Day 036 - 13 Oct 94 - Page 15
1
2 Q. How do you grade your social classes in America -- do you
3 give them numbers or letters?
4 A. There is no standard practice in that regard.
5
6 Q. I am not sure we have a standard practice, but there are
7 usually, I think -- I am not an expert -- about five
8 grades attributed to socioeconomic status in this
9 country. Do you suppose that the customers of McDonald's
10 are mostly from the top one or two socioeconomic grades,
11 Dr. Barnard?
12 A. I would have no idea.
13
14 Q. Nor should you; it is, perhaps, an unfair question. Let
15 me finish the paragraph, if I may: "These data are not
16 consistent with the hypothesis that high dietary fat
17 intake increases breast cancer risk. Indeed, they suggest
18 a possible protective effect of high-fat intake, but this
19 result may be influenced by methodologic problems with the
20 dietary assessment. These results certainly indicate the
21 need for further exploration". Do you agree with that
22 last sentence as at 1987?
23 A. I would agree that there is a continuing need for
24 exploration. This study again recalls the analogy, if
25 I may express it again -- I may have expressed it more
26 than once before -- that when one is looking at cohort
27 studies, one cannot -- one is making a serious error if
28 one looks at them as, say, a poll, and if half the cohort
29 studies or more than half identify a dietary culprit, if
30 you will, and a minority do not, then one simply goes by
31 the weight of the numbers. That is not the way that these
32 results can be used because of the methodologic problems
33 which you cited on page 469.
34
35 Q. Yes, I deliberately drew that to your attention.
36 A. One will expect that many, many cohort studies will
37 find no significant effect at all, even when a very
38 important effect is present. That is why I tried to use
39 the analogy, in the hope it would be helpful, that if one
40 sends, say, 20 search parties into the woods and 15 of
41 them come up empty handed, but five of them find exactly
42 what you are looking for, one does not ask the 15 to vote
43 and the five to vote and then conclude that there is
44 nothing whatsoever in the woods. That would be a terrible
45 error, and any scientist who discounts negative cohort
46 studies as evidence that this relationship simply does not
47 exist is making a misinterpretation.
48
49 Q. Oh, yes, Dr. Barnard, perhaps you have not properly
50 understood what may be the issue in this court and
51 eventually what the issue is will be determined by his
52 Lordship's rulings on various matters (with which you need
53 not be concerned). Assume for the moment that it is not
54 part of my purpose to establish positively a negative;
55 assume that it is not my purpose to establish that the
56 links between diet and cancer, such as the evidence shows
57 them to be, are not causal.
58 A. Yes.
59
60 Q. Do you understand?
