Medicine's Ten Greatest Discoveries

Steven B. Harris sbharris at ix.netcom.com
Sun Sep 6 16:18:10 EST 1998


In <6stu7f$mpu$1 at nina.pagesz.net> henryj at nina.pagesz.net (George
Conklin) writes: 
>
>In article <6stijp$lf8 at sjx-ixn5.ix.netcom.com>,
>Steven B. Harris <sbharris at ix.netcom.com> wrote:
>>In <EQrI1.2130$c3.3697154 at tor-nn1.netcom.ca> "David Lloyd-Jones"
>><dlj at pobox.com> writes: 
>>
>>>In somewhat the same sense, multiple regression analysis, which was
>>>destroyed epistemologically by Locke and Hume a couple of centuries
>>before
>>>it actually came into general use, has no claim whatsoever to
Platonic
>>>logical rigor -- but is one of the most useful forms of explanation
>>that we
>>>have.  Usually right, on rare occasions wrong.
>>
>>
>>   In your dreams.  Logistic regression is mainly good for ruling out
>>causation (in which case it really is usually right, and only on rare
>>occations wrong).  But for cases where variables correlate, the usual
>>case is that direct causal relationships cannot be inferred, and
quite
>>often do not exist. 
>
>   Any old excuse to justify the medical establishment's
>fear of real facts. When we have vastly different medical
>practice in the USA, such as radical prostate operations
>common in some areas and not in others, but have similar
>RESULTS, then badly-trained MDs (trained, NOT educated)
>mouth that this means nothing becasue to admit it would
>interfere with the cash flow.



    Alas, your example only proves my point.  A *lack* of 
correlation between two things (rate of radical prostate opperations
and cancer outcome) is a far stronger result from epidemiological
statistics than correlations are.

   Again, LACK of correlation between variables is strong (though not
perfect) evidence against causation.  This, simply because if the
variables are causally related, it is difficult not to see the
correlation statistically (something else must be making up for it, and
doing it nearly perfectly-- that's rare).  Lack of correlation is thus
the strongest use of epidemiology.  Correlations from epidemiology,
however, are WEAK evidence, and mere SUGGESTIONS of a causal
relationship, and have to be followed up by all kinds of other kinds of
studies before they can be taken as firm knowledge of causation.  For
example, if my stats showed that in states where they did a lot of
radical prostate surgery, people with prostate cancer lived longer,
you'd have a perfect right to be skeptical that this said anything very
firm about the usefulness of the technique.  Since, of course, you and
I can both easily think of many ways in which this could come out to be
true, without this surgery truely being any better, as a treatment (for
appropriate stages of this cancer).

                                           Steve Harris, M.D.  



More information about the Bioforum mailing list