Thursday, January 31, 2008

Larry's Curb Your Anticipation, Part 11:
Methods v. Application

Up to the Introduction; back to part 10; on to part 12.

By Commenter Larry

We're going to start winding down, making some observations about the things we've discussed in previous parts with application to the Landis case. These aren't long, and each could easily become their own extended discussions, but TBV cut it off, so here they are as bullet points, a few at a time.


Let’s make some brief observations:

  • One of our goals in this article is to place the ISL rules into a proper context. Here on TBV, we’ve frequently discussed ISL criteria such as specificity and interference, but we have not done so in the context of method validation. We need to understand that under the ISL, these criteria are method validation criteria. They speak to the fitness for purpose of a lab’s methods. So, we must be careful how we use these criteria. If we say, for example, that the chromatograms in the FL case lack the specificity required under ISL, we’re not saying that the lab failed to perform its test method correctly. We’re saying that there’s something wrong with the method itself. We’re saying that the method needs to be reevaluated, revised, or possibly thrown out and replaced with a better method.

  • Let’s use two examples to illustrate this point. Both Duck and Mr. Idiot have argued that LNDD’s testing for exogenous testosterone should include analysis of complete mass spectrum data. This is an argument that goes to specificity; in effect, Duck and Mr. Idiot are questioning whether any method for testing for exogenous testosterone could have been properly validated if the method does not use mass spectrum data to determine specificity. In contrast, Ali has pointed out the wide variation between the results found by LNDD during its initial testing of the S17 sample, and the results reached later upon reanalysis of the S17 electronic data files (EDFs). Ali’s point here does not relate directly to method validation – Ali is not expressly questioning LNDD’s methods, but instead is pointing out an apparent failure of the lab to produce good results with its methods. The data pointed to by Ali may or may not indicate that LNDD’s method is flawed; the problem may lie elsewhere (for example, in LNDD’s sampling procedures, or with the training of LNDD’s personnel, or with LNDD’s equipment).

Up to the Introduction; back to part 10; on to part 12.