To screen or not to screen: an Australian “discrete choice experiment”

The question of whether men should be regularly screened for risk of prostate cancer through the use of the PSA test is (at least) controversial — and a matter that some in the prostate cancer community are almost willing to “go to war” over.

This makes a study by Howard et al., just published in the journal Health Expectations, particularly interesting — albeit difficult to understand in detail if (like your sitemaster, and unlike the authors) you have limited expertise and experience in the fields of health decision modeling, statistical analysis, and applied mathematics.

Basically, what Howard and her colleagues did was to construct a relatively complex model including six key risks (properly referred to as “attributes”) that might occur over the next 10 years as a consequence of having a PSA test at a particular point in time:

  • The risk that one would actually die of prostate cancer
  • The risk that one would be diagnosed with prostate cancer of any type (including cancers that never really needed to be diagnosed at all)
  • The risk that one would have an unnecessary biopsy (effectively a “false alarm” based on the PSA result)
  • The risk that one would have ongoing impotence after treatment for prostate cancer
  • The risk that one would have ongoing urinary incontinence or moderate to severe bowel dysfunction after treatment for prostate cancer
  • The amount of  money that one might have to spend “out of pocket” to address prostate cancer-related issues

Through a careful modeling exercise, the authors assigned a variety of risk levels (properly referred to as “attribute levels”) — by age group — to each of these risks. For example, the table below shows the possible attribute levels assigned for the attribute that one would have an unnecessary biopsy:


After conducting a two-stage pilot study to ensure that men would be able to understand the way that the real study was designed and worked, the actual decision choice experiment was conducted as follows:

  • Australian men aged between 40 and 69 years with no known family history of prostate cancer were asked to complete an on-line “decision choice” questionnaire.
  • The questionnaire included 15 questions, for each of which there were three possible answers, and one additional question that allowed the researchers to understand whether respondents really understood the attribute levels that they were being asked to make judgements about.

In other words, men were asked, in a series of different ways, questions that were designed to elicit their levels of comfort with whether they would accept a specific risk level if they had a PSA test compared to the risk level if they remained unscreened.

Here are the basic findings reported by Howard et al.

  • 793 Australian males started to respond to the decision choice questionnaire
  • 662/793 men (83.5 percent; the “respondents”) completed the questionnaire.
    • The 662 respondents had an average (mean) age of 55 years.
    • 70 percent of respondents had a current spouse or partner.
  • About 35 percent of respondents consistently gave a positive response to accepting a screening option (as opposed to remaining unscreened).
  • 8.5 percent of respondents consistently elected not to be screened.
  • 63 percent of respondents chose a screening option in answer to > 10 of the 15 questions.
  • 15 percent of respondents chose a screening option in answer to 6 to 10 of the 15 questions.
  • 21 percent of respondents chose a screening option in answer to 5 or fewer of the questions (inclusive of the 8.5 percent of men who never elected a screening option).
  • The ability to avoid more prostate cancer-specific  deaths increased the respondents’ preferences for screening, and so
    • Preference for screening varied with age.
    • Younger men valued the potential mortality benefit more than older men.
  • Respondents were less likely to prefer screening as opposed to no screening when
    • The chance of needing a biopsy was increased
    • The chance of experiencing side effects after treatment increased
    • The chance of significant out of pocket expenses increased
  • Higher age (60 to 69 years) was associated with a lower level of preference for screening compared to no screening.
  • To avoid a single prostate cancer-specific death, men aged 40 to 49 years were willing to accept that, compared to no screening, screening would cause
    • An extra 65/10,000 men to have unnecessary prostate biopsies.
    • An extra 31/10,000 men to experience bowel problems or incontinence
  • Older men were willing to accept significantly higher trade offs; for example, among the 50- to 59-year-old age group, men would accept that screening would cause
    • An extra 233/10,000 men to have unnecessary prostate biopsies
    • An extra 72/10,000 men to experience bowel problems or incontinence

The paper includes a great deal more detail correlating the respondents characteristics to their ages and their choices. For example, Table 3 of the paper provides highly detailed information about whether respondents had ever previously had a PSA test or a prostate biopsy, their educational level, their employment, their household income, their perceptions about their risk for prostate cancer, etc.

Now it is important to understand that this paper doesn’t tell us anything about whether screening is “right” or “wrong”, but it does tell us a great deal about the perceptions of Australian males and their views on risks associated with PSA-based screening, and from that perspective it is a fascinating study. The “New” Prostate Cancer InfoLink would love to know whether the same decision choice experiment, if administered to comparable American males and (for example) Scandinavian males in exactly the same manner, gave us closely analogous or very different results.

It is also important to appreciate that this study has significant limitations, which the authors have addressed in detail in the paper itself. However, even with these limitations, Howard et al. deserve congratulations for  this deep probe into how Australian men actually think about whether to have PSA tests or not.

Editorial comment: The “New” Prostate Cancer InfoLink wishes to thank Prof. Kirsten Howard for kindly providing us with a copy of the full text of this very detailed and interesting paper.

5 Responses

  1. What men really need are better “clues” as to their true risks of developing an aggressive and thus potentially life-threatening prostate cancer. Needed tools include better biomarkers, better imaging techniques, and a revised prostate risk scoring system that starts at 0 instead of the current Gleason grades that essentially start at 3 of 5 or the scores starting at 6 of 10. And, of course, less biased urologists. …

  2. Notable that participants were less likely to prefer screening when the chance of experiencing morbidity of treatment/side effects increased.

  3. In 2004, at age 75, the PSA test raised the red flag for me, leading to a biopsy that revealed a Gleason 9 prostate cancer. This was certainly a potentially fatal situation, if ignored. My treatment was external radiation (IMRT) plus hormonal therapy and I am still here. Regular PSA testing shows no recurrence to date. The test very likely saved my life, I am happy to say.


    I’m not sure I understand the study adequately, but it would seem that in real life “The risk that one would actually die of prostate cancer” for either screened or unscreened men would be a key factor in each man’s decision regarding the other five factors. In other words, this would or should be (were it known) critical context for each factor explored.

    At this point we have only tentative information to this question – how many men would need to be screened to save a life, from the European Randomized Screening Study for Prostate Cancer; at least this is the most credible information to me, along with some Scandanavian research. Though the ERSPC is still quite premature, the trend of screenings-to-save-a life has plummeted as just a few years of follow-up have been added to the original 2009 report. Moreover, some of the Scandanavian research suggests a strikingly low number of screenings needed. I’m curious how the researchers tried to convey that risk to the participants in the study.

  5. Jim:

    If you e-mail the lead author, I am sure she will send you the full text of the paper (since she was happy to send a copy to me). Do remember that this is a model designed to assess opinions. The study was in no way designed to assess the rightness or wrongness of PSA testing. And you should also remember that “The risk that you would actually die of prostate cancer” is impossible to assess for any one individual. One can only assess this for a defined group of individuals, and that risk clearly varied from group to group in the ERSPC trial.

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

This site uses Akismet to reduce spam. Learn how your comment data is processed.

%d bloggers like this: