Day 022 - 12 Sep 94 - Page 63
1
2 Q. 12, it says.
3
4 MR. RAMPTON: My Lord, it is the early analysis one, it is the
5 1987 one.
6 A. Yes, the first report.
7
8 MR. MORRIS: Right. "It is not likely that the difference is
9 due chance. However, two important findings of that study
10 in our analysis are in agreement. Total calorie intake is
11 not associated with increased risk of breast cancer.
12 Secondly, there is little evidence for an association
13 between fat intake and increased risk of breast cancer in
14 premenopausal women. Our finding of an association
15 between fat intake and post-menopausal breast cancer is
16 important for two reasons. First, if the association is
17 causal, prevention of even a small percentage of breast
18 cancer is important in terms of absolute numbers because
19 of the high incidence of this cancer in a number of
20 countries. Secondly, if the association with adult and
21 diet simply reflects an underlying causal association with
22 dietary in adolescence, as has been suggested by a number
23 of authors, identification of the latter association would
24 directly affect the public health.
25
26 Therefore, it seems a matter of some urgency to acquire
27 further analytic epidemiological data to help resolve the
28 apparent difference between results of our analysis and
29 those of the cohort studies performed by Willett and
30 Jones. Our identification of an inverse association with
31 various markers of fruit and vegetable consumption, led
32 support to the hypothesis that increased consumption of
33 these foods may reduce breast cancer risk. A protective
34 effect of fruit and vegetables has" -- I cannot read the
35 word?
36 A. "... been observed".
37
38 Q. ... "also been observed in a number of studies of other
39 cancers, including cancer of the colon, lung and
40 pancreas. This appears to be one of the more consistent
41 dietary findings from analytic epidemiological studies.
42 Further investigation of such association seems urgent."
43
44 So, you may disagree with his conclusions based upon his
45 analysis, but that was a survey anyway of all the
46 pre-existing fat and breast cancer studies?
47 A. One of the problems of these so-called meta-analyses,
48 in other words, looking at, lumping together a large
49 numbers of trials, is something which statisticians argue
50 about very strongly; to know whether these techniques
51 which have been in the past used mainly for looking at
52 clinical trials where you are comparing results of
53 treatment, for example, can they be applied to this kind
54 of analysis? There is great debate about whether, in
55 fact, you are able to do that and produce what are
56 meaningful results. So, there is a great deal of
57 criticism about this particular study and the way in which
58 the trials were lumped together and analysed. But -----
59
60 Q. Can we just go to one final -----
