QQuestion:  How is prostate caner mortality defined?  I  thought it was
the ratio of deaths to number of cases reported, but perhaps I'm wrong.
 If it is just the ratio of deaths to the vulnerable population, it
would depend crucially on the age structure of the population and how it
was modeled.   The answer you came up with might be very dependent on
the model.

I see I will have to actually look at your references and see what how
the terms are defined and what the model is.

Using the scientific sense of the word, I think he is correct in
saying that the PSA test has not been "proven" useful.  Yet, as we
move along the path, gathering more data to test the PSA hypothesis,
many demographic experiments have shown that PCa death rates fall
significantly in areas where PCa testing is implemented, while
neighboring areas where testing was not introduced do not show the
same drop in PCa death rates.  Most of these experiments are seriously
flawed.  We have learned from them and have designed and are running
better experiments.  While it is important to continue to collect data
that will lead to the acceptance or rejection of widespread PSA
testing as a way to reduce PCa mortality, it is just as important to
read the signs along the way so that the PCa vulnerable population can
make real-time decisions today...ron

the trouble is that such a statement can be interpreted in various ways,
many of which are completely misleading.  As ron points out, what it
means is that no suitably randomized prospective study has shown a
statistically significant difference in some specific outcome between a
group of men who are tested and anothre group, identical in all relevant
characteristics, as far as we can tell to the first, which has not been
tested.  There are a whole bunch of suggestive studies comparing men who
have been tested to men who have not been tested which seem to show an
advantage with respect to mortality for testing, but we can't be sure
the different groups in these studies don't differ in some other way,
which might also explain the difference in mortality.

It is very easy to read the above statement as saying that testing has
been proven not to be useful (again, presumably in some relevant sense).
 But it doesn't say that at all.

If your expert wants to be objective, all he can say is that the solid
scientific evidence is indifferent with respect to PSA testing.  There
is no evidence that he is willing to accept that it is useful, but there
is also no evidence that, using the same criteria, that it is not
useful.   So he might very well decide not to be tested because he sees
evidence, which doesn't meet the strict criteria for acceptability.  But
at the same time, he can't tell you or me that we should not use other
evidence which doesn't meet that criteria to justify to ourselves being
tested.

He might argue that it is wrong for an expert to recommend testing as a
public health matter, but he also can't say an expert should recommend
not being tested.

Let me point out that we often base decisions on evidence which doesn't
meet the strictest criterion.   For example, there has never been a
statistically valid prospective study which shows that smoking causes
lung cancer.  For a long time, tobacco companies made a big deal of
that.  But the other evidence linking smoking to lung cancer, although
it doesn't meet the strongest criteria, is so strong that no seriou
scientist would claim that it is still and opne question whether smoking
causes lung cancer or not.

Note that PSA testing is by itself innocuous.   The procedure has no
serious risks, and it is not very expensive.  The problem is what you do
after testing suggests a biopsy.  Biopsy is more risky, but still not
terribly so, and it is not awfully expensive as medical procedure go.
So the real problem is that biopsy may show you have cancer.  That is
really where the  possibility of significant damage to the patient
arises.  If the patient chooses aggressive treatment, it may kill the
patient, although that risk is very small, and it may damage the patient
in other ways. (Note that these other side effects are far from certain
also.) So the question is to determine how many cancers discovered this
way really need to be treated.   Now, on this matter, there has been one
prospective randomized study of Swedish men diagnosed with cancer which
compares RP to WW (presumably, followed by HT if necessary).  And it
showed statistically significant benefits for RP in a variety of
categories, including overall mortality, number of cases of metastatic
cancer, and cancer mortality.   Still, that doesn't necessarily prove
the same would hold true in the US where circumstances are different,
which shows that even a rigorous study may not provide a definitive answer.

Then why does his decision give you pause.  Presumably, you believe that
in your case the cancer really did need to be treated aggressively and
moreover there is a good chance it would not have been detected in time
to be cured without PSA testing.  In retrospect, you made the right
deicison.  If you have followed the expert's advice,  you would have
forgone testing with likely unpleasant effects.  He may argue that if we
took a large number of men there is no evidence that on the average,
those that weren't tested did worse than those who were.   But he would
have to admit that some men, like you, would have been helped.

Let me outline the skeptics response to that---although I don't find it
convincing.  The argument is that the testing reveals many cases which
then end up being treated but in fact never needed treatment in the
first place.  The Swedish study seems to suggest that, after detection,
a comparison of men treated to men not treated do better on the averge
with respect to total mortality. prostate cancer specific mortality, and
the development of metastatic cancer.  Unfortunately, it seems unlikely
that the cancers in the Swedish study were detected by PSA testing, so,
while suggestive, the evidence doesn't show that testing by itself is
beneficial.

Of course, there is a prospective randomized study going on in the US
which, in pfinciple, could settle the question.  Unfortunately, this
study also suffers from some procedural problems.  To be definitive,
such a study would have to be carried on until all the men in the study
are dead, from whatever cause.  That is particularly the case since few
uorlogists recommend aggressive treatment for older men in whom prostate
cancer is diagnosed.   Younger men are more likely to be treated
aggressively, and their cancer could take 25, 20 or more years to
develop.  That means that any study which just takes a snapshot at some
specific number of years, like 10 years, might consider many cancers
innocuous just because they hadn't had enough time to develop further.

Another problem is that medicine isn't likely to sit still for the long
time periods necessary to establish the hypothesis being tested.   It is
quite possible that in the not too distant future, some other test will
be developed which, either alone or in conjunction with PSA testing, is
much more definitive than PSA testing by itself.  Or, treatment methods
may evolve so that the side effects are so minimal that they can be
ignored.  In that case, no one would have anything to lose by treating a
prostate cancer, and worrying about whether or not it might prove
innocuous would be pointless.  We have already seen this sort of thing
happening.  The dividing line for when to biopsy used to be 4 ng/ml.
Today age adjusted stricter criteria are used and they are combining
with other methods, such as free PCA and PSA velocity.  When the current
study began, the criterion was simply PSA 4.  By the time the study is
completed, no one will care about the result.