TECHNICAL NOTE
ORP/CSD-76-2
ESTIMATE OF THE CANCER RISK
DUE TO NUCLEAR-ELECTRIC
POWER GENERATION
OCTOBER 1976 '
'
U.S. ENVIRONMENTAL PROTECTION AGENCY
OFFICE OF RADIATION PROGRAMS
WASHINGTON, D.C. 20460
-------
TECHNICAL NOTE
ORP/CSD-76-2
ESTIMATES OF THE CANCER RISK
DUE TO NUCLEAR-ELECTRIC
POWER GENERATION
by
W. H. M. Ellett, Ph.D.
and
A. C. B. Richardson
October 1976
Environmental Protection Agency
Office of Radiation Programs
Washington, D. C., 20460
-------
PREFACE
This paper was presented at the meeting on "Origins of
Cancer," held at the Cold Spring Harbor Laboratory, September 17
through September 14, 1976. The meeting was arranged by J. D.
Watson, Director, Cold Spring Laboratory and H. Hiatt, Dean,
Harvard University School of Public Health. The proceedings of
the meeting will be published in 1977 as the fourth volume in the
Cold Spring Harbor Cell Proliferation Series.
-------
TABLE OF CONTENTS
Page
Introduction 1
Technical Considerations 3
Environmental Dose Commitment
Health Effects from Planned Releases 5
Occupational Exposures 10
Accidents as a Source of Population Dose Commitment 13
Adequacy of NAS-BEIR estimates of Radiocarcinogenesis 16
Conclusions 26
-------
Introduction
Exposure to ionizing radiation results in an increased risk
of cancer. The degree of risk incurred per unit dose can be
determined, however, only from data obtained mainly at high doses
and dose rates; therefore it is necessary to assume a theoretical
dose response relationship to estimate the degree of risk from
small doses of radiation. The most generally accepted procedure
is to estimate cancer risk on the basis of a linear non-threshold
hypothesis; i.e., to assume that the number of cancers which may
occur per unit dose at low doses and dose rates is the same as
observed at high doses and dose rates. Following this practice,
we have used the radiation risk estimates for fatal cancer given
in the 1972 "BEIR" report by the National Academy of Science (NAS
1972). In this analysis, we have considered the risk of fatal
cancer for the entire fuel cycle due to the radiation exposure
associated with the annual production of one gigawatt (1000
megawatts) of nuclear electric power. A gigawatt-year (GW(e)-y)
is the amount of energy produced annually by a single nuclear
reactor of the type and size currently being constructed.
Three sources of radiation exposure are examined:
occupational exposure to radiation workers, exposure of the
general population from routine releases of radioactivity to the
environment and, finally, exposure due to accidental releases of
radioactivity. The quality of data available for risk assessment
-------
varies depending on the source of exposure being considered; good
for routine releases, fair for occupational exposure and
relatively poor for accidents.
Cancer risks due to nuclear electric power generation are
somewhat unique in that the release of some of the radioactive
materials implies a dose commitment to future generations which
persists over such long periods of time (millenia) that the risk
cannot be quantified meaningfully. If, however, the assessment
is confined to health effects expected to occur within a few
generations, usable estimates of the number of fatal cancers per
gigawatt-year can be made. In this analysis, exposures to
radiation due to the uranium fuel cycle are projected for a
period of 100 years following the production of the electricity.
The impact over longer time periods is discussed only
qualitatively.
Routine effluent releases from nuclear reactors are not the
focal point of this assessment since, except for occupational
exposures, they are relatively insignificant sources of
radioactive pollutants. Rather, the risk assessment should
include the entire uranium fuel cycle, which involves the mining
and processing of uranium ores, uranium enrichment, fuel
fabrication, fuel reprocessing, and finally waste storage and
disposal. Unfortunately, options for the disposal of radioactive
materials following fuel reprocessing are as yet poorly developed
-------
and the environmental risk from uranium mining has received
little attention. Therefore, the risks to the general population
associated with waste disposal and mining are not accounted for
in this analysis.
Technical Considerations
The sequence of operations occurring before and after the
fissioning of fuel at the power reactor is shown schematically in
Figure 1. Uranium ore, which usually contains approximately 0.2%
uranium, is first mined and then milled to produce a concentrate
called "yellowcake" which is about 75 percent uranium. The
greatest portion of the feed ore becomes a plant residual
depleted in uranium but containing radium-226 and its long half-
life parent thorium-230. These "tailings11 are a source of radon-
222 emissions which can last for several hundred thousand years.
Following the manufacture of yellowcake, a conversion step
purifies and converts the uranium oxide to uranium hexafloride,
the chemical form in which uranium is supplied to an enrichment
plant. At this plant, the isotopic concentration of uranium-235
is increased from its natural abundance, about 0.7* uranium, to
the design specification of the power reactor, usually 2-431, by a
differential gaseous diffusion process. The enriched uranium
hexafloride is converted into uranium oxide pellets at a fuel
fabrication plant and then loaded into thin zircalloy or
-------
stainless-steel tubes which are fabricated into fuel element
bundles.
Current plans call for the reprocessing of spent fuel after
a three-year "burnup" in the reactor. At this point the primary
physical barriers to the release of radioactivity are removed and
residual gaseous fission products, mainly tritium and krypton-85,
are released into the environment. Control of krypton-85
effluents is technically feasible and is the subject of a current
Federal rulemaking. Most of the remaining radioactive materials,
with the notable exception of carbon-14, are recovered and either
recycled or stored permanently as indicated in Figure 1.
Environmental Dose Commitment
Planned releases of radioactive materials occur in every
step of the uranium fuel cycle. However, the major potential
impact on human health is from radionuclides intentionally
released into the environment during the milling of ores and at
fuel reprocessing. The importance of ore milling and fuel
reprocessing to health considerations is not always recognized
since most risk assessments are confined to a consideration of
the annual dose rate delivered to persons living in the vicinity
of the release point. Such temporally and spatially limited
studies do not address adequately the dose committed from the
discharge of long half-life radioactive materials into the
biosphere. Krypton-85 released from a fuel reprocessing plant is
-------
a case in point. Given an adequate stack height for the release
of this noble gas, individual annual doses are not large in terms
of currently permitted annual dose levels. However, its
relatively long half-life (10.7 years) allows world-wide
atmospheric diffusion and, since there are no environmental
sinks, krypton-85 persists in the atmosphere for several decades.
In other words, the production of a given amount of nuclear
electricity commits to the environment a material which will
deliver an impact in subsequent years due to the persistence of
radioactive materials. The total dose considered from this
perspective is termed the "environmental dose commitment" and,
simply defined, is the sum of the doses to all individuals over
the entire time period the material persists in the environment
in a state available for interaction with humans (Richardson
1974).
Health Effects from Planned Releases
Health effects due to 100-year environmental dose
commitments from the uranium fuel cycle have been considered in a
number of recent EPA studies published in support of proposed
Federal environmental radiation standards, as cited below. These
analyses are based on the expected performance of average
facilities operating under current control requirements. They do
not take into account the effect of any additional control
measures which may be instituted in the future.
-------
Table 1 indicates the estimated number of fatal cancers due
to the 100-year environmental dose commitment for that portion of
the fuel cycle that preceeds the generation of electrical energy
(EPA 1973-1). Clearly, of all the sources considered, radon
emissions from tailings piles dominate the risk estimate. This
is due to several factors that are often not appreciated.
Typically, the duration of transcontinental transport in the
troposphere ranges from four to eight days. Therefore, the 3.6
day half-life of radon is not too short to allow the widespread
diffusion of this radionuclide into populated areas. Studies by
the Environmental Protection Agency indicate a factor of about
0.4 for radon decay between points of emission in the Rocky
Mountain states and the eastern seaboard of the United States.
Atmospheric transport beyond the Atlantic Coast has not yet been
studied in detail. Our preliminary estimate indicates that about
half of the projected cancers due to these emissions would be
expected to occur in the U.S., the balance occurring in the rest
of the northern hemisphere.
-------
Table 1
Estimated Cancer Fatalities due to the Extraction
and Fabrication of Uranium Fuel
(100-year environmental dose commitment)
Industry
Uranium Milling
UF5 Conversion
UFs Enrichment
Fuel Fabrication
Transportation
Principal Radionuclides
radon -2 2 2
uranium
uranium
uranium
uranium
Fatal Cancers per GW(e)-y
5 x 10-»
1 x 10-z
1 x 10-3
6 x 10-s
2 x 10-*
E = 0.5
The duration of radon emission from tailings piles is not
controlled by its parent isotope radium-226, which has a 1602
year half-life. A 1600 year half-life could at least be dealt
with conceptually; but, unfortunately, the radium-226 in tailings
piles is in near equilibirum with its parent, thorium-230, which
has an 80,000 year half-life. Granted that a 100-year dose
commitment seems particularly inapplicable to this case, it seems
equally unwise to predict that a tailings pile will continue to
emit radon into the biosphere over a time period so long that
geological stability cannot be assured. This brings to mind a
third point. Some investigators argue that since the radon
emissions are from a naturally occurring radioactive material,
the health effects would occur anyway and, therefore, should not
be charged to the nuclear fuel cycle (Cohen 1976) . Such a
-------
viewpoint fails to recognize that without man(s intervention this
radon would decay deep within the earth and not enter the
biosphere. Under these circumstances it would seem reasonable
not to differentiate between the release of radon from tailings
piles and the release of krypton-85 during fuel reprocessing. It
can be seen from Table 1, that compared to the potential health
impact due to radon, planned releases from other portions of this
part of the fuel cycle have much lesser impact.
Estimated numbers of fatal cancers from the 100-year
environmental dose commitment due to the reprocessing of the
spent fuel which has produced one gigawatt-year of electrical
energy are shown in Table 2 (EPA 1973-2). Until fairly recently
it had been assumed that krypton-85 and tritium were the most
important radionuclides involved. However, Magno and coworkers
have pointed out, rather surprisingly, that the dose commitment
from carbon-14 exceeds that from all other radionuclides combined
(Magno, Nelson and Ellett 1971). Carbon-14 is not a fission
product. It is produced via an n,p reaction on nitrogen-11
impurities in uranium fuels and to lesser extent from an n,
reaction on oxygen-17 in reactor cooling water (Fowler, et. al.,
1976).
-------
Table 2
Estimated Cancer Fatalities due to the Reprocessing of Spent Fuel
(100-year environmental dose commitment)
Principal Radionuclides
Fatal Cancers per GW(e)-y
Carbon-14
Tritium
Krypton-85
2.3 x 10-»
1.8 x
4.6 x 10-2
Z=0.3
Carbon-14 dioxide injected into the atmosphere during fuel
reprocessing moves from inorganic reservoirs to living systems.
The concentrations of this carbon-14 in man has been calculated
using a world transport model for carbon-14 developed by Machta
(Machta 1973). This model is a multi-reservoir exchange model
consisting of seven compartments with exchange rates based on the
analysis of carbon-14 produced in nuclear weapons testing. The
model takes into consideration projected increased levels of
tropospheric carbon due to the combustion of fossil fuels which
are depleted in carbon-14. The 100-year environmental dose
commitment from carbon-14 has been calculated on the basis of its
specific activity in tissue. This assumes an equilibrium between
tissue and the troposphere and may underestimate individual doses
slightly in the vicinity of a fuel reprocessing facility.
The estimated number of fatal cancers due to the 100-year
environmental dose commitment from a reactor producing one
gigawatt-year of electrical energy are listed in Table 3 (EPA
-------
1973-3, EPA 1976-2, Fowler, et. al., 1976). Except for carbon-
14, these dose commitments are negligible compared to these from
the balance of the fuel cycle.
Table 3
Estimated Fatal Cancers due to Nuclear Reactor Effluents
(100-year environmental dose commitment)
Principal Radionuclide Fatal Cancers per GW(e)-y
Carbon-149 x 10~z
Krypton-85 7 x 10-s
Tritium 8 x 10-s
Short half life effluents <1 x IQ-z
z=io-»
Occupational Exposures
Contrary to the case for exposures from planned effluent
releases, occupational doses due to nuclear energy production are
dominated by the exposure of nuclear reactor operators and
maintenance personnel. Recently, Golden and Pavlick have
reviewed occupational doses as a function of plant age at the two
largest and oldest multiple reactor complexes in the United
States, Quad Cities and Dresden (Golden and Pavlick 1974, Golden
1976). These studies show that the early operating experience at
a reactor is not indicative of later occupational doses.
Occupational exposure increases with plant use as a result of the
activation of structural materials and equipment as well as
increased inventories of radioactive materials in storage areas.
10
-------
Occupational exposure also occurs more frequently due to the need
for increased plant maintenance with plant age. Indeed,
maintenance operations, both scheduled and unplanned, lead to
about half the total occupational dose and impact mainly on
persons, such as welders, not routinely employed in the operation
of reactors. The mean annual occupational dose per reactor at
Dresden, the oldest of these reactor complexes is 1050 man-rem
and at Quad city 800 man-rem. Even though these reactors produce
less than a gigawatt year annually, occupational doses are not a
sensitive function of the reactor's generating capacity and a
rather conservative estimate of the population dose due to
occupational exposure is judged to be about 800 man-rem per
gigawatt-year (electric) as shown in Table 4. Nuclear Regulatory
Commission (NRC) estimates for occupational exposures at nuclear
power reactors and for the balance of the fuel cycle are also
included in Table 4 (AEC 1974).
11
-------
Table 4
Estimated Cancer Fatalities due to Occupational Exposure
Industry
Mining
Reactors
Fuel Fabrication
Fuel Reprocessing
Transportation
Waste Management
Total Body Man-rem
EPA NRC Fatal Cancers per GW(e)-y
2 x 10-2
800 (235) 1.5 x 10-»
40 — >
12
0.5
2
1-2 x 10-2
Z = 0.2
The dose to uranium miners, the only class of workers in the
fuel cycle for which there is epidemiological evidence of
radiogenic cancer, has not usually been included in occupational
risk estimates, since underground miners are not considered to be
radiation workers under the Atomic Energy Act. Radon exposures
to uranium miners are controlled by the States, under Federal
guidance. Current limits for radon daughters (218Po, 21£*Pb, 2IlfPo)
are expressed in terms of cumulative working levels. In a single
year an underground uranium miner may be exposed to -four working
level months (WLM). Average exposures probably range between one and
three WLM annually (MESA 1975). On the basis that 53 man-years
of mining effort are required to obtain the ore necessary to
produce one gigawatt-year (electric), the occupationally induced
12
-------
lung cancer death for the ore needed to generate one gigawatt-
year of electricity is estimated as about 0.02.* Though this
last calculation is imprecise, due to the lack of reliable data
on exposure to miners, it is probably realistic to assume that
the nuclear reactor is the most important source of occupational
exposure in the nuclear energy industry.
Accidents as a source of population dose commitment
While unplanned releases of radioactivity into the biosphere
as a result of accidents at uranium fuel cycle facilities are a
potential source of radiation-induced cancers, little can be said
with much confidence about their likelihood or the effects of
such accidents on the general population. Only for nuclear
reactors has a significant start been made towards predicting
this risk. Two factors are involved in such analyses: a
consequence model, which yields the distribution of doses that
would occur on the average from a postulated accident, and a
fault tree analysis that yields the joint probability that a
given sequence of events will actually occur, given the
probability of individual component failure, power failure, etc.
In the final analysis these probabilities are essentially
engineering judgments. Therefore, such risk assessments are
dependent on "Bayesian estimates" reflecting degrees of faith in
particular engineering practices rather than on "Fisher
likelihood estimates," which draw on an actual sample of observed
*It is assumed that two-thirds of the uranium ore is obtained by
underground mining.
13
-------
events. Not suprisingly, knowledgeable individuals have
different degrees of faith based on their own prejudgments.
The most serious reactor accidents are those involving
meltdown of the reactor core where containment is more or less
lost and substantial quantities of some of the reactor contents
enter the environment. In addition, there are non-core accidents
where the quantity of material released is generally much
smaller. The recently published Reactor Safety Study, more
commonly known as the "Rassmussen Report," evaluates numerically
the population dose due to core related accidents (NRC 1975).
The cancer risks reported in the Reactor Safety Study are not
directly comparable to those given in this analysis since the
estimated number of cancers per unit population dose utilized in
that Study is approximately a factor of four smaller than those
given in the NAS BEIR Report (EPA 1976-1). Table 5 lists the
projected number of cancers per gigawatt year (electric) from
large reactor accidents based on the Reactor Safety Study, after
taking this factor of four into account.
14
-------
Table 5
Estimated Cancer Fatalities from Accidental Release of Nuclear Material
Source Fatal Cancers per GW(e)-y
Nuclear Reactor (core failure) B.H x 10- *
Nuclear Reactor (non-core related) unknown
Fuel Reprocessing unknown
Transportation 10~7 (EPA 1975)
Waste Storage unknown
Major accidents in the balance of the fuel cycle should have
a much smaller impact on the general population. For example,
the amount of short half- life gaseous radioactivity involved at
any one time at a fuel reprocessing operation is much less than
in the case of an operating reactor. There remains the question
of the health impact from less serious reactor accidents. To
date only individual doses, not the population dose commitment,
have been examined for this class of accidents. These lesser
accidents are of some interest, since their estimated likelihood
of occurrence is about one hundred times greater than that for
accidents involving core meltdown.
Current regulations for identified non-core related
accidents impose design guidelines that will limit individual
exposure at the facility boundaries to doses ranging from 500 to
2000 mrem. Weighting these doses by their expected frequency of
15
-------
occurrence yields a total body dose of 35 mrem per gigawatt-year
(electric) to the most highly exposed individual in the general
population. This is about 10 to 20 times larger than the maximum
total body dose of 2-4 millirem per year expected from routine
releases (EPA 1976-2). Even though maximum individual doses are
not directly proportional to the total dose commitment to
populations, it would appear to be worthwhile to evaluate the 100
year environmental dose commitment caused by minor reactor
accidents. Table 5 is an admission that, in reality, very little
is known about the potential dose commitment and health effects
due to accidents in the uranium fuel cycle. Until recently,
accident analysis for the licensing of nuclear facilities has
been exclusively concerned with consequences in the immediate
vicinity of the facility so that appropriate emergency plans may
be implemented if needed. There is a great deal of interest in
expanding this limited viewpoint, and it is reasonable to expect
significant changes in the amount of information that will be
available within the next few years.
Adequacy of NAS-BEIR estimates of radiocarcinoqenesis
Before considering the total potential health impact from
one gigawatt-year of nuclear electric energy production it is
worthwhile to consider the degree of confidence that can be
placed in current estimates of the cancer risk due to radiation.
Although there is general agreement in both the United States and
16
-------
in international scientific communities that the NAS-BEIR report
is a comprehensive analysis of the radiation effects data for
human exposures available in 1972, it is being argued by some
that in their calculation of the cancer risks from ionizing
radiation, the BEIR Committee did not give sufficient weight to
animal data on radiation carcinogenesis. More specifically, this
criticism is based on the observation that the incidence of some
radiation induced cancers in rodents is reduced at low doses
(and/or low dose rates) of low LET* radiation. The hypothesis is
that repair of precancerous damage reduced the cancer incidence.
The revisionists' contention is that this postulated repair of
radiation damage should be taken into account when estimating the
risks from nuclear energy. While there is little or no evidence
for such repair in large mammals or in man, it has some
theoretical basis in the model of radiation injury proposed by
Kellerer and Rossi (Kellerer and Rossi 1971). Their theory of
dual radiation injury holds that initial radiation injury, i.e.,
damage to critical sites, has both linear and quadratic
dependence on dose.
y = a D + b D2
Where the coefficients a and b correspond to one and two "hit"
damage respectively. For high LET radiations the first
coefficient, a, is so much greater than the second, b, that the
dose squared term can be neglected at all dose levels. This is
*Lightly Ionizing.
17
-------
not true for low LET radiations, for which the relative
importance of each term depends on the dose range of interest. A
quadratic component in the response to radiation has been
observed for genetic damage from ionizing radiation and, if one
step in the development of a cancer is defined as a somatic
mutation, reasonable arguments can be made in support of the
applicability of this hypothesis to the initiation of
radiocarcinogenesis.
Rather than discuss the theory in critical detail, the point
to be made here is that its general acceptance would not have
much effect on the conclusions reached in the NAS-BEIR report
concerning cancer risk from ionizing radiation. The Rossi-
Kellerer theory predicts a linear response for low doses and a
supra-linear response at high doses, the dose squared term
dominating at doses that exceed the ratio a/b. Obviously, if the
human effects data used by the NAS-BEIR Committee were obtained
only at high doses, a linear approximation to the dose response
data would greatly overestimate the risk. This has been
suggested by several groups, including the National Council on
Radiation Protection and the Nuclear Regulatory commission (NCRP
1975, NRC 1975). More recently, a concensus has developed that
if the NAS-BEIR risk coefficients do indeed overestimate
carcinogenesis, it would be only by a factor of two, an amount
comparable to other uncertainties in the application of these
18
-------
data (Brown 1976, Upton 1976, EPA 1976-2). The basis for this
consensus is that current information en the values of a and b
(based on genetic studies) indicate that the coefficient a is
about 100 times greater than b so that the linear term dominates
the response function at doses less than 100 rad or so. This is
about the low end of the range of doses considered by the NAS-
BEIR Committee. Since, in general, the Committee interpolated
linearly between zero and the lowest dose level where excess
cancer was observed, it is probably unlikely that their
conclusions were heavily biased by a dose response dominated by
higher power terms.
Indeed, it is by no means clear that any modification need
be made to NAS-BEIR risk estimates. There are many points in
favor of using a linear assumption for cancer risk prediction.
One is the lack of a supra-linear response in much of the
available human data. Figure 2, which is taken from the BEIR
Report, indicates human breast cancer increases in a linear
manner for a range of doses between 50 to 7000 rads (NAS 1972).
Even though this example is not as clear cut as one would like
(the higher doses were accompanied by increased fractionization)
it seems suprising, in view of the Kellerer-Rossi theory, that so
little departure from linearity is observed at doses as large as
7000 rad.
19
-------
Turning to the other end of the scale, the low dose range,
again the BEIR risk estimates seem to be holding up rather well.
The 1972 BEIR estimates of the radiation risk per rem for thyroid
cancer in children ranged from 1.6 to 9.3 cancers per 10* persons
years at risk (NAS 1972). These results were obtained from a
variety of epidemiological series in which the average thyroid
dose exceeded 200 rem. More recently, Modan and coworkers
(Modan, et. al., 1974) have reported on a follow-up study of
11,000 Israeli children treated with x-rays for tinea capitis, a
fungal scalp infection. Because a great deal of care was
exercised in beam direction and in the use of patient sheilding,
the estimated thyroid doses, compared to those considered by the
BEIR Committee, were quite low, varying from 6.2 to 7.4 rem with
a mean of 6.5 rem. The observed excess thyroid cancer incidence
per rem, using siblings as controls, was 6.9 per 10* person years
at risk. This finding is in agreement with the upper range of
the NAS-BEIR estimate, cited above, and does not support a
hypothesized reduced response at low doses. Rather than argue
that thyroid and breast cancer are unrepresentative of radiogenic
cancers in man, and therefore exceptions to an apparent recovery
from precancerous radiation damage observed in rodents, it would
appear more prudent and indeed advisable to limit exceptions to
the NAS-BEIR risk estimates to those human cancers where
supporting epidemiological data is available.
20
-------
This is not to say the NAS-BEIR estimates are precise. They
are not. Not only is the human data on which they are based
quite limited/ but there are also critical questions as to how
the BEIR risk estimates should be applied to heterogeneous
population groups, since they are based, in large part, on adult
data. The NAS-BEIR Report provides two estimates of radiation
risk. One is an estimate of the absolute risk, that is the
number of cancers that will occur per unit dose in an exposed
population. The other estimate is of relative risk, the
percentage increase in the excess cancer mortality per unit dose.
When lifetime exposure to a population containing a mixed age
distribution is considered, the number of fatal cancers per
person rem estimated on the basis of relative risk exceeds the
number estimated on the basis of absolute risk by a factor of
about four. In the previous analysis, averages of the relative
and absolute risk were used to predict the cancer risk from
nuclear generated electricity. This procedure will underestimate
the actual risk if radiation-induced carcinogenesis increases in
proportion to the cancer incidence in an unirradiated population.
There is some limited data on this point from the Japanese atomic
bomb casualties. Figure 3, taken from the NAS-BEIR Report,
illustrates the difference between these two models (NAS 1972).
The two curves compare the age-specific cancer death rate in a
21
-------
Japanese control group with the age specific absolute risk for
fatal solid tumors in atomic bomb survivors.
If the absolute risk model is an accurate predictor of
cancer mortality from atomic bomb radiation, the incidence of
radiation cancers would be independent of age at exposure, i.e.
it would appear as a horizontal line parallel to the abcissa. On
the other hand, if the risk increases in proportion to the age-
dependent incidence of 'spontaneous1 tumors, the observed age-
specific risk from bomb exposures would parallel the age-specific
mortality rate of the control population. As can be seen, the
correlation between radiation induced and non-radiogenic cancer
is high, particularly if the increased sensitivity of children to
radiation is taken into account. One should note, however, the
large uncertainty associated with radiation cancer mortality for
the 0 to 9 age group in Figure 3. The fact is that as yet there
have been very few deaths due to solid tumors in this age group,
only nine for exposures greater than 100 rads. As this group
ages* it will be of interest to observe to what extent their
cancer frequency reflects that of the general Japanese population
and their radiation exposure history. Perhaps in twenty years or
so we will be able to exclude either the relative or absolute
risk model, for the most common cancers.
In view of the emphasis that is placed on the Japanese
studies in some assessments of radiation risk, the uncertainties
*The persons at risk are between 31 and 41 years old now.
22
-------
associated with these data should be more widely appreciated,
particularly the unreliability of these data for elucidating the
form of the dose response function for low LET radiation.
Recently, a number of investigators (Mays, Lloyd and Marshall
1975; Rossi and Kellerer 1971; Jablon 1974) have suggested that
the incidence of leukemia observed after the atomic bombing of
Nagasaki can be used to assess the risk from low LET radiation
since, unlike the situation at Hiroshima, the neutron dose to
Nagasaki survivors is negligible. A tentative conclusion that
has been drawn from such studies is that the incidence of
leukemia varies as the square of the dose for low LET radiations
rather than linearly.
An important question that has not been considered
explicitly in any of these studies is whether or not the leukemia
data obtained on Nagasaki survivors is sufficient to permit any
meaningful discussion of the dose-response relationship at doses
below 100 rad or so, since the number of persons at risk in the
Nagasaki study population is relatively small compared to that
from Hiroshima. Table 6 compares the number of leukemia cases
reported in the Nagasaki Cohort Study with that expected from
Japanese national death rates published by the Japanese Ministry
of Health and Welfare. All of the data in Table 6 is taken from
ABCC Report 10-71 (Jablon and Kato 1972).
23
-------
Table 6
Observed & Expected Leukemia Mortality in Nagasaki Survivors1
Mortality
Observed (O)
Expected (E)
no radiation
Excess (0-E)
2.4
(0-9)
11
5.6
5.4
Mean
21.3
(10-19)
2
1.8
0.2
dose (rads)
70.3 114.3
(50-100) (100-200)
0 3
0.6 0.6
-.6 2.4
329
(200-600,
15
0.7
14.3
ICohort Study, Jablon and Kato 1972.
From Table 6 it is seen that when the expected "background"
cancer incidence is subtracted from the observed, there is no
dose response data for total body doses below 100 rad*. It is
apparent that the net number of excess effects observed precludes
a realistic differentation between linear and dose squared
response patterns in the dose range of interest.
Tne paucity of effects information shown in Table 6 does
give rise to at least one more guestion. In view of the lack of
excess cancer below 100 rad, is the Nagasaki leukemia data
consistent with risk estimates given in the BEIR Report, which
are based on an assumed linear nonthreshold dose response
relationship? BEIR risk estimates for leukemia separate the
irradiated populations into two risk groups; those over ten years
of age, and survivors who were less than ten at the time of the
bombing. The BEIR absolute risk estimate (based on all leukemia
results and not just the Japanese studies) assume for adults one
*Use of other control groups to estimate the exposed mortality, such
as persons not in the city at the time of the bombing, does not
alter this finding.
24
-------
case per rad per 106 person years at risk and for children, twice
this number of cancers per rad (NAS 1972). Tables 7 and 8
compare the leukemia incidence in the Nagasaki Cohort Study for the
older and younger age groups (Jablon and Kato 1972} with that
predicted using the risk coefficients given in the BEIR Report. They
indicate that for doses less than 200 rads the net leukemia mortality
among Nagasaki survivors is about what would be expected on the basis
of BEIR estimates. Certainly, the Nagasaki data does not appear to
provide an adequate basis for rejecting a linear dose response
relationship.*
Table 7
Nagasaki Leukemia Mortality in Persons Over 10 Years
of Age at Time of Bomb1
Mortality
ersons at risk
Observed (O)
Expected (E)
Excess (O-E)
BEIR Estimate
2.4
'(0-9)
5158
8
4.5
3.5
.3
Mean dose (rads)
21.3 70.3
(10-49) (50-100)
3719 983
2 0
1.4 .5
.6 -.5 1
1.0 1.0
144.3
(100-200)
1030
2
.5
.5
2.7
329
(200-600)
1110
8
.6
7.4
9.8
Cohort Study, Jablon and Kato 1972
*Note added in proof (second printing)
At the recent 1976 Radiation Research Society Meeting, Beebe,
Kato and Land pointed out that when all leukemia fatalities in the
Nagasaki Register are considered, and not just those in the ABCC
Cohort Study subset, the incidence of leukemia increases monotonically
with dose at all dose levels. Although these data are still
insufficient to determine a functional relationship between leukemia
incidence and dose, they do illustrate the absence of excess leukemia
deaths shown in Table 6 is a statistical artifact inherent in the
small number of cases recorded in the Cohort Study.
25
-------
Table 8
Nagasaki Leukemia Mortality in Persons Under 10 Years
of Age at Time of Bomb1
Mortality
Persons at risk
Observed (0)
Expected (E)
Excess (0-E)
BEIR Estimate
2. a
(0-9)
1592
3
1.1
1.9
.1
Mean
21.3
(10-49)
981
0
.4
-.4
.8
dose (rads)
70.3
(50-100)
248
0
.1
-.1
.7
144.3
(100-200)
199
1
.1
.9
1.0
329
(200-600)
200
7
.1
6.9
2.6
Icohort Study, Jablon and Kato 1972,
Conclusions
Having estimated the cancer risk from the uranium cycle and
then indicated to you some of the issues concerning the degree of
uncertainty in such risk estimates, it can be understood that the
projected health risk from nuclear power is just that, an
estimate. Table 9 sums up the estimated risk of fatal cancer for
the various portions of the uranium fuel cycle. If the risk per
gigawatt from accidents and waste disposal are as low as seems
likely now, about one cancer death could be committed for each
gigawatt-year of electrical energy produced.
26
-------
Table 9
Estimate of the Total Cancer Fatalities for one
Gigawatt-year of Electrical Energy
Planned releases079
Occupational Exposure 0.2
Accidents 0.1_
z = l. a.
In our opinion* it is beyond tbe state of the art to compare
these risks to the cancer risks from other forms of electrical
energy production. We simply do not know the long term health
risk from particulates, SO2 and other chemical emissions which
are inherent with the use of fossil fuels to produce electrical
energy. Not as a surrogate for such data but merely to give some
indication that no source of energy is without risk. Table 10
compares the occupational risk due to accidents in other fuel
cycles for equivalent amounts of electrical energy (CEQ 1973).
Obviously, most fossil fuels are not appreciably safer and coal
is considerably more hazardous. Having said that, it is also
fair to point out that if the cancer estimates for the uranium
fuel cycle were based on a 1000-year environmental dose
commitment, it would shift the balance in favor of coal if only
the occupational risks were counted. However, until we know more
about the health effects due to the long term environmental
pollution caused by other given energy cycles, comparisons of the
health risk for various fuel cycles is premature.
27
-------
Table 10
Occupational Fatalities for Other Energy Sources
Boiler Fuel Accidental Fatalities per GW(e)-y
Coal O
Oil 0.42
Gas 0.27
It is worthwhile at this point in the discussion to step
back and consider not so much what has been stated, but rather
that part of the risk analysis which has not been explored.
Numerical estimates of risk from radiation are but one aspect of
an overall assessment. Equally important are issues not
discussed in this paper. These include such ethical questions as
whether or not the risks are voluntary, and the degree of
correlation between the distribution of benefits and the
distribution of risk among the various segments of the
population. Consideration must also be given to the extent to
which the risks are subject to control and the cost of such
controls. Progress is being made in developing a macroscopic
viewpoint for risk analysis so that various energy sources can be
compared in a consistent manner (Rowe 1977). Nevertheless the
achievement of this goal is still ahead of us.
Finally, it is important to recognize that the implications,
for the Nation, inherent in the development and widespread use of
any energy source transcend public health considerations alone.
28
-------
Health risk estimates are easily transcribed by the media into a
news story and may perhaps be overemphasized for that reason.
Less easily imparted is information on how the selection of prime
energy sources will impact on other areas of national life. To
name but a few, the continued use of imported oil as a prime
energy source will influence foreign policy considerations as
well as have implications for the continued integrity of the
marine environment. Utilization of coal as an energy resource
effects the allocation of large areas of land for periods of time
that are comparable to that of long half-life nuclear materials.
The institutionalization of domestic security procedures and the
commitment of a large portion of the Nation's capital
expenditures are inherent in the development of a nuclear
economy, particularly if it includes plutonium as an energy
source. All of these factors concern alternative benefits rather
than health risk. They are not amenable to quantification and
can be evaluated only partially on technical grounds. It follows
that societal and political judgements must provide the actual
basis for any final decisions on the acceptability of risk.
From this broader perspective the importance of public
health considerations is not diminished, even though it is
obviously only one input to a comprehensive decision-making
process. Rather, it points out the need for health risk
estimates to be as free as possible of bias concerning the
29
-------
benefits from any given energy source. The balancing of health
risks against societal benefits is not an appropriate
responsibility for the scientific community, which by training
and perhaps by inclination has little special competence for
evaluating societal benefits. Technical judgments on risk should
be separated from societal and political judgments concerning
benefits, so that decisions can be made on the basis of visible
preferences. Unfortunately* there appears to be a growing
tendency to include societal considerations in the assessment of
risks, particularly nuclear risks. While both critics and
proponents of nuclear energy have valid viewpoints which
transcend health risk considerations, the discussion of this
energy source is often colored by risk assessments that are based
on preconceptions of either benefit or harm to society. The
discussion of nuclear energy would be more enlightened if these
other issues were addressed directly and not weighted into the
assessment of health risk.
30
-------
The Light Water Reactor Nuclear Fuel Cycle
URANIUM MINES
AND MILLS
CONVERSION
TOUFg
ISOTOPIC
ENRICHMENT
FUEL
FABRICATION
RECOVERED
URANIUM
WASTE STORAGE —
REPROCESSING
i
/
FUEL STORAGE
Figure 1: The uranium fuel cycle for light water moderated
nuclear power reactors. The material transfers indicated by
the dotted lines have yet to be implemented commercially.
31
-------
700 200 300
Number of f/uoroscopies
400
Figure 2: Incidence of breast cancer per thousand person year at
risk (PYR). The error bars represent 90* confidence
intervals, and the line is the best fitting weighted least
squares regression line (NAS 1972).
32
-------
10.0-
jo
«5
Q)
«o
o
5
-------
REFERENCES
Beebe, G. W., H. Kato and C. E. Land, 1976. "Mortality
Experience of A-bomb Survivors, 1950-1974, presented at the
24th Annual Scientific Meeting of the Radiation Research
Society in San Francisco, June 29, 1976.
Brown, J. M., 1976. Linearity vs Monlinearity of Dose
Response for Radiation Carcinogenesis, Health Physics vo^
pp. 231-245.
Conen, B. L. 1976. "Environmental Impacts due to Radon
Emissions11, Bull. Atomic Scientists, 32, 2, pp. 61-64.
Council on Environmental Quality (CEQ) 1973. Energy ana the
Environment, pp. 42-53, U.S. Government Printing Office,
Washington, L.C. ,20402
Fowler, T. W., R. L. ClarK, J. M. Gruhlke, 'and J. L.
Russell, 1976. Public Health Considerations of Carbon-Ht
Discharges rrom t'he Lignt-Water-Cooled Nuclear Power Reactor
Industry, Technical Note ORP/TaD-76-3, U.S. Environmental
Protection Agency, Washington, D.C. 20460
Golden, J. C., 1976. "Effect of a Change in Occupational
Radiation Exposure Limits on Manpower and Total Dose at a
Nuclear Power Station". Paper read at the 21st Annual
Meeting of the Health Physics Society, Health Pnysics
(Abstract in press).
Golden, J. C. and R. A. Pavlick 1974. "A Review of
Effluents, General Population Doses and Occupational Doses
Resulting from Commonwealth Edison's Use of Nuclear Power",
Symposium on Population Exposures, CGNF-741018, pp. 199-206,
National Technical Information Services, Springfield,
Virginia 22161.
Jablon, S., 1974. "Environmental factors in Cancer
Induction: Appraisal of Epideiriologic Evidence Leukemia,
Lymphoma and Radiation," Proceedings of the XI International
Cancer Congress, Florence, Excerpta tfedica Congress Series
No. 351, Amsterdam
Jablon, 5. and H. Kato, 1972. "Mortality Among A-Borne
Survivors", Report 6, ABCC TR-01071, Japanese National
Institute of health - Atomic Bomb Casualty Commission
Hiroshima and Nagasaki, Japan.
Kellerer, A. M. and H. H. Rossi 1971. "RBE and the Primary
Mechanism of Radiation Action", Had. Res. 47, 15.
34
-------
Machta, L., 1973. "Prediction of CO2 in the Atmosphere",
Carbon and the Biosphere. CONF-720510, G. M. Woodwell and E.
V. Pecan, editors, pp. 21-31, National Technical Information
Services, Springfield, Virginia 22161.
Magno, P. J., C. B. Nelson and W. H. Ellett, 1974. A
Consideration of the Significance of Carbon-14 Discharges in
the Nuclear Power Industry, Proceedings of the 13th AEC Air
Cleaning Conference. CONF-740807, Vol. 2, pp. 1047-1055,
National Technical Information Services, Springfield, Virginia
Mays, C. to., Lloyd, R. D. ana Marshall, J. K., 1975.
"Malignancy risx to humans from total-body gamma-ray
irradiation," CONF-730907, pp. 417-428, Proceedings of the
3rd International Congress, International Radiation
Protection Association, U.S. Atomic Energy Commission, Oak
Ridge, Tennessee.
Mining Enforcement and Safety Administration (MESA) 1975.
"Administration of the Federal Metal and Nonmetallic Mine
Safety Act, P.L. 89-577, Annual Report to Congress of the
Secretary of the Interior."
Modan, B., D. Baidatz, H. Mart, R. Steinitz and 3. G. Levin,
1974. Radiation-induced Head and Neck Tumours, Lancet, vol.
1, p. 7852.
National Academy or Sciences (NAS) 1972. The Effects on
Populations of Exposure to Low Levels of Ionizing Radiation,
Report of the Advisory Committee on the Biological Effects
of Ionizing Radiations, Government Printing Office, Washington, D.C
National Council on Radiation Protection and Measurements
(ISCRP) 1975. "Review of the Current State of Radiation
Protection thilosopny," Report No. 43, NCRP, Washington, D.C. 20014
Ricnardson, A. C. B., 1974. The historical Development of
Concepts of Radiation Dose Commitment, Symposium on
Population Exposures, CONF-741018, pp. 1-8, National
Technical Information Services, Springfield, Virginia
22161.
Rossi, H. H. and A. M. Kellerer 1974. "Tne validity of risk
estimate of leukemia incidence based on Japanese data", Rad.
Res. 58, 131
Rowe, W. D. Anatomy of RISK (in press) , Wiley, New York,
1977.
Upton, A. C., 1976. Radiological Etfects of Low Dose:
Implications for Radiological Protection, Rad. Res. (In
press.)
35
-------
U.S. Atomic Energy Commission (ALC) 1974. Draft Generic
Environmental Statement - Mix Oxide Fuel, WA3H-1327, Vol. 3,
Chapter IV, p. J-12, U.S Nuclear Regulatory Commission,
Bethesda, Maryland.
U.S. Environmental Protection Agency (EPA) 1973-1.
Environmental Analysis of the Uranium Fuel Cycle, Vol. I, p.
11, EPA-520/9-73-003-13, USEPA, Washington, D. C.
EPA, 1973-2, Ibid. Vol. Ill, p. 12.
EPA, 1973-3, Ibid. Vol II, p. 10.
U.S. Environmental Protection Agency (EPA) 1975.
Transportation Accident Bisks in the Nuclear Power Industy
1975-2020, EPA-520/3-75-023, Washington, B.C. 20U60.
U.S. Environmental Protection Agency (EPA) 1976-1. Reactor
Safety Study (WASH-1400), A Keview of the Final ^e^ort^ EPA-
520/3-76-009, pp. 2-1 - 2-7^ USEPA, Washington. IK C..
U.S. Environmental Protection Agency (EPA) 1976-2.
Environmental Radiation Protection Requirements for Normal
Operations of Activities in the Uranium Fuel Cycle - Final
Environmental Statement, vol. I, pp. 5i-62f EPA-520/U-76-
016, USEPA, Wasnington, D. C. 20460.
EPA 1976-2, Ibid. pp. 187-188
U.S. Nuclear Regulatory Commission (NRC) 1975. Reactor
Safety Study, An Assessment of Accident Risks in U.S.
Commercial Nuclear Power Plants, Appendix VI, WASH-1400
(NUREG-75/014), National Technical Information Services,
Springfield, Virginia 22161.
U. S. Environmental Protection Agency (EPA) 1975. Transportation
Accident Risks in the Nuclear Power Industry 1975-2020,
EPA-520/3-75-023, Washington, D.C. 20460.
36
------- |