Re: Medical Research-Evidence

In article <k4Eeg.6437$921.356@xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx>,
George Conklin <georgeconklin1@xxxxxxxxxxxxx> wrote:

"Skeptic" <bcs002b@xxxxxxxxx> wrote in message
news:cikag.733050$084.383622@xxxxxxxxxxxx

[Long line of back-and-forth deleted.]



Even if the results get no better than they are now, their is a
signficant
difference, clincally and statistically.

What Herman was trying to tell you is that this is not so.

Herman is not a a doctor and a physician is in a better situation to
determine what is CLINICALLY significant.

On that, I agree, provided that the full information
is known. One can have either type of significance
without the other. Unfortunately, too many medical
people do not realize this.

....................

Herman's comment that he would rather go with a study where 3 out of
3
are cured than with a large study showing very small differences. It is
a
valid point. He is in favor a Bayesian methods. Even the New York
Times
had an article comparing the approaches.

I'm not disagreeing, which you don't seem to understand. However, we
don't
have such results. What we have is a larger study showing a smaller
response, but nonetheless, it showed that patients were cured from
surgical
intervention for prostate cancer. Small numbers? Yes. But lives are
lives
so meaningful numbers.



Very small
differences must be evaluated carefully even if statistically
significant.
The original authors state differences are small. Why dispute
them?

I'm not disputing them. I support them. You just don't seem to
understand
that a relatively small difference can be a very important one...
especially if you're amonge the extra 20 lives saved in that study.

The problem is that when you rerun the study, with low levels of
significance you might just as well find the opposite, as the HRT
studies
showed or seemed to show.

you're living in hypotheticals now. Fact - the study showed a survival
advantage for surgery vs. watchful waiting. It's the best study we
have
right on that topic.

Actually according to what Herman has posted over the years, it is not
hypothetical at all, but a possible artifact.

Yes, an unlikely possibiltiy, but nonetheless a possibility. This is why
multiple studies are being done and why the same study in question will
continue to evaluate their results over the years. The quality of the
study
from this one study was excellent and I find the results believable. If
held true over time and confirmed by other studies, the issue will have no
doubt left. The doubt at this time is minimal.



Here is a comment from UK:

Prostate therapy benefits doubted
Many men diagnosed with low-grade prostate cancer do not benefit from
radical treatment, research suggests.
The researchers calculated that, even without treatment, only about 1% of
men aged 55-59 with diagnosed low-grade disease would die within 15 years.

This is NOT a small number. What are the other factors
involved? We need to find more of the relevant factors,
and this is not going to be done by the type of medical
studies now being done. We cannot look at one or two
things at a time, but must look at dozens.

This is the opposite problem to the usual data mining.
It is hard to come to valid conclusions in high dimensional
problems without making assumptions, and few realize what
is involved in making probability assumptions. As one of
the too few who have looked into the problem of how much
one can get, there are things which can be done, and things
which cannot.

Side effects of radical treatment such as surgery and radiotherapy can
include incontinence and impotence.

This involves both probabilistic considerations and
considerations of personal utility.

It also requires the use of idiosyncratic variables.
Population figures are inadequate; they only look at
the type of cancer, not who has it. It is possible
to estimate probabilities without observing "identical"
cases, and the medical profession, as well as others,
needs to realize this and act on it.

The Department of Health said its advisers would consider the Institute of
Cancer Research findings.

The study appears in the British Journal of Cancer.


The decision whether to have radical treatment can be tremendously
difficult for the patient
Dr Chris Parker

Of course. The information made available to the
patient is inadequate.

[Large quoted section delete.]
--
This address is for information only. I do not claim that these views
are those of the Statistics Department or of Purdue University.
Herman Rubin, Department of Statistics, Purdue University
hrubin@xxxxxxxxxxxxxxx Phone: (765)494-6054 FAX: (765)494-0558
.

Relevant Pages

  • Re: Acupuncture for Headaches
    ... patient should *not* know. ... etc.) and seeing if acupuncture does better than the others. ... I never said that "all unblinded studies are statistically meaningless." ... but not because of statistics. ...
    (misc.health.alternative)
  • Re: Medical Research
    ... which does not understand how to use statistics. ... benefits of any available treatment, ... the individual patient. ... hormone treatment probably does help in a selected patient population. ...
    (talk.politics.medicine)
  • Re: Medical Research-Evidence
    ... probabilities of the various results for each type of ... and let the patient make the decision. ... preferences for various outcomes, decide on the treatment. ... Herman Rubin, Department of Statistics, Purdue University ...
    (talk.politics.medicine)
  • Re: Medical Research
    ... which does not understand how to use statistics. ... of a treatment, and especially so if qualitative results ... the individual patient. ... In actual practice, patients most ...
    (talk.politics.medicine)
  • Re: Medical Research
    ... which does not understand how to use statistics. ... of a treatment, and especially so if qualitative results ... the individual patient. ... In actual practice, patients most ...
    (talk.politics.medicine)