Validation is a process that establishes the extent to which a technique generates meaningful data when it is applied, for a specific purpose, in a specific experimental context. This review develops a framework for the performance of validation studies of the application of PCR-based techniques to the detection of fish pathogenic bacteria in the environment. It is argued that the potential for the application of these techniques to environmental microbiology will only be realised if adequate attention is paid to their validation. The framework developed here advocates the evaluation of the quantitative, qualitative and reliability criteria of a technique at four levels of experimental complexity. The problems encountered in assessing each of these criteria of validity at each level of complexity are discussed. Three of the levels of experimental complexity, in vitro studies, studies in seeded matrices and studies in incurred matrices can be performed within the laboratory. Consideration of the problems inherent in these laboratory-based validation studies demonstrates that they cannot provide evidence of adequate validity for field applications but it is suggested that their performance will be cost-effective. Their major role is in providing rational grounds for rejecting techniques whose application to environmentally derived matrices has been shown to be invalid. It is argued that, at the fourth level of experimental complexity represented by field studies, comparative and predictive validation provide powerful tools for establishing the value of any application of a PCR-based technique. Such studies are, however, time consuming and their performance cannot be envisaged with techniques that could have been rejected during laboratory-based studies. (C) 1998 Elsevier Science B.V.
