Intended for healthcare professionals

Rapid response to:


Environmental tobacco smoke and tobacco related mortality in a prospective study of Californians, 1960-98

BMJ 2003; 326 doi: (Published 15 May 2003) Cite this as: BMJ 2003;326:1057

Rapid Response:

Re: Professor Repace has gone quiet

I must admit that, although I regard myself as reasonably
intelligent, I cannot be very clever because most of the statistical stuff
above is beyond me. I mean, I understand that if there are 10 horses in a
race, the probability is 9:1 that one of them will win the race. However,
it is fairly common for 'dead heats' to occur where 2 horses cross the
finishing line so closely together that no difference in position can be
seen. I would suppose that, if one found in reality that the occurrence of
'dead heats' was, say, 1 in 1000, the probability of a 9:1 ratio would
have to be reduced a little. I am not quite sure of my maths, but perhaps
8.999:1 would be correct? But one might go on to ask what is the
probability of a three way dead heat? It must happen occasionally - or
must it? Even if it has never happened in the past, it is still possible,
isn't it? In fact, it is always possible that all 10 horses could be
involved in a 'dead heat'.

As regards Passive Smoking, the statistical problem that arises in my
mind (and I very much stand to be corrected) is this. If we take the
probability of a smoker to be significantly affected by smoking to be
100%, do we see the probability of a passive smoker to be, say, 1%? Or
should that probability be 0.1%? Or even, 0.01%? Or even 0.001%?

Further, in the same way that it is always possible that 10 horses in
a race could possibly cross the finishing line at the same time, is it
right to project a minuscule possible percentage probability by virtue of
some sort of population multiplier to project an actual number of deaths
based on minuscule probabilities?

I must admit that Mr Atherton's sources bother me a little. If it is
true that the sources he quotes are funded by the tobacco industry, then
they may be suspect. But I did not see in Atherton's submission anything

In Repace's submission however, I did see a logical inconsistency.

Now, because I am afraid of what I might do to this response if I try
to go back, I will have to try to remember. Did Repace say that the the
incidence of harmful chemicals from tobacco smoke INSIDE enclosed premises
is 2 or 3 orders of magnitudes greater than the incidence OUTSIDE enclosed
premises? Words to that effect.

If we reasonably assume that, in most circumstances, the harmful
incidence (if I may reverse the logic a little) of tobacco smoke OUTSIDE
enclosed premises is unutterably minuscule for all intents and purposes
(bearing in mind the minuscule amount of carcinogenic chemicals in tobacco
smoke anyway), then '2 or 3 orders of magnitude' is still virtually zero.
It follows, therefore, that in Repace's opinion, the incidence of harmful
chemicals resulting from tobacco smoke INSIDE enclosed premises (in
normal, everyday circumstances) is zero, for all intents and purposes
(that is, as regards damage to health).

Repace is a very clever Professor. I am not.

Competing interests:
None declared

Competing interests: No competing interests

24 February 2010
James Watson
Retired Bank Officer