J.
27 th Gctcter 1976
CIGAnE7TE SMOKING AND CAV5AL
The public positicn cf tobac=o co-panies with
respect to causal explanations of 'Ile assccaL.on
of cigarette
sncking and di seases is dominated ty legal
consIderat ions.
In the ..;Itirnate conpanies wish to te able to
ditpute'that a
part!c-j:ar product was the cause of injury to a
particular
perscn. By repudiation df a causal role for
cigarette
s.~ckinq in general they hope to avoid liability
In particular
cases. This domination by le3al consideration thus
leads
the Ind.;stry into a public rejecticn in toltal of
any causal
relati:-ship between smoking and disease and puts
the
industr-... in a peculiar position with respect to
product
safety tiscussions, safety evaluations,
collaborative
researz- etc. C~:--ipanies are acti7ely see~lra to
make
products acceptable as safer while denying
strenuously the
need tc ~o so. To -any the industry appears
intranstgent
and irresponsible. The problen of causal.ity has
been
Inf2ate:: to enormous proportions. The Industry
~Rs
ret,_at-i~ b'-,hihd' impossibl-e -derpixids
-fo;r"sc,"entifiba.
whereas ~s-icA proof has rye er
v e,. reiquired as a basis for
action i-i the legal and poll-tical fields. Indeed
If the
doctrine were widely adopted the results wculd be
disastrous.
I believe that with a better understanding of the
nature of
causality it
is plain that while epidemlolcgical evidence
does in,~:-ate a cause for concern and action it
cannot form
a basis ~n which to claim damage for Injury to a
specific
ind!vId-.;a1.
In the Surgecn General's Adviscry Committee
Firs t F2::`rt the q-jestion of causality is
raised. They
reccgnlse~ the dif.-Iculties but agreed that
causal 51cr-ifi-
cance cf an association Is a matter of judgment
and noted
a number of criteria which may be utillsed
together, not
one of -_~~!ch alone would be sufficient, e.q. the
consistency,
strength and specificity of the assc-ciatlo~, the
te,,,pc.,ral
relatIcns1!!p and the cohprence of the
association. Th. e y
consideret characterisation of the assessment of
an
asscciati:n as a "factor", a "deter,inant" or a
"cause".
They rec:;gnised "factor" as a source of variation
but not
necessarily as a cause. Further there can be
co-existence
of several factors required for the occurrence of
a disease
but one factor may play a determinant role. The
word cause
was held to convey the notion of a significant
effectual
M506 :w
Czntd.
ok
c0
r--.)
d\
BATCo document for Province of British Columbia I
November 1999


2
relationsn!:) bet--cen an a~;ent and an
ass-:,:iated disorder in
the host. All the nembers acceptet a ~.ultlple
etllioloqy
vle. of b;:Iogical processes. In the end they
defined
"cause" merely to cs~flrm ~7e!r c-_~-Icti:)ns.
The Surgeon Car~eral's Co,.n!ttQe had such great
d'fficultf kith the problem of causality b~-cause
the tests
for causality we "Instinct,ely" a,~sot. are
Lhose-of "necessity
and sufficiency sptlt out I:.- J-S-Mllfl,~hereas
epidemiology
requires - as the Ccnmittee demonstrates In
practice
different 'reatment. It is my thesis that the
tests Of
Hill (.h!c" I will call tests for scecial
causality) apply
only to -e:nanistic syste-s -hereas epidelology,
for
example, requires a probabilistic a=proach.
In the nineteenth ce-~tury It was assumed that
e,,ery eff,!~t must have a.cs-se. 7z pr-zve that A
caused 8
it would have to be de,~cnstrated that e-.-ent A
pre~_edes event
8 and that event A is both ~.ecessarv and
sufficient to
ensure that event 5 will fs1low. The view of
science
itself was mechanistic, it.-3s fell, that If
everything
wa',s.),no.n t" t t - ti#~e,of that system
P"._ a sys. em,!'tlep he - f u
rt a W imi'ts'; this
~Wou'14, be pzed'iciabl it
y
concept is. still va z: f- a'.ppearances the
In spite
behaviour of billiard balls is entirely
plre~ictable and,
given the d3ta, is calculat:e. T I, 19 c:-~ncept
still
dominates legal thinking. Althouq'i the extension
of
c,:-ntrIbut::y negligence insreasingly recognises
multiple
causes the concept Is nevertheless echa~istic -
there
Is an underlying assumpticn that, qlven the facts,
the
results are predlct3!~,.le, ca:sulable and legally
apportionable.
This mec~.ani.stlc cDr=ept of causality Is still
also the
p:;ular c:ncept. :, the 7:~ern :rld cf advertising
even
L~%e most c=~plex systems ireginable (hu-nan
emotional
situatior,s) are presented as predictably
controllable by
simple events such as using Brand X.
Julian Huxley in 19SB wrote "The conclusion to
be dra-,~n f::~m the e,.-idence is definite,
Increased smoking
Increases the prcbatillty cf devel,~Pinq lung
cancer.
Unfortunately the significance of such a
sLatenient is not
clear to iamy people. Obsessed by the naive Idea
of
finding a single cause for every effect, they
shake their
heads and say that the evidence is cnly
statistical$ as
If that Invalidated It. In point of fact, every
scientific
law is statistical, and all that physicists can
tell us
about electrons within the atom Is the probability
of their
occupying any particular pcsition".
Contd.
BATCO document for Province of BritiSh Columbia 1
November 1999
C)
CD
CN
--_j
cc
r1,) . f
rQ
,_n


3
J,
Thus ;u!ian 14ux!ey be-,am to szell out a
probabilistic ~!.causal !ty (~h.Ch 1 -111 call
general
causality) based on twent.leth ceitiry science.
When we
have a science ~vhl=h eTibraces the uncertainly
principle -
the utiknowability of the total data for so-.ie
systems - and
probability to describe the pr,;p'erties of large
populations
the., we car-ot expect s-pecial causality to be
very,useful.
Hcwe,er, just as lle~tcmlsl.a-s are true enough
for everyday
experience and relativity would be superfluous the
tests
for special causality can still be applied to very
simple
unified systenis. out for c-z.~plex systems
Involving the
properties of larce numbers e.,tperlnentally
established
relatiomshizs may be used to predict the behaviour
of the
population with accuracy.
r! Fcr eia-nple frc7 ttie e-iidence we have that
smoking is a factor in. multiple correlations and
Is strongly
associated -.ith sc'-.,e diseases, this; may be
sufficient to
substantiate a cla-lim that.smoking Is a cause of
the disease
or caluses ah!' I.ncre~se in ;the-, Ir':cAde`mce:
of~the-dlsease. . if.
A..t 'can be re'l, i a'h p r 4!d c-t1gd
t*11!at"If'sm'6k'1'm9. Is d6treased
in a populati-on ~-so W i I'l be':the Inc
idence,-~oj. this 6e,
that
4) .
: . ,
.
-
ti
disease then src~k i c is a cause In he genera I
or
probab
sense.
But the e~ide,ce obtained f,rorn populations is
not relevant to the Individual -as far as the
Individual
is concerned. aeneral causality has no validity
and it would
be quite Improper to imply predictability. And
this applies
to the lecal question "Could sm.-king have caised
damage to
this -am?" as ell as to the q_,est!:;n "Did this
brand do
th s
to this an7"
(71
In an Individual case, to prove an agent is
a cause f2r an elfect one must fall back to
establishing
that the particular agent was fzr that Individual
in all
his special circu--istances both necessary and
sufficient.
But these sz~eclal circumstances lncl,.;~e genetic
constitution,
abnormal phenotype or genotype, social status,
exposure to
other known agents (e.g. carcinogens) together
with a host
of relevant considerations. Diseases In fact are
not
caused merely by the presence of a particular
agent. if
such a simple cause were assumed then there is a
necessity
to explain why It only occurs in a minority of
"susceptible"
Individuals.
C.mtd.
M 5 06
A
CO
r1 j
BATCo document for Province of British Columbia I
November 1999


71,
q
In Sul-ary, f-;r s:cIal policy purposes It is
sensible and totally r-lel-lt to use the
experimental
evidence pertalning to large 9~rt-jps and also to
select
the simplest ~%ypotheslls. It ,ay therefore be
concluded
that for cera.n -grcu=s of pecple smoking causes
the
Incidence of certain dise3ses ts be higher than It
would
otherwise b e. But nc valld c:,clusions may be
dra,n from
the epidemicl:,gical st-.--dies with respect to
any particular
individual. In this case, at present, there is no
way
of knowing or of calculatino t~'e Drobability
whether smoking
will cause disease In any specific Individual or
whether
by giving up smoking he will a-icid disease. Thus
I find
it completely consistent Lcth t: accept Lhat the
evidence
de.,nonstrates the need for same action on
sinoklng as a social
problem the individual Smcker
M:ndnalsouto~re'ect a message. to
that ki ca ses cancer-, etc.
Fcotnote .6
1. Mill "was concerned t .aestablish methods
employed by
sclentLs ts,. methcds of induction, which would
act as
an .1 ndd.pond~'
C-t~eck on ob-servat'ion:
a) Ift.; 6#"M~re instances of''the ~he~nomenon
under
lnve~s.tijati`on have only one.cIrcumstance in
common
.....,hisAs U~e ca.-.,se.
b) If an instance In which the phenomenon occurs
and
an instance In which It does not occur have every
c1 r:--nstance in cz7non except one, tha.t one
only
"zcjrring in the f:rmer, this one differing
clrz-imstance is the cause ... (or an
Indispensible
part of it).
C) If t~o or more instances in which the
phenomenon
ocz.;:s have c~nly cie circumstance in common,
while
two cr more Instanzes in which it does not occur
ha---e nothing I.-, cc-7~-non (save the absence of
that
circ--;nstance) the circunstance in which alone
the
two sets of instances differ is the cause.
d) Whatever phenznencn varies In any manner
whenever
anzther phencrnenon varies In scme particular way,
Is either cause or an effect of that phenomenon
or is concerned with It through some fact of
causation.
Mill was never clear whether his methods were
of discovery or of proof but they did influence
the. approach
of experimental scientists. Ne*vertheless his
Inductive
method necessarily assumes that every event must
have a cause.
SJG/NW
M 5 0 6
ON
--j
CO
rj
j-,
C7-:)
(-;,4
BATCO document for Province of BritiSh COIUMbia 1
NOvernber 1999


22,-d Cctober !--76
M 5 0 6
J,
UIALUAT!Ctl Cr CICARE77ES
,he eval,--ation cf safety of s.-~ki~g cr =f
cigarettes is inpcsed on two rnain bodies, '.he
ciga:ette
manufacturers and the public a-ithorities. It
.!,;~t te
said that such an evaluation is entirely the
respcns!!~ility
of every adult smoker and for the -anufact-.jrer
thi! is a
te~mpting proposal. But if this vied were to be
sustained
then cigarettes would have tc !;a regarded
entirely d.fferently
from most other items - agro cnemicals,
cos,,etics, f:,,cd
additives, motor cars, etc. In any case thi3
arg-.;-nent
merely transfers the on-us of decision - gu!dance
on the
evaluation would still be req-jired frori
manufacturers and
public authority. It is unlikely then that
the~cigarette
manufacturer can avoid involve:nent ~.Ith safety'
eval.:azion
and It might be helpful to lc~:k at the factors
Invc:ved.
Safety is not a measured or indeed a measurable
factor. Safety is best def' ned as. the de-,ree.to
which
atterYctanE,'.risks,-a~re ju,e-;'e-d. to bellacce
able.. ~nd'~ r i sk Is
the prob.ability and protiaple severley of harrm
which is.
measurable. 6ecause I h. is reasurable' the
'evaluation of
risk Is sUscePtible to'a 5-c ien t ~ Uc zpproach.
Judging or
evaluating safety ( the acceptati I I ty, 0! risk)
however is a
normative', political activity and is not
therefore s,--sceptible
to a scientific approach. It is to be expected
then that.
there will be wide variance in safety Judgment
between, say
government and Interested Ind---stry or between
one well
informed smoker and another. There should not be -
and
in truth there is not - as muc'- variance of the
vie.s on
risk bet-een scientists whether in Industry or
public
service.
As far as smoking is concerned there is riuch
confusion caused by scientists and particularly by
dcctors
bec-~ming emotionally involved and making
jud;-ients in the
political area %,;here they may r-ot be ccrpeteit.
C, ',a
other hand politicians and ir-d-.;strialists
venture into the
scientific area of risk where t%ey may be clearly
in=:.npetent.
The position is not helped by scme doctors belief
that they
are trained to be competent In both areas.
Whereas, in my
judg,ent, their general training quite properly
leads them
to preoccupation with Individual (i.e. unique)
decisions
and away from both social or scientific
generall3aticn.
C---Itd.
CD
CD
C71
BATCo document for Province of British Columbia I
November 1999


5 6
n
2' -
But even in the risk area the postt!cn !s
not clear cut for smoking. 'In spite of the
general
sinilarlty in the results of various
epidemiological st-idles
there are no precise measurements of the var!ous
ris-A
in-vclved In smoking. There is also considerable
sc:~e for
different interpretations of data a~d some fairly
f~r7nldable
prpble-is with self seleoted sa~nples In
epidemiological
studies. Thus although the evidence on which risks
of
s-nokinc may be estimated' Is well known, perhaps
the znly
cz-nclusicn supported by all scientists is that
the risk
is'above zero.
When it comes to safety evaluation, although it
Is true that the same sort of factors are involved
whether
Judgments are made for social or business
purposes, -.;hen
presented with a variance in risk probabilities
industrialists
are justifiably inclined (at least publicly) to
choose the
lower and those held responsible for public safety
s.,Ilarly
to choose the higher risk.. There is thus an
understandable
an d justifiable diyergenc,4 I r% the evaluation
of t-he safety Of'.
cl;arettes. : 'In a hiqhl~ ti" the
Vi'lls-ed sociq~y it is in
interests of all to aim to ge,t these, evaluations
to c:nverge.
'more research.' -to get,-cl'oser agreement
This can be done by
on risk. However, value udg:rents must still be
made and
it may be worthwhile Indicating some of the
factors hich
are involved in making such jud:ments. One factor
is
concerned with benefits - and it- must be made
clear that
benefits are value judgments. Scientific research
can
Indicate efficacies In achieving measurable
results but
it cannot indicate or meas--:re benefits.
The sort of q-_,estjorls which must be asked are
as fcll:,s:-
Is the risk voluntary c: in~_-Iuntary7
Is the effect immediate zr d-?1ayed7
Are there alternatives7
Is the risk known with ce:tainty7
Is Ure exposure essent!a!7
Is the risk encountered occuzationally7
Is the hazard common or "dread"7
Does it affect average people or sensitive people?
Will the product be-used as intended or is it
likely
to be misused?
Are the consequences reversible or irreversible7
We are thus concerned with considering, as well
as risk, social cost, cu3tz-9, reasonableness,
best available
practice, benefits etc.t and in making a decision
thesa
several or many. components --st be appraised,
compared and
11 Ccntd.
C-3
(71
_Ij
C-0 r.
NJ TCD
BATCo document for Province of BritiSh Columbia I
November 1999


A
,-J
"L..
~i
f
r
M 0 6
3
--?,;!%ed aca-~nst each other.
The difficulties in fcod safety eva..;.itlon
t:e s1.,1:ar to those for cigarette's. T-ere are
C:n31der3ble
t!f!eren:es In judgment on food and It is to be
e-pected that
'-nere will be wide differences In the jud~;~ents
cf smokers,
;--1-'tic!ans, and ind.s trial I sts.
The ciga:rette manufactjrers, in my --:!lion,
a--e entitled to say: "As long as there is
insuffIcient
::nsensus an risk (but not until there Is clinical
proof
zf the mechanism) we are not able to make useful
safety
'--~;ments" - this 15 in effect the t1ght :-.pe.
Cz-jpled
.4-th my previous argument on general (pr~-babi
list!=) and
s.=ecial (.Techanistic) causality which er..'zies
the
-~~=-.nufacturer to say, even if there ever is
consens:js on
rLsk, this can only be construed as a general
causil
:*-1a'1-'cns.-,'p applying to the incideice cf
disease in
W~~Pulati--ns and is not applicable to any unique
individual
`-.ese are very effective d4fensive P.R. and legal
positicns.
Czuole~!- f:~-rther with s6ppo~'t tbrreseafc- a
recd;lnltion
".at this cnly can establ'ish (.o rreduceY: Rick
etc. and
t'~at in the end safety, benefits, fairness '4.nd
balance are'
'I value judgme
nts our public position is logically quite
respectat~le. if at the same time we recocnise..
that those
--'th wider responsibilities to the public are
entitled t=
na.~e different value judgments with respect to
the
aczeptabil!ty of risk and the loss of berefit and
to act
cn them, I believe we have the best pabli:
position
:`tainable in the circumstances. In su7nrnary this
Is
(a) 9-1sk Is the business cf science a--~ there
r,0t consensus.
(b) Salety Is dependent on value jud:-ents and
ide differences are to be expe--led.
(c) research at all le-iels shoult help
evaluations to converge.
(d) there is evidence that s,oklnq is ass:~clated
with higher incidences of some diseases and
general causality may be claimed, special
causality can never be proved.
(e) Vie slogan Is informed responsibility 6f
adults
governments and Industry.
C::1
C7*,
-,j
CO
NJ
CO
U-
C:)
BATCO doCUrnent for Province of BritiSh COIUMbia I
Novernber 1999