Just about every lab has embraced the modern molecular testing age. But when you buy that new bit of equipment do you stop to consider the implications on the test sensitivity and specificity? Every time the manufacturer "updates" the software, do you re-standardise the test? Do you check for failure over time of the optical system behind that melt curve? (lamps decay over time). Do you re-calibrate the test EVERY time you open a new batch of chemicals? Do you cross-check the sensitivity and specificity of all of the machines in your lab? Some manufacturers machines are better at detection than others. Do you routinely test for lab contamination of reagents (especially problematic in University labs, I found)?
I used to do lab audits - but see https://www.cochrane.org/news/how-accurate-are-routine-laboratory-tests-diagnosis-covid-19 as an example. Of 67 tests, only three of the tests had both sensitivity and specificity over 50%. That's "heads or tails" territory.
If you want to lift your game - read Bustin S.A., Benes V., Garson J.A., et al. The MIQE guidelines: Minimum information for publication of quantitative real-time PCR experiments. Clinical Chemistry. 2009;55(4):611–622 and Smith M. Validating Real-Time Polymerase Chain Reaction (PCR) Assays. Encyclopedia of Virology. 2021:35–44.
I didn't confine my question to medical testing, indeed most testing is non-medical import-export, environmental, and food quality testing - and a vast pile of research papers using PCR to identify stuff. I also (as I said) did lab audits and I became interested in science fraud. I observe from published papers that nobody these days mentions any quality control steps - many don't even mention a control. If you go back to first principles and do PCR properly, you will find, as Bustin writes, that you can inadvertently (or not) manipulate the results. I have seen a manufacturers PCR thermocycler that will not amplify the OIE recommended primers for whitespot syndrome virus in prawns yet a competing manufacturers thermocycler on the same bench, using the same mix, will do so. I have seen a lab (a government testing lab) fail to get a positive PCR amplification because they did not use the published method specified taq polymerase. They at least, recognised and fixed the problem - but would a university lab? I have also seen a commercial lab deliberately manipulate the melt curves to provide either positive or negative results as the client requires. Then I see the standard university lab, the sloppy practices, the out of date reagents, the lack of controls and wonder why anyone accepts the (published) result? Just as a benchmark, I suppose, the last lab I actually worked in, we spent well over 1/3 of our time just checking the specificity and sensitivity of our tests as reagents came and went. When did you EVER see a paper from a university team that routinely report the specificity and sensitivity of the test they used? More topically - after the last manufacturer "software upgrade" how do you know that your machine still works?
- Badges
- Science topic
- Similar topics
- Filing