A Prostate Screening Picture Worth A Thousand Words

Prostate cancer early detection

The Max Planck Institute for Human Development has just kindly given us permission to post this excellent chart depicting the effects of prostate screening on men over 50. It goes with this post from yesterday — Analyzing those widespread feelings of ‘Hands off my PSA test‘ – and illustrates the reasoning behind recent recommendations against routine prostate screening for healthy men.

I love the feeling of dawning clarity as my eye passes slowly over the data and the point comes across: Hmmm, 1,000 men in each group. With prostate screening, the same number of men die of prostate cancer as without screening (red circles with a P inside). But with screening, a couple of hundred get “false positives” that worry them but turn out okay. And 20 are treated unnecessarily for prostate cancer, with all the downsides of the treatment, to no benefit (blue circles with an X inside). Okay, I think I get it now…

A couple of points from the folks at the Max Planck Institute:

In light of the most favorable evidence, the number of men dying from prostate cancer could be reduced from 8 to 7 out of 1,000, but in light of the less favorable evidence, there is no such reduction. In any case, overall mortality was shown not to be reduced at all in all trials.

Credit is due to Lisa Schwartz and Steven Woloshin at Dartmouth Medical School for the conceptual development and evaluation of drug fact boxes; they also provided an idea of how a fact box could look in the case of PSA screening. While the Harding Center has developed this particular facts box, it is standing on the shoulders of giants.

They also note that the graph is based on epidemiological data from a 2010 BMJ paper.

Please follow our community rules when engaging in comment discussion on this site.
  • LeoK711

    I think their equating “harm” with having a biopsy is ridiculous for a procedure that is more uncomfortable than truly painful. Of course this is subjective but it’s my personal experience with the procedure – my initial fears were quite overblown.

    Granted, something like 5 of 1000 biopsies has medical complications, so that’s one out of their 200 “harmed”, not all 200.

    This graphic is very seductive, but I question its validity. It seems very skewed in one direction in its presentation and content.

    • Linehanm01760

       “I question its validity. It seems very skewed in one direction in its presentation and content.”
      Opinions based on what?
      Please describe with SOME detail how the “presentation and content” is ‘skewed’.

      • LeoK711

        Well, for one thing just what I mentioned: a definition of harm minimal enough to be questionable. And for the other points raised on the page, mine and others.

  • http://profiles.google.com/tim.bartik Tim Bartik

    Several points here. First, the point about “no effects on all-cause mortality” reflects a statistical misunderstanding of what empirical reserach can and can’t show, and should not be given any emphasis in news coverage. For example, given the magnitude of prostate cancer mortality reduction estimated in the European study, which is about 0.1% (1 in a thousand men screened), there is no way with an overall mortality rate of 20% in this sample to detect effects on all-cause mortality unless we had a sample size of several million. In other words, existing studies and all feasible studies are “underpowered” to detect all-cause mortality effects of prostate cancer screening with the needed precision. So the lack of statistically signficant effects on all-cause mortality is meaningless. Actually, the point estimate in the European study is that all-cause mortality declines due to screening by 0.2% (2 in 1000 men screened), but this estimate is too imprecise to rule out a zero effect, or,for that matter to rule out an effect of 0.4% (4 in 1000 men screened).  

    Second, this chart obscures the real issue:is reducing the risk of death by a certain percentage worth it if there are significant side effects. As this chart outlines, if one believes the European study’s results, screening leads to treatment in which this treatment saves 5% of those treated (1 saved out of 20 treated as mentioned in the story). However, as the USPSTF concluded, perhaps 30% of those treated will have serious side-effects with impotence or incontinence. What is worth more to you: reducing your risk of death after 10 years by 5%, or avoiding a risk of impotence or incontinence of 30%? Not a no-brainer choice. Different men will have different answers. Of course, if one believes the U.S. study, then there are no benefits of screening-induced treatment in terms of reduced risk of death. So it depends on which study you believe to be better. In any event, before men are screened, they should ask themselves whether they would choose to be treated if diagnosed. If they wouldn’t, then there’s no point in screening.

    Third, as this reveals, the real issue is the effectiveness of treatment. If more men with low-risk prostate cancer would choose active surveillance after diagnosis, we could probably reduce the number of men with serious side-effects without a significant sacrifice in lives saved. It is not the screening that is the problem. It is what people do with the screening and biopsy information. 

    • Cate Prato

      Thank you Tim, for the most rational response I’ve read to this US study and the recommendation that no screening take place. It seems irresponsible at best to say there is absolutely no benefit to screening.

    • LeoK711

      I agree with your bottom line – that the issue is not over-screening but over-treatment.

      But I don’t see where they say 1 of 20 treated is saved. If this were the case, then their number for treated unnecessarily should be 19, not all 20 treated.

      The bigger problem with this graphic: They’re saying 20 were unnecessarily treated, yet they don’t give the TOTAL number treated. It’s not all the blue dots, because those are labeled unnecessary biopsies – yet they also don’t give the TOTAL of biopsies.

      It’s almost as if they’re saying all treatments and all biopsies were unnecessary! How can they know the number of unnecessary procedures without knowing – and showing – the total?

      I love the general idea of showing something like statistics as intuitively as possible, but I question the validity of this graphic because important information is omitted.

      • http://profiles.google.com/tim.bartik Tim Bartik

         The particular graphic assumes that screening leads to a 20% biopsy rate, and a 2% treatment rate (10% of those biopsied). The graphic also assumes that no lives are saved due to these biopsies and treatment.

        However, contrary to this graphic, the European study implies that about 1 out of those 20 treated would have otherwise died within an 11 year followup period. The other 19 were treated “unnecessarily”, with two caveats to that conclusion: (1) This is as of 11 years after screening has begun — perhaps more of those 19 treatments will prove to be life-saving after another 5 or 10 years; (2) in the absence of better info, we can’t know in advance whether the treatment will save any particular man’s life. All we can say is that the odds are that 5% of the men treated would have otherwise died of prostate cancer after 11 years.