[Durgical](https://durgical.com/en/) › [DurgiPedia](https://durgical.com/en/durgipedia/) › [Epidemiology](https://durgical.com/en/epidemiology/) › [Bias, confounding and causation](https://durgical.com/en/epidemiology/areas/bias/) › Verification (work-up) bias

# Verification (work-up) bias

- Bias

If only people who test positive get the reference test, sensitivity looks higher and specificity lower than they really are.

Updated Nov 5, 2026

## Key points

- Partial verification: only some, mostly the positives, receive the reference standard.
- Differential verification: positives and negatives receive different reference standards.
- With partial verification of positives, sensitivity is overestimated and specificity underestimated.
- Prevented by applying the same reference standard to everyone, or to a random sample of the negatives.
- If a random sample of negatives was verified, the result can be corrected by extrapolation.

## What it is

Verification (work-up) bias occurs in diagnostic accuracy studies when not all participants receive the reference standard, or receive different reference standards, and the decision depends on the result of the test under study. Typically nearly all positives are verified and few negatives: false negatives are missed, sensitivity is inflated and specificity is underestimated.

## How it arises

- The new test is applied to all participants.
- The reference standard (biopsy, surgery) is invasive or costly and is done mainly in the positives.
- Unverified negatives are excluded or assumed healthy, and the false negatives are lost.

## Worked example

- 1 000 people take a screening test: 200 positive and 800 negative.
- All positives get a biopsy: 120 have the disease and 80 do not. Of the negatives, only 80 (10%) get a biopsy: 8 are diseased.
- Using only the verified: sensitivity = 120 / (120 + 8) ≈ 94%; specificity = 72 / (72 + 80) ≈ 47%.
- Correction: if 8 of 80 verified negatives are diseased (10%), 80 diseased and 720 healthy are expected among the 800 negatives.
- Corrected sensitivity = 120 / (120 + 80) = 60%; corrected specificity = 720 / (720 + 80) = 90%.
- Reading: verifying almost only the positives inflated sensitivity (94% versus 60%) and sank specificity (47% versus 90%).

## Markers

- Partial verification
- Reference standard

## How to prevent or control it

- When reading a diagnostic study, check whether everyone got the same reference standard regardless of the result.
- When the reference cannot be applied to all, clinical follow-up of the negatives can serve as an alternative reference, if reported.

## Common pitfalls

- Excluding the unverified from the analysis without saying so.
- Assuming all unverified negatives are healthy: the hidden false negatives vanish from the calculation and sensitivity stays inflated.

## Memory aid

If only the positives are confirmed, the false negatives never make it into the picture.

## Test yourself

Think of the answer, then tap the question to see it.

## What is verification bias?

The error that occurs when the decision to apply the reference standard depends on the result of the test under study, for example verifying only the positives.

## If only positives are verified, what happens to the estimated sensitivity?

It is overestimated, because false negatives go undetected and are left out of the calculation.

## How is it prevented?

By applying the same reference standard to all participants, or at least to a random sample of the negatives so the result can be corrected.

[Add to my cards (3)](https://durgical.app/estudiar/?add=/en/epidemiology/verification-bias/)

## Connections

## Related 5

- [Spectrum bias](https://durgical.com/en/epidemiology/spectrum-bias/)
- [Sensitivity](https://durgical.com/en/epidemiology/sensitivity/)
- [Specificity](https://durgical.com/en/epidemiology/specificity/)
- [Reference standard (gold standard)](https://durgical.com/en/epidemiology/reference-standard/)
- [Principles of screening](https://durgical.com/en/epidemiology/principles-of-screening/)

[← All of epidemiology](https://durgical.com/en/epidemiology/)

Educational content on epidemiological and statistical methods; the examples use made-up numbers to teach the calculation. Not a clinical or legal guide: current Colombian rules are in the official sources.

Sources: [Verification bias | Catalog of Bias](https://catalogofbias.org/biases/verification-bias/) (Catalogue of Bias Collaboration, 2019); [Spectrum bias | Catalog of Bias](https://catalogofbias.org/biases/spectrum-bias/) (Catalogue of Bias Collaboration, 2019).

## **Spotted a mistake?** We are always improving for the community. Tell us what is wrong and our team will review it. Report a mistake

Representative model, for educational purposes only.

---
Source: https://durgical.com/en/epidemiology/verification-bias/
License: CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/). Cite as: DurgiPedia (durgical.com), https://durgical.com/en/epidemiology/verification-bias/