Combining the outcomes of endoscopy, laboratory testing, and professional judgement in gastroenterological decision-making.

Abstract

The need to combine the results of multiple separate tests or make decisions based on the judgement by multiple experts permeates the clinical and professional practice of a gastroenterologist. The present analysis is aimed at delineating four different means to combine results of multiple tests and discuss their applicability and limitations. In serial testing, the overall test outcome is rated as being positive if the outcome is positive in all individual tests applied in series. Serial testing increases the overall specificity at the expense of decreasing sensitivity. In parallel testing, the overall test outcome is rated as being positive if the outcome is positive in any of the multiple tests. Parallel testing increases the overall sensitivity at the expense of decreasing specificity. In majority testing, the overall test outcome follows the rating by the majority of testers. Majority testing avoids the trade-off between sensitivity and specificity associated with serial and parallel testing and leads to relatively high overall values for both test characteristics. For majority testing to function well, however, it requires many independent testers, which can render this method costly and impractical in the clinical setting. Sequential testing applies to situations in which the output of a previous test provides the input for a later test. Sequential testing generally results in an overall test with lower sensitivity and specificity values than any of the individual tests and should be avoided if possible. All methods to improve test performance by combining the results of multiple tests have limitations and need to be regarded with skepticism. In the long run, the best solutions are provided by improvements of the individual test or performance by the individual diagnostician/reviewer through training and acquisition of new skills.