CIG/',RETTC AND
The public position ci! tcb,-i.,:,:o cu~;,;,;~n.'
c~;
re5p~,-:t to causal explanations of the
as~,ociation of
smckin; and diseases is dominated by legal cons
idQra tion-T.
In t"e ultimate companies wish to be able to
dispute tha: a
particular product was the cause of injury to a
particul-:7
perscn. By repudiation of a causal role for
cigarette
smok_`ng in general they hope to avoid liability
in partic-_,lar
cases. This domination by legal consideration thus
lea--s
the industry into a public rejection in total of
any causn-1
relationship between smoking and disease and puts
the
indus--ry in a peculiar position with respect to
product
safet-, discussions, safety evaluations,
collaborative
research etc. Companies are actively seeking to
make
products acceptable as safer whil' t-
e denying strenuously -
need to do so. To many the industry appears
intransigen-z
and irresponsible. The problem of causality has
been
infla-:~d to enormous proportions. The industry
has
retreated behind impossible demands for"scientific
proof"
whereas such proof has never been required as a
basis for
action in the legal and political fields. Indeed
if the
doct:,--ne were widely adopted the results would
be di.sast:-::L:s.
I believe that with a better understanding of the
nature off
Causality it is plain that while epidemiological
evidence
does indicate a cause for concern and action it
cannot fz---.i
a bas-'s on which to claim damage for injury to a
specif_:=
individual.
In the Surgeon General's Advisory Committee
First Report the question of causality is raised.
They
recocnised the difficulties but agreed that causal
signif-'-
cance of an association is a matter of judgment
and noted
a nu-_-er- of criteria which may be utilised
together, not
one cf %,ihich alone would be sufficient, e.g. the
conr1J,t.-.r-.cy,
stre:-!zth and specificity of the association, the
tempz~ral
relationship and the coherence of the association.
The-..-
considered characterisation of the assessm-lent
of' an
associat:_o.i as a "factor", a "determlnant" or a
"cause".
The,, rezognisud "factor" as a source of variation
but noz
nece,-:arily an a cause. Further there can be
co-existc.-z-_%
of s~;r~rjl fact.-)rs requir-f-7! f,)L, Lht-~
')L~:Urrcnce of a disr--s-
but --,e fazt,.:c .-y play a detcrm'n,)nt role.
The word
~~,s ~-_Id to convey tho. notion uf a
ficInifi,:ant nffc~:tual
C_
BATCo document for Province of British Columbia 4
November 1999


relat-:-:~n--hip betvieen an agcilt and an
iL;5ociaLed disorder in
the ~izst. All the members accepted a multiple
etiology
vie. of biological procesnes. In the end they
defined
"causa" mcrely to zonfirm their convictionz.
The Surgeon General's Committee had such great
diff-;zulty with the problem of causality because
the Lasts
for causality we "instinctively" adopt are those
of necessity
and sufficiency spelt out by J.S.Milf1w)hcreas
epidemiology
requires - as the Committee demonstrates in
practice -
different treatment. It is my thesis that the
tests of
Mill (which I will call tests for special
causality) apply
only to mechanistic systems whereas epidemiology,
for
example, requires a probabilistic: approach.
In the nineteenth century it was assumed that
every effect must have a cause. To prove that A
caused B
it would have to be demonstrated that event A
precedes event
B and that event A is both necessary and
sufficient to
ensure that event B will follow. The view of
science
itself was mechanistic; it was felt that if
everything
was known about a system then the future of that
system
would be predictably certain. Within its limits
this
concent is still valid. In spite of appearances
the
behaviour of billiard balls is entirely
predictable and,
given the data, is calculable. This concept still
dominates legal thinking. Although the extension
of
contributory negligence increasingly recognises
multiple
causes the concept is nevertheless mechanistic -
there
is an underlying assumption that, given the facts,
the
results are predictable, calculable and legally
apportionable.
This mechanistic concept of causality is still
also the
popular concept. In the modern world of
advertising even
the most complex systems imaginable (human
emotional
situations) are presented as predictably
controllable by
simple events such as using Brand X.
Julian Huxley in 1958 wrote "The conclusion to
be dra.n from the evidence is definite: increased
smoking
increases the probability of developing lung
cancer.
Unfortunately the significance of such a statement
is not
clear to many people. Obsessed by the naive idea
of
finding a single cause for every effect, they
shake their
heads and say that the evidence is only
staListical, as
if that invalidated it. In point of fact, every
scientific
law is statistical, and all that physi,:ists can
tell us L-j
CZD
about ~!lectrons withn thr atoin is the
probability of their
p.-i r
occupying any ticulir ponitiDn".
BATCo document for Province of British Columbia 4
November 1999


Thus- Julian Huxley be-zan to spell out a
vici of causal-;Ly Ginich I will call gerieral
causality) basod on Lwuntieth century science.
When .,c
ha~c a science whicli embraces Lhc uncertainLy
principle -
the unknoviability of the total data f )r some
systems - and
pro'--ability to describe the properties of large
populaticns
*
thcr. -.~e ca
not expect special causality to be very useful.
However, just as Newton'sLaws are true enough for
everyday
ex perience and relativity would be su.perfluous
the tests
for special causality can still b e applied to
very simple
unified systems. But for complex systems involving
the
properties of large numbers experimentally
established
relationships may be used to predict the behaviour
of the
population with accuracy.
For example from the evidence we have that
smoking is a factor in multiple correlations and
is strongly
asscciated with some diseases, this may be
sufficient to
substantiate a claim that smoking is a cause of
the disease
or causes an increase in the incidence of the
disease. if
it can be reliably predicted that if smoking is
decreased
in a population so will be the incidence of this
or that
disease then smoking is a cause in the general or
probabilistic
sense.
But the evidence obtained from populations is
not relevant to the individual - as far as the
individual
is concerned general causality has no validity and
it would
be quite improper to imply predictability. And
this applies
to the legal question "Could smoking have caused
damage to
this man?" as well at to the question "Did this
brand do
this to this man?"
In an individual case, to prove an agent is
a cause for an effect one must fall back to
establishing
that the particular agent was for that individual
in all
his special circumstances both necessary and
sufficient.
But these special circumstances include genetic
constitution,
abn--rmal 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 s-,nple cause were assumed then there is a
necessity
to explain why it only occurs in a minority of
"susceptible"
in--:-viduals.
Contd-
BATCo document for Province of British Columbia 4
November 1999


In Su.,mary, for social P7;licy purp)scs it i~~
sensible and totally relevant to use the
expurim--ntal
evidence pertaining to large groups and also to
nrlect
the si-,plest hypothesi s. It may th,-ref-c be
c,nclud-i
that for certain groups of people smoking causcs
the
incidence of certain diseases to be higher than it
%~auld
otherwise be. But no valid conclusions may be
Jra--i fr~3ra
the epidemiological studies with respect t3 any
particular
individual. In this case, at present, there is no
way
of knowing or of calculatino the orobabilitv
whether smoking
will cause disease in any specific individual or
whether
by giving up smoking he will avoid disease. Thus I
find
it completely consistent both to accept that the
evidence
demonstrates the need for some action on smoking
as a social
problem and also to reject a message to the
individual smoker
that "smoking causes lung cancer", etc.
-
z
Footnote
1. Mill was concerned to establish methods
employed by
scientists, methods of induction, which would act
as
an independent check on mere observation:
a) If two or more instances of the phenomenon
under
investigation have only one circumstance in common
... this is the cause.
b) If an instance in which the phenomenon occurs
and
an instance in which it does not occur have every
circumstance in common except one, that one only
o::curring in the former, this one differing
circumstance is the cause ... (or an
indispens2Lble
part of it)..
c) If two or more instances in which the
phenomenon
occurs have only one circumstance in common, while
two or more instances in which it does not occur
have nothing in common (save the absence of that
circumstance) the circumstance in which alone the
two sets of instances differ is the cause.
d) Whatever phenomenon varies in any manner
whenever
another phenomenon varies in same particular way,
is either cause or an effect of that phenomenon
or is concerned with it through so-ne fact of
causation.
Mill was never clear wheUier his method:; were
of discovery or of proof but they did influence
the approz;,.-h
of experimental scientists. Nevcrthf~l,-ss his
indu,-tivu-
C)
methcc nc,:cssarily assumes that cv-ary vvpnt
mur;!- linvo a caii,;-
Xz-
SJG/N`-'
BATCo document for Province of British Columbia 4
November 1999