In statistics, verification bias is a type of measurement bias in which the results of a diagnostic test affect whether the gold standard procedure is used to verify the test result.
[1][2] In clinical practice, verification bias is more likely to occur when a preliminary diagnostic test is negative.
Because many gold standard tests can be invasive, expensive, and carry a higher risk (e.g. angiography, biopsy, surgery), patients and physicians may be more reluctant to undergo further work-up if a preliminary test is negative.
[citation needed] In cohort studies, obtaining a gold standard test on every patient may not always be ethical, practical, or cost effective.
[citation needed] In most situations, verification bias introduces a sensitivity estimate that is too high and a specificity that is too low.