A brief clarification on PSA testing

In light of the new recommendations on PSA testing, here’s a repost.  Originally posted on 10/16/2011–PalMD

The recent recommendation released by the U.S. Preventative Services Task Force against using the PSA blood test to screen for prostate cancer has left many patients confused, and many doctors unsurprised.  We’ve known for a while that in the aggregate, the data do not show that screening for prostate cancer using the PSA test saves lives. We’ve also known that it may lead to complications of treatment.

Learning how to screen for and successfully treat this cancer is an ongoing project.  For example, some of the negatives of screening (that it leads to invasive procedures with significant side effects) can be mitigated as our procedures continue to improve.

But that’s a discussion for another time.

What I’m finding my patients asking is “Should I get a PSA test and why?”

The most important clarification before answering this question is explaining the difference between a screening test and a diagnostic test.  Screening, by definition, looks at people without symptoms for any signs of disease.  If a man is having urinary symptoms—difficulty emptying his bladder, frequent urination, decreased stream, and others—checking a PSA is no longer a screening test: it is a diagnostic test, and the entire current debate is largely irrelevant.

If you have urinary symptoms you are not using a PSA to screen for cancer, but to look for it in someone who has symptoms pointing toward a prostate disorder.  That’s a very different situation requiring a different set of questions.


  1. D. C. Sessions

     /  October 16, 2011

    Some of the downsides that come into the analysis worry me a bit. Yes, it’s possible (and I’ll even grant likely) that when confronted with a scary test result men might be led to overdo treatment. On the other hand, I confess to a bias in the direction of, “knowledge is good.”

    Having been in the “small and soft is now a good thing” set for more than a decade, I’m still getting routine PSA checks along with my annual bloodwork. And if it comes back elevated, my main response (which I’ve discussed several times with my PCP) will be, “that’s interesting. Let’s see what it looks like next year.”

    Funny-looking moles are another matter altogether.

  2. Because the new recommendations are largely based on the belief that the harms due to treatment outweigh the benefits of treatment, I do not think that whether the PSA testing is prompted by symptoms or is simply a general screening really makes any difference. There might be a difference in the benefit-cost calculation if the PSA testing was prompted by a family history of fatal prostate cancer, which would increase the probability that any prostate cancer would be aggressive.

    However, I disagree with your statement, which agrees with the Task Force, that “the data do not show that screening for prostate cancer using the PSA test saves lives”. The data is no doubt mixed, but the European study, which is arguably the best study, does estimate that screening reduces prostate cancer specific mortality.

    The European study’s basic finding is that the average difference between the screening and control group was sufficient that for every 48 diagnoses of prostate cancer due to screening, 1 prostate cancer death was averted. But this is as of an average follow-up of 9 years after the experiment. More recent analyses of the European data suggest that the “number needed to treat” to reduce one prostate cancer death is 18 to 1 as of 12 years after the screening began. The NNT might continue to go down after 12 years, but 12 years is as far as it is currently feasible to go with the European data.

    I don’t think a 5% reduced risk of prostate cancer death due to treatment induced by screening is a small reduction in the probability of death. And this is as of 12 years after screening was begun. The time interval from actual treatment would be less.

    Of course, this does not mean that all men would or should choose treatment, or should choose screening. The tradeoff is a 5% reduced risk of death versus a considerably larger probability of side-effects. Different men will value this tradeoff differently.

    Now, perhaps you are referring to the finding that we can’t find a screening effect on overall mortality. But given the sample sizes currently available, there is no way the relatively small changes in prostate cancer mortality out of the ENTIRE screening group over a 9 year follow-up can be detected in overall mortality statistics for which the average will be 10% or more in these samples of men. You would need experiments with millions of men to reliably detect effects on overall mortality. The relative effect is too small to be detectable. You can detect the effect on prostate cancer specific mortality, because it is a larger relative effect.

    If we are to use OVERALL mortality results to judge all screening tests, we are likely to be unable to find conclusive effects in any sample period for which the probability of death from the screened condition is of modest size relative to overall mortality. But as the major benefits and costs of many screening mechanisms, such as PSA screening, are due to the TREATMENT benefits and harms, not the screening itself, it can still be optimal to use screening even if the incidence of deaths to the condition over the follow-up period is small relative to overall mortality.

  3. Samantha

     /  May 22, 2012


    A screening test is a shot in the dark to see if a particular person may have a particular disease. In the case of the PSA you’re testing an otherwise healthy person to see if they may have prostate cancer. If they test positive on the PSA you may continue with further tests such as a biopsy, which may have significant side-effects.

    A diagnostic test is an attempt to rule out a particular etiology with regards to an observed symptom. If your patient presents with a symptom that could indicate prostate cancer you’re going to need to rule it out somehow and if you didn’t have the PSA you’d have to go straight to biopsy. In this case, the PSA prevents an unnecessary procedure.

    • The costs and side effects of biopsies is NOT the main reason that the USPSTF is arguing against screening. Rather, they are arguing that screening leads to biopsies and consequent prostate cancer diagnoses that then leads to treatment decisions in which, the USPSTF argues, the harms outweigh benefits.

      I disagree with their reading of the evidence on treatment harms and benefits. For many men, it is by no means obvious whether or not surgery is warranted if it reduces the risk of death after 10 years by 5%, but in 30% of the cases you have serious side-effects after surgery. Some men will choose to minimize risk of death, others will choose a higher risk of death and fewer side-effects.

      Now, this relies on the European study’s estimates. If one instead believes in the U.S. study’s estimates, then screening has no benefits. But, in the face of this conflicting evidence, I do not see how the USPSTF can reasonably argue that there is “moderate or high certainty” that harms outweigh benefits or that there are no net benefits. There is nothing close to moderate certainty for this policy issue.

  4. atpants

     /  May 22, 2012

    Yeah, the USPSTF requires pretty strong evidence to be impressed, and PSA is looking worse and worse for screening. The American Urological Association predictably disagrees.

  5. I’d say if you present symptoms for anything potentially fatal, it might be time for a test.

    18’s a pretty high NNT. I’d look at risk factors first. How old is the patient? Does he smoke? If he is balding, when did he start balding? (Early pattern baldness can indicate high levels of dihydrotestosterone.) Does his family have a history of prostate cancer or cancers found on the same loci? (Once we’ve gotten this whole human genome diversity project done, finding people at risk for cancer will be a lot easier.)

    Of course, then there’s the legal issue: Even if the test is next to worthless, it’ll still be done because some people still say it should be done. Diets with a 60% or greater carbohydrate ratio are still recommended by many physicians for diabetes(!) because that’s the American Heart Association’s standard, though even they’re changing their view, mostly because that means pumping in more insulin, leading to more insulin resistance, leading to an increased risk of an ischemic event. Why do doctors still recommend this? Because if they don’t, their patients (or their patients’ families) might sue if the patient does have a heart attack. Never mind if the patient actually STUCK to his low-calorie regimen. Lawsuits make every industry just a tiny bit more conservative in mentality.

%d bloggers like this: