• The harms are just as important as the benefits

    I’m not sure I can say it any better than this:

    Although 1 in 6 men (16.7%) will receive a diagnosis of prostate cancer in their lifetime, only 2.9% will eventually die of the disease.

    That means that 97 out of 100 men who are diagnosed with prostate cancer die of something else. And this:

    The proportion of men who are diagnosed with prostate cancer but never have associated clinical symptoms is difficult to estimate, but it may range from 23% to 66%.

    In other words, many prostate cancers are slow-growing, and as many as two-thirds would have no effect on a man’s health or lifetime if left undiagnosed. Finally, this:

    The small potential benefit of prostate cancer screening corresponds to preventing, at most, 1 death caused by prostate cancer per 1000 men screened after 11 years of follow-up.

    The harms, on the other hand, are well-known. Men who get diagnosed with prostate cancer and who undergo a radical prostatectomy have an 11% increased risk for urinary incontinence and a 37% increased risk for erectile dysfunction.

    All of this comes from the new American College of Physicians’ guidelines on screening for prostate cancer. They have two recommendations. First, they advise clinicians to inform men between the age of 50 and 69 years that the potential benefits for prostate cancer screening are limited and that the harms are substantial. Furthermore, they stress that screening should only occur in patients who express a clear preference for screening after being fully informed. Second, they recommend that clinicians not use a PSA test in average-risk men younger than 50 years, older than 69 years, or who are expected to live less than 10 to 15 years. In other words, the only men who should get a PSA are those age 50-69 years who have above-average risk and a life expectancy of 15 years or more.

    This would seriously reduce the number of men who get screened, who have unnecessary surgery, and who suffer harms from such procedures. It is evidence based, thoughtful, and still allows for patient preferences. Nonetheless, I expect the cries of “death panels” to start soon.


    UPDATE: The language is unclear in the guideline, but SEER statistics seem to state that 2.9% of all men will eventually die of prostate cancer.

    • I’m in agreement that we need to reduce prostate cancer screening. Obviously skewed economic incentives and a general societal obsession with advanced medical technology play large roles in causing overtreatment.

      I’m a bit torn on how to correct things. Numerous authors have written again and again that aggressive treatments for prostate cancer have harms that may outweigh the benefits, and that watchful waiting upon diagnosis is a perfectly safe strategy for most men.

      And yet the public is not getting the message. I think partly because it’s a complex message: most cancers are scary, but prostate cancer is not scary because it’s very slow growing; only high-risk men should consider prostate cancer screening; only men with relatively high life expectancies should consider prostate cancer screening; aggressive treatment won’t help most men. Your layperson will ask, what’s high risk? Most men? Life expectancy??

      Big questions for me are how paternalistic do we have to get to reduce the amount of unnecessary prostate cancer treatment? At the maximum level of paternalism that Americans are willing to tolerate, how much unnecessary treatment is there? How much unnecessary treatment is in the UK, which is relatively individualistic, but has a national healthcare system and treatment guidelines for prostate cancer?

      FYI, the providers’ quick reference guide is here:

    • Great point, but I’m confused on your first set of numbers–from what you wrote, it looks like about 1 in 5 or 6 men who are diagnosed will die of prostate cancer (2.9%/16.7%). Overall, only 3 in 100 will die of it. Is that right, or is the death rate 2.9% of men who are diagnosed?

    • I think that the ACP guidelines are more reasonable than the USPSTF recommendations not to screen at all, but I wonder whether the recommendation to not screen unless you are above average in risk corresponds to most men’s preferences.

      The statement made in the report that PSA screening may save only 1 life per 1000 men screened is as of 11 years of followup, which they state, but I think it’s important to understand that the picture is likely to change as one considers longer-run follow-up periods. The same European study that leads to that figure also leads, with reasonable simulation models, to much larger effects on prostate cancer mortality of screening over longer periods.

      For example, using the results from the European study, Gulati, Gore, and Etzioni do simulations that imply that for every 1000 men screened, under various screening strategies, prostate cancer deaths are reduced by 4 to 8. Annal of Internal Medicine, February 2013. https://annals.org/article.aspx?articleid=1567368 .

      The real issue is the ratio of harms to lives saved (or years of harms experienced vs. years of life saved would perhaps be even better). The Gulati/Gore/Etzioni piece suggests that for various screening strategies, the ratio of the extra men treated to the extra lives saved ranges from 3 to 7, depending on the screening strategy. If half the men treated suffer significant harms in either erectile problems or incontinence, then the issue is, how much risk of such harms is one willing to undergo for how many years for a certain probability of reducing one’s risk from dying some years hence?

      I think if one looks more deeply into the benefit-cost analysis, many men of average risk will feel that the likely benefits of at least SOME screening outweigh the likely risks. This is particularly true if one is willing to consider active surveillance if the screening reveals a relatively low-grade prostate cancer.

      • And indeed, the NICE guidelines I linked to state that for low-risk men, active surveillance is the preferred option. Active surveillance is still an option for intermediate-risk men.

        In this guideline, risk is defined by a combination of PSA level, Gleason score and clinical stage of the cancer. Gleason score is a grading score specifically for prostate cancer. If prostate cancer is slow-growing, one could expect the majority of cancers to be low-risk.

      • I want to make my comment a bit more pointed.

        It is accurate to say that the European study implies that as of 11 years of followup, the prostate cancer deaths that might be averted are 0.1% of men screened or 1 out of 1000.

        It is also accurate to say that out of the additional men actually treated, there is an 11% risk of incontinence, and a 37% risk of impotence.

        But the problem is that people want to compare that 0.1% reduced death risk with the 11%/37% risks of harms. But that’s inaccurate because these figures use totally different denominators.

        To compare these numbers, they need to be compared on a common basis.

        In the European study, if one looks at Appendix Table 9B in the 2012 paper in the New England Journal ( http://www.nejm.org/doi/suppl/10.1056/NEJMoa1113135/suppl_file/nejmoa1113135_appendix.pdf ), screening results in a 2.3% increase in the probability of receiving some treatment other than active surveillance/watchful waiting as of an 11 year followup, or 23 out of 1000.

        If we assumed that 11% of those treated suffered from incontinence, and 37% from impotence, then for every life saved as of an 11 year follow-up, we have 2 extra cases of incontinence, and 9 extra cases of impotence.

        Now, this is an unfortunate ratio of 11 harms to 1 life saved. But this is far different then comparing 11% and 37% to 0.1%, which is a no-brainer. If the ratio of harms to life saved was 480 to 1, then we wouldn’t hesitate to say that the harms outweigh the benefits.

        In the case of the ratio of 11 harms to 1 life saved, it is certainly arguable that many men will feel that the benefits outweigh the costs when judged from an ex ante perspective. Other men will feel differently. It certainly is not obvious that a man of average risk will not want to be screened.

        Furthermore, the research by Gulati/Gore/Etzioni implies that if we go beyond an 11-year followup, screening strategies result in a greater reduction in prostate cancer deaths, whereas some additional prostate cancers end up being detected even without screening. So the ratio of additional prostate cancer treatments to prostate cancer deaths goes down, and also the ratio of harms to live saved goes down. In their various screening strategies, they get long-run ratios of additional prostate cancers detected to lives saved running from 3 to 7. If half of those treated are harmed, we have ratios of harms to lives saved of 1.5 to 1 up to 3.5 to 1.

        Of course, especially as we consider a longer-run perspective, we need to consider that the harms occur earlier then the change in prostate cancer mortality.

        Furthermore, all of this goes out the window if we don’t use the European results, and instead rely on the U.S. screening study’s results, which find no statistically significant benefit of screening. However, Etzioni and her colleagues argue in another paper, previously mentioned on this blog ( http://theincidentaleconomist.com/wordpress/is-broad-psa-screening-justified-by-what-standard-of-evidence-do-we-decide-this/ ) that the U.S. study’s results are not statistically significantly different from what would be predicted by their simulation model, given that many of the persons in the U.S. control group were also screened. This ends up becoming a difficult issue of what you take to be the proper null hypothesis.

    • “Although 1 in 6 men (16.7%) will receive a diagnosis of prostate cancer in their lifetime, only 2.9% will eventually die of the disease.”

      Isn’t the 16.7% inflated by the seniors (I hesitate to set an age!) who are diagnosed but will die of something else before prostate cancer progresses?

      • Yes, I think that’s the point. A good number of people will die with prostate cancer, but not of prostate cancer.

    • At present treatment of prostate cancer is in a transition phase and involves a learning process so the diagnosis and treatment becomes more selective. Cancer of the prostate does kill and even if something else kills the patient first, it can be painful and debilitating. Thus the 2.9% count is significant along with the others that might not die but live with various degrees of morbidity. There was no discussion of PSA 2 which might be helpful in being more selective along with the size of the prostate, sonogram, digital rectal exam or the rate of increase of the PSA or the ratio of PSA 2 to total PSA. All of these things among other things might make the selection process more accurate and lead to preventing morbidity and mortality for more people making screening and management options more valuable. The side effects of surgery can be ameliorated by physicians with special training to reduce the problems already mentioned.

      In the end I think as long as the patient is given a choice along with balanced information I don’t think the term ‘death panels’ will be used by serious people.

    • Aaron
      A somewhat related comment.

      I waited for the news coverage.
      …still waiting.

      Funny how mammograms, even if costs and implications equal to PSA, get disproportionate attention.

      Not a peep here. Clearly, the inferior sex becomes plain 🙂