Error rates in cervical cytological screening tests.

In an attempt to clear up the confusion evident in the literature concerning some aspects of cervical cytological screening tests, the principle is stated upon which the data acquired in a series of tests should be tabulated. The table is used to define several rates or probabilities, most of which express the rates at which errors occur. Certain rates are distinguished as of basic importance, others playing only a secondary role. The inter-relations of the rates are displayed as equations and reference is made to a set of conversion tables constructed from the equations. As an illustration, the data from a particular published paper is treated in detail, showing how the various rates can be calculated or estimated, and, in passing, also demonstrating their high degree of uncertainty.