Half of my career was as a chemical analyst, and my training included how to consider variability when interpreting test results. For all the discussions on PSA I have not seen any information on how variable the nature of the test is.
Although results might be reported to three decimal places this is an indication of how sensitive the test is but does not indicate how much fluctuation the nature of the test might introduce.
This variation is statistically determined by doing multiple tests on the same sample, then working out the statistical standard deviation of the results. Normally we report the results to the 3 standard deviation level, so a result would formally look like:
0.245 +/- 0.x at the 99.7 confidence level, where x is the variability caused by the test and which we do not know. If it was something like 0.05 then we would say we are 99.7% confident that the result would lie somewhere between 0.195 and 0.295 ug/l .
So when we compare two results without knowing how variable the test is you should be cautious about drawing too many conclusions especially for results close to the detection level. This concern can be ignored though for higher levels of PSA and for sets of results that are hugely different in value.
This situation becomes even more complex because (as far as I know) we do not have a good handle on how much natural variation of PSA level there is with an individual after radical prostatectomy caused by natural/environmental factors.
If you want more of an eye-opener, have a look at lab to lab variations reported in Australia by this article:
https://www1.racgp.org.au/newsgp/clinical/psa-testing-dilemma-a-pathologist-s-perspective
Edited by member 21 Jan 2025 at 21:51
| Reason: Not specified