A sad manipulation of data to support a point of view

Dr. William Catalona has long been one of the strongest advocates of prostate cancer screening. He has been a key individual in the development and promotion of PSA testing for some 20 years, and he believes passionately that prostate cancer screening saves lives. We respect his beliefs and his right to express those beliefs.

This year Dr. Catalona will give the Ramon Guiteras Lecture to the assembled attendees at the annual meeting of the American Urological Association (AUA) in Washington, DC. The Ramon Gutieras Lecture is the most prestigious annual lecture given at the annual meeting of the AUA. Unsurprisingly, the subject of his lecture will be how PSA screening lowers prostate cancer mortality. An outline of Dr. Catalona’s proposed lecture appears in this month’s issue of AUA News, which is now available on line.

In the outline of his lecture, Dr. Catalona writes as follows:

In the United States there has been a 75% reduction in metastatic disease at diagnosis and a 40% decrease in the age-adjusted PC mortality rate during the “PSA era,” which means that in the U.S. more than 22,000 fewer men die of PC each year than in 1992 ….

Now it is absolutely true that, based on information available in the Surveillance, Epidemiology and End Results (SEER) database, there was a 40 percent reduction in the age-adjusted, prostate cancer-specific mortality rate between 1992 and 2007. (2007 is the last year for which such data are available in that database.) Specifically:

  • In 1992, the age-adjusted prostate cancer-specific mortality rate was 39.22 per 100,000.
  • In 2007, the age-adjusted prostate cancer-specific mortality rate was 23.50 per 100,000.

This is most certainly a 40 percent decrease.

We are also very willing to take Dr. Catalona’s word for the 75 percent reduction in incidence of metastatic disease at the time of diagnosis. This has been a well-understood and major benefit of widespread PSA testing.

Where we have a big problem is with Dr. Catalona’s assertion that PSA testing has reduced the number of men dying of prostate cancer in America each year by 22,000. This is a statement that has no evidence to support it. It is an extrapolation of the reduction in age-adjusted mortality from 1992 to 2007 that cannot be justified by any available data.

The development and early use of the first commercially available PSA test (the Hybritech Tandem PSA test) dates back to the late 1980s. By 1992, the PSA test had been widely available and used to test for risk of prostate cancer for about 3 years. As a consequence, the incidence of prostate cancer and the mortality rate from prostate cancer were both at their highest-ever known levels because (finally) physicians were actually seeking out and trying to treat men with this disease early. Prostate Cancer Awareness Week had been initiated and funded by a major pharmaceutical company with the very real goal of identifying men with late stage prostate cancer in order to maximize sales of the first antiandrogen (flutamide), and the age-adjusted prostate cancer-specific mortality rate had risen from 35.88 per 100,000 in 1988 to reach its highest ever level at 39.31 per 100,000 in 1993.

In other words, between 1990 and 1993, the PSA test was being widely used to find men who actually had not modern, low- and intermediate-risk prostate cancer, but men with PSA levels of 10 ng/ml and higher who were commonly found to have metastatic forms of prostate cancer at diagnosis. These men died distressingly quickly, and many did so within 3 years. But we had become a lot better at knowing what they died of … and the words “prostate cancer” came out of the shadows (like the words “breast cancer” before them). Men started to go for PSA tests. Thankfully we started to be able to diagnose these men earlier and earlier and manage them better and better than had been in the case in the past. Gradually, the incidence of metastatic disease at the time of diagnosis dropped, and so did the age-adjusted mortality rate.

What did not drop so much, however, was the total number of men dying of prostate cancer. Between 1992 and 2010, the actual numbers of prostate cancer-specific deaths reported by the American Cancer Society (see Jemal et al., 2010, Table 8) have ranged from a high of  34,902 (in 1994) to a low of 28,372 (in 2006). For 2010, the projected number of prostate cancer-specific deaths is 32,050. What is absolutely certain is that no one has ever even suggested it would drop by 22,000 from the 34,240  reported in 1992 to an annual figure of about 12,000 (as implied by Dr. Catalona).

The truth is that despite the clear reduction in the age-adjusted prostate cancer-specific mortality rate since the early 1990s, the actual number of men dying of prostate cancer has remained, over the same 20-year period, within a remarkably stable range from about 28,000 to 35,000 since the introduction of the PSA test. There are all sorts of possible reasons for this, and no one (to date) has been able to offer a clear explanation. Part of the reason is likely that we are all living longer. Another part, however, is clearly associated with the way in which data are recorded and subsequently manipulated by statisticians. Data can be right or wrong. They are also subject to the assumptions made by those who manipulate them.

What distresses us, however, is that a man of Dr. Catalona’s stature should feel the need to suggest that PSA testing is now “saving” 22,000 men from prostate cancer-specific mortality each year. He even uses a photograph of a baseball park full of fans to make his point about the size of this accomplishment. We consider this suggestion to be outrageous. It certainly is not justified by any available facts.

15 Responses

  1. Mike,
    After reading the article in question, I think that you are not interpreting what Dr. Catalona is saying correctly. In the absence of a PSA test, in the 1990s, the mortality rate was increasing at a clip of 1.8% a year. If such trend had continued, more men would have been dying presently. He calculated that number to be 22,000 more men than actually die now.

    In reality, more men are dying than necessary because PSA use is not as widespread as claimed and sadly because of the ongoing controversy it is going to be used less and less.

    What Dr. Catalona is saying is supported by current mortality rates in countries in which the use of PSA testing is low. The fact that he is using a visual effect in showing a stadium with 22,000 men might shake up some men that are now swayed the other way to ignore the risk of prostate cancer. And that is really sad…

  2. Dear Ralph:

    If that is what Dr. Catalona means, then: (a) it is certainly not what he wrote and (b) it is based on the assumption that nothing else has happened over the past 20 years to affect prostate cancer mortality rates, which is demonstrably not the case. To give you just one example, the survival times of men from the initial observation of metastatic disease was well understood to be between 18 to 36 months in 1990. It is now significantly longer than that and may well be more like a median of 5 to 7 years.

    If Dr. Catalona has data to support the argument you are making on his behalf, then he should have referred to such data. In fact, he very carefully linked his projection of 22,000 lives a year saved to the reduction in age-adjusted mortality rates (which is the only source for a 40% reduction in the rate of prostate cancer deaths).

  3. Mike:

    This man is not stupid. How can he claim 22,000 lives saved when the mortality remains close to 30,000?

    He talked about 40% less deaths than in the ’90s, and if that death rate trend in the ’90s had remained as it was (without PSA testing) more men would be dying today. Hey, it is no secret that in countries in which PSA has not been used as extensively more men are dying now than in the 1990s. I think that is what he is pointing at by calculating a number based on a trend and a higher life expectancy. Is that number correct? We have to ask him how he arrived at that figure, but there is no question that the actual number here would be higher without PSA testing.

  4. Dear Ralph:

    Respectfully, Dr. Catalona doesn’t mention anything about the death rate trend in the 1990s remaining as it had been. That is your inference. He very specifically ties his claim of a reduction of 22,000 lives per annum to the 40% reduction in age-related prostate cancer-specific mortality. That’s why I was so careful to give the precise quotation. I left nothing out — not a word.

    If I knew how he could make such a claim, I would never have written the commentary above.

    It is a reasonable assumption that the number would be higher if neither the PSA test nor any other new test had ever been developed, but it doesn’t justify the claim, because if the PSA test hadn’t been developed we don’t know what would have happened over 20 years.

    No one at this end is arguing that the PSA test is a bad test or shouldn’t be used, but it won’t help the advocacy community when people like Dr. Catalona make this sort of unsubstantiated statement. To the contrary, it will only make people like Dr. Brawley point out (with great accuracy) that Dr. Catalona has had a 20-year relationship with the manufacturers of PSA tests, and that they have supported a considerable proportion of his research over that time period.

  5. Mike,

    He said: “In the United States there has been a 75% reduction in metastatic disease at diagnosis and a 40% decrease in the age-adjusted PC mortality rate during the“PSA era,” which means that in the U.S. more than 22,000 fewer men die of PC each year than in 1992.”

    Isn’t he commenting on the mortality rate in the 90s? Without that 40% reduction, more men would be dying now so instead of 30,000 men dying now, there would be 52,000 men dying. It blows my mind that you think that be is saying what you interpreted.

    We do not know what it would have happened if the PSA test or any other test would not have not been used? Don’t we have examples of low use in other countries?

    I really can’t believe that you think that Dr. Catalona is in the pocket of the PSA manufacturers … and uses this interpretation of yours to support the premise …

  6. Mike,

    Thanks for the putative explanation of the rise in prostate mortality rates that was associated with the use of PSA for diagnosis. An interesting point of view and the first time anyone has responded to my repeated requests over the years for some kind of rational explanation for the very sharp rise to a peak from which any decrease is measured.

    In considering the oft-repeated claims of reduced mortality resulting from the War on Cancer which was declared by President Nixon in the 1970s I took a look at the World Health Organisation statistics, which assembles data in a somewhat diffeeent format to the SEER data.

    The first available year is 1979, by chance early on in the War on Cancer and also fortuitously just before the PSA test was approved. In that year there were 1,044,959 male deaths in the USA. 22,240 of these deaths were attributed to prostate cancer, so we can say that 2.12% of male deaths in 1979 were due to prostate cancer.

    The latest available year is 2005. In that year there were 1,207,675 (increase 16%) male deaths in the USA. 28,905 (increase 30%) of these deaths were attributed to prostate cancer, so we can say that 2.39% of male deaths in 2005 were due to prostate cancer.

    It seems to me that this could be expressed as an increase of 13% in the relative number of men who died from prostate cancer between 1979 and 2005.

    It is often said that the first casualty of any war is the truth and, to quote Mohandas K. Gandhi, “An error does not become truth by reason of multiplied propagation, nor does the truth become error because nobody will see it. “

  7. Mike,

    You said: “Respectfully, Dr. Catalona doesn’t mention anything about the death rate trend in the 1990s remaining as it had been. ”

    In the AUA News article the caption of the stadium image reads: “Baseball stadium with approximately 22,000 fans, which is number of men in U.S. who would die of prostate cancer each year if there were no PSA screening.”

    I think this qualifies as, “the trend remains as it was before the PSA test era.”

  8. Dear Ralph:

    If the caption below the baseball park picture read: “Baseball stadium with approximately 22,000 fans, which is number of additional men in U.S. who would be dying of prostate cancer each year if there were no PSA screening,” then it would match the argument you are making. But that still wouldn’t provide any factual evidence to support Catalona’s claim. And your assumption that the 1.8% annual increase in the prostate cancer death rate from 1975 to 1990 would have continued form 1990 through 2010 also has no basis in fact because it is what it is — an assumption.

    Remember … I am not bashing the PSA test at all. I think it is the best test we have and we need to use it. My issue is with the hype. There is no evidence at all to support Catalona’s very clearly stated claim that “in the U.S. more than 22,000 fewer men die of [prostate cancer] each year than in 1992” and that this is a consequence of PSA screening. If that wasn’t what he meant, then why did he write it? We both know it’s not true.

  9. Sorry Ralph … Your comment from 04/05/11 at 1:51 pm ended up in the spam file, and I just “rescued” it …

    (1) It “blew my mind” that he wrote the comment at all.

    (2) If he is commenting as you suggest, it is not what he wrote, which is extremely clear (at least to me and apparently to Terry Herbert too).

    (3) If you can show me data from another country in which prostate cancer deaths have continuously increased on a steady path from 1975 to 2010 without any significant use of the PSA test, please feel able to do so. I don’t actually think such a country exists any more.

    (4) I did not say that Dr. Catalona was “in the pocket” of the PSA manufacturers. I said that people like Dr. Brawley would be likely to imply that point and that the PSA manufacturers have funded a great deal of Dr. Catalona’s research over the years. The first of those statements is speculation on my part. The second is absolutely true.

  10. Mike,

    I know you are not bashing the use of PSA. Why would he say that 22,000 men would die if the PSA was not in use? Every one knows that some 30,000 are dying now. Is he numerically impaired? He is simply trying to show that without PSA testing more men die. This is the same case in countries where the use of PSA is very low. I think you are exceedingly critical of Dr. Catalona in your interpretation of what he is trying to show graphically. His premise is what would prostate cancer mortality be today without the PSA marker. He followed the trend that existed before the test became available and came with up with a higher number of deaths.

    There has not been a national screening program here. This reduction in mortality happened through individual testing with PSA. The debate about screening has gone long enough. A national screening program will never happen. Hell, mammograms are recommended and is there a national, across the board use on mammography here?

  11. Dear Sitemaster,

    I was and still am not certain that I would have the time to comment on the article below on a related subject. I mentioned that the use of the word “power” was unclear to me and you directed me to the right place. This article encourages me to read that article. Still no promises though, since I am undergoing treatment for prostate cancer. It’s tiring.

  12. You said ”If you can show me data from another country in which prostate cancer deaths have continuously increased on a steady path from 1975 to 2010 without any significant use of the PSA test, please feel able to do so. I don’t actually think such a country exists any more.”

    I hope you are being cute by asking for “any significant use of the PSA test data.” I am using the WHO database and some of the data doesn’t go back to 1975. Here is a list of countries with an increased prostate cancer mortality even when PSA has been available. Determining the use of PSA is not only difficult, but also probably irrelevant because if testing with PSA is not associated with a treatment (that in some way alters the natural history of the disease) mortality rates are what they are.

    Trinidad and Tobago
    1979: 70 deaths; Mortality rate 12.8/100K
    2002: 277 deaths; Mortality rate 42.9/100K

    Cuba (Michael Moore’s medical paradise)
    1979: 730 deaths; Mortality rate 14.7/100K
    2005: 2260 deaths; Mortality rate 40.1/100K

    1987: 1830 deaths; Mortality rate 44.1/100K
    2005: 2451 deaths; Mortality rate 54.8/100K

    1987: 537 deaths; Mortality rate 22.5/100K
    2006: 814 deaths; Mortality rate 31.6/100K

    1986: 875 deaths; Mortality rate 42.5/100K
    2005: 1041 deaths; Mortality rate 45.4/100K

    1979: 5419 deaths; Mortality rate 19.9/100K
    2006: 10052 deaths; Mortality rate 33.9/100K

    These are data found about PSA use in these countries. I searched widely without much success …

    Norway 1999 Age 50-65; PSA use: 7%

    United Kingdom 1999-2001 Age 45-84; PSA use: 7%

    As you can see when PSA is available but not used “widely” as it is here in the USA, mortality tends to continue to rise. Surprise! Dr. Catalona is right after all… ;-)

  13. Ralph:

    Please read the article I sent you. There are a hundred reasons why there might be these sorts of increases in prostate cancer-specific mortality between the late 1970s and the early 2000s (in any country), starting with the actual death attribution. It is of interest to note that in the two countries you list that are generally considered to have the best health data registries (Sweden and Norway), the changes over time are, in fact, relatively small. Their numbers are nowhere even close to those suggested by your interpretation of what Dr. Catalona may have been implying, which is a 65% increase in mortality since 1992. The Swedish number is only about 25% since 1987.

  14. Mike,

    Rather than continue our exchange, I feel that waiting for Dr. Catalona’s presentation in May will be helpful in supporting either your interpretation or mine. After the presentation we will be in a better position to understand how he arrived at his figure.

    In the mean time I stand by my interpretation as much as you stand by yours.

    As far as the small changes you mention in both Norway and Sweden please note that the data span is only 8 and 9 years.

  15. Ralph: Agreed … No problem.

Leave a Reply

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

WordPress.com Logo

You are commenting using your WordPress.com 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: