risk estimation and decision making: the health …...genesis; and the health effect, developmental...

PEDIATRIC DENTISTRY/Copyright © 1982 byAmerican Academy of Pedodontics

Special Issue/Radiology Conference

Risk estimation and decision making: Thehealth effects on populations of exposure tolow levels of ionizing radiation

Jacob I. Fabrikant, MD, PhD

IntroductionMy assignment this morning is to try to give you

some general background for an understanding of thepotential health effects in populations exposed to low-level radiation. To do this, I have decided to place ourdiscussions within the framework of the scientificliberations and the scientific controversies that arosedu~ng the preparation of the current ReportI of theCommittee on the Biological Effects of Ionizingdiation of the National Academy of Sciences-NationalResearch Council (1980 BEIR-III Report). I shall tryto explain how ce~w.in of the areas addressed by thepresent BEIR Committee~ have attempted to dealwith the scientific basis for establishing appropriateradiation protection guides, and what effect this mayhave on evaluation of radiation risks and on decision-making for the regulation of societal activities con-cerned with the health effects in human populationsexposed to low-level radiation. I speak only as an indi-vidual on what I consider important in these discus-sions, and in no way do I speak for the BEIR Commit-tee, or for any of its members, whose deliberations arenow available as a comprehensive report: The Effectson Populations of Exposure to Low Levels of IonizingRadiation: 1980.1 It would be difficult for me not to besomewhat biased in favor of the substance of theBEIR Reports,1-3 since as an individual I have beensufficiently close to the ongoing scientific delibera-tions of agreement and disagreement as they havedeveloped over the past 10 years.

I think it would be best for me to review, verybriefly, why we have advisory committees on radia-tion, and why the BEIR Committee, and its currentReport,1 may be somewhat different than the others. Ishall discuss what we know and what we do not knowabout the health effects of low-level radiation. Fur-ther, I shall comment on how the risks of radiation-induced cancer and genetically-related ill-health inman may be estimated, the sources of the scientific

aCommittee on the Biological Effect~ of Ionizing Radiation,National Academy of Sciences-National Research Council, Wash-ing~n, D.C., USA

and epidemiological data, the dose-response modelsused, and the uncertainties which limit precise esti-mates of excess risks from radiation. Fi~aally, I shouldlike to conjecture with you on what lessons we havelearned from the BEIR-III Committee experience,and especially on what the implications might be ofnumerical risk estimation for radiation :protection anddecision-making for public health policy.

What is the purpose of advisory committees onradiation and health?

For more than three-fourths of a century, scientificand medical observations have led to responsible pub-lic awareness of the potential health effects of ionizingradiations, initially from medical and industrial expo-sure, then from nuclear weapons and weapons testing,and now from the production of nuclear energy. Suchawareness has called for expert scientific advice andguidance for protection of the public health. Advisorycommittees on radiation of international and nationalscientific composition have, for these many years, metand served faithfully and effectively to discuss, toreview, to evaluate, and to report on three importantmatters of societal concern: 1) to place into perspec-tive the actual and potential harm to the health ofman and his descendants in the present and in the fu-ture from those societal activities involving the use ofionizing radiations; 2) to develop quantitative indicesof harm based on dose-response relationships to pro-vide a scientific basis for the evaluation of somaticand genetic risk so as to better protect human popula-tions exposed to low-level radiation; and 3) to identifythe sources and levels of radiation which could causeharm, to assess their relative importance, and to pro-vide a framework on how to reduce unnecessary radia-tion exposure to human populations.

To a greater or lesser extent, each advisory commit-tee On radiation -- such as the UNSCEAR,b theICRP,¢ the NCRP,~ the NRPB,e and others in France,

bUnited Nations Scientific Committee on the Effects of AtomicRadiation, United Nations, New York, U.S.A.

400 RISK ESTIMATION AND DECISION MAKING: Fabdkant

Canada, and elsewhere in Europe and Japan, and theBEIR Committee -- have dealt with these matters.But significant differences occur in the scientificreports of these various bodies. We should expect dif-ferences to occur because of the charge, scope, andcomposition of each committee, and probably mostimportant, because of public attitudes existing at thetime of the deliberations of that particular committee,and at the time of the writing of that particular re-port. The BEIR Report~ is different. However, themain differences are not so much due to new experi-mental or epidemiological data or new interpretationsof existing data, but rather because of philosophicalapproach and appraisal of existing and future radia-tion protection. This results from an atmosphere ofconstantly changing societal conditions and publicattitudes.Why is the 1980 BEIR-III ~ report different?

The Report~ of the Committee on the Biological Ef-fects of Ionizing Radiation is the record of the deliber-ations of an expert scientific advisory committee ofthe National Academy of Sciences-National ResearchCouncil. It deals with the scientific basis of the healtheffects in human populations exposed to low levels ofionizing radiation. The 1980 ReporV broadly encom-passes two areas: 1) It reviews the current scientificknowledge -- epidemiological surveys and laboratoryanimal experiments -- relevant to radiation exposureof human populations and to the delayed or latehealth effects of low-level radiation; 2) It evaluatesand analyzes these late health effects -- both somaticand genetic -- in relation to the risks to health fromexposure to low-level radiation. The Committee con-sisted of 22 members, selected for their expertise inareas of biology, biophysics, biostatistics, epidemiol-ogy, genetics, mathematics, medicine, physics, publichealth, and the radiological sciences. The reports ofthe BEIR Committeet-~ have, in the past, become valu-able texts for the scientific basis for development ofappropriate and practical radiation protection stan-dards and for decision-making for public healthpolicy.

The 1972 BEIR-I Report2 and the 1980 BEIR-IIIReportI may differ from one or more of the otherradiation advisory committee reports of theUNSCEAR,4.5 the ICRP,~7 the NCRP,8,~ and of othernational councils and committees, in a number ofimportant ways.

First, the BEIR Reports~~ are fashioned and writ-ten as readable, usable scientific documents for those

c International Committee on Radiological Protection, Sutton, Sur-

rey, England.dNational Council on Radiation Protection and Units, Washington,D.C., U.S.A.

e National Radiological Protection Board, United Kingdom, Hat-well, Oxon, England.

societal activities concerned with radiation health.The conclusions, recommendations, and detailed ap-pendices are written in a straightforward scientificmanner, to be read and understood by scientists, phy-sicians, and government decision-makers alike.

Second, the BEIR Committee~~ does not set radia-tion standards or public health policy. The commit-tee’s report are presented to be useful to those re-sponsible for the evaluation of risks and for decision-making concerning regulatory programs and publichealth policy involving radiation. There is no intent toset the direction for those decision-makers who mustconsider the strengths and limitations of science andtechnology along with the relevant societal and eco-nomic conditions in the development and execution ofsuch regulatory programs. In this regard, the BEIRReports1~ suggest that those responsible for setting ra-diation protection standards take into account thecurrent societal needs, so that such standards estab-lish levels of radiation exposure which reflect society’sneeds at any given time- particularly for generalpopulation and occupational exposure from medicalapplications and from nuclear energy -- not necessar-ily those which are absolutely safe.

Third, available epidemiological surveys and labo-ratory animal data are reviewed and assessed for theirvalue in estimating numerical risk coefficients for latehealth effects -- particularly cancer and geneticallyrelated ill-health -- in human populations exposed tolow-level radiation. Therefore, the BEIR Reports~.2 usea practical format for decision-makers. The numericalrisk coefficients estimated are presented in terms ofprobability, with probable upper and lower bound-aries derived solely from the scientific facts, epi-demiological and experimental data, and the scientifichypotheses and assumptions on which they are based.

Finally, the BEIR Reports1~ address the continuedneed to assess and evaluate both the benefits and risksfrom those activities involving radiation. In our so-ciety, such assessment is essential for societal decision-making when establishing appropriate and achievableradiation protection standards based on risk evalua-tion. Decisions can and must be made about the valueand cost of technological and societal programs forrisk reduction by decreasing levels of radiation expo-sure. This includes decisions concerning alternativemethods involving nonradiation activities, comparingtheir costs to human health and to the environment~

with other methods.What are the important biological effects of low-level radiation?

Here, I shall discuss primarily those delayed or latehealth effects in humans following exposure to low-LET radiation,* X rays, and gamma rays from radio-active sources, and, to a much lesser extent, to high-

* (linear energy transfer

PEDIATRIC DENTISTRY: Volume 3, Special Issue 2 401

LET neutron and alpha radiations, since these are theionizing radiations most often encountered in thenuclear industry and in medicine. Briefly, low-levelradiation can affect the cells and tissues of the body inthree important ways.

First, if the macromolecular lesion occurs in one ora few cells, such as those of the blood-forming tissues,the irradiated cell can occasionally transform into acancer cell, and, after a period of time, there is an in-creased risk of cancer developing in the exposed indi-vidual. This biological effect is carcinogenesis; and thehealth effect, cancer. Second, if the embryo or fetus isexposed during gestation, injury can occur to the pro-liferating and differentiating cells and tissues, leadingto abnormal growth. This biological effect is terato-genesis; and the health effect, developmental abnor-mality in the newborn. Third, if the macromolecularlesion occurs in the reproductive cell of the testis orthe ovary, the hereditary genome of the germ cell canbe altered, and the injury can be expressed in thedescendants of the exposed individual. This biologicaleffect is mutagenesis; and the health effect, geneti-cally related ill-health.

There are a number of other important biologicaleffects of ionizing radiation, such as induction of cata-racts in the lens of the eye or impairment of fertility,but these three important late effects -- carcino-genesis, teratogenesis and mutagenesis m stand out asthose of greatest concern. This is because a consider-able amount of scientific information is known fromepidemiological studies of exposed human populationsand from laboratory animal experiments. Further-more, we believe that any exposure to radiation, evenat low levels of dose, carries some risk of such deleteri-ous effects. And as the dose of radiation increasesabove very low levels, the risk of these deleterioushealth effects increases in exposed human popula-tions. It is these latter observations that have beencentral to the public concern about the potentialhealth effects of low-level radiation, and to the task ofestimating risks and of establishing standards for pro-tection of the health of exposed populations. Indeed,all reports of expert advisory committees on radiationare in close agreement on the broad and substantiveissues of such health effects.

What is known about the important health effectsof low-level radiation?

A number of very important observations on thelate health effects of low-level radiation have now con-vincingly emerged, about which there is general agree-ment. These observations are based primarily onevaluation of epidemiological surveys of exposedhuman populations, on extensive research in labora-tory animals, on analysis of dose-response relation-ships of carcinogenic, teratogenic and genetic effects,

and on known mechanisms of cell and tissue injury invivo and in vitro.

First, cancer-induction is considered to be the

most important late somatic effect of low-dose ioniz-ing radiation. Solid cancers arising in the various or-gans and tissues of the body, such as the female breastand the thyroid gland, rather than leukemia, are theprincipal late effects in individuals exposed to radia-tion. The different tissues appear to vary greatly intheir relative susceptibility to cancer-induction byradiation. The most frequently occurring radiation-in-duced cancers in man include, in decreasing order ofsusceptibility: the female breast; the thyroid gland,especially in young children and in females; the blood-forming tissues; the lung; certain organs of thegastrointestinal tract; and the bones. Influences affect-ing the cancer risk include: age at the time of irradia-tion, and at the time of expression of the disease, sex,and radiation factors and types -- LET and RBE -- af-fecting the cancer risk.

Second, the effects of growth and development inthe irradiated embryo and fetus are related to the ges-tational stage at which exposure occurs. It appearsthat a threshold level of radiation dose and dose ratemay exist below which gross teratogenic effects willnot be observed. However, these dose levels would varygreatly depending on the particular developmental ab-normality and on the radition types and qualities.

Third, estimation of the radiation risks of geneti-cally related ill-health are based mainly on laboratoryanimal, observations -- primarily from laboratorymouse experiments -- because of the paucity of dataon exposed human populations. Our knowledge of fun-damental mechanisms of radiation injury at the ge-netic level is far more complete than, tbr example, ofmechanisms of radiation carcinogenesis, thereby per-mitting greater assurance in extrapolating informa-tion on genetic mutagenesis from laboratory animalsto man. With new information on the broad spectrumand incidence of genetically related ill.health in man,such as mental retardation and diabetes, the risk ofradiation mutagenesis in man affecting future genera-tions takes on new and special consideration.

What is not known about these health effects oflow-level radiation?

In spite of a thorough understanding of these latehealth effects in exposed human populations, there isstill a considerable amount we do not know about thepotential health effects of low-level radiation.

First, we do not know what the health effects are atdose rates as low as a few hundred millirem per year,that is, a few factors above natural background radia-tion exposure. It is probable that if any health effectsdo occur, they will be masked by environmental or

402 RISK ESTIMATION AND DECISION MAKING: Fabrikant

other competing factors that produce similar healtheffects.

Second, the epidemiological surveys of exposedhuman populations are highly uncertain in regard tothe forms of the dose-response relationships for radia-tion-induced cancer in man. This is especially the casefor low-level radiation. Therefore, it has been neces-sary to estimate human cancer risk from low radiationdoses primarily from observations of relatively highdoses, frequently greater than 100 rads. Estimates ofthe cancer risk at low doses appears to depend moreon what is assumed about the mathematical form ofthe dose-response function than on the available epi-demiological data. However, it is not known whetherthe excess cancer risk observed at high-dose levels alsoapplies to low-dose~levels.

Third, we do not have reliable methods for estimat-ing the repair of injured cells and tissues of the bodyexposed to very low doses and dose rates. And further,we do not know how to identify those persons whomay be particuarly susceptible to radiation injury,perhaps on the basis of genetic predisposition.

We also have only very limited epidemiologicaldata on the precise radiation doses absorbed by thetissues and organs of persons in irradiated populationsexposed in the past. Furthermore, we do not know thecomplete cancer incidence in each study population,since new cases of cancer continue to appear with thepassing of time. Accordingly, any estimation of excesscancer risk based on such limited dose-incidence infor-mation must necessarily be incomplete, until the en-tire study population has died from natural or othercauses.

Finally, we do not now know the role of competingenvironmental and other host factors -- biological,chemical or physical n existing at the time of expo-sure or following exposure, which may influence andaffect the carcinogenic, teratogenic, or genetic effectsof low-level radiation.

What are the uncertainties in the dose-responserelationships for relation-induced cancer?

In recent years, a general hypothesis for estimationof excess cancer risk in irradiated human populations,based on theoretical considerations, extensive labora-tory animal studies, and limited epidemiological sur-veys, suggests various and complex dose-response rela-tionships between radiation dose and observed cancerincidence, m5 Among the most widely considered mod-els for cancer-induction by radiation, based on avail-able information and consistent with both knowledgeand theory, takes the complex quadratic form: I(D) (a0 + alD + a2D~) exp (-blD-b2D~), where I is thecancer incidence in the irradiated population at radia-tion dose D in rad, and a0, al, a2, bl and b2 are non-negative constants (Figure 1).

Dose-response model forradiation carcinogenesis

I(D)=(ao+a, D+a2 D2 )e(’~’ D-~ D~ )

Dose, D(rad)Figure 1. Dose-response model for radiation carcinogenesis.

This multicomponent dose-response curve con-tains: 1) initial upward-curving linear and quadraticfunctions of dose, which represent the process ofcancer-induction by radiation; and 2) a modifying ex-ponential function of dose, which is generally consid-ered to represent the competing effects of biochemicaland molecular processes at the subcellular level, lead-ing to cell-killing at high doses, a0 is the ordinate in-tercept of 0 dose, and defines the natural incidence ofcancer in the population, al is the initial slope of thecurve at 0 dose, and defines the linear component inthe low-dose range, a2 is the curvature near 0 dose,and defines the upward-curving quadratic function ofdose. bl and b2 are the slopes of the downward-curv-ing function in the high-dose range, and define theprocesses involved in the cell.killing function.

Analysis of a number of dose-incidence curves forcancer-induction in irradiated populations, both inhumans and in animals, has demonstrated that for dif-ferent radiation-induced cancers only certain of theparameter values of these constants can be theoreti-cally determined.~ However, the extent of the varia-tions in the shapes of the dose-response curves derivedfrom the epidemiological or experimental data doesnot permit direct determination of any of these pre-cise parameter values, or even of assuming theirvalues, or of assuming any fixed relationship betweentwo or more of these parameters. Furthermore, in thecase of the epidemiological surveys, this complex gen-eral dose-response form cannot be universally applied.Therefore, it has become necessary to simplify themodel by reducing the number of parameters whichhave the least effect on the form of the dose-responserelationship in the low-dose range. Such simpler mod-els, with increasing complexity, include the linear, thepure quadratic, the quadratic (with a linear term), and


Figure 2. Shapes of dose-response curves: linear (up-per left); pure quadratic(upper right); quadraticwith a linear term (lowerleft); and multicomponentquadratic with a linearterm and with an exponen-tial modifier (lower right).

.-:::

finally, the multicomponent quadratic form with alinear term and with an exponential modifier (Figure2).

Three limitations constrain precise numerical esti-mation of excess cancer risks of low-level radiation inexposed human populations. First, we lack an under-standing of the fundamental mechanisms of cancer-induction by radiation. Second, the dose reponse datafrom epidemiological surveys are highly uncertain,particularly at low levels of dose. Third, experimentaland theoretical considerations suggest that variousand different dose-response relationships may exist fordifferent radiation-induced cancers in exposed humanpopulations. Nevertheless, these limitations do not re-lieve decision-makers of the responsibility for guidingpublic health policy based on appropriate radiationprotection standards. Accordingly, not only is it essen-tial that quantitative risk estimation be calculated,based on the available epidemiological and radio-biological data, but in addition, for any authoritativecommittee report, such as for the current BEIR-IIIReport,~ it is equally essential that precise explana-tions and qualifications of the assumptions, proce-dures, and limitations involved in the calculation ofsuch risk estimates must be clearly provided.

This has been done explicitly, but not without

much discussion and disagreement among the Com-mittee members, in the current BEIR-III Report~ con-taining the estimates of excess cancer risk. In its finalanalyses, the majority of the members of the BEIRCommittee preferred to emphasize that some experi-mental and human data, as well as theoretical consid-erations, suggest that for exposure to low-LET radia-tion, such as X rays and gamma rays, at low doses, thelinear model probably leads to overestimates of risk ofmost radiation-induced cancers in man, but that themodel can be used to define the upper limits of risk.Similarly, a majority of the members of the Commit-tee believed that the pure quadratic model may beused to define the lower limits of risk from low-dose,low-LET radiation. The Committee generally agreed,that for exposure to high-LET radiation, such asneutrons and alpha particles, linear risk estimates forlow doses are less likely to overestimate the risk andmay, in fact, underestimate the risk.

What is the controversy over low-level radia-tion?

The estimation of the cancer risk of exposure tolow-level radiation is said to be clouded by scientificdispute. In particular, there appears t~) be disagree-ment among some scientists as to the effects of very


low levels of radiation, even as low as our natural ra-diation background. Some say this was the centralissue of controversy within the BEIR-III Committee,which had been highlighted in scientific periodicals,such as Nature and Science, and in the news media,such as The New York Times.

While there is no precise definition of low-level

exposure, many scientists would generally agree thatlow-level radiation is that which falls within the doserange considered permissible for occupational expo-sure. According to accepted standards,16 5 rem peryear to the whole body would be an allowable upperlimit of low-level radiation dose for the individual ra-diation worker. With this as the boundary conditionfor occupational exposure, it could very well be con-eluded that most of the estimated delayed cancercases which might be associated with a hypotheticalnuclear reactor accident, or even long periods of occu-pational exposure among radiation workers, would beconsidered by some scientists to be caused by expo-sures well below these allowable occupational limits.Furthermore, if it is assumed that any extra radiationabove natural background, however small, causes ad-ditional cancer; some extra cancers will inevitably re-sult if millions of people are exposed. Other scientistsstrongly dispute this, and £Lrmly believe that low-levelradiation is nowhere near as dangerous as their col-leagues contend. Central to this dispute is the factthat cancers induced by radiation are indistinguish-able from those occurring naturally; hence, their exist-ence can be inferred only on the basis of a statisticalexcess above the natural incidence. Since such healtheffects, if any, are so rarely seen under low-level radia-tion because the exposures are so small, the issue ofthis dispute may never be resolved B it may bebeyond the abilities of science and mathematics todecipher.

It is just this type of controversy that was at theroot of the division among scientists within the 1980BEIR-III Committee.17,18 There is little doubt that theCommittee’s most difficult task was to estimate thecarcinogenic risk of low-dose, low-LET, whole-bodyradiation. Here, to the disquiet of some of the mem-bers of the Committee, emphasis was placed almostentirely on the limited number of human epi-demiological studies, since it was felt by the majorityof the members that little information from labora-tory animal and biophysical studies could be applieddirectly to man. Therefore, as the earlier 1972 BEIR-IReport~ had done, some scientists of the 1980 BEIR-III Committee considered it necessary to adopt alinear hypothesis of dose-response to estimate thecancer risk at very low-level radiation exposure whereno human epidemiological data are available. It isassumed the same proportional risks are present at

low levels as at high levels of radiation. This positionimplied that even very small doses of radiation arecarcinogenic, a finding that could force the U.S. Envi-ronmental Protection Agency to adopt stricter healthstandards to protect against occupational and generalpopulation exposure for one example. Other scientistsin the Committee did not accept this position, and be-lieve this was an alarmist approach. When there is nohuman epidemiological evidence at low doses of low-LET radiation, these scientists preferred to assumethat the risks of causing cancer are proportionallylower.

Let us look at some of the problems. In its delibera-tions, the BEIRoIII Committee concluded two impor-tant observations: 1) It was not yet possible to makeprecise low-dose estimates for cancer-induction by ra-diation because the level of risk was so low that itcould not be observed directly in man; and 2) therewas great uncertainty as to the dose-response functionmost appropriate for extrapolating to the low-dose re-gion. In studies of exposed animal and human popula-tions, the shape of the dose-response relationships forcancer-induction at low doses may be practically im-possible to ascertain statistically. This is because thepopulation sample sizes required to estimate or test asmall absolute cancer excess are extremely large. Spe-cifically, the required sample sizes are approximatelyinversely proportional to radiation dose, and if 1,000exposed and 1,000 control persons are required in eachgroup to test this cancer excess adequately at 100 rads,then about 100,000 in each population group are re-quired at 10 rads, and about 10,000,000 in each groupare required at 1 rad. Thus, it appears that experimen-tal evidence and theoretical considerations are muchmore likely than empirical epidemiological data toguide the choice of a dose-response function forcancer-induction.

In this dilemma and after much disagreement

among some of its members, the majority of the mem-bers of the 1980 BEIR-III Committee chose to adoptas a working model for low-dose, low-LET radiationand carcinogenesis the linear quadratic (i.e., quadratic function with a linear term in the low-doseregion) dose-response form with an exponential termto account for the frequently observed turndown ofthe curve in the high-dose region. However, in apply-ing this multicomponent model, only certain of its de-rivatives, including the linear, the linear-quadratic,(i.e., the quadratic with linear term), and the purequadratic functions, could prove practical for purposesof estimation of cancer risk (Figure 2). For the final re-port, in estimating the excess cancer risk from low-dose low-LET radiation, a majority of the BEIR-IIICommittee members preferred the linear-quadraticdose-response model, felt to be consistent with epi-


demiological and radiobiological data, in preference tomore extreme linear or pure quadratic dose-responsemodels.

In the 1972 BEIR-I Reports the cancer risk esti-mates for whole-body radiation exposure were derivedfrom linear model average excess cancer risk per radobserved at doses generally of a hundred or more rads.These risk estimates were generally criticized on thegrounds that the increment in cancer risk per rad maywell depend on radiation dose, and that the truecancer risk at low doses may therefore be lower orhigher than the linear model predicts. 9 In laboratoryanimal experiments, the dose-response curves for ra-diation-induced cancer can have a variety of shapes.As a general rule, for low-LET radiation, the slope ofthe curve increases with increasing dose. However, athigh doses, the slope often decreases and may even be-come negative. Dose-response curves may also varywith the kind of cancer, with animal species, and withdose rate. On the basis of the experimental evidenceand current microdosimetric theory, therefore, thecurrent BEIR-III Committee could quite reasonablyadopt as the basis for its consideration of dose-response models the quadratic from with a linear termin the low-dose region, and with an exponential termfor a negative slope in the high-dose region (Figure 1).

On the other hand, in large part, the availablehuman data from the large body of epidemiologicalstudies fail to suggest any specific dose-responsemodel, and are not sufficiently reliable to discriminateamong a priori models suggested by the experimentaland theoretical studies. However, there appears atpresent to be certain exceptions from the human expe-rience. For example, cancer of the skin is not observedat low radiation doses2 Dose-response relationshipsfor the Nagasaki leukemia data appear to have posi-tive curvature. ~ The incidence of breast cancer in-duced by radiation seems to be adequately describedby a linear dose-response mode2~

In the Committee’s attempts to apply derivativesof the multicomponent, linear-quadratic dose-response model to the epidemiological data, simplifi-cation was necessary to obtain statistically stable riskestimates in many cases. Certain members of theBEIR-III Committee were passionately divided onthis matter; some strongly favored the linear model,others favored the pure quadratic form.~7.~8 A furthermodification of the linear-quadratic form was as-sumed with the linear and quadratic components to beequivalent at some dose -- which was consistent withthe epidemiological data and the radiobiological evi-dence -- and avoided dependence on either of the twoextreme forms.~*~6

What are the uncertainties in estimation of thecarcinogenic risk in man of low-level radiation?

The quantitative estimation of the carcinogenic

risk of low-dose, low-LET radiation is subject to nu-merous uncertainties. The greatest of these concernsthe shape of the dose-response curve. Others includethe length of the latent period, the RBE for fastneutrons and alpha radiation relative to gamma andX radiation, the period during which the radiationrisk is expressed, the model used in projecting riskbeyond the period of observation, the effect of doserate or dose fractionation, and the influence of differ-ences in the natural incidence of specific types ofcancer. Uncertainties are also introduced by the bio-logical risk characteristics of humans, for example, theeffect of age at irradiation, the influence of any dis-ease for which the radiation was given therapeutically,and the influence of length of observation or follow-upof the study populations. The collective influence ofthese uncertainties diminishes the credibility of anyestimates of human cancer risk that can be made forlow-dose, low-LET radiation.

What are the sources of epidemiological data forthe estimation of excess cancer risk in exposedhu~nan populations?

The tissues and organs about which we have themost reliable epidemiological data on radiation-induced cancer in man {obtained from a variety ofsources from which corroborative risk coefficientshave been estimated), include the bone marrow, thethyroid, the breast, and the lung. The data on boneand the digestive organs are preliminary at best, anddo not approach the precision of the others. For sev-eral of these tissues and organs, risk estimates are ob-tained from very different epidemiological surveys,some followed for over 25 years including adequatecontrol groups. There is impressive agreement whenone considers the lack of precision inherent in the sta-tistical analyses of the case-finding and cohort studypopulations, variability in ascertainment and clinicalperiods of observation, age, sex and racial structure,and different dose levels, and constraints on data fromcontrol groups.

The most reliable and consistent data have beenthose of the risk of leukemia which come from; theJapanese atomic bomb survivors, ~ the ankylosingspondylitis patients treated with X-ray therapy inEngland and Wales, ~ the metropathia patientstrea~ed with radiothrapy for benign uterine bleeding,~

the tinea capitis patients treated with radiation forringworm of the scalp, ~ and early radiologists.~

There is evidence of an age-dependence and a dose-dependence, a relatively short latent period of a fewyears, and a relatively short period of expression, some10 years. This cancer is almost always fatal.

The epidemiological data on thyroid cancer aremore complex. These surveys include the large seriesof children treated with radiation to the neck and


mediastinum for enlarged thymus,~ children treatedto the scalp for tinea capitis, ~.~ and the Japaneseatomic bomb survivors ~ and Marshall Islanders ~ ex-posed to nuclear explosions. There is an age-depend-ence and a sex-dependence -- children and females ap-pear to be more sensitive. Although the induction rateis high, the latent period is relatively short, and it isprobable that no increased risk will be found in futurefollow-up of these study populations. In addition,most tumors are either thyroid nodules, or benign ortreatable tumors, and only about 5% of the radiation-induced thyroid tumors are fatal.

The epidemiological surveys on radiation-inducedbreast cancer in women~3.~1 primarily include womenwith tuberculosis who received frequent fluoroscopicexaminations for artificial pneumothorax,~ postpar-tum mastitis patients treated with radiotherapy,~ andthe Japanese atomic bomb survivors in Hiroshima andNagasaki.~ There is an age-dependence and dose-dependence, as well as a sex-dependence, and thelatent period is long, some 20 to 30 years. About halfof these neoplasms are fatal.

The epithelial lining of the bronchus and lung is acomplex tissue as regards radiation dose involving pa-rameters of the special physical and biological charac-teristics of the radiation quality. The epidemiologicalsurveys include the Japanese atomic bomb survivors,~

the uranium miners in the United States and Cana-da,3L~ and the ankylosing spondylitis patients inEngland and Wales.~= There is some evidence of anage-dependence from the Japanese experience, and rel-atively long latent period. This cancer is almostalways fatal.

The risk of radiation-induced bone sarcoma, basedprimarily on surveys of the radium and thorium pa-tients who had received the radioactive substances formedical treatment or ingested them in the course oftheir occupations,=,~ is low. For all other tumors aris-ing in various organs and tissues of the body, valuesare extremely crude and estimates are preliminary atbest.

There is now a large amount of epidemiologicaldata from comprehensive surveys from a variety ofsources. These data indicate that leukemia is nolonger the major cancer induced by radiation -- thatsolid cancers are exceeding the relative incidence ofradiation-induced leukemia.5 That is, in view of thelong latent period after some 30 years or more follow-ing radiation exposure, the risk of excess solid cancersis many times the risk of excess leukemia. But theserisk estimates must remain very crude at the presenttime, since they do not take into account any lack ofprecision in certain of the epidemiological studies, par-ticularly as regards radiation dose distribution, ascer-tainment, latency periods, and other physical and bio-logical parameters. The BEIR,~,~ the UNSCEAR,4,5 and

the ICR1~.7 Reports have estimated the risk from low-LET, whole-body exposure in different ways. Basedon the epidemiological surveys carefully followed,with adequate control study populations, a crude fig-ure of the total lifetime absolute risk of radiation-induced cancer deaths can be derived. This estimatefor low-LET radiation, delivered at low doses, wouldbe less than about 100 excess cases per million personsexposed per tad. But this figure could very well be anover-estimate of the true risk, and the actual numberof excess cancer cases may be much lower.~.5 Althoughany such numerical estimate must be considered un-reliable, it does provide a very rough figure for compa-rison with other estimates of avoidable risks, or volun-tary risks, encountered in everyday life.What are the risk estimates of radiation-inducedcancer in man?

The chief sources of epidemiological data currentlyfor risk estimation of radiation-induced cancer in manare the Japanese atomic bomb survivors exposed towhole-body irradiation in Hiroshima and Nagasaki,s°

the patients with ankylosing spondylitis =.= and otherpatients who were exposed to partial body irradiationtherapeutically, ~-~.~ or to diagnostic radiographs andthe various occupationally-exposed populations~~

such as uranium miners and radium dial painters.Most epidemiological data do not systematically coverthe range of low to moderate radiation doses whichare fairly reliable in the Japanese atomic bomb sur-vivor data. Analysis in terms of dose-response, there-fore, necessarily rely greatly on the Japanese data.The substantial neutron component of dose inHiroshima, and its correlation with gamma dose, limitthe value of the more numerous Hiroshima data forthe estimation of cancer risk from low-LET radiation.The Nagasaki data, for which the neutron componentof dose is small, are less reliable for doses below 100rads.

The 1980 BEIR-III Report~ chose three exposuresituations for illustrative computations of the lifetimecancer risk of low-dose, low-LET whole-body radia-tion: 1) a single exposure of representative (life-table)population to 10 rads; 2) a continuous, lifetime expo-sure of a representative (life-table) population to 1 tadper year; and 3) an exposure to 1 rad per year overseveral age intervals approximating conditions of oc-cupational exposure. These three exposure situationswere not chosen to reflect any circumstances thatwould normally occur, but to embrace the areas ofconcern -- general population and occupational expo-sure, and single and continuous exposure. These doselevels were substantially different from the only expo-sure situation chosen for the illustrative computationby the 1972 BEIR-I Committee, where 100 mrem peryear was the level selected.~ Some members of the cur-rent BEIR-III Committee strongly felt that below

PEDIATRIC DENTiS~RY: Volume 3, Special Issue 2 407

these three dose levels, which were arbitrarily chosenfor the 1980 Report,I the uncertainties of extrapola-tion to very low dose levels were too great to justifyany attempt at risk estimation. Other members feltjust as strongly that risk estimates for cancer-induc-tion by radiation could be reliably calculated at doselevels of 1 rad or even much less. These differenceswere never satisfactorily settled. The selected annuallevel of chronic exposure of 1 rad per year, althoughonly one-fifth the maximum permissible dose for occu-pational exposure, is nevertheless consistent with theoccupational exposure experience in the nuclear in-dustry. The 1969-1971 U.S. life-table was used as thebasis for the calculations. The expression time wastaken as 25 years for leukemia and the remaining yearsof life for other cancers. Separate risk estimates weremade for cancer mortality and for cancer incidence.

In the absence of any increased radiation exposure,among one million persons of life-table age and sexcomposition in the United States, about 164,000 per-sons would be expected to die from cancer accordingto present cancer mortality rates. For a situation inwhich these one million persons are exposed to a singledose increment of 10 rads of low-LET radiation, thelinear-quadratic dose-response model predicts in-creases of about 0.5% and 1.4% over the normal expec-tation of cancer mortality, according to the projectionmodel used. For continuous lifetime exposure to 1 radper year, the increase in cancer mortality, according tothe linear-quadratic model, ranges from 3% to 8% overthe normal expectation, depending on the projectionmodel (Table 1).

Table 2 compares the cancer risk following expo-sure to 10 rads, calculated according to three differentdose-response models, viz., the linear-quadratic, thelinear, and the quadratic. The upper and lower limitsof these cancer mortality risk estimates suggest a very

Table 1. Estimated excess mortality per million personsfrom all forms of cancer, linear-quadratic dose-responsemodel for Low-LET radiation.1

Absolute-Risk Relative-RiskProjection Projection

Model Model

Single exposure to 10 fads:Normal expectation 163,800 163,800Excess cases: number 766 2,255

% of normal 0.47 1.4

Continuous exposure to1 rad/yr, lJ£etime:

Normal expectation 167,300 167,300Excess cases: number 4,751 12,920

% of normal 2.8 7.7

wide range or envelope of values which may differ byas much as an order of magnitude or more. The uncer-tainty derives mainly from the dose-response modelsused, from the alternative absolute and relative pro-jection models, and from the sampling variation in thesource data. The lowest risk estimates u the lowerbound of the envelope -- are obtained from the purequadratic model; the highest -- the upper bound of theenvelope -- from the linear model; and the linear-quadratic model provides estimates between these twoextremes.

Table 3 compares the 1980 BEIR-III Report~ cancermortality risk estimates with those of the 1972 BEIR-I Report 2 and the 1977 UNSCEAR Report. 5 To dothis, it was most convenient to express them as cancerdeaths per million persons per rad of continuous life-time exposure. For continuous lifetime exposure to 1rad per year, the linear-quadratic dose-response model for low-LET radiation yields riskestimates considerably below the comparable linear-model estimates in the 1972 BEIR-I Report;~ the dif-


Table 3. Comparative estimates ofthe lifetime ~ of cancer mort~lityinduced by Low-LET radiation --excess deaths per million, average Source of Estimatevalue per tad by projection model,dose-response model, and type of BEIR, 1980bexposurey

1972 BEIR report factors

UNSCEAR 1977

Projection

Dose- Single Exposure to Continuous LifetimeRepor~se 10 ra~ Exposure to i rad/yrModels Absolute Relate’re Absolute Relat£ve

LQ-L, LQ.L 77 226 67 182

Linear 117 621 115 568

Linear 75-175

a) For BEIR 1980, the first model is used for leukemia, thesecond for other forms of cancer. The corresponding esti-mates when the other models are used (thereby prodding anenvelope of risk estimates)b) The values s_re average values per tad, and are not to betaken as estimates at only I tad of dose.

ferences mainly reflect changes in the assumptionsmade by the two BEIR Committees almost a decadeapart. The 1980 BEIR-III Committee preferred alinear-quadratic rather than linear dose-responsemodel for low-LET radiation, and did not assume afixed relationship between the effects of high-LETand low-LET radiation (which was based on the Japa-nese atomic bomb survivor studies). Furthermore, the1980 BEIR-III ReportI cancer risk estimates do not,as in the 1972 BEIR-I Report,~ carry through to theend of life the very high relative-risk coefficients ob-tained with respect to childhood cancers induced inutero by radiation.

There is a good deal of reluctance by some scien-tists to introduce cancer-incidence data for purposesof radiation-induced cancer risk estimation. Cancermortality data are considered far more reliable thancomparable incidence data, and thus, cancer incidencerisk estimates are less firm than mortality estimates.However, the incidence of radiation-induced cancer isconsidered by many scientists and by decision-makersalike, to provide a more complete expression of thetotal social cost of radiation-induced cancer in manthan does mortality. The 1980 BEIR-III Committeechose to introduce cancer-incidence data for risk esti-mation for the first time in any report, and also ap-plied a variety of dose-response models and severaldata sources. For continuous lifetime exposure low-LET, whole-body, to 1 rad per year, for example, andbased on the linear-quadratic model, the increasedrisks expressed as percent of the normal incidence ofcancer in males were about 2% to 6%, depending onthe projection model. The various dose-response mod-els produced estimates that differed by more than anorder of magnitude, whereas the different data sourcesgave broadly similar results. Risks for females weresubstantially higher than those for males, due prima-rily to the relative importance of radiation-inducedbreast and thyroid cancer.

L-L, L-L 167 501 158 430Q-L, Q-L 10 28

Estimates of excess cancer risk for individualorgans and tissues depend in large part on partial-body irradiation and use a much wider variety of epi-demiological data sources. Except for leukemia andbone cancer, estimates for individual sites of cancercan be made only on the basis of the linear model, andall risk coefficients are estimated as the number of ex-cess cancer cases per year per million persons exposedper rad. For leukemia, the linear-quadratic modelyielded about 1.0 to 1.4 excess leukemia cases, forfemales and males respectively. For example linear-model estimates for solid cancers were: for thyroid inmales, about 2, and in females, about 6; for femalebreast, about 6; and for lung, about 4. These risk coef-ficients derive largely from epidemiological data inwhich exposure was at high doses. These values may,in some cases, overestimate risk at low doses.

What is known about the teratogenic effects oflow-level radiation?

Developing mammals, including man, are particu-larly sensitive to radiation during their intrauterineand early postnatal life. The developmental effects ofradiation on the embryo and fetus are strongly relatedto the stage at which exposure occurs. Most informationcomes mainly from laboratory animal studies, but thehuman data are sufficient to indicate qualitative corre-spondence for developmentally equivalet stages.~,~1

Radiation during preimplantation stages probablyproduces no abnormalities in survivors, owing to thegreat developmental plasticity of very early mam-malian embryos. Radiation at later stages may, how-ever, produce morphologic abnormalities, general orlocal growth retardation, or functional impairments, ifdoses are sufficient. Obvious malformations are par-ticularly associated with irradiation during the periodof major organogenesis, which in man extends approxi-mately from the second through the ninth week from


conception. More restricted morphologic and func-tional abnormalities and growth retardations domi-nate the spectrum of radiation effects produced dur-ing the fetal and early postnatal periods. Some ofthese effects can be apparent at birth, and others mayshow up later; and subtle functional damage cannotbe adequately measured with available techniques.

Because the central nervous system is formed dur-ing a relatively long period in human development,such abnormalities as microcephaly and mentalretardation figure prominently among the list ofradiation effects reported in man.

In laboratory animals, developmental abnor-malities (CNS injury and oocyte killing) have beenobserved at doses below 10 rads. ~° The experimentaldata can be used with some confidence to fill in gapsin the human experience, particularly with respect toextrapolations to low exposure levels, where it is verydifficult to obtain direct evidence in human popula-tions. Atomic-bomb data for Hiroshima show that thefrequency of small head size was increased by acute airdoses in the range of 10-19 rads kerma (average fetaldose, gamma rays at 5 rads plus neutrons at 0.4 rad)received during the sensitive period, and suggest thatit was also increased in the 1-9 kerma range (averagefetal dose, 1.3 rads gamma plus 0.1 rad neutrons). AtNagasaki, where almost the entire kerma was due togamma rays, there was no increase in the frequency ofsmall head size at air doses below 150 rads kerma.~

Because a given gross malformation or functionalimpairment probably results from damage to morethan a single target, the existence of a threshold radia-tion dose below which that effect is not observed maybe predicted. There is evidence of such thresholds, butthey vary widely, depending on the abnormality. Low-ering of the dose rate diminishes the damage. Further-more, exposure protraction can reduce dose effective-ness by decreasing to below the threshold the portionof the dose received during a particular sensitiveperiod.

What is known about the genetic effects of low-level radiation?

Because radiation-induced transmitted genetic ef-fects have not been demonstrated in man, and becauseof the likelihood that adequate information will notsoon be forthcoming, estimation of genetic risks mustbe based on laboratory animal data. This entails theuncertainty of extrapolation from the laboratorymouse to man. However, there is information on thenature of the basic lesions, which are believed to besimilar in all organisms. Some of the uncertainties inthe evaluation of somatic effects are absent in the esti-mation of genetic risk.

The genetic disorders that can result from radia-tion exposure are: 1) those which depend on changes

in individual genes (gene mutations or small dele-tions); and 2) those which depend on changes chromosomes, either in total number or in gene ar-rangement (chromosomal aberrations). Gene muta-tions are expected to have greater health consequencesthan chromosome aberrations. At low levels of expo-sure, the effects of radiation in producing either kindof genetic change is proportional to dose. Risk esti-mates are based either on experimental[ findings at thelowest doses and dose rates for which reliable datahave been obtained or on adjustment of the observeddata obtained at high doses and dose !rates by a dose-rate reduction factor. For low doses and dose rates, alinear extrapolation from fractionated-dose and low-dose-rate laboratory mouse data continues to consti-tute the basis for estimating genetic risk to the generalpopulation.1,2 Genetic-risk estimates are expressed aseffects per generation per rem, with appropriate cor-rections for special situations, such as exposures ofsmall groups to high-LET radiation.

Two methods may used to estimate the incidence ofdisorders caused by gene mutations2 One method esti-mates the incidence expected after the continuous ex-posure of the population over a large number of gener-ations. The other method estimates the incidence ofdisorders expected in a single generation after the ex-posure of the parents. By the first method, it is esti-mated that about 1-6% of all spontaneous mutationsthat occur in humans are due to background radia-tion. A small increase in radiation exposure abovebackground leads to a correspondingly small relativeincrease in the rate of mutation. The numerical rela-tionship of rates of induced and spontaneous mutationis relative-mutation-risk factor, that is, the ratio ofthe rate of mutations induced per rem to the spon-taneous rate. The reciprocal of the rd[ative-mutation-risk factor is the "doubling dose," or the amount of ra-diation required to produce as many mutations as arealready occurring spontaneously. The estimated rela-tive mutation risk for humans is 0.02-0.004 per rem (ora doubling dose of 50-250 rem). After many genera-tions of increased exposure to radiation, it is expectedthat human hereditary disorders thai; are maintainedin the population by recurrent gene mutation wouldshow a similar increase in incidence.

Table 4 lists the current 1980 BEIR-III Report1

risk estimates of the potential genel~ic effects of anaverage population exposure of 1 rem per 30-year gen-eration. In the first generation, it is estimated that 1rem of parental exposure throughout the general pop-ulation will result in an increase of 5-75 additionalserious genetic disorders per million liveborn off-spring. Such an exposure of 1 rem received in eachgeneration is estimated to result, at genetic equilib-rium, in an increase of 60-1,100 serious genetic disor-ders per million liveborn offspring. The ranges of the


Table 4. Genetic effects of an average population exposureof i tern per 30-year generation.I

Type of Current IncidenceGenetic in i Million Live-

Disordera born Offspring.

Effect of rem per Generationper Million Livebom Offspring.First Generationb Equilibziumc

Autosomal dominant 10,000and X-linked

Irregularlyinherited 90,000

Recessive 1,100

Chromosomalaberrations(congenitalmalformations)

40-200

5-65 20-900

Very few Slowlyincreases

Less than 10 Increasesslightly

a) Includes diseases that cause serious handicap at some timeduring lifetime, b) Estimated directly from measured phenotypicdamage or from observed cytogenetic effects, c) Estimated by therelative-mutation-risk method.

risk estimates emphasize the limitations of current un-derstanding of genetic effects of radiation on humanpopulations. Within this range of uncertainty, how-ever, the risk is nevertheless small in relation to cur-rent estimates of the incidence of serious human dis-orders of genetic origin -- roughly 11% of livebornoffspring, that is, approximately 107,000 cases permillion liveborn.

Genetic risk estimates are based on induced disor-ders judged to cause serious genetic ill-health at sometime during life. Some disorders are obviously moreimportant than others. In contrast with somatic ef-fects, which occur only in the persons exposed, geneticdisorders occur in descendants of exposed persons andcan often be transmitted to many future generations.The major somatic risk estimates are concerned withinduced cancers. Although many of these are fatal,some are curable, such as most thyroid cancers, butentail the risk and costs of medical care and disability.Somatic effects also include developmental abnor-malities of varied severity caused by fetal or embry-onic exposure. Comparisons of genetic and somatic ef-fects must take into account ethical or socioeconomicjudgments. It is extremely difficult to compare the so-cietal impact of a cancer with that of a serious geneticdisorder.1

What are the implications of numerical risk esti-marion and decision-making for radiation protec-tion and public health policy?

In its evaluation of the epidemiological surveys andthe laboratory animal data, the national and inter-national committees on radiation carefully review and

assess the value of the available scientific evidence forestimating numerical risk coefficients for the healtheffects in human populations exposed to low-level ra-diation. Such devices require scientific judgment andassumptions based on the available data only, andnecessarily and understandably lead to some disagree-ment not only outside the committee room, butamong committee members as well. But such disputesand disagreements center not on scientific facts andnot on existing epidemiological or experimental data,but rather on the assumptions, interpretations, andanalyses of the available facts and data.

The present scientific evidence and the interpreta-tion of available epidemiological data can draw somefirm conclusions on which to base scientific publichealth policy for radiation protection standards. Thesetting of any permissible radiation level or guide forlow-level exposure remains essentially an arbitraryprocedure. Any lack of precision minimizes neither theneed for setting responsible public health policies, northe conclusion that such risks are extremely smallwhen compared with alternative options, and thoseaccepted by society as normal everyday hazards. Sincesociety benefits from the necessary activities of energyproduction and medical care, it is apparent that itmust also establish appropriate standards and con-trolling procedures which continue to assure that itsneeds and services are being met with the lowest pos-sible risks.

In a third century of inquiry, including some of themost extensive and comprehensive scientific efforts onthe health effects of an environmental agent, much ofthe important information necessary for determina-tion of radiation protection standards is now becom-ing available to decision-makers for practical and re-sponsible public health policy. It is now assumed thatany exposure to radiation at low levels of dose carriessome risk of deleterious health effects. How low thelevel may be, the probability or magnitude of the riskat very low-levels of dose still are not known and mayremain so. Radiation and the public health, when itinvolves the public health, becomes a broad societalproblem and not solely a scientific one. Decisions con-cerning this problem will be made by society, mostoften by men and women of law and government. Ourbest scientific knowledge and our best scientific adviceare essential for the protection of the public health,for the effective application of new technologies inmedicine and industry, and for guidance in the pro-duction of nuclear energy. Unless man wishes to dis-pense with those activities which inevitably involveexposure to low-levels of ionizing radiations, he mustrecognize that some degree of risk to health, howeversmall, exists. In the evaluation of such risks from ra-diation, it is necessary to limit the radiation exposureto a level at which the risk is acceptable both to the


individual and to society. A pragmatic appraisal of

how man wishes to continue to derive the benefits ofhealth and happiness from activities involving ioniz-

ing radiation -- considering the everchanging publicattitudes and our resource-limited society -- is the

present and future task which lies before each expert

advisory committee on the biological effects of ioniz-

~ng radiation concerned with risk assessment and deci-

sion-making.

Research supported by the Office of Health and EnvironmentalResearch of the U.S. Department of Energy under ContractW-7405-ENGo48 and the Environmental Protection Agency.

Dr. Fabrikant is professor of radiology, University of CaliforniaSchool of Medicine, San Francisco, and Staff Senior Scientist, Law-rence Berkeley Laboratory, University of California, Berkeley. Hismailing address is Donner Laboratory, University of California,Berkeley, California 94720. Requests for reprints should be sent tothe author.

References

1. Committee on the Biological Effects of Ionizing Radiations.National Academy of Sciences-National Research Council. TheEffects on Populations of Exposure to Low Levels of IonizingRadiation. Washington, D.C., 1980.

2. Advisory Committee on the Biological Effects of Ionizing Radia-tions. National Academy of Sciences-National Research Council.The Effects on Populations of Exposure to Low Levels of Ioniz-ing Radiation. Washington, D.C., 1972.

3. Advisory Committee on the Biological Effects of Ionizing Radia-tions. National Academy of Sciences-National Research Council.Considerations of Health Benefit-Cost Analysis for ActivitiesInvolving Ionizing Radiation Exposure and Alternatives. Wash-ington, D.C., 1977.

4. United Nations Scientific Committee on the Effects of AtomicRadiation. Ionizing Radiation: Levels and Effects. UnitedNations, New York, 1972.

5. United Nations Scientific Committee on the Effects of AtomicRadiation. Sources and Effects of Ionizing Radiation. UnitedNations, New York, 1977.

6. International Commission on Radiological Protection. TheEvaluation of Risks from Radiation. ICRP Publication 8. Perga-

mon Press, Oxford, 1966.7. International Commission on Radiological Protection. Recom-

mendations of the International Commission on RadiologicalProtection. ICRP Publication 26. Pergamon Press, Oxford, 1977.

8. National Council on Radiation Protection and Measurements.NCRP Report No. 39. Basic Radiation Protection Criteria.Washington, D.C., 1971.

9. National Council on Radiation Protection and Measurements.NCRP Report No. 43. Review of the Current State of RadiationProtection Philosophy. Washington, D.C, 1975.

10.Brown, J. M. Linearity vs. non-llnearity of dose response forradiation carcinogenesis. Health Phys 31:231, 1976.

11.Brown, J. M. The shape of the doee-response curve for radiationc~rcinogenesis. Extrapolation to low doses. Radiation Res 71:34,1977.

12.Upton, A. C. Radiobiological effects of low doses: Implicationsfor radiobiological protection. Radiation Res 71:51, 1977.

13.Upton, A. C., Beebe, G. W., Brown, J. M., Qnimby, E. H., andShellabarger, C. Report of NCI ad hoc working group on therisks associated with mammography in mass screening for thedete~ion of breast cancer. J Natl Cancer Inst 59:480, 1977.

14.Fabrikant, J. I. Perspectives of decision-making and estimationof risk in populations exposed to low levels of ionizing radiation.(In) AAAS Annual Meeting. Symposium on Epidemiology Stud-ies of Low-Level Radiation Exposure, AAAS Annual Meeting,January 5, 1979. Report LBL-8667, pp 1-40, Lawrence BerkeleyLaboratory, University of California, Berkeley, California, Janu-ary 1979.

15.Fabrikant, J. I. The BEIR-III Report and the health effects oflow-level radiation. (In) Symposium on Nuclear Reactor Safety:A Current Perspective. AAAS Annual Meeting, San Francisco,January 1980. Science, 1980. To be published.

16.International Commission on Radiological Protection. Protec-tion Against Ionizing Radiation from External Sources, ICRPPublication 15. Pergamon Press, Oxford, 1970. Data for Protec-tion Against Ionizing Radiation from External Sources: Supple-ment to ICRP Publication 15. ICRP Publication 21. PergamonPress, Oxford, 1973.

17.Fabrikant, J. I. The BEIR III Controversy. Radiat Res 84:361,1980.

18.Fabrikant, J. I. The BEIR III Report: Origin of the controversy.A JR 136:209, 1981.

19.Shore, R. E., Albert, R. E., and Pastemack, B. Follow-up studyof patients treated by x-ray epilation for tinea capitis. ArchEnvironment Health 31:21, 1976.

20.Beebe, G. W., Kato, H., und Land, C. E. Mort~lity Experience ofAtomic Bomb survivors 1950-1974. Life Span Study Report 8.Radiation Effects Research Foundation Technical ReportRERF TR 1-77, Nation~ Academy of Sciences, Washington,D.C., 1977.

21.Mole, R. H. The sensitivity of the human breast to cancer induc-tion by ionizing radiation. Brit J Radio151:401, 1978.

22.Court-Brown, W. M. and Doll, R. Mortality from cancer andother causes from radiotherapy for ankylosing spondylitis. BritMed J 2:1327, 1965.

23. Smith, P. G. and Doll, R. Age and time dependent changes inthe rates of radiation induced cancers in patients with ankylos-ing spondylitis following a single course of x-ray treatment. (In)International Symposium on the Late Biolo~/cal Effects of Ion-izing Radiation, IAEA-SM-224/771, Vienna, 13-17 March 1978,In press.

24.Smith, P. G. and Doll, R. Late effects of x-irradiation in patientsfor metropathia haemorrhagica. Brit J Radio149:223, 1976.

25.Shore, R. E., Albert, R. E., and Pasternack, B. S. Follow-up ofpatients treated by x-ray epilation for tines capitis. Arch Envi-ron Health 31:21, 1976.

26.Modan, B., Baidatz, D., Maxt, H., Steinitz, R., and Levin, S. G.Radiation-induced head and neck tumours. Lancet 1:277, 1974.

27.Hempelmarm, L. H., Hall, W. J., Phillips, M., Cooper, R. A., andAmes, W. R. Neoplasms in persons treated with x-rays in in-fancy: Fourth survey in 20 years. J Natl Cancer Inst 55:519,1975.

28.Conard, R. A. Smmnary of thyroid findings in Marahallese 22years after exposure to radioactive fallout in radiation-associ-ated thyroid carcinoma. (In) DeGroot, L. J., ed. Radiation-Associated Thyroid Carcinoma, pp 241-257, Academic Press,New York, 1977.

29.Myrden, J. A. and Quinlan, J. J. Breast carcinoma followingmultiple fluoroscopies with pneumothorax treatment of pulmo-nary tuberculosis. Ann Roy Coil Physicians (;an 7:45, 1974.

30.Hempelmann, L. H., Kowaluk, E., Mansttr, P. S., Pastemack, B.S., Albert, R. E., and Haughie, G. E. Breast neoplasms in womentreated with x-rays for acute post-partum mastitis. J NatlCancer Inst 59:813, 1977.

31.Health effects of Alpha-emitting Particles in the RespiratoryTract. Report of ad hoc Committee on "Hot Particles" of theAdvisory Committee on the Biological Effects of Ionizing Radia-tions. National Academy of Sciences, Washington, D.C., 1976.


32.Hum, J. M, Report of the Royal Commis~on on the Health andSafety of Workers in Mines. Ministry of the Attorney General,Province of Ontario, Toronto, Canada, 1976.

33.Mays, C. W. and Spiess, H. Bone sarcoma risks to man from224Ra, 226RA and 239pu. (In) Biological Effects of 224Ra.Benefit and Risk of Therapeutic Application. (Miller, W. A. andEbert, H. G., eds.) pp 168-181, Nyhoff Medical Division, TheHague, 1978.

34.Radium-induced malignancies. (In) Argonne National Labora-tory Report ANL-77-65, Part II, 206-210, 1977.

35.Matanoski, G. N., Seltzer, R., Sartwell, P. E., Diumond, E. L.,and Elliott, E. A. The current mortality rates of radiolog£sts andother physician specialists. Deaths from all causes and fromcancer. Amer J Epidemiol 101:188, 1975.

36.Matanceki, G. N., Seltzer, R., Sartwell, P. E., Diumond, E. L.,and Elliott, E. A. The current mortality rates of radiologists andother physician specialists: Specific causes of death. Amer JEpidemiol 101:199, 1975.

37. Miller, R. W. Delayed radiation effects in atomic bomb sur-vivors. Science 166:569, 1969.

38.Miller, R. W. and Mulvihill, J. J. Small head size after atomicradiation. Teratology 14:355, 1976.

39.DeKaban, A. S. Abnormalities in children exposed to x-irradia-tion durihg various stages of gestation. Tentative timetable ofradiation injury to human fetus. J Nucl Med 9:471, 1968.

40.Kumeyama, Y., Hoshino, K., and Hayashi, Y. Effects of low-dose x-radiation on the matrix celia in the telencephalon 6fmouse embryos. (In) Proceedings of Symposium on Develop-mental Toxicology of Energy-Related Pollutants. U.S. Depart-ment of Energy Series (in press).

41.Wood, J. W., Johnson, K. G., and Omori, Y. In utero exposure tothe Hircehima atomic bomb. An evaluation of head size andmental retardation. Pediatrics 39:385, 1967.

42.Awa, A. A. Cytogenetic study. (In) Thirty Years Study Hiroshima and Nagasaki Atomic Bomb Survivors. Okada, S. etal., Eds. Rediat IRes 16 (Suppl):75, 1975.

43.Neel, J. V., Kato, H., and Schull, W. L. Mortality in the childrenof atomic bomb survivors and controls. Genetics 76:311, 1974.

44.Russell, W. L. The genetic effects of radiation. (In) PeacefulUses of Atomic Energy. Vol. 13, pp 487-500. IAEA, Vienna, 1972.

45.Russell, W. L. Mutation frequencies in female mice and the esti-mation of hazards of radiation in women. Proc Natl Acad SciU.S.A. 74:3523, 1977.


risk estimation and decision making: the health …...genesis; and the health effect, developmental...

Documents