One factor affecting how well autoverification works is the proportion of abnormal values. Autoverification works best when:
(1) the number of abnormal values is low.
(2) the analytical method and instrumentation are reliable.
A result may be autoverified if it meets a series of rules:
(1) the quality control is within range
(2) the result is within the normal reference range
(3) the result passes delta check
A result should not be autoverified if:
(1) the result is a critical value (this would fail item 2 above)
(2) the result is an absurd value (this would fail item 2 above)
(3) the result is used to reflex other tests (unlikely for a normal value)
If the result does not meet the criteria for autoverification, then it is reviewed by the technologist who may decide to:
(1) release the result
(2) repeat testing on the same specimen
(3) repeat testing on a new specimen
(4) contact a supervisor or clinician
Autoverification needs to undergo validation, periodic review and troubleshooting.