Day 017 - 25 Jul 94 - Page 25


     
     1        breast cancer and compared these to the intake of women
              who were found not to have malignant disease.  No
     2        relationship with fat was found.
 
     3        Hirohata and his colleagues have studied fat intake in
              relation to breast cancer in three groups of women with a
     4        differing incidence of the disease.  Japanese women in
              Fukuoka, Japan (low risk), Japanese women in Hawaii
     5        (intermediate) and Caucasian women in Hawaii (high risk).
              No significant differences were found between cases and
     6        controls, whether surgical or neighbourhood, in any of the
              three groups in their mean intake of total or saturated
     7        fat".
 
     8        Given what Kolonel thought he had found earlier on in
              relation to Hawaii, does that surprise you?
     9        A.  No, because, in fact, Kolonel himself admitted that
              the association he detected was a very weak one, and this
    10        is -- we were talking about statistics a little earlier
              on, and whether these changes or associations could have
    11        occurred purely by chance.  What one is seeing in many of
              these studies is that there is a wide range of what we
    12        call confidence limits.  In other words, you cannot comb
              down the information enough to be sure that the difference
    13        could not have arisen purely by chance.
 
    14        So these two studies that you have been quoting are
              overlapping each over; they are still within the range of
    15        variability as determined in these two separate studies.
 
    16   Q.   Then I will go on:  "Overall, the results of these
              case-control studies provide relatively little support for
    17        the hypothesis that fat is a cause of breast cancer.
              Commentators have been quick to defend the hypothesis by
    18        stressing the limitations of dietary case-control studies.
                Not only may the presence of disease add to the obvious
    19        difficulties of recalling past diet, but also, by
              affecting current diet, it may unconsciously influence
    20        memory of previous dietary habits.  It has also been
              suggested that those groups studied with negative results
    21        may have contained insufficient variation in fact intake
              for any real effect to be detected.  Reference to the data
    22        presented, however, shows this criticism to be unfounded
              if the relationship shown in Figure 1", that was the
    23        linear graph, "is genuine".  Does that seem sense to you?
              A.  I think that is a very fair assessment of the problems
    24        associated with these studies.
 
    25   Q.   Can I ask you this, Dr. Arnott:  I do not know about the
              medical world, but is there sometimes a tendency for 
    26        people who have hit upon the thought they have found an 
              attractive looking hypothesis or theory to be, what shall 
    27        I say, somewhat inpenetrable to intelligent criticism?
              A.  I think it is a general problem in medicine, not just
    28        in these sorts of studies, but in reporting results of
              treatment.  It is always nice to be able report something
    29        positive whereas, in fact, in reality, reporting a
              negative result is equally as important.  I would just
    30        allude to the study that I was involved with on nutron
              therapy.

Prev Next Index