Abstract
This 1984 British American Tobacco (B.A.T.) document was used recently in an ongoing case against the tobacco industry (the City of St. Louis et al. v. American Tobacco et. al). In response to tobacco company claims that they had nothing to do with a B.A.T Co. document entitled a "Compendium of Epidemiological Studies" authored by Ray Thornton of B.A.T.Co, an attorney pointed this document out to the judge whereby B.A.T Industries is given pointers on how to maintain the controversy using the document. Here is a key quote from the document:
"The Royal College of Physicians claims that 90% of all lung cancer deaths can be attributed to smoking. There can be no doubt that this is widely believed to be true and that lung cancer is the most emotive single issue. If we can cast doubt on the relationship between smoking and lung cancer then we have cast doubts on the entire case against smoking..."
Page 1: 00041582
Log in for more options!
To
Re~
D~. R.E. Thornton -
SOUTHAMPTON. -" :"
G EF~/.I'C.. .....
Dear Ray, .: -
BAT IrtDUSTi~IES
Thank you very much for your letter of 6th April.. I "
tried to call you but I gather you were not well th~s week. I -
trustyou are now feeltug bet~ar and let me offer you some ~
thot~hts. " " - .
I like your outline very much. ]But even now I feel thaZ
the conceptual difficulties are very great and it Ls v~lly tmporU~nt
that we lead people gentiy into th~s vezD, difficult subject.
I also feel there is so much to say about lung cancer that
it would be very deshzable to concemtrate on this single issue
alone.
I could tmagine an outltue for an extremely t~te~est~g
shorZ book which might be on the followhng l~nes.
(I)
There exists a major controversy about smoking and .
' health. But mlthough many people believe that the •
"case for the prosecution" is absolutely clear cuz and
proven, the hard fact is thet there are ma/ly anomalies
tu the evidence. The more we apply modern techniques
of research, the more anomalies we find. It is the
of this book to examine some of the anomalies and to show
that the whole subject needs far more objective in~esd
(2) • .The peo~&~ a~tack ~he ~bacco industry shU'~ their .....
ground fr~quent]y~ but it is fatr to say that lung can&er. .....
is a key issue. The Roya/College of Physicians claims
that 90~ of.all lung cancer deaths can be attributed to "
: " smokhug~., There can be no doubt that this is widely .........
believed to be true end that lung cancer is the mostV .........
So let us look at the e~/dence on lung cancer in more ......
BAT INDUSTRIES - OKLAHOMA TOBACCO LITIGATION
Page 2: 00041582
Log in for more options!
(~}
(~)
C7)
.... (s) ..
- 2 -
In the UK the number of deaths from lung cancer is
declining fast. Give a chart illustrating the crude ....
figures'of deaths per year. This alone casts some--
doubt on the re/afionship. But we also know that-the
rate of smoking is declining and that the tar content .
of cigarettes is declining. Therefore the vital question
we need te ccmsider is this. When precisely did the
lung cancer down-turn begin? Only then can we con-
sider how ths~ down-turn relates to the down-turn in
the ~moking habits.
We introduce the idea of epidemblogy - a statistical
analysis of possible associations between deaths and
other factors. ('We use this moment to stress the lack
of direct causal evidence, so that statistics are all we
• h~ve to go on. ) We also introduce the role of the
computer in analysing masses of data.
We introduce the Government publication "Trends in
Cancer Mortality". We explain who its authors are,
what they are trying to do, and why it is analysing by
the year of birth rather than by the year of death - in
other words, we explain the concept of cohort analysis.
The authors are tr3-[ng to show what are the real trends in
each different type of cancer. With reference to lung
cancer they are in a position to show the real date of
change of trend. Remind the reader why this was the
crucial question.
We introduce the simple charts on pages 28 and 29 ~ by
which I mean the bottom right-hand chart showing the
cohort value only~ a simple one-line chart in each case.
We quote the conclusions given by the authors and intepret
them to mean that the change in trend applied to the cohorts
born in 1900 (men) and 1925 (women). "'
We then introduce the other charts on the page and explain
them. Then we move on to the more complicated chart
shown on page 3.(figure I) on your own paper and cite what
Osmond and Gardner said about this. In this way we have
now established the lung cancer trend with absohrte clarity.
we t==~ the evidence on smoking and show ~ ~ ~i:
yearly smoking 'figures. But again we explain why th~se
are not an adequate guide. What matters is the quantity/
of smoking over a period by each cohort. So wenow
chart the amount of smoldng up tothe age of 33~ as done
in your paper~ but explaining the figures in full. " " [- "
BAT INDUSTRIES - OKLAHOMA TOBACCO LITIGATION
Page 3: 00041582
Log in for more options!
- 3 -
(9)
Only now, for the furst time, do we have comparable
evidence which gives the lung cancer death rates and .
also the smoking rates for the same groups of people,
cohort by cohort. Emphasise the imporU%nce of this
as the only method which can give reliable results. ~-..
(10) Then superimpose the smoking and the cancer charts
for both men and women. "" "
We conclude that the l~nes on the chart are very far
apart and suggest no relationship whatsoever - to
many people this is an astonishing result which is why
it is so imporJu%nt ~o lead up to this point only very
slowly~ building up the argument step by step.
(II) But all this only confirms what many experts alr~a~i, y
believed, that there were far too many anomalies"in[~he
evidence.
(12)
We now examine some of uhe other anomalies, ~[ving
de~ailed figures, charts and maps tn each case.
(i) M~ny non-smokers get lung cancer.
(ii) Many heavy smokers don't get lung cancer.
(iii) Lung cancer rates are lower among inhalers.,
than ardong non-inhalers.
(iv) The number of non-smokers who get lung cancer
may be rising.
(v) There are continuing errors in the diagnosis of
lung cancer~ especially when these are checked "
agahlst autopsy results going back to the 19th
c entity.
(vi) 30~ of lung cancer is secondary and has no
statistical cor~ala~o~ w~th 'smoking.
(vii) Lung cancer rarely occurs in both lungs although
they are equally exposed to smoke. " "
(~x) . Lung cancer r~sks ~~ccorduug to :; -~
occupation, diet, race, constitution and psymhology.
(x) Lung cancer risks ~.ary a=cording to place ~
residence.
./.
BAT INDUSTRIES - OKLAHOMA TOBACCO LITIGATION
Page 4: 00041582
Log in for more options!
- 4 -
( 13 ) Examim the experience of other countries in as much
detail as possible. How do lung cancer death rates
match smoking rates? What is the correlation between
them? What are the trends? Is any cohort axmlysis
available on any of them? Draw appropriate conclusionso
(14) All this~,,suggests there is no valid relationship. But
how does it square with other bits of evidence? Examine
some of the most telling arguments cited in the RCP Report,
etc., and show some grounds for doubt, (such as non-
randomness of the data base and other statistical failings ).
(~6)
(tT)
(18)
(19)
(20)
But if smoking is not the cause what is the cause?
Explain the general problem of the biology and causation
6f diseases.
Modern research into cancer causation. What we know
about the mechanics. Why is causation such a mystery?
What exactly is the problem?
An alternative analysis. The Butch theory that cancer
has little or nothing to do with any environmental factors.
We draw only this modest conclusion. Ttmt the evidence
is still inadequate and the case is not proven either way.
Hence the need for more research and how it might be
organised.
But it would help the cause of good health if doctors under-
stood the facts and if everybody stuck to the truth.
BAT INDUSTRIES - OKLAHOMA TOBACCO LITIGATION
