other findings, which is work for a clinician rather than a comparison against a printed range.