For test developments, one should have very precise ideas about the purpose they will serve and in what kind of environment they will be used. It should be self-evident that diagnostic goal(s) have to be clearly defined and an appropriate technique carefully chosen before a costly test development is started. Only a precise definition of the diagnostic goal(s) allows a relevant validation.

It is obvious that one single diagnostic test cannot fulfil all requirements. The prime test parameters, sensitivity and specificity, must be carefully adapted to optimise the predictive value for either a positive or for a negative test result. One should not forget that predictive values depend on prevalence. This means that the threshold value of a quantitative test must be chosen according to the level of prevalence and to the purpose that the test is to serve. For example, for HIV diagnosis, maximal specificity (excluding false-positive results) is a high priority for the diagnosis of individual cases. On the other hand, one would aim at a high predictive value for negative results, accepting false-positives, for screening blood products for transfusion. For many diagnostic issues, the diagnostic strategy combining a sensitive screening test to be followed by a more specific confirmatory test might be an acceptable solution.