Skip navigation

Screenwatch and PSA

Bandolier has received a number of letters saying how helpful readers have found our Screening Whitelist (approved screening programmes) and Screening Blacklist (programmes and tests which should not be used). We have also been taken to task for using "blacklist" as a pejorative term. Perhaps we were over-simple, for, as a radiologist once said of mammographic screening, "it is not all black and white in radiology".

Scottish law is better

In English law one is innocent or guilty. Scottish law has a third category, not proven, which is as useful to epidemiologists as it is to lawyers. The three verdicts for screening epidemiologists are:-
  • innocent - doing more good than harm
  • guilty - doing more harm than good
  • not proven - of unknown effectiveness

Only the innocent should go freely into the health service, providing, of course, that they are also affordable at reasonable cost.

We have hitherto divided screening tests into black and white, guilty and innocent. The new Executive Letter on Clinical Effectiveness, signed by Alan Langlands, emphasises in an appendix that a number of interventions are currently the subject of research and should not be used except if they are part of a well-designed research project, and a number of these are screening tests.

Not proven, not innocent

Interventions are sometimes introduced into the service under the disguise of "research" and then slip into routine clinical practice. This is certainly the case with some of the new genetic tests. The category of "not proven" allows us clearly to mark interventions which should be introduced only as part of a research study which has been through an Ethics Committee and is fully costed and supported. Most of these tests were on our Screening Blacklist, and in a simple binary view of the world should remain there and not be introduced as services. Only the innocent, interventions which have been shown to do more good than harm, should be introduced to the service.

PSA and prostate cancer screening

There may well eventually be useful screening programmes for prostate cancer, combined with effective remedies. The situation now is far from that. Just looking at a few of the issues, as in the figure, gives a taste of the problems.
For instance, even though the figure of 4 µg/L for PSA is commonly taken as an action limit, the 95% confidence interval for that measure in a typical laboratory is about 3.2-4.8 µg/L. If a sample with a "true" level of 4 µg/L were tested in every laboratory in the UK, the range might be as wide as 2.5-5.5 µg/L. This is not to denigrate the laboratories, since this would be typical within assay variation and between laboratory performance for this sort of test, with many different reagent kits and methods available.

It is also interesting to see what a literature search produces. Though there are perhaps over 1,000 papers published which include PSA measurements in the last three years, only two (so far as Bandolier could find) specifically addressed the effect of ejaculation.

One [1] involved men under 40 years (so may be of questionable value in older men) and showed that PSA values fell from a pre-ejaculation average of 1.4 µg/L to a one-day post-ejaculation value of 0.2 µg/L.

Another paper [2] found no significant relation between PSA levels and ejaculation; it looked at men in their '20s whose median PSA was 0.6 µg/L (and mode 0.4 µg/L). It used a test which was not designed to measure PSA accurately below about 1 µg/L. So this study almost certainly lacked sensitivity to see any change, yet this reference has been taken by some urologists as a positive indication that ejaculation had no effect on serum PSA concentrations.

So what about men over 50 years, with higher PSA values? - apparently no evidence on that one, yet that is exactly the patient group where evidence is most needed. It is interesting to relate this to the article on diagnostic tests in this edition of Bandolier.

So without wanting to preempt considerable primary study and overview work presently being undertaken in this area, it is easy to see why one possible screening programme should have the status of 'not proven', despite the considerable pressures to introduce it.


  1. R Simak, S Madersbacher, Z-F Zhang, U Maier. The impact of ejaculation on serum prostate specific antigen. Journal of Urology 1993 150: 895-7.
  2. WJ Glenski, GG Klee, EJ Bergstralh, JE Oesterling. PSA: establishment of the reference range for the clinically normal prostate gland and the effect of DRE, ejaculation and time on serum concentrations. The Prostate 1992 21:99-110.

previous or next story in this issue