TITLE
StatiStical Analysis of Mortality Rates of
Cigarette, Pipe,
arid Cigar Smokers
AUTHORS
J. A. E. Bell NIASc. , PhD- , Senior Research Eng.
International Nickel Company of Canada, Lin-dted,
J. Roy
Gordon Res~arch Laboratory, Clarkson, Ontario.
D. F. Laing Senior Technologist, University of
Tororto, Department of Metallurgy and Material
Science,
Toronto, Ontario.
ACKNOWLEDGMENTS
Non e
CZ)
-Ij
BATCo document for Province of BritiSh Columbia 5
November 1999


J. A. E. Bell, PhD D. H. Laing
ABSTRACT
Statistical analysis of recent data on the
mortality rates
of pipe, cigar and cigarette smokers showed that
the mortality ratio
of smokers could be best described in tern-is of
the dose rate (volunae
of snioke per day), length of exposure and
toxicity of the smoke. When
the toxicity of cigarette smoke was defined as
one, the toxicity of pipe
smoke was found to be 0. 07 11 and of cigar smoke
0. 28 13. The
hypothesis that the toxicity of the smoke is
controlled by the combus-
tion conditions during pyrolysis of the tobacco is
shown, for iha first
time, to be the only reasonable explanation for
this difference in
toxic ity.
CD
4 NJ
BATCo document for Province of BritiSh Columbia 5
November 1999


- I -
Statistical Analysis of Mortality Rates of
Cigarette, Pipe, and Cigar
Smokers
Introductioa
A proper correlation between the mortality rates
of cigarette,
pipe, and cigar smokers as a function of the three
exposure variaLlcs,
amount smoked, depth of inhalation, and age at
%hich srnoking started,
has not been carried o-ut. The purpose of this
investigation is to develop
the statistical "best fit"equations relating the
mortality ratio and the
exposure variables. Such an analysis permits a
comparison between
the toxicity of smoke from cigars, pipes, and
cigarettes. In addition,
the possibility of explaining the mortality ratios
in terms of a more con-
ventional dose rate and length of exposure concept
is investigated. The
development of these equations allows, for the
first tirne, evaluation of
hypotheses concerning the reason for the lower
mortality ratios of pipe
and cigar snackers compared to cigarette smokers.
Calculations
For a proper investigation the mortality rate
should be tab-
ulated as a function of two of the exposure
variables at every level of
the third variable. For instance, the mortality
rate of cigarette
smokers as a function of the number of cigarettes
smoked per day and
the depth of inhalation should be shown for every
age of started smok-
ing for a given age group i. e. 40 to 69 years.
BATCO document for Province of BritiSh COIUMbia 5
November 1999


- 2 -
However, the mortality rates (1), (2), (3) and (4)
have not been
so tabulated. Table 3 reference (3) shows the
mortality of cigarette
smokers as a function of the depth of inhalation
and the amount smoked at
the average age of starting to smoke (19 years).
From this table the best
fit equation of the mortality ratio (easily
calculated from the mortality
rate) can be found by multinomial regression.
To convert the depth of inhalation which is
expressed by (1), (2),
(3) and (4) as none, slight, moderate, or deep, to
a more mathematical
form; the volume of air inhaled by smoking through
a cigarette filter was
measured when ten male subjects were requested to
give the above depths
of inhalation ten times. The rounded off ratios
(compared to the average)
were; none - 0. 54, slight - 0. 80, moderate - 1.
00 (average smoker) and
deep - 1. 25. It should be noted that it is
physically impossible to inhale
more deeply without taking a larger volume of
smoke into the body. While
it can be argued that these values do not
represent the population, they
are almost certainly more representative than the
subjective classification
by smokers into one of four depth of inhalation
groups. In addition the re-
lationships derived below are valid for any
reasonable assumption of the
relative values. The standardized value used in
all smoking machines is
35 cc which corresponds to the average cigarette
smoker (2). The puff
volume for any inhalation is calculated by
multiplying the above ratio times
35 cc.
41-
BATCo document for Province of British Columbia 5
November 1999


- 3 -
Thus the best fit multinomial equation relating
the
mortality ratio* (MR), to the number of cigarettes
smoked per
day (NC), and the depth of inhalation (DR) at the
average age of
starting to smoke for men aged 40 to 69 was found
to be:
MR = 0. 983 + 0. 028(NC) + 0. 0155(DR) - 0.
00031(NC)2 + 0. 11
where 0. 11 is the standard error. Each of the
above terms is
significant at the 99% confidence level. The
multiple correlation
coefficient of this equation is 0. 97 (a perfect
correlation is l. 0)
and the F value for analysis of variance is 86.
While equation (1) is the best fit equation for
this data
it does not have a physical interpretation. A more
meaniful method
of analyzing the results would be to postulate
that the mortality
rate was proportional to the "dose rate" of the
smoke. Thus for
cigarette smokers the daily dose rate would
correspond to the number
of cigarettes per day multiplied by the number of
puffs per cigarette,
multiplied by the volume inhaled per puff or; in
other words, the
volume of smoke inhaled per day. Analysis of the
data in Table 3
reference (3) for male smokers aged 40 to 69
indicates the best fit
equation is of the form:
MR = 1.01 + 0.95(DoseRate x 10-4113 + 0.13
whereDoseRate = (NC) x 12 x (DR) ccSmoke/Day. The
multiple correlation coefficient for this
relationship is 0. 96, the
mortality ratio is the ratio of death rates of
smokers to non-
smokers in the same age group.
(1)
CD
BATCo document for Province of BritiSh Columbia 5
November 1999


dependence with (dose rate) 1/3 is significant at
a con-fidence level of over
99. 99%, and F value for analysis of variance is
105. Three extra data
points for all the cigarette smoking relationships
at MR = 1. 0, DR = 0,
NC = 0, were added as it is the best known point
and represents the zero
value for each variable. Thus expressing the
mortality ratio as a func-
tion of the dose rate gives as good a fit to the
data as the best Ht multin-
ornial equation.
As mentioned a proper multinornial regression on
all three ex-
posure variables is not possible because the
information is not tabulated
in a convenient form. However, the mortality rates
are expressed as a
function of the three variables when the other two
are averaged (1), (2)
and (3). A multinomial regression on this grouped
data is thus 'possible.
If a comparison between smokers of cigars, pipes,
and cigare"--es is to
be made, it is necessary to consider the mortality
of men over 4:5) years
of age in order to have enough observations in
each grouped data point.
However, the mortality rates have not been
summarized in this age group
in any of the reference works. The mortality rates
were thus calculated
from tables presented in reference (2) in the age
group 45 - 65 as thlis
group contains a higher number of excess deaths
from smoking than an
older grou?.
The mortality rates were normalized on the
standart-2 population C=)
C=)
_Z:t-
distribution (Table 1, Appendix of reference (2))
as was done ir. that work.
The avarage values for amount smoked, age started
smoking a:-.d depth of
inhalation were averagcd from the appropriate
tables i*n (2) -ovc:, the 45 to ON
BATCo document for Province of British Columbia 5
November 1999


5
65 age group. The mortality ratio as a function of
the amount smo'z-,ed, depth
of inhalaticn, and age started smoking for
cigarette, pipe and cigar smokers
are shown in Table I for the age group 45 to 65.
The best fit equations were found to be:
Male Ci!Zarettc Smokers Only Aced (45 - 65)
2 2 +
MR = 3. 18 0 156 (NC) 134 (AS) + . 000 17 (AS) +
0003 3 (DR) 0 87 (3)
multiple correlation cooff. = . 99 F = 126. 5
where AS = age started smoking.
Male Cigare.--te + Cigar + Pipe Synokerv- Aged (45
65)
MR = 2.05 - .0101(NC) 113(AS) + .027(DR) +
.00023(NC) 2 + .00171(AS) z
096 (4)
multiple correlation coeff. = .98 F = 60. 5
The corresponding best fit dose rate equations
are:
Male Cigarette Smokers Only Aged (45 - 65)
MR = .93 - .031(NC x 12 x DR x 10- 4)1/3 , (55 -
AS) + . 13
where NC x 12 x DR = dose rate = volume smoke/day
multiple correlation coeff. = .97 F = 207.
Male Cigarette Plus Pipe and Cigar Smokers Aged
(45 - 65)
-4 1/3 +
MR = .92 + .026 ((NC) x 12 x (DR) x 10 x (55-AS) -
.19
multiple correlation coeff. = .91 F 66.
Far pipc and cigar smokers there are not enough
data
points to evaLuate the best fit multinomial
equations so only the
mortality ra:io as a function of the dose rate
equations can be cal-
classifi-:d by nun-iber of cigarettes smoked per
&.y
(5)
(6)
BATCo document for Province of British Columbia 5
November 1999


- 6 -
culated. No data is available for the age started
smoking for pipe
and cigar smokers.
Male Pipe Smokers Only Aged (45 - 65)
MR = 1. OZ + 0. 16
multiple correlation coeff. 08 F = .01.
Cigar Smokers Only Aged (45 65)
MR = .82 + .208 (NCIG x 60-xDR x 10- 3)1/3 + .13
where NCIG = number of cigars per day
multiple correlation coeff. = 0. 7 F = 1. 82.
The average number of puffs per cigar is 60 and
for pipes 90.
Discussion
An excellent correlation (correlation coefficient
99)
relating the mortality ratio of cigarette smokers
and the exposure
variables of amount smoked, depth or volume of
inhalation, and age
started smoking, has been found by multinomial
regression. The
mortality ratio is linearly related to the dose
rate (volume of smoke
per day) to the one third power times the average
length of exposure
to the smoke.
If the mortality ratio of men who smoke cigars and
pipes
in addition to cigarettes is compared to that of
men who smoke only
cigarettes by utilization of the respective best
fit equations (3) and
(4), an important conclusion can be drawn. The
best value of the
mortality ratio for male cigarette smokers aged 45
to 65 at 24 cigarettes
(7)
(8)
U,I
BATCo document for Province of BritiSh Columbia 5
November 1999


- 7 -
per da,., an inhalation volume of 3 5 cc, and an
age of started smoking
of 19, is 2. 00 087, while the corresponding value
for cigarette
plus cigar and pipe smokers is 1. 80 096. Thus one
of the two
followi=g conclusions can be made depending
whether the smokers
smoke--: cigars and pipes concurrently with
cigarettes.or at some
time switched mode of smoking:
(al Cicrar and pipe smoking in addition to
cigarette smoking
produces a significant lowering of the mortality
ratio of the smoker;
or:
(b) Cigar and pipe smoking did not induce
mortality as read-
ily as cigarette smoking over the period when
those devices were
smoked.
The most important conclusion is derived from the
analysis
of pipe smoking; namely, pipe smoking is not
dangerous to health and
there is no correlation between the dose rate or
exposure and the mor-
tality ratio. Cigar smoke is only slightly toxic
and the relationship between
the mortality ratio and the dose rate is not
highly significant.
The lower mortality ratio of pipe and cigar
smokers compared
to cigarette smokers is often linked to a lower
volume of inhalation. For
example; 70% of pipe smokers (45 to 65 years)
inhale only approximately
19 cc ?er puff, 80% of cigar smokers inhale only
19 cc per puff, while the
average cigarette smoker inhales approximately 35
cc per puff (see
(D
_Ct::.
NJ
BATCo document for Province of British Columbia 5
November 1999


- 8 -
Table 1. W-Hen converted in terms of dose rates
the average daily dose
of these three modes of smokin,, are: cigarette -
12, 000 cc/day, pipe -
15, 000 cc/day, cigar - 4. 000 cc/day. Calculation
of the mortality ratio
from equations (5, 7, 8) at equivalent dose rates
clearly indicate, at a
confidence level of better than 99. 9076, that the
lower mortality ratio of
pipe and cig=-r smokers cannot primarily be caused
by a smaller depthof
inhalation or dose rate. No data is available for
the age started smoking
of pipe and :;_'gar smokers so the effect of this
variable cannot be check-
ed. However, even if the average age of starting
to smoke cigarettes
was 24 years instead of 19 years the same
conclusion would result.
A more fundamental way of expressing the data is
to calculate
the relative toxicity of pipe and cigar smoke
compared to cigarette smoke.
If the toxici--y of the smoke is defined by the
equation:
MORTALIT--i- RATIO (MR) = a + b (TOXICITY) (DOSE
RATE) 1/3
x (LENGTH OF EXPOSURE)
then the rels-tive toxicity of pipe and cigar
smoke can be easily calculated
from a com-jarison of the respective mortality
ratio versus dose rate
equation (5,7, 8).
Let the toxicity of cigarette smoke be 1. 0; then
the mortality
ratio for all male smokers aged 45 to 65 is
expressed by the equation: C":)
BATCo document for Province of British Columbia 5
November 1999


- 9 -
Bell
4 1/3
MR =0. 93 +0-031 (TOXICITY) (CC SMOKE/DAY x 10- x
(5 5 AS) (9)
where AS is the age started smoking
Comparison of eqn. (9) with eqn. (7) and (8)
yields
the following toxicity values.
DEVICE SMOKED TOXICITY OF SMOKE
CIGARETTE 1.0 + .07 (Defined)
CIGAR .28 + .13
PIPE .07 + .11
Placing the appropriate toxicity value in equation
(9) allows the
calculation of the mortality ratio of a male
smoker ages 45 to 65
for any smoking device.
Conclusions
The mortality ratio of cigarette smokers is highly
correlated (multiple correlation coefficient =
.97) to the dose
rate of the smoke (volume of smoke inhaled per
day) and the
length of exposure to the dose. The increase in
mortality ratio
with dose rate is large for cigarette smokers and
represents a
doubling of the death rate of male smokers from 45
to 65 at
average dose rates.
The mortality ratio of pipe smokers is independent
of the dose rate of the smoker.
NJ
C)o
4
BATCo document for Province of BritiSh Columbia 5
November 1999


- 10 -
The mortality ratio of cigar smokers is correlated
to the dose rate of the smoke but the increase in
mortality with
dose rate is not large.
The mortality ratio of cigarette smokers who also
smoke pipes and cigars is significantly lower than
smokers who
smoke only cigarettes at the same cigarette smoke
dose rate
and length of exposure.
The lower mortality ratio of pipe and cigar
si-nokers
compared to cigarette smokers cannot be explained
by a differ-
ence in amount smoked, depth of inhalation, or age
started
smoking.
If the relative toxicity of smoke from a cigarette
is
+
defined as 1. 0 then the toxicity of pipe smoke is
0. 07 - . 11 and
cigar smoke .28 't .13. No factors mentioned in
current re-
ports on smoking and health (1), (2), (3) and (4)
can satisfactorily
account for the lower toxicity of pipe and cigar
smoke. The only
hypothesis which might satisfactorily explain the
smoke toxicity
differences is a difference in smoke composition
produced by
different combustion conditions in the pipe, cigar
and cigarette.
The combustion of tobacco is a complex process CD
CD1
involving both destructive distillation and
oxidation reactions. 4-1.
Two major phases are present in the smoke - a
gaseous phase
CC)
BATCo document for Province of BritiSh Columbia 5
November 1999


and a liquid particulate phase. From
thermodynamiWd kinetic
considerations it is well known that the time -
temperature re-
lationships which exist during the combustion are
the important
variables governing the composition of both the
particulate and
gaseous phases. Tests were undertaken to see if
differences in
the time - temperature relationship in the
combustion of tobacco
in a cigarette, pipe, and cigar did occur and if
the smoke from
these devices was different.
Results of preliminary tests indicate for instance
that
the velocities of the burning interfaces are for
cigarettes, cigars
and pipes 1. 9, 1. 1, and 0. 25 in/min
respectively, and the tar
content on a relative per puff basis is 100, 45,
and 8 respectively.
These results indicate indisputably that there are
differences in
both the time - temperature relationship during
combustion and in
the products of the combustion. A detailed
description of a more
complete investigation will be the basis of a
future communication.
The hypothesis that the differences in the
combustion
conditions of the pipe, cigar and cigarette are
responsible for
their different effect on mortality, has not
received the widespresd
attention it would appear to merit and this fact
is somewhat disturbing.
CO
4 QI_4
BATCo document for Province of British Columbia 5
November 1999


TAB LE I
Mortality Ratios For American MaIcs A--~s 45 to 65
As A Function Of Amount
Smoked, Depth of inhalation, Age Started Smoking
And Smoking Device
Mode of Smoking Mortality Average Nurnber Of Draw
Age Of Starting
Ratio Cigarettes Per Dav Volume To Smoke
Cigarettes Only 1. 64 5 35 19
2.04 15 35 19
2. 18 30 35 19
2.44 50 35 19
1.74 28 19 19
2.05 28 27 19
2. 14 28 35 19
2.36 28 44 19
1.35 28 35 33
1.77 28 35 27
1.86 28 35 22
2.24 28 35 17
2. 56 28 35 13
Cigarettes Plus 1. 50 5 34 19
Other 1. 65 15 34 19
1.86 30 34 19
2.54 50 34 19
1.43 24 19 19
1.72 24 28 19
1.81 24 35 19
2. 10 24 44 19
1.66 24 34 33
1.37 24 34 27
1.62 24 34 22
1.84 24 34 17
2.24 24 34 13
Number of Cigars Or
Pipe Fulls Per Day
Cigar 1.09 2 20 20
1.03 7 20 20
1.02 3 19 20
1.28 3 28 20
Pipe 1.08 5 21. 5 20
.92 13 21. 5 20
.98 8 19 20
1. 21 8 'Z8 20
cc
BATCo document for Province of British Columbia 5
November 1999


- 12 - Bell
References
Advisory Committe to the Surgeon General on
Smoking and Health, U.S.A. Public Health
Service Publication #1103.
(2) E. C. Hammond Smoking in Relation to the
tp
Death Rates of One Million Men and Women,
National Cancer Institute Monograph No. 19
P. 127-204, 1966.
(3) E. C. Hammond Smoking in Relation to Mortality
and Morbidity, Journal of the National Cancer
Institute Vol. 3 2, No. 5, May 19 64.
(4) E. C. Hammond and L. Garfinkel The Influence
of Health on Smoking Habits, National Cancer
Institute Monograph, No. 19, P. 269-285, 1966.
C=)
rQ
C10
BATCo document for Province of British Columbia 5
November 1999