Day 022 - 12 Sep 94 - Page 52


     
     1        follow-up, they had 150 incident cases of colon cancer
     2        which was documented?
     3        A.  This is the problem.  The numbers of cases are very
     4        small in this.  You are only dealing with 150 events out
     5        of 89,000 people.
     6
     7   Q.   Whereas when you get to tab 5, you have approximately 10
     8        times as many incident cases of breast cancer with the
     9        same number of women over an eight-year period rather than
    10        a six-year period.   Your confidence limits interval is
    11         .67 to 1.08?
    12        A.  That is right.
    13
    14   Q.   So, it is statistically, pretty well related to the number
    15        of incidents you have?
    16        A.  It is.
    17
    18   Q.   With more or less the same number of subjects; is that
    19        right?  I want to make sure I have understood it.
    20        A.  It is.  The problem is that the smaller the number of
    21        events that take place the less reliable are the data,
    22        unfortunately.
    23
    24   Q.   Or the larger the prospect that it could be chance and not
    25        related to any identified factor?
    26        A.  That is right.
    27
    28   Q.   Is my understanding right?
    29        A.  Yes.
    30
    31   MR. JUSTICE BELL:  I am certainly no statistician.  Yes,
    32        Mr. Morris?
    33
    34   MR. MORRIS:  This study was really over a period of six
    35        years -- well, yes.  The people were originally healthy,
    36        so they developed colorectal cancer in that time?
    37        A.  Yes.
    38
    39   Q.   So really -----
    40        A.  One would argue that the follow-up period of time is
    41        too short.
    42
    43   Q.   Yes, but this is going to be a problem with any cohort
    44        study of cancer?
    45        A.  It is.
    46
    47   Q.   That even a period of eight or 10 years?
    48        A.  Yes, that is right.  It is what we were talking about
    49        earlier on as being a problem with these prospective
    50        studies, that you need to carry on looking at them over a 
    51        long period of time.  Because, you know, it is quite 
    52        possible that in a few years time, another five years 
    53        time, these results will change.
    54
    55   Q.   If we look at the last page before the references,
    56        internal 377 or file No. 377, the top paragraph on the
    57        right-hand side, it says that the absence of a positive
    58        association between dietary fat and breast cancer in this
    59        same cohort has been questioned, because of a possibly
    60        inadequate variation of fat intake in the studied

Prev Next Index