Method validation
It is almost universally accepted that any method used in forensic work must undergo some type of validation to prove that it is fit for purpose, that is that it will provide anaccurate answer for the types of samples being analysed and is sufficiently robust to be used by different analysts. However, there is considerable disagreement about the extent of validation required. Perhaps arguably, the extent of validation depends on the specificity and sophistication of the assay, whether the assay is in routine use and the potential consequences of producing an inaccurate result. For example, the legal consequences of a quantitative error in measuring an endogenous analyte, such as adrenaline (epinephrine), are obviously more serious than for, say, strychnine, which ordinarily should not be present in any amount. Similarly, to report an amfetamine analogue as present instead of a decongestant, such as pseudoephedrine, can have seriousconsequences. Qualitative methods, such as immunoassay, should be validated for specificity and limit of detection (LOD). It is accepted that most immunoassays cross-react to some extent with analytes other than those targeted, and it is important to know the extent of that cross reactivity, particularly for structurally related compounds. LOD is important because a class assay (e.g. opiates, amfetamines) may be far less sensitive for some drugs than for others, which allows the possibility of a false negative (the drug not being detected when, in fact, it is present). Where an assay is used to analyse matrices other than those for which it was designed, appropriate validation should be performed. For GC-based and other chromato graphic drug screens, it is not usually practical to determine the LOD for every analyte expected to be detected, but the LOD can be determined for representative examples. If the laboratory is asked to determine whether a particular drug is present in a specimen, the laboratory should have some idea what the sensitivity of the assay is, and that it can at least detect potentially toxic concentrations.
A quantitative assay should be validated for accuracy, precision, linearity and LOD. However, it can be argued that some assays are, by their design, self-validating. For example, if a GC-MS SIM assay uses a good internal standard (ideally a deuterated one), numerous (e.g. six) matrix matched calibrators, at least one independently prepared matrix-matched control and appro priate acceptance criteria for the calibrators, the calibration as a whole and to the control, the assay could be described as self-validating. If the assay was not accurate, the control result would be out of range. If the assay had poor precision, one or more of the calibrators would not read within an acceptable percentage, when read against the calibration. Specificity can be demonstrated by the appropriate choice of ion ratio qualifiers and lack of chromatographic interference with those ion chromatograms. For some analytes, determination of LOD or limit of quantification (LOQ) is irrelevant if a cut-off is used, or where the analyte concentration is only accepted if within the valid calibration range. Demonstration of that cut-off may be satisfied if it represents the value of the lowest calibrator or control.
