George conklin keeps repeating this same argument.  He asserts that
surely there would be something to show by now if treatment were
effective.   There is of course no logic to that assertion.  Diseases
with a long time horizon may show no significant difference for long
periods of time but then show quite significant results after a longer
time.  I keep pointing out that if one did a completely randomized study
comparing smokers to no-smokers to see if smoking caused lung cancer,
there would be little or no effect for quite a long period of time.  If
one limited the time horizon of the study, one would conclude that
smoking doesn't cause lung cancer.   Of course, no such study has ever
been done, and no such study will ever be done.   The link between
smoking and lung cancer is well established without such gold standard
studies.

Similarly, the early Swedish studies in the 80s showed that aggressive
treatment was no more effective than watchful waiting followed by
hormone therapy when necssary.  That was true when patients were
followed for ten years.   Later studies showed significant differences
developed in the same patients as time progressed.

The latest Swedish studies by Homberg, et. al. show significant
differences in prostate cancer mortality in well less than ten years.
They also show significant differences in other aspects of disease
progression.  However, the Swedish population is different from the US
population because of differences in screening.  Men in the US will tend
to have prostate cancer detected five or more years earlier than the men
in the Swedish study.  There are several rather obvious effects this
might have.   First of all, one would expect higher cure rates for early
treatment.  Second, there could be more men treated for cancers which
would never bother those men during their lifetimes.  Whether this would
make a difference between populations treated by surgery or treated by
watchful waiting followed by hormone therapy when necessary is not clear
since innocuous cancers would be detected in both groups.  Finally, it
seems clear, everything would be extended by at least five years.

There are some importnat thing to remember about statistical studies.
First, any study only tells you something about the population being
sampled.  If a study were to show that within ten years for men 65 or
older there was no difference in survival rates between two treatment
protocols, then if you were over 65, similar in characteristics to the
men in the study,  didn't expect to live longer than ten years, and were
concerned only with survival, then you could reasonably use the results
of the study to guide your choice.

Second, no sensible study is going to give one up or down answer for
diseases like prostate cancer.  We should expect there to be significant
differences among different parts of the study population segregated by
various characteristics.  For one, age should be a significant factor as
should the aggressiveness of the cancer.   When might be appropriate for
a man in his late 70s with a Gleason 6 cancer might not be appropriate
for a man in his 50 with the same cancer or for a man in his late 60s
with a Gleason 7 cancer.  It would be very surprising if such
differences were not to emerge from such a study.