A 45 percent reduction in prostate cancer-specific mortality? Not exactly.


According to a report yesterday in Consumer Affairs, “American men with prostate cancer were 45 percent less likely to die from the disease in 2006 than they were in 1999.”

To be fair, what the federal government actually said was that, “American men with prostate cancer were 25 percent less likely to die from the disease in 2006 than they were in 1999″ (but even then, the actual reduction was only 24.2 percent; see below.) Consumer Affairs compounded a typographical transcription (changing 31 to 13) into a mathematical error.

The government’s original statement appears in the September 29 issue of News and Numbers from the Agency for Healthcare Research and Quality (AHRQ) and has been picked up in a number of online media. The data have been extracted from Table 4.2.1 in in the Appendix to the 2009 National Healthcare Disparities Report (also issued by AHRQ, back in March).

Specifically, AHRQ says that the rates for prostate cancer-specific mortality (PCSM) were:

  • 31/100,000 in 1999 and 23.5/100,000 in 2006 overall (a 24.2 percent decrease)
  • 29/100,000 in 1999 and 22/100,000 in 2006 for White men (a 24.1 percent decrease)
  • 69/100,000 in 1999 and 50.5/100,000 in 2006 for Black men (a 26.8 percent decrease, but still more than twice the overall rate)
  • 23/100,000 in 1999 and 18/100,000 in 2006 for Hispanics (a 21.7 percent decrease)
  • 17/100,000 in 1999 and 14/100,000 in 2006 for Native Americans (a 17.6 percent decrease)
  • 14/100,000 in 1999 and 10/100,000 in 2006 for Asian Americans and Pacific Islanders (a 28.6 percent decrease)

They also stated that:

  • The rate of PCSM in men aged 65 years or more dropped from 205/100,000 in 1999 to 164/100,000 in 2006 (a 20.0 percent decrease).

AHRQ makes no comment on why PCSM might have dropped by 24.2 percent from 1999 to 2006. Nor do they comment on the growth in the absolute numbers of men projected to die from prostate cancer by the American Cancer Society after 2006 (from 27,350 in 2006 to  32,050 in 2010; a 17 percent increase).

It is not at all clear to The “New” Prostate Cancer InfoLink why AHRQ should have decided to make this statement several months after issuance of the original data. Maybe it had something to do with Prostate Cancer Awareness Month (which ended in September). It does seem likely that there was a significant decline in PCSM between 1999 and 2006. It is much less clear what the real size of that decrease in mortality might have been or whether it has been sustained since 2006.

A major problem in the interpretation of all data like these is that we do not have a reliable nationwide cancer database in the USA — we have a representative one (the SEER database) based on cancer information from selected regions of the country. This is used to make estimates and projections which are highly dependent on the assumptions and models used in making those estimates and projections.

10 Responses

  1. How I wish that someone could tell me (a) why the mortality rate increased so dramatically from 1989 to 1999 and (b) why the measurement in the difference between the current rate and the “old rate” doesn’t mention that it took a number of years to get down to the pre-PSA mortality rate.

    All very curious! Has there been a change in the definitions perchance? It was reported that no less a personage than Dr Patrick Walsh put down a similar drop in Britain to the fact that prostate cancer-related pneumonia deaths were no longer counted!!

    Ah well, spinning makes the world go around.

  2. First of all, I assume you talk about the mortality rate in the USA. If that is the case, the mortality rate here went down between 1989 and 1999. It actually went further down in the following years and by 2003 it was as low as it was in the pre-PSA era. This in spite of the fact that male life expectancy had increased by 3.6 years during those years and the alleged widespread use of PSA testing barely edged 50% in the aged population at risk.

    It is clear that in countries in which the use of PSA testing is more prevalent, fewer men are dying of prostate cancer. The following prostate cancer mortality rates are from the WHO database:

    USA: 1989, 25.2/100,000; 1999, 23.8/100,000; 2003, 20.7/100,000
    Australia: 1989, 24.1/100,000; 1999, 26.7/100,000; 2003, 28.7/100,000
    UK: 1989, 31.0/100,000; 1999, 32.5/100,000; 2003, 34.9/100,000

    This should answer your questions. If you have an explanation for these differences I would like to hear them.

  3. Well, I’m still confused. The SEER data (obtainable via the SEER Fast Facts pages)
    show a mortality rate of 30.97 in 1975. This was followed after a few years by an increase which peaked at 39.34/100,000 in 1993. That’s an increase, by my calculation, of 27%.

    The rates then started falling, and by 2000 (7 years later) had reached a level similar to that in 1975 — 30.31/100,000.

    After 2007, the rate fell further to 23.5/100,000, which by my calculation is 75% of the 1975 rate.

    These rates are so different from those posted above that I find them difficult to reconcile. No doubt someone will point out why I am wrong to consider the SEER figures.

  4. The numbers are different because the two databases use different parameters to express the rates. Here are the definitions:

    SEER database: Rates are per 100,000 and are age-adjusted to the 2000 US Std Population (19 age groups – Census P25-1130).

    WHO database: Number of deaths (N) at ages (in years) and age-sex-specific death rates (rate) per 100 000 population.

    The reality is that the trends for the years you mentioned originally (1989 to 1999) indicate a reduction in deaths in the US and is still holding through 2007 (based on the SEER database). The reduction in the death rates between 1975 and 2007 is not 75%. It is 24.1% and this is with an increase in male life expectancy of almost 5 years.

    The bottom line is that there has been a reduction in deaths since the introduction of PSA testing (still unexplained by the experts). Both databases show a rate reduction in the US while the WHO database showed an increase in deaths in countries where PSA use is lower.

    BTW, I meant the significant differences between rates in the countries mentioned.

  5. As ever, the question of why the mortality rate increased in the years prior to 1993 but after the introduction of PSA testing is swept aside. Why did that happen? I believe that the explanation of that spectacular rise might give some insight into the reason for the subsequent reduction. And I think it may well have something to do with definitions — and spin.

    The graphic illustration of the peak in mortality rates is shown clearly in the Welch-Albertsen study entitled “Prostate cancer diagnosis and treatment after the introduction of prostate-specific antigen screening: 1986–2005.

    Is the increase in male life expectancy germane when the median age for death has remained fairly static at 80?

    Incidentally, I did not say that the reduction in deaths from 1975 – 2007 was 75%. I said that the rate in 2007 was 75% of the rate in 1975 i.e. a 25% reduction.

  6. Dear Terry:

    There is absolutely no doubt that the rise in prostate cancer mortality between about 1980 and 1993 had something to do with the assignation of “prostate cancer” as a cause of death. With the availability of the PSA test it was relatively easy to know that a man had advanced prostate cancer without giving him a bone scan! That doesn’t mean there was a conspiracy. After all, people were dying of AIDS long before anyone was able to say they were dying of AIDS.

    It is easy to argue that, in fact, the prostate cancer mortality rate in the period immediately prior to the early 1980s was an undercount because many men actually dying of prostate cancer were being assigned to some other cause of death. if that were the case (which it may be), then the prostate cancer mortality rate didn’t climb at all in the period you refer to. It simply wasn’t accurate to begin with!

    :O)

  7. Mike,

    First up can I say that I did not use the word “conspiracy” nor did I intend to imply that there was a “conspiracy.” But the fact remains that definitions can and do change over time, either by agreement or by other forces, including market forces.

    Perhaps the best recent example of a signficant change in the PCa industry was the decision by pathologists to no longer label GS 5 atypical cells as “cancer.” Although this agreement was only formalized in January this year, it had in fact been acted on for at least 5 years prior to that — and one of the consequences was the so-called Gleason migration to higher scores. I raised questions concerning this change well before the official anouncement and was told that there was no such “conspiracy” — yet is was perfectly obvious to me from the lack of men submitting their stories to my Yana site that GS 5 had disappeared from the prostate cancer scene.

    I am aware of the fact that there are protocols concerning the cause of death, to establish what we in the insurance industry refer to as the “proximate cause.” So, for example, a man who dies on the operating table whilst undergoing prostatectomy would not be marked down as a prostate cancer death; presumably a man who contracted golden staph during his stay at hospital following surgery and subsequently died would also not be a prostate cancer statistic. But what of the man who dies a week after his surgery from a heart attack? I’m sure there is a protocol covering this, but don’t know what it might be.

    Dr Patrick Walsh was quoted in a news report when challenged on his claims regarding the reduction in mortality rates due to screening by a reporter pointing out that there had been a similar drop in Britain, where there was no screening. Walsh was reported as saying that the British had excluded prostate cancer-associated pneumonia deaths from their database and this was the reason that their mortality rates had declined.

    So that was what I was referring to. It seemed clear to me that something created the mortality surge that peaked at the same time as the incidence rate peaked. But why? If the mortality rate change was due to inaccuracy at the start, why did it climb and fall in the same time frame as the incidence rate? This trend is observed in other countries as well as the US.

    That’s why I speculated at the possibility of a change in definitions and have wondered if we are truly comparing apples and apples or if there are perhaps baskets of mixed fruit.

  8. See this table from the SEER database. It shows that between 1975 and 2007 the reduction in prostate cancer mortality was 30.3% for men of all races younger than 65 and 23.6% for men of all races older than 65. Since prostate cancer is more prevalent in men as they age, an increase in life expectancy is an important factor in the calculation of mortality rates.

    I thought we were discussing mortality rates … and now you jumped to incidence. I think I am getting nowhere in this discussion, so I give up!

  9. Dear Terry:

    Of course we aren’t comparing apples to apples, for all of the reasons that you and Ralph constantly raise every time this discussion comes up. There are a thousand minor changes in definition all the time, down to the level at which one physician’s assignation of death from pneumonia in a hospitalized patient with metastatic prostate cancer in the ICU is pnuemonia and another might be prostate cancer. Arguably they are both correct because the real cause of death was “complications of pneumonia in an immune-compromised patient with terminal metastatic prostate cancer.” This descriptor is probably not an “assignable” cause of death for mortality statistics, however. So the epidemiologists may assign it as prostate cancer or as pneumonia depending on the day of the week (for all I know).

    We are quibbling over how many angels can fit on the head of a pin. A 25% decline in prostate cancer mortality over the past 20+ years is hardly a great achievement, especially when we can’t really tell what has driven this mortality reduction (earlier detection, better treatment, better nutrition, less complications of treatment, all of the above), and I can’t get overly excited about the details because they are so highly subject to fruit analysis in the first place. This is why the American Cancer Society’s annual projections of incidence and death in the USA come with the warning that they can’t be compared from years to year (even though we all do so).

  10. Oooooer … sorry Squire!! (tugs forelock deferentially).

    I can’t guarantee I’ll never try to discuss this again, but believe me or believe me not this is the FIRST time anyone has accepted that the results are not comparable and that the claims of enormous numbers of lives saved may not be entirely realistic.

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 )

Google photo

You are commenting using your Google 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: