Field screening test methods: performance criteria and performance characteristics

Field-portable test methods may be quantitative, semi-quantitative, or qual itative and screening methods are often used in the field to determine if t he concentration of a toxic substance exceeds regulatory or recommended sta ndards or action levels. For on-site analysis, accurate quantitative tests for held measurements may not be available, depending on the analyte(s) or specific field situation. Thus, in lieu of more definitive test methods, sc reening tests which are based on qualitative or semi-quantitative methods a re often used for making immediate decisions in the field, e.g. for complia nce or risk assessment. Also, quantitative methods may be used for screenin g purposes in many instances. To ensure the quality of these screening test s and the decisions that are made based upon their results, screening metho ds need to be evaluated with sufficient data and should meet basic performa nce criteria prior to their being employed for decision-making purposes. Al though quantitative, semi-quantitative and qualitative methods demonstrate different characteristics, it is desired that the performance criteria for all three method categories be consistent. If there is consistency, then on e can have a sound basis for selecting the most appropriate test(s) for a g iven application. In order to unify the performance criteria for the differ ent types of methods, a performance function is used to characterise both q ualitative and semi-quantitative methods; in turn, this performance functio n is related to that for quantitative methods. False negative rates, false positive rates, sensitivity and specificity are key characteristics of scre ening methods that can be determined from the pertinent performance curves. The performance characteristics of each method are related to the uncertai nty region that is associated with each method and the applicable uncertain ty regions can be gleaned from the performance curves. Also, various option s for using multiple test results to improve decisions based on test result s are provided. Published by Elsevier Science B.V.