Day 040 - 21 Oct 94 - Page 20

     
     1        A.  The Butler and Conning paper?
     2
     3   Q.   1983, Ford, Butler and Gaunt?
     4        A.  No.  It has not been published, to the best of my
     5        knowledge.
     6
     7   Q.   It says it is an unpublished report; I was wondering
     8        whether you had had some entree to it?
     9        A.  I have obtained through the British library some of
    10        these unpublished reports, but by no means all of them and
    11        I am not certain that that is one I have scrutinized.
    12
    13   Q.   That is one you could get hold of, is it?
    14        A.  It may be.  Insofar as it is a document submitted to
    15        the Committee on Toxicity prior to the introduction of new
    16        arrangements in the late 80s, it might fall outside of the
    17        provisions of those arrangements and I may not be able to
    18        obtain it.  I cannot say without further investigation
    19        whether it is publicly available.
    20
         But what I would have been more comfortable with would have
    22        been if they had not simply said:  "It fell short of
    23        statistical significance", but gave a measure of the
    24        statistical significance and, better still, if they had
    25        included a table giving the incidence within these groups,
    26        in the control group and the low dose group.  Because it is
    27        important to know whether it fell short of statistical
    28        significance, not only to a marginal extent or a
    29        substantial extent, but what was the incidence of these
    30        lesions in the control group.  Because if there was a
    31        particularly high incidence of relatively historical
    32        controls in the concurrent controls, then it might have
    33        fallen short of statistical significance for that reason.
    34
    35        I think this brings me to a point that arose that I wanted
    36        to make towards the end of the discussion yesterday
    37        afternoon that I was not able to make because we had to
    38        stop at 4.30, which is this:  Since, typically, we are
    39        dealing with animals of groups of 50 animals per gender per
    40        dose group, and the incidence of the relevant lesions in
         the control group is usually greater than zero, these
    42        experiments are relatively insensitive because it typically
    43        means, when one takes into account the variation within the
    44        controls and the requirement that demonstrates statistical
    45        significance, that using groups of these sizes it is not
    46        possible in practice to be confident of anything more than
    47        the real incidence of these lesions in these groups is
    48        unlikely to have been greater than, say, five or 10 per
    49        cent of the population.
    50 
    51        So, I think it is a mistake to say there was no effect at 
    52        50 milligrams; what would be better to say is, the 
    53        incidence of effects at the 50 milligram dose level was not
    54        greater than a specific figure.  But the specific figure is
    55        not given.
    56
    57        So there are many respects in which there is scope for
    58        different interpretations of these figures, and I would not
    59        necessarily assume, as I think you are assuming, that if
    60        and when the Scientific Committee for Food, or some other