Document r6DdgZQbp8ZGZNwkkNLyxEvMV
Sources and Interpretations of Asbestos Exposure Data by Michael Gough. Office of Technology Assessment, United States Congress
The attached paper was presented at a workshop "Exposure Assessment: Problems and Prospects" Sponsored by the Task Force on Environmental Cancer and Heart and Lung Disease National Institutes of Health September 14, 1982
UCC 015221
A('0151
2
understanding of the relationship between asbestos and cancer. Individual case studies and reviews of case series began to appear in the literature in the 1930's. However, according to Selikoff (1981) the establishment of an association in populations between occupational exposure to asbestos and lung cancer depended on the classic study of Doll (1955).
From the standpoint of cancer, there was no reason to measure exposure levels to asbestos until an association between the fibers and disease was shown. Not unexpectedly, there are few references in the literature to measured exposure levels during the 1930*8, 40's, and 50's. Of course, given the long latent period for cancer, those are the exposures that are associated with today's cancer cases.
Despite the fact that asbestos was positively identified as a cause of lung cancer in the 1950's and that exposure to it was known to be widespread, no published estimate of its Impact on nationwide mortality was were until 1978. In that year, two estimates were made. Selikoff (according to a footnote in the paper that is next mentioned) estimated that the annual number of asbestos-related cancer deaths was about 50,000. His estimate elicited little public attention.
The other 1976 estimate, which was titled ''Estimates of the Fraction of
Cancer in the United States Related to Occupational Factors," was prepared by
the National Cancer Institute (NCI), the National Institute for Environmental
Health Sciences (NIEHS), and the National Institute for Occupational Safety
and Health (NIOSH). Ten employees of those institutes were listed as
contributors to the estimates paper. Because no one was identified as the
author of the paper, it is difficult to make reference to it in the usual
shorthand method used for scientific publications, and it is now often called
UCC 015222
A 00152
4
exposed to asbestos since the beginning of World War II. Of those, 4 million were classified as "heavily exposed," and an additional 1.5 to 3.5 were classified as "significantly exposed." No details or documentation was provided for these estimates.
Thirty-five to 44 percent of the heavily exposed population were projected to die from asbestos-related cancers. This proportion of asbestoscaused cancer death was reported from a study of asbestos insulation workers by Sellkoff, Hammond, and Seldman (1979; 1980). Assigning that proportion of deaths to asbestos resulted in an estimate of 1,600,000 abestos-caused cancer deaths among the 4,000,000 heavily exposed workers (Table 1).
The significantly exposed population was projected to suffer one-quarter the asbestos-related mortality estimated for the heavily exposed workers. The estimates paper associated 400,000 to 700,000 deaths in that population with asbestos.
Thirteen to 18 percent of all cancer mortality being associated with a single substance is a frighteningly large percentage, but, at the same time, it holds great promise. It suggests that reducing asbestos exposure would significantly impact on overall cancer mortality rateB.
From the standpoint of this conference, the estimates paper is especially interesting because it did not include a single estimate of the amount of asbestos that workers had been exposed to. It did depend, as does nearly every estimate that I will mention, on Sellkoff and his colleagues' measurements of risks and mortality in the heavily-exposed insulation workers.
The estimates paper has beeu faulted because it Included all exposed UCC 015223
6
striking difference between the number of workers classified as heavily exposed.
Hogan and Hoel cite a paper by Cochrane et al (1980) that reports that 27 of 92 senior staff members of a large South African asbestos/cement company had been heavily exposed to asbestos. They suggest that this 18.5 percent heavily exposed subset might be representative of the whole asbestos Industry. I think that most of us would agree that extrapolating from exposure levels in a single company to estimates of exposure in a population of 7 to 8 million workers is not something to be done lightly. Hogan and Hoel apparently shared that view, and they also made a projection from data of Enterline (1978) that 3.4 percent was a best estimate for the proportion of all asbestos workers that are heavily exposed. The range around Enterline's best estimate was very wide, from 0.5 percent to something less than 25 percent, or about 50-fold. Combining estimates of 18.5 and 3.4 percent, in a way that is none too clear to me, Hogan and Hoel settled on a best estimate that 15 percent of the 7 to 8 million workers were heavily exposed. The remaining "less-heavily" exposed workers were assumed to have been exposed at one-eighth to one-quarter the rate of the heavy exposure group.
The number of workers estimated to be heavily exposed by Hogan and Hoel is between 1.06 and 1.23 million as compared to the 4.0 million from the estimates paper. This reduction by a factor of almost 4 has, of course, a pronounced effect on the the estimated number of asbestos-related cancers. Further reducing the estimate is the elimination of the 50 percent of the exposed workers who were already dead from the projections of current and future death rates. Hogan and Hoel's best estimate is that 3.1 percent of current cancer deaths are caused by asbestos with a range from 1.4 to 4.4
UCC 015224
A 00154
8
al'e data
0.63 X (7.1 - 8.2 X 106) - number of exposed workers " 6.5 - 5.2 X 106
(4.5 - 5.2 X 10*) X (1.34 - 1.0) (0.2) - 310,000 - 350,000
excess
cancer deaths.
The calculation from Kolonel et al's data agreed even better with Hogan and
Hoel's estimate when the relative risk of 1.34 was corrected for the relative
risk observed in the nonexposed shipyard workers, rr 0.88. When the
calculation waB done on that basis, there were 470,000 to 540,000 estimated
excess cancer deaths. The range of thi9 calculation neatly brackets Hogan and
Hoel's estimate of 497 thousand excess cancer deaths.
Hogan and Hoel used many of the same techniques as the estimates paper. However their best estimate for the annual asbestos-caused death rate was between 0.18 and 0.24 of the number from the estimates paper.
The Hogan and Hoel paper and the estimates paper demonstrates the absence of readily available data for estimating exposures. As is apparent from comparing the two studies, estimates of excess deaths depend on projections of the percentage of workers who were heavily exposed. There are few data available for identifying that population.
The Hogan and Hoel paper was circulated as a preprint and accepted for publication during 1980, but it was not published until a year later. During that year, a number of other papers dealing with estimates of asbestos-related cancer were presented at a Banbury House Conference at the Cold Spring Harbor laboratory on Long Island (Peto and Schnelderman, 1981). Those papers will now be discussed.
Enterline (1981)
UCC 015225
10
Enterline's best estimate, based on three sources of data. Is that 80,000 heavily-exposed workers were alive In 1981. That number, 80,000, Is very much lower than the corresponding number In the estimates paper (4,000,000) and In Hogan and Hoel (about 500,000 assummlng that half of the heavily exposed workers In their study were alive In 1981). Enterline borrows an estimate from Sellkoff's work that heavily-exposed workers were exposed to 8-12 fiber per cubic centimeter (f/cc), as Is shown on Table 1.
Enterline cites three sources of data for lung cancer death rates among workers heavily exposed to asbestos. Selikoff et al (1980) found that 22 percent of insulation workers died from lung cancer. Measurements of asbestos-caused lung cancer death rates are lower among workers in asbestos factories. Fourteen percent of retired workers from a Johns Manvllle plant died from asbestos-caused lung cancer (Henderson and Enterline, 1979); 15 percent in a British plant (Feto, et al, 1977). Enterline takes 20 percent as a reasonable average for death rates from asbestos-caused lung cancer and calculates that there are now 530 annual deaths from that cause among primary asbestos products workers and insulators and retirees from those trades.
Enterline applied Kolonel et al's estimate that about 60 percent of shipyard workers were exposed to asbestos and thereby reduces Hogan and Hoel's estimate of 2.5 million living persons having been exposed in World War II shipyards to 1.5 million. He also argues that exposure levels were lower in the shipyards than among the most heavily exposed workers. J. Thorton, who is not otherwise identified in Enterline's paper, la credited with making an educated guess that World War II shipyard workers were exposed at a rate about one quarter of that observed in primary asbestos workers. As is shown on the table, working through those numbers produces an annual number of 900
UCC 015226
12
mesothelioma and lung cancer, between the Insulators and other workers, (2) measured concentrations of asbestos, and (3) the prevalence of X-ray abnormalities associated with asbestos. In some cases, none of these measures was available, and then relative rl6k estimates were based on the number of mesotheliomas Identified in a nationwide survey (McDonald and McDonald, 1980).
As is shown on Table 1, the projected percentages of deaths caused by asbestos were all related to the risks found for asbestos workers Using those risk estimates and estimates of the number of people exposed (I could not easily derive those numbers from data presented in the paper), Nicholson et al calculate that the current annual number of deaths from asbestos-caused cancer is about 8,500. This number fits neatly between the 4,000 estimated by Enterline and the 12,000 by Hogan and Hoel.
It is a relatively simple matter to go through the categories of occupations in those four papers and to see where the estimates vary. The variations depend on different estimates of the numbers of workers exposed and on different estimates of exposure levels. The most remarkable thing about all the estimates may be that each depends heavily on Selikoff's estimates of asbestos-caused mortality in Insulators. Almost all the other estimates are fractions of that estimate.
McDonald and McDonald (1981)
McDonald and McDonald have been leaders in collecting and analyzing data
UCC 015227
14
asbestos-caused female cancer deaths Is lower by an order of magnitude.
The McDonald and McDonald estimate Is dependent on the percentage of mesotheliomas caused by asbestos and the ratios between asbestos-caused mesotheliomas and other cancers. They argue that the fraction of mesotheliomas caused by asbestos cannot be less than 0.5 and. of course, no greater than 1.0. Also the ratio between other cancers and mesotheliomas must lie between 1.0 and 5.0. These reasonably tight ranges produce a lower limit for their calculation of 0.5 percent of all male cancer deaths (about 1,000 annually), and an upper limit of 2.9 percent (about 6,000 annually).
J. Peto, Henderson, and Pike (1981)
This paper Is a critical look at the time of appearance of mesothelioma following exposure to asbestos. The authors report that combined pleural and peritoneal mesotheliomas increase as a function of the time after first exposure raised to the 3.5 power. The appearance of mesothelioma in an exposed population Is independent of age at first exposure, length of exposure, and type of asbestos fiber.
The paper presents estimates of the number of mesotheliomas to be expected in males from exposure In World War II shipyards (5,000), from other exposures during World War II (3,000), and from all exposures prior to 1965 (37,500). As Is shown on Table 2, multiplying those figures by 3 and adding 37,500 mesotheliomas results in an estimate of about 150,000 cancers from all pre-1965 exposures.
The current annual number of asbestos-caused mesothelioma deaths among males is about 930. Assuming that the death rate from other asbestos-caused
UCC 015228
16
cancer deaths annually because It is known to cause a few hundred mesothelioma deaths. They provide no other information about their calculations.
The agreement between the McDonald and McDonald estimate based on mesothelioma mortality and Enterline's estimate did not go unnoticed at the Banbury Bouse Conference. I am of two minds about the agreement between the estimates. The fact that both studies project that asbestos is responsible for about 1 percent of cancer deaths can be taken as evidence that that number is, in fact, a close approximation of the true magnitude. However, I don't think more than a moment's reflection is necessary to generate a note of caution. There are a large number of assumptions in all the estimates and the congruent numbers may be partially a matter of chance. The fact that all four estimates based on mesotheliomas -- those of McDonald and McDonald, J. Feto et al, Hlgginson et al, and Doll and R. Peto -- are similar reflects that the same sources of data and methods of analysis were used.
Whatever reservations exist about the accuracy of the estimate that 1 percent of cancer mortality is due to asbestos, the participants at the Banbury House conference agreed that the lower estimates were more likely correct than the much higher ones reported in the estimates paper. The lower number means that we can expect a barely discernible impact on total cancer mortality as asbestos exposures decrease. Nevertheless, even 1 percent of cancer mortality represents 4,000 deaths annually. Elimination or reduction of those deathB should accompany decreasing exposures.
Exposure Assessment
The estimates of asbestos-related cancer are a clear example UCC 015229
M0159
References
Cochrane, J. C., I. Webster and A. Solomon. 1980. Prolonged variable exposure to asbestos fiber. Current Cancer Research Occupational and Environmental Carcinogenesis. Washington, D.C.: Department of Health Education and Welfare.
and on
Doll, R. 1955. Mortality from lung cancer in asbestos workers. Br. J. Ind. Med. 12: 61.
Doll, R. and R. Peto. 1981. The causes of cancer: Quantitative estimates of avoidable risks of cancer in the United States today. J. Natl. Ca. Inst. 66:1191-1308.
Enterline, P. E. 1978. Memorandum. Comments on "Estimates of the Fraction of Cancer in the United States Related to Occupational Factors." Washington, D.C.:Asbestos Information Association.
1981. Cancer caused by asbestos. in Peto and Schneiderman. pp 19-33.
Estimates of the fraction of cancer in the United States related to occupational factors. Prepared by the National Cancer Institute, the National Institute of Environmental Health Sciences, and the National Institute for Occupational Safety and Health. September 15, 1978.
Henderson, V. and P. E. Enterline. 1979. Asbestos exposure; Factors associated with excess cancer and respiratory disease mortality. in I. J. Selikoff and E. C. Hammond (ed) Health Hazards of Asbestos Exposure. Ann. N. Y. Acad, of Sci. 330:117 .
Higglnson, J., J. C. Bahar, J. Clemmesen, H. Demopoulos, L. Garflnkel, T. Hlrayama, and D. Schottenfeld. 1980. Proportion of cancers due to occupations. Prev. Med. _9_: 180.
Hogan, M. D. and D. G. Hoel. 1981. Estimated cancer risk associated with occupational asbestos exposure. Risk Analysis _1_: 67-7 6.
Kolonel, L. N., T. E. Harris. 1980. workers in Hawaii: 743.
Hlrohata, B. V. Chappell, F. V. Viola and D. Cancer mortality in a cohort of naval shipyard
Early findings. J. Natl. Ca. Inst. 64:739-
McDonald, J. C. and A. D. McDonald. 1980. Malignant mesothelioma in North America. Cancer 46:1650
__________ 1981. Mesothelioma as an index of asbestos impact. in
Peto and Schneiderman. pp 73-82 .
A
UCC 015230
Table 1
Estim ates o f Cancer M o rta lity from Exposure to Asbestos Based on E stim a te s o f Numbers o f Workers Exposed and E s tim a te s o f R isks
Cc30 oPVx:
< Cu
P O0 g9
*4u) 'O">
*-9eS
X
02u 5
po 0ooo0ooo* v CmM
<3 vp
"pC0o000"Pl
O .zP03eo
J= a
ooo 9k
ov*
9*9
M 03 0
3
Q
O 4J
P0 49=
U0->
P00
*o
<
OP Q o* uoh
o
*3 >3
O
u m
fA
vOw 4P03B9
bKil ^P0P0
00OC3M
OO o
ooo
3
p0 0p 0 0 3 -M
B
p
0OO. 0>0
0 >< >J
60 W
X > Xc
oeH 0O. *a4. p3O u3
p0
4J b
3
S
*4
a 0u
< *)
mI
po po
0ooCoM0oo<on> -h CM
ooo a
oino
ooo rin-
po P0
gg
ooo
ooo
9k 9k
-oo
ofM
ooO*
oPo"> 9k
N CM
O dJ IT) CA
og * oat gg m ia
H *C tt W (O
0B Uod Se 4J 4J -o oe
*S
e o S
o
00
t0o) oaK41
JU8 S
Id B H 4o->
4J
H4(B0J
*cr
<3
^o
3.
12,200
497,000
8CM* 00
S3 ** uo i<aa
-3* 00
0 O 00
*3
S3 00
o
O 4J
Oo o oo O O
4J 4J
4J
p u 4-1
o lA <r lA m mt m tA <o CA <A oo CA fA
po ooom0ooo. oo v CO
Ooopo0oOat Kcon* s>r*
0at 0oin-4
t
a>>C>
tX s >
<4002
tiV-t
s:
0oo* 0at rC~M 1CfAtt
1 tX>
:00>C
X
E0 -P .3 <aC &>
MO*-t OX
(w0
UCC 015231
(1.4 t
(estim ated to ta l number o f workers exposed 7 to 8 m illio n ; about h a lf a liv e a t tim e o f a n a ly s is .)
vA / V*;
Table 2
E stim ates o f Cancer M o r ta lity from Exposures* to Asbestos
E stim a te s Based on Numbers o f M esotheliom as and a R e la tio n s h ip Between th a t Number and th e Number o f T o ta l Cancers
A 3 Vi A
CCU91 M
< C <0
iow U
O*
CM
0** o Li
^m
Voi
o
.u -<
tUnl C<39 U 25
0i
o DO
VyI w vi
DD
?*s a
Li 9 CL 25
e o a4 Li tt 9 1H3 o IM o tt aH
CD A
tt Li u
5
oo moo on
tt
Li
9
x
tt o
I tt
CL
mK
tt
o
0 0
H
o0
X ^4
c
Li o
e0
rt
oa
Li
00
Li O
a
<0 u
9
*-< o
1u
u
D. Ps ox
tt 0
o
'
a* 0 X
o 5 x
X tt p4
tt ^4 9
e0
tt tt
tt 0 X o
0o
Li 9
00
0 tt
u
o0
*4 50
tt 00
Li
H H 44 tt
s-l tt 0 O
0 tt
e
X Li u 0
tt 0
0
Li O tt
tt tt
V
X 0 *4
5 H
tt 0
tt 0
9H 00
tt
H v*/ o
44 A
Li
O Li cn o tt
Li tt
o u tt CO Vi X
tt L D tt
0 9+ 0
0 oo
b
l
tt mmt
e
3 B
o 4J
*9 9
*3
-4 0
0e
9S
UGV
X 0X
m
ClCAMO oo* CM cn
ooo4* moo
moo ***s
ooo* *5*
oo
o
O
m m
(M o
in
1?
A A 3 O
> D
L4 *9
L4 C
55*
0
e
<<-N
o
Lo
44 o
*9 fn
tt A
Li O
O tt
1 VI
tt tt
A
CL Li
D
K9
A
V tt
AA
O eA
tt a o
0X
5i
tt n H D
4*4 M O tj
*-4 M in n II
tt X
J5*
I-.
O A
A 3
Li o-l 4) A
o |4 w B U
tt tt
tt X A <U A 0 Li O Vi
M
D
44 3 Vi U O ADC
14 o
O V B
A U
tt o K 3
X B
oV
DC y
CO c
9 in m
3
C w iO X *"
O'
*9 tt h nlu
tt -9 1 on o
Li M D
0 0 Vi VI W
0 X V 3 91
*4 a.
O
Li ^4 o* g
tt X iH A 3
M tt A ^ C
O Li tt ^4 a. 0
Ltti
14 tt tt X LI Li tt oo
G tt 0 0o o tt 9 w tt
tt +9
0 U tt | 0 tt 0o O Li 4H tt -t tt tt X X tt Li 0 o tt >4 tt tt 0 6 44 9 oc
Li <~t Jl 2
00 9u 0 wt /-s m rs ^4 Li
tt +O
c *4 0 wO
c o tt e l4 <-4 bo 0 bo m-4 Li X tt
:
1 tt
o
Li
tt 9 tt X e tt Li 0 X 0 tt tt
tt 9 0
tt
B
W
Li
44
0
> tt
tt -4
tt 44 X
9 0
tt 0 Li
Li 9
LI
5
C a 4-t
9 0u
X tt Li X Li
tt 0
0 Li
o00
tt o
tt
tt JS
o
o
Li
tt
r- tt 9
tt 0
^"4
tt
tt 0 Li
Li 0
0o9 Li 0
tt V tt
XLi
9 9
Li tt
Cl
C tt c
H |4 0
S <J
tt 9
O Li
C tt tt
H 0 X
tt 0 LJ
S 5o
S
0 H
m4 Ltti fr.
UCC 015232
v
v*
<
