Choosing a diagnostic test threshold

A diagnostic threshold turns a continuous test result into a decision, such as whether to investigate a patient further. Changing the threshold usually changes the balance between missed cases and false-positive results.

There is no universally best cut-off. The useful threshold depends on the intended use, patient population and consequences of the decision.

Understand the trade-off

Sensitivity describes the proportion of people with the condition who test positive. Specificity describes the proportion without the condition who test negative. These measures require a suitable reference standard and representative study population.

A receiver operating characteristic (ROC) curve shows sensitivity against the false-positive rate across thresholds. It helps describe test performance, but a statistically attractive point is not automatically the best clinical or economic choice.

Follow the clinical pathway

Ask what happens after each result. A false positive might lead to further testing, anxiety or an unnecessary procedure. A false negative might delay treatment. The size and likelihood of these consequences affect the preferred threshold.

Also consider prevalence: positive and negative predictive values depend on how common the condition is in the tested population. Evidence from one setting may not transfer directly to another.

Evaluate and validate

Compare candidate thresholds using clinically relevant outcomes, resource use and costs. Define the selection method clearly and validate performance in data not used to choose the cut-off.

For adoption decisions, modelling can explore the downstream effects of different thresholds and identify uncertainties worth investigating. It complements clinical validation rather than replacing it.

Further reading

FDA: reporting results from studies evaluating diagnostic tests.