REFLECTIONS ON THE WORK OF THE ATOMIC BOMB … bg/Beebe GW 1979... · The atomic bombs dropped on Hiro- ... Japanese National Institute of Health; NAS, National Academy of Sciences;

EPIDEMIOLOGIC REVIEWS Vol. 1, 1979Copyright © 1979 by The Johns Hopkins University School of Hygiene and Public Health Printed in USA.All rights reserved

REFLECTIONS ON THE WORK OF THE ATOMIC BOMBCASUALTY COMMISSION IN JAPAN

GILBERT W. BEEBE'

The atomic bombs dropped on Hiro-shima and Nagasaki in 1945 led to theorganization of the Atomic Bomb Casu-alty Commission (ABCC), whose historycontains important lessons for epidemi-ologists concerned with large-scale inves-tigations of problems requiring long-term,prospective study. The history is perhapsall the more meaningful because it is nota linear process starting from a definitiveprotocol and moving through well-plannedstages of development to a fulfillmentclearly visualized at the outset. Rather,it is an adaptive process, the record of astruggle against difficulties of many dif-ferent kinds, its continuation dependenton the vision of relatively few people, onthe growing social need for informationabout the health hazards created by aburgeoning nuclear technology, and on abelated infusion of epidemiologic conceptsand methods without which the studies ofsomatic effects seemed doomed to failure.But it is also a bifurcated history, inthat the genetic studies do not fit thisgeneral pattern. They stand apart assuperbly planned and efficiently conduct-ed and yet with little evident influenceon the somatic studies which have cometo dominate the program.

The lessons to be learned from the his-

1 Clinical Epidemiology Branch, National CancerInstitute, Bethesda, MD 20205.

Abbreviations: ABCC, Atomic Bomb CasualtyCommission; AEC, Atomic Energy Commission;AHS, Adult Health Study; ATB, at the time ofthe bomb; BEIR, biological effects of ionizing radia-tion; BGS, Biochemical Genetics Study; F,, conceivedafter the bomb; LET, linear energy transfer; LSS,Life Span Study; JNIH, Japanese National Instituteof Health; NAS, National Academy of Sciences; NRC,National Research Council; ORNL, Oak RidgeNational Laboratory; RBE, relative biological effec-tiveness; RERF, Radiation Effects ResearchFoundation; T...D, tentative dose.

tory of the ABCC concern research strat-egy, specific methodologic issues, and howto maintain an effective research organi-zation dedicated to a major long-term,prospective study. Just as its problems ofstrategy and method are not specific tothe investigation of the effects of ionizingradiation, so its problems of survival andmanagement are not confined to the op-eration of a research enterprise far fromits institutional base or in a foreigncountry.

For epidemiologists it is important tounderstand that the ABCC studies wereset in motion without substantial epide-miologic input, that epidemiologic con-cepts and principles rescued the effortafter it had reached low ebb in 1955,and that its entire history underscoresthe value of the epidemiologic approach.Some illustrious names in epidemiologyand statistics aided in the introductionof epidemiologic thinking that so greatlycontributed to its success: William G.Cochran, Harold F. Dorn, Thomas Francis,Jr., John E. Gordon, Alexander D. Lang-muir, and Thomas Parran.

CHRONOLOGY

The atomic bombs dropped on August6 and 9, 1945, were followed by theJapanese surrender on August 14. USArmy, Navy, and Manhattan District in-vestigating teams began arriving in Sep-tember, and on October 12 a "Joint Com-mission for the Investigation of the Effectsof the Atomic Bomb in Japan" was formedby authority of the Supreme Commanderof the Allied Powers to perform a co-ordinated study in cooperation withJapanese groups already on the scene.From that study of acute effects (1) arosethe proposal for a continuing study under

184

ATOMIC BOMB CASUALTY COMMISSION 185

civilian, non-governmental auspices. OnNovember 26, 1946, President Trumanapproved a directive to the NationalAcademy of Sciences-National ResearchCouncil (NAS-NRC) "to undertake a longrange, continuing study of the biologicaland medical effects of the atomic bombon man." In the planning documentsspecific interest is expressed in cancer,leukemia, shortened life span, reducedvigor, altered development, sterility,modified genetic pattern, changes invision, "shifted epidemiology," abnormalpigmentation, and epilation.

A team was sent to Japan to explorethe situation, establish an interim organ-ization, and make recommendations for along-term program. An NAS-NRC Com-mittee on Atomic Casualties was organ-ized under the chairmanship of ThomasM. Rivers. (Other members were GeorgeW. Beadle, Detlev W. Bronk, Austin M.Brues, George M. Lyon, C. P. Rhoads,Shields Warren, Stafford L. Warren,George H. Whipple, and Raymond E.Zirkle.) At its first meeting on March 25,1947, the Committee heard an encour-aging report from Brues and Paul S.Henshaw on their investigative mis-sion to Japan and on the steps beingtaken there to establish an interim or-ganization that soon became the AtomicBomb Casualty Commission. The newlyformed Atomic Energy Commission(AEC) was to provide the necessary fundsunder a contract with the NAS. Thefindings of the studies were expectedto be useful in industrial, medical, andmilitary uses of atomic energy.

1947-1950. At the outset the Commis-sion, with the help of US occupationauthorities, was very much involved inbuilding an organization, providing forits logistic support in impoverished Japan,responding to the interests of US inves-tigators with specific interests, e.g., inhematology, growth and development,and cataracts, and laying the groundworkfor a systematic medical examinationprogram. Of greatest importance, in ret-

rospect, were Neel and Schull's genetic,study (2), presented to the Committee onAtomic Casualties at its second meetingon May 1, 1947, and an agreement withJapanese authorities to supplement theOctober, 1950, population census with an.inquiry about exposure to the atomicbombs. The records of the Joint Com-mission had proved to have little valuebecause of inadequate personal identifica-tion, and the need for a firm samplingbase for screening studies had becomeevident. Partly because of concern over theunknown importance of fallout and resi-dual radiation, Kure was chosen as a con-trol city for Hiroshima, and mention wasmade of Sasebo as a control for Nagasaki.By the end of the period, ABCC hadfunctioning departments of genetics, cyto-genetics, obstetrics and gynecology,pediatrics, internal medicine, surgery,ophthalmology, radiology, pathology, bio-chemistry, microbiology, and biometrics.The newly formed Japanese National In-stitute of Health (JNIH) was asked toprovide an interface between ABCC andthe Japanese community, and arrange-ments were made for pathological speci-mens to be sent to the then (US) ArmyInstitute of Pathology.

1950-1955. During the next few yearsNeel and Schull's genetic study (2) wasbrought to a successful conclusion, definiteevidence of the leukemogenic effect wasobtained (3), and studies of the in uteroexposed revealed the existence of smallheads and mental retardation in the mostproximally exposed (4). Ophthalmologicsurveys were continued, as were thegrowth and development studies. Shield-ing surveys were begun in an effort toget closer to a physical measure of radia-tion exposure. A medical examination pro-gram for adults, begun in 1950, falteredprogressively in the face of negativefindings and declining participation.Under fiscal pressure, and with AECassurances that fallout and residual radia-tion were unimportant, efforts to employKure and Sasebo as control cities were

186 GILBERT W. BEEBE

abandoned. Continuity of leadership couldnot be maintained. As the war in Koreaprogressed, and inflation in Japanworsened, a fiscal crisis loomed and late in1954 there was even talk of closing theoperation entirely. By 1955 there wereserious doubts about prospects for thefuture, examination programs were failing,little that was new was being seen, andstaff was discouraged. The validity of thepreliminary findings of a study of mortal-ity was seriously questioned at an ad hocNRC conference. (Conferees includedLangmuir, Chairman, Dorn, Gordon, A. E.Brandt, Felix E. Moore, H. M. Luykx,and Seymour Jablon.)

Late in 1955, Francis, together withMoore and Jablon, made a site visit toABCC at the request of R. Keith Cannan,Chairman of the NRC Division of MedicalSciences. Their mission was to make anobjective scientific appraisal of the in-stitution and its program, and to makerecommendations for the future. In theirreport they provided not only a thoroughassessment of the situation, but also anover-all strategy for the future withemphasis on the integration of differentscreening and observational modalitiesthrough the mechanism of a fixed cohortof exposed and non-exposed subjects (5).The specific components of the recom-mended program, soon adopted as theUnified Study Program, were: a fixedsample; epidemiologic detection or con-tinuing morbidity survey; clinical detec-tion; post-mortem detection; and deathcertificate study.

1955—1961. Adoption of the recommen-dations of the Francis Committee was theturning-point for ABCC. Although thestaff voiced numerous objections, exceptfor the epidemiologic detection plan(weekly reporting of illness or death bylay monitors each responsible for 15-20survivors in his neighborhood), the Uni-fied Study Program was put into effect,first by Cannan, who served as actingdirector in Japan for five months in 1957,and then by George B. Darling of the

Yale University School of Medicine, whoserved as director from 1957 until 1972.Ties were formed with the Yale Univer-sity School of Medicine, the Universityof California at Los Angeles (UCLA)School of Medicine, and the Follow-upAgency of the National Research Councilto provide leadership and staff for theABCC departments of medicine, path-ology, and statistics, respectively. In 1958these groups began sending personnel oftheir own, including department heads.The clinical program was reorganizedfirst, as the Adult Health Study (AHS)(6), and put into operation in 1957—1958,in large part through the efforts of ArthurJ. McDowell, who later organized anddirected the Health Examination Surveyof the US National Center for HealthStatistics, and James W. Hollingsworthof the Yale University School of Medicine.The AHS sample of 20,000 became thefirst segment of the mortality samplefor what the Francis Committee calledthe death certificate study, later namedthe Life Span Study (LSS); the sampleof 100,000 LSS subjects was completedin 1961 (7) and then extended to 109,000in 1967 (8). A second mortality study ofabout 2800 in utero exposed and controlswas designed in 1960 (9), and a third(Fi, conceived after the bomb) mortalitystudy of 54,000 children of survivors andof non-exposed parents was designed inthat same year to detect any geneticdamage sufficient to influence mortality(10).

In lieu of the Francis plan for layobservers to report illness among smallpanels of survivors, consideration wasgiven to health surveys of industrial pop-ulations containing significant numbersof A-bomb survivors and to home visits.These techniques were not found to becost-effective, however, and were soonabandoned in favor of routine history-taking at the biennial AHS examination.

The pathology program was the mostdifficult to integrate with the new clini-cal and mortality studies, as both post-


mortem and surgical specimen examina-tions had for many years been performedas a free service to the medical com-munity with little regard for their re-search value. In 1961 such examinationswere restricted to the several fixed co-horts under study (LSS and AHS, F,, andin utero), and by 1963 the autopsy ratein the LSS sample had risen from 7 to 44per cent regardless of place of death (11).

The Adult Health Study was designedto detect changes in the incidence ornatural history of disease, physiologic orbiochemical changes or "markers" (shortof actual disease), new disease entities,and nonspecific changes in vigor or ac-celeration of aging (6). Three levels ofobservation were defined: 1) a standardhistory, physical, and laboratory exam-ination repeated periodically; 2) supple-mentary observations serving ad hocinquiries into particular diseases ororgan-systems; and 3) supplementaryobservations testing specific ideas aboutpossible effects.

With the completion of the design forthe LSS study (12), Darling inaugurateda series of bilingual protocols in technicalreport format that became institutionalimperatives and ensured continuity ofeffort in the face of turnover of personnel.With notable exceptions, US personnelserved two-year terms. Although therewas far more stability among the Japa-nese staff, especially JNIH personnel,many of these also served tours of oneto two years. The protocols for majorprograms (LSS, AHS, Fu in utero, pathol-ogy) included agreements with the JNIHfor joint sponsorship and conduct. Proto-cols for substudies were less formal butwere also published in bilingual format.In 1959 also the Director began thebilingual Technical Report series intowhich all ABCC work, whether intendedfor publication or not, was entered priorto publication.

Shortly after the Francis Report, theOak Ridge National Laboratory (ORNL)began a determined effort to develop a

physical dosimetry program by meansof which survivors might be classified byabsorbed dose rather than by distancefrom the hypocenter and acute radiationsymptoms (13). By 1958, this program(Ichiban) of the Health Physics Division,ORNL, combined with ABCC documenta-tion of each survivor's location andshielding configuration at the time of thebomb (ATB), had produced a first tenta-tive dosimetry plan (T57D) that was usedto screen for effects and to begin theinvestigation of dose-response relation-ships.

Tumor registries were established bythe city medical association in Hiroshimain 1957 and in Nagasaki in 1958 (14), andthe leukemia detection program begunin 1947 became a registry operation about1959. Initially the numerator data werecombined with estimates of the survivorpopulations resident in the two cities,distributed by distance from each hypo-center.

1961 -1975. Reorientation of the pa-thology program within the framework ofthe Unified Study Program made it possi-ble to develop much-needed informationon the reliability of the Japanese deathcertificate (15), which was customarilyfilled out before any autopsy could bedone at ABCC, on ascertainment bias,and on the influence of radiation ontumors not well diagnosed on the deathcertificate. Pathologists also investigatedthe hypothesis of radiation-acceleratedaging (16-18) and contributed impor-tantly to the studies of cardiovascularand cerebrovascular diseases, especiallythe Ni-Hon-San study (19).

Further experimental work on shield-ing and the discovery that the yield ofthe Hiroshima weapon had been grosslyexaggerated led to calculation of the sec-ond-generation (T65D) dose estimates in1966-1967 (20). Absorbed organ dosesbased on the T65D values have recentlybeen provided by Kerr (21). Hashizumeand Maruyama (22) of the National In-stitute of Radiological Sciences in Japan


have also developed parallel air-dosecurves for each city, and factors for con-verting kerma estimates to absorbed or-gan- or tissue-dose estimates.

Early in the 1960s it became clear thatestimates of the survivor population resi-dent in the two cities had become un-reliable because information on migra-tion was inadequate. Accordingly, onlycases falling within the fixed cohorts(LSS, AHS, F,, in utero) were henceforthused in analyses of the effect of radiation(23). Although the Nagasaki Tumor Reg-istry remained strong and reasonablycomplete, the Hiroshima University ma-terial generally did not find its way intothe Registry, and in 1973 the HiroshimaPrefectural Medical Association estab-lished a "tissue registry" with similargoals to which the University medicaldepartments do contribute. In 1974 theNagasaki City Medical Association com-menced the operation of its tissue regis-try (24). A combination of the tworegistries in each city guarantees vir-tually complete ascertainment of malig-nant tumors for investigative purposes.

In the mid-1960s the three-day periph-eral leukocyte culture technique of Moor-head et al. (25) became established underBloom et al. (26), and a major cytogenet-ic program was begun to investigate theeffect of radiation on the leukocytes ofA-bomb survivors (27). In recent yearscytogenetic techniques have also beenused by Awa (28) and his associates inthe investigation of Fi subjects for somat-ic-cell changes that would reflect radia-tion-induced germ-cell changes in theirexposed parents.

Throughout the 1960s inflation pro-ceeded at a brisk pace in Japan, withtotal ABCC costs rising from $1.9 millionfor a staff of 905 in 1959 to $4.2 millionfor a staff of 695 in 1970. In 1971 theJapanese yen was revalued, so that dollarcosts rose even higher. By FY 74 thetotal was up to $6.2 million with noincrease in staff, and a substantial inputfrom the US National Cancer Institute

was required to fend off disaster. In hisannual report for 1967-1968, Darlingmade a plea for a stronger commitmentby the government of Japan to the sup-port and direction of the institution, butit was not until 1973 that the two govern-ments began the dialogue that led finally,in November, 1974, to an agreement ona new organization, a Japanese founda-tion established by the Health and Wel-fare Ministry, to succeed ABCC. The ob-jectives and program remained un-changed, but management was vested ina binational board of directors for thenewly created Radiation Effects Re-search Foundation (RERF) which re-placed ABCC on April 1, 1975. The actof endowment provided that control beshared equally by the two governmentsthrough their representatives on theboard and on an advisory scientific coun-cil, and through their budgetary support.The US support by the Department ofEnergy continues to be provided throughthe National Academy of Sciences asbefore.

In February, 1975, in anticipation ofthe organizational change, the NAS sentto Japan a Committee for Scientific Re-view of ABCC under the chairmanshipof James E. Crow. (Other members wereHenry S. Kaplan, Paul A. Marks, RobertW. Miller, John B. Storer, Arthur C.Upton, and Jablon.) The Committeerecommended, inter alia, that: 1) thebasic elements of the ABCC programcontinue under the Foundation; 2) theLife Span and F, mortality studies bestrengthened by linkage to the tumor andtissue registries; 3) the Adult HealthStudy be modified to give greater empha-sis to objective measurements and auto-mated histories, and less emphasis tophysical examination; 4) autopsies nolonger be actively solicited and pathol-ogists concern themselves with the exam-ination of tissues and participation instudy teams investigating specific dis-eases; 5) the cytogenetic program becontinued and strengthened as needed;


6) the in utero sample be studied not onlyvia death certificates but also throughperiodic examination; and 7) the newtissue registries be actively supportedand used in research, and initiation ofcancer screening programs be considered(29). A major new research proposal byNeel, the Biochemical Genetics Study(BGS), was also recommended for earlyimplementation. Pilot work had shownfeasibility and personnel had beentrained in the search for protein variantswith such techniques as starch gel elec-trophoresis and high-speed centrifugalanalysis.

At the outset the new foundation facedthe important programmatic issues de-fined in the Crow report (29). By early1979 all the necessary decisions had beenmade, the decision to discontinue routineautopsy contacting being the most con-troversial and difficult. With the autopsyrate below 20 per cent, most autopsiesbeing performed elsewhere in medicalcare facilities, and repeated efforts tostimulate the program having failed, dis-continuation of the very expensive autop-sy contacting process that had evokedresentment in the community eventuallybecame inevitable.

RESEARCH GOALS AND METHODS

GoalsThe scientific goals of ABCC/RERF

have evolved and matured over the threedecades of its history in response to thegrowth of knowledge in the field and thesocial need for specific information tosupport regulatory policy and procedures.Because of its isolated position in Japan,however, and the absence of any strongradiobiological emphasis in its staffing,the institutional goals have seemed lessinfluenced by specific radiobiological con-siderations than one might expect.

From the outset the goals have beendominated by a desire to identify the ef-fects of ionizing radiation rather than tomeasure them or to explain them in termsof fundamental mechanisms. The process

of identification has been fueled by in-terests in specific effects and by the beliefthat a broad screening program was re-quired to catalogue the late effects ofradiation. Interest in the measurement ofeffects had to wait upon their identifica-tion, and upon the development of dosim-etry adequate to serve it.

Until the Francis Report, and exceptfor the genetic study, clinical conceptsand goals determined program. Conceivedas an opportunity to exploit a uniquesituation, the program quickly focusedon specific interests in cataracts, growthand development, hematologic effects,and genetic effects. As the institutiontook shape, however, its broad purposewas also expressed in non-specific screen-ing of representative samples of heavilyexposed survivors and of lightly exposedor non-exposed controls. These weremainly clinical efforts by internists andpediatricians with laboratory support.If ionizing radiation produced uniquelesions, perhaps biochemical or patholog-ic investigation would have proved morefruitful; but in the absence of uniquelesions, investigation necessarily de-pended on statistical comparisons whosepower of detection depended, in turn, onrelative size of effect, length of latentperiod, and size of sample.

Adoption of the Francis Report greatlystrengthened the program but did littleto change the goals: detection of effectsremained the dominant theme, and syste-matic screening the dominant method.An interest in measurement appearedfirst in the work of Brill et al. (30) onleukemia 10 years after the leukemogeniceffect was first demonstrated (3). In theearly 1960s an excess of thyroid cancerwas shown (31) and later in the decadebreast cancer (32) and lung cancer (33)were added to the list without, however,any real attention to the quantitativeaspects of these effects. It was in thereports on the Life Span Study (8, 34)that the interest in quantitating effectsfirst became fully established, with the


latter report (34) providing much of thebasis for the estimates in the NAS BEIRreport of 1972 (35).

The interest in screening is now nolonger dominant, although the expecta-tion is that the nearly general carcino-genic effect of ionizing radiation willeventually be illustrated by a widerarray of specific sites and cell types thanis presently recognized. Also, there re-mains some interest in the hypothesisof accelerated aging (36), and a greatinterest in the possibility of demonstrat-ing the presumed genetic effect of ionizingradiation (24). But, in the main, the goalsnow center on the measurement of effectsrather than on their identification, ondose-response estimates, on the func-tional form of the dose-response relation-ship, on the quantitative aspects oflatency and duration of effect, on theevaluation of the relative biological ef-fectiveness (RBE) for neutrons, and onthe influence of both host factors andenvironmental factors on the dose-response relationship (24). In addition,there is a greater interest in mechanisms,especially in the possibility that immuno-logic changes may underlie or mediatethe carcinogenic effects that are beingdemonstrated by epidemiologic methods.

StrategiesOne must conclude that no explicit re-

search strategy governed the work ofABCC until the recommendations of theFrancis Committee were adopted, al-though in its report to the NAS theFrancis Committee made it clear thatthe recommended Unified Study Programshould be regarded as an extension andrationalization of the program to date (5):

In an effort to provide increased stability,integration of purpose and effort, and cross-stimulation, the Committee presents withrecommendation a Unified Study Program.It is believed that the program retains thevaluable features of present procedures butshould provide greater consistency and in-crease the sensitivity of detection at everylevel.

The proposed strategy, of course, wasprimarily a strategy for screening, whichwas the task of the time, well before therewas serious talk of developing a physicaldosimetry. It was by no means inconsis-tent with the present emphasis on mea-surement of effects and analysis of theirdependence upon host and environmentalfactors other than radiation, but did notmake explicit provision for this reorienta-tion of goals. It stressed 1) multiple ob-servational systems; 2) interlocking ofobservations through the use of a commoncohort of subjects; and 3) continuity ofeffort through institutionalization ofresearch plans and recruitment of re-search personnel from an establishedagency in the US. In its specifics it ig-nored the continuing need for informationon the genetic effects of ionizing radiationand failed to foresee the critical role to beplayed by physical dosimetry. But theproposed strategy, coupled with Darling's15-year tenure as Director, and institu-tional ties with Yale, UCLA, and the NRCFollow-up Agency, made it possible forABCC to endure beyond the latent periodfor many of the tumors caused by ionizingradiation, the existence of which nowlends so much force to the program.

Although the Unified Study Programis now outmoded, there has been no newsynthesis to replace it other than the1975 report of the Crow Committee thatis less a new blueprint for the future thanan evaluation of the older strategy withrecommendations for changes in empha-sis. In many ways the Unified Study Pro-gram is now an anachronism, and some ofits heavily institutionalized elementsdeter its progress. This has been par-ticularly true of the clinical detectionprogram, the Adult Health Study, thatplayed such an important role in theinitial detection of solid tumors and innarrowing the scope of probable late ef-fects to carcinogenesis. But the AHSsample is too small any longer to carrythe burden of the investigations into


carcinogenesis, a burden for which eventhe LSS sample is none too large, and itsascertainment of tumors is incompletewithout the aid of the tumor and tissueregistries. Pathologic detection, especial-ly via post-mortem studies, remains aviable concept in the abstract, but inpractice is no longer possible; the com-munity will no longer tolerate an aggres-sive autopsy contacting program. Thedeath certificate study (the LSS) remainscentral to the effort, and with the T65Ddosimetry, supplemented by conversionfactors for absorbed doses to individualorgans, will undoubtedly remain thecenterpiece of RERF strategy. But eventhe LSS cannot meet the total need withrespect to carcinogenesis, because thedeath certificate is too imprecise in itsattribution of death to individual tumors.For a few tumors the sensitivity and spec-ificity ratios are reasonably high, andfor all malignant neoplasms as a class,very high (15), but merely to completethe catalogue of tissues vulnerable tothe carcinogenic force of radiation moreprecision is needed than the death cer-tificate can provide. It must be supple-mented by information from tumor andtissue registries, or by ad hoc surveys(37), for which the LSS sample continuesto provide a most valuable samplingframe.

The Adult Health Study was alwaysthe most expensive part of the UnifiedStudy Program, in some years requiringmore than half the operating budget.Now, although no longer cost-effective,it cannot be dropped, but only attenuated,because it is the only visible direct con-tribution to the welfare of the A-bombsurvivors. Without some such service tothe community it is doubtful that RERFcould survive. And, as long as the AHSsample is kept under surveillance, itprovides a useful sampling frame forspecial, ad hoc studies requiring clinicalexamination of subjects. There are otherelements in the current strategy, ele-

ments that involve possible mechanismsby which ionizing radiation producesbiologic effects. One is the cytogeneticprogram, in which it has been shownthat somatic cells exhibit dose-dependentchromosomal changes decades after theinsult. But whether these changes arenow merely a biologic record of the radia-tion injury, or portend biologic changes,remains unknown. Finally, there is anactive program in which immunologicimpairment is being sought as a late ef-fect of ionizing radiation (38).

Sampling considerations

The population and sources of informa-tion. The population has three distinctparts: those exposed prenatally, thoseexposed postnatally, and those conceivedonly after the bomb (the Fi). Each com-ponent posed problems for representativesampling. The Joint Commission sampleof examined patients was of unknown rep-resentativeness and, in any case, therecords contained insufficient identifyinginformation. In 1949 ABCC scientists tooka Radiation Census in both cities, identi-fying about 86,000 survivors in Hiro-shima, 94,000 in Nagasaki, and drewupon the results in setting up their earlyscreening programs. In connection withthe first postwar national populationcensus in 1950, 284,000 survivors wereidentified throughout Japan, 98,000 inHiroshima, and 97,000 in Nagasaki,from responses to the question: "Was anymember of this household in HiroshimaCity or Nagasaki City at the time of theA-bomb?" It was this sample that theFrancis Committee recommended as thebasis for the fixed cohorts in the UnifiedStudy Program. As it provided no informa-tion on the events of 1945-1950, or onthe possibly different health status ofsurvivors who were no longer in thebombed cities in 1950, however, use waslater made of the 1946 Hiroshima AtomicBomb Casualty Census to investigate


these problems (39). Supplementary 10per cent sample censuses were conductedby ABCC in 1950 and 1951 to identifyrepresentative non-exposed subjects, andexposure status has generally been afeature of subsequent official censuses inHiroshima and Nagasaki.

The in utero exposed were defined onthe basis of local birth records in the ninemonths after the bombs plus field investi-gation of the exposure status of theirparents. In developing his Ft sample forthe first genetic study, Neel had access topregnant women registering for supple-mentary food rations during 1948—1952.Later this sample was augmented frombirth registrations in the 1946—1958period, to create the present Fi mortalitysample (10). The sample for Neel's Bio-chemical Genetics Study of Fi subjectsdraws upon these samples but additionalFi subjects were obtained from the familyregistries of the LSS sample.

Contrasts. Of particular interest is thestructure of the statistical contrastsestablished by ABCC to detect and mea-sure effects. After a brief flirtation withKure and Sasebo as control cities forHiroshima and Nagasaki, this approachwas abandoned in favor of intrinsic con-trols. Virtually all of the research per-formed at ABCC has made use of internalcontrols, much of it in dichotomies of highdose vs. non-exposed, or high dose vs. lowdose, comparisons. Increasingly, however,as the dosimetry improved, regressionmethods have been used both to identifyeffects and to estimate their size. Theregression approach is well served bythe stratified sampling plan based on therecommendations of the Francis Com-mittee for the LSS, for example, in whichall eligible survivors under 2.0 km fromthe hypocenter ATB were to be selectedand group-matched by age and sex withincity to distally exposed (2.5 to 10.0 km)and to non-exposed subjects, with theaddition of all eligibles within the 2.0 to2.5 km zone. Of about 480,000 individualsthought to be in the two cities at the time

of the Francis Report in 1955, about 8.5per cent were survivors exposed <2.0 km;5.0 per cent between 2.0 and 2.5 km; 29.4per cent between 2.5 and 10.0 km; andabout 57.0 per cent were non-exposed.When it became possible to calculate indi-vidual doses for survivors in the LSS sam-ple, there was good coverage of the entiredose range, including 0.

In all analyses, whether regression ornot, the non-exposed occupy an ambiguousposition because of differences in healthhistory and their more favorable mor-tality experience in the early 1950s (40),perhaps because of the selective influenceof migration. In time, however, as theirmortality rates mounted to the level ofthe zero-dose group, the non-exposed havebeen used to augment it. The non-exposedinclude so-called "early entrants," thosewho entered the city only after the burstto perform rescue work, see relatives, andthe like. This subgroup has long beencontroversial because some Japaneseinvestigators have claimed that the earlyentrants have experienced an undulyhigh incidence of leukemia (41). Withinthe non-exposed group of the LSS sample,they are identifiable and ABCC studieshave revealed no suggestion of excessmortality (34); the dosimetry researchsuggests that their average exposure wasprobably not biologically significant (42).

Demographic variables. Migration inand out of the city has been somewhattroublesome from the start. The firstattempt at a mortality study, in whichall deaths among A-bomb survivors in thecity were related to estimates of theirnumber in the city population, founderedon uncertainties as to the effects of migra-tion. The integrity of the fixed cohortsrecommended by the Francis Committeehas removed much of the impact of migra-tion on ABCC studies, but observationalmethods such as the biennial clinicalexamination of AHS subjects, and thevarious registries, can be applied only tothose who remain in the area. Fortunate-ly, out-migration has been independent of


dose (43), so that such programs have notbeen seriously impaired by migration.Migration does not affect the ascertain-ment of mortality, as Japan has a nationalsystem of family registration, one featureof which is a permanent address (honseki).Vital events occurring anywhere in Japanmust be reported to the office having cus-tody of the family registry (koseki).

Sample size. Although the fixed co-horts recommended by the Francis Com-mittee were thought by some NAS ad-visors to be unmanageably large, theyproved to be too small for some purposes,especially in Nagasaki. Dependence uponthose exposed <2.0 km ATB as the key-stone of the LSS sampling plan resultedin a Nagasaki sample much smaller thanthe Hiroshima sample, the implicationsof which were not appreciated until itbecame plain that what was most wantedfrom these studies was the dose-responsebehavior of the Nagasaki sample becauseits exposure was almost entirely to gammaradiation. The strong admixture of neu-trons in Hiroshima, and their high corre-lation with the gamma dose, make itimpossible to separate their effects. Evenwhen a two-cities analysis is made andseparate estimates derived for gammaand neutron radiation, it is the Nagasakiexperience that appears to control thegamma risk estimate. At the time theLSS sample was being put together (12),it was thought that the air-dose curves forthe two cities were very similar. In themid-1960s it was realized that the yield ofthe Hiroshima weapon was about 12.5kilotons, not 20 as previously believed(44). In consequence, the 2.0 km cut thatwas to have been at about 10 rads in eachcity actually divided the distributions atabout 17 rads in Nagasaki and 2 rads inHiroshima; at 2.5 km the dose was about3 rads. Now, when the importance ofthe cut had been made at 2.5 km in Naga-saki, the Nagasaki sample would havebeen perhaps 70 per cent larger, but stillmuch smaller than the Hiroshima sampleand no stronger in the region beyond

3 rads. Now, when the importance ofestimates in the low-dose range is soacute, it is clear that the Nagasaki sam-ple is too small, and especially so at thelow-dose end.

The Nagasaki leukemia data have beensubjected to several analyses to deter-mine an appropriate dose-response curveand contrasted with the Hiroshima datato estimate the RBE of neutrons (45—48). It is of more than passing interest,therefore, that of the 231 NagasakiLeukemia Registry cases for the 1946—1974 period, only 34 could be used inanalyses based on the LSS sample (49).The uncertainty attached to any dose-response analysis of the leukemia casesin the LSS sample for Nagasaki is sogreat as to reduce it almost to the levelof a tour de force, but at present there isno reliable substitute for the LSS sample.

Although the number of A-bomb sur-vivors is not small by ordinary standards,some of the risks for which better esti-mates are needed are probably so smallthat even upper confidence limits arelikely to be quite high. This is especiallytrue of the genetic risk; even with morethan 25 different biochemical markersbeing examined in perhaps 10,000 off-spring of irradiated parents and 10,000controls in Neel's Biochemical GeneticsStudy, it seems very doubtful that a sig-nificant increase in the mutation rate canbe shown. But the study should neverthe-less provide quantitative informationnow lacking, and an empirically deter-mined upper limit on the radiation-induced increment in mutation ratewould have great value. Somewhat simi-lar considerations pertain to the esti-mation of the carcinogenic risk of lowdoses of low-linear energy transfer (LET)radiation, i.e., those in the neighborhoodof background and of the present protec-tion standard for the general population.The chance of demonstrating non-zerorisks at this level seems quite small, forthe signal/noise ratio is likely to be quitelow, and the Nagasaki sample does not


have strong representation in the low-dose range. In this instance it seemsdoubtful that useful estimates of the low-dose carcinogenic risk will be forth-coming from the Nagasaki data in theabsence of prior knowledge of the bestfunctional form for the dose-responsecurve. The mortality data through 1974were insufficient to provide statisticallysignificant evidence of any relationshipbetween dose and all forms of cancerexcept leukemia in Nagasaki, even whenthe entire range of dose was taken intoaccount (49).

A very useful sampling aid in the AHShas been the subdivision of the sampleinto 24 schedule groups, one for eachmonth of the two-year examination cycle,so that a representative sub-sample ofpatients would be called in each month.Short-term studies could be undertakenwith the assurance that the examineeswere representative and covered the fullrange of exposure.

Observational methodsThe observational methods employed

in the studies of the A-bomb survivorsspan many levels of biologic organizationand range from the biochemical and sub-cellular to the gross morphologic and thebehavioral. They vary from the routineand commonplace, such as height, weight,and blood pressure determinations, tounusual or new methods such as starchgel electrophoresis and high-speed cen-trifugal analysis to detect protein variantsin red cells and serum. This variety re-flects, in the first instance, the breadthand depth of the search for the effects ofionizing radiation on human health and,in the second, the rapidly changing tech-nology of the times, particularly in bio-chemistry. Some of the change, of course,has been at the expense of comparabilityfrom one time to the next, so that evenhere, where the intent was to examinepatients in the same way cycle after cycle,close comparability has proved elusive,especially in the history and physical

examination. Even for routine laboratorydeterminations, e.g., serum cholesterol,it has not been easy to re-calibrate valuesobtained at different times with differentmethods to permit the investigation ofchanging trends accompanying the greatdietary changes occurring in Japan.

Dosimetry. Radiation dose being thefundamental independent variable, thehistory of ABCC has been in no small parta progressive refinement in the approachto its determination, a process that isstill going on. The essential elementsare 1) determination of the epicenter andhypocenter of the explosion (50); 2) deter-mination of the precise location of thesurvivor and the configuration of hisenvironmental shielding (51); 3) deter-mination of the shape of the air-dosecurves from weapons tests (44); 4) nor-malization of the air-dose curves to theyield of each weapon (44); 5) estimationof attenuation factors for environmentalshielding (52); 6) estimation of attenua-tion factors for specific body tissues (21);and 7) estimation of fallout and residualradiation (42). For all but the first twoelements the distinction between gammaand neutron radiation is fundamental.

The dosimetry system developed byORNL in conjunction with ABCC hasenabled individual dose estimates to becalculated for 97 per cent of the LSSsample. The remainder represent deathsin the early 1950s for whom shieldinghistories could not be obtained and sur-vivors whose shielding situations havethus far defied quantitative evaluation,e.g., those in streetcars. Estimates of theerror of a dose determination have beenmade (53), but are necessarily rather ap-proximate. Validation studies of the loca-tion of survivors ATB have been generallyencouraging, but some of the calculateddoses are impossibly high and have beenarbitrarily reduced to 600 rads (34). TheT65 air-dose curves of Auxier (13) havebeen reasonably well confirmed by Hashi-zume and Maruyama (22) from the radio-activity and thermoluminescence in-


duced in building materials at variousdistances from the hypocenter. The atten-uation factors for converting rads kermato critical organ doses (21) have only be-gun to be used and it is too early to knowhow stable these factors are. There isfairly good agreement between the fac-tors generated by the ORNL group andthe group at the Japanese National Insti-tute of Radiological Sciences.

Mortality. Observations as to mortalityhave primarily to do with cause of death,but the extensive series of post-mortemexaminations (4353 in 1970) (15) hasmade it possible to do organ surveys andbiochemical determinations as well. Theprincipal autopsy diagnosis has been com-pared with the certified cause of death inorder to determine the validity of deathcertificate information. These compari-sons have provided useful perspective onthe underlying cause of death obtainedfrom the death certificate and have dem-onstrated the importance of supplement-ing that information with tissue diag-noses available through the cancerregistries.

Mortality ascertainment is one of thestrongest features of the ABCC studies.Initially ABCC copied all death certifi-cates filed in each city for case-findingand for collation with samples understudy. With the actualization of the Uni-fied Study Program, however, eligibilityfor membership in the fixed cohorts gen-erally depended on knowledge of thehonseki, or permanent family address.Inquiry could then be made at any timeat the corresponding city office to deter-mine survival. Knowing the fact of death,and with appropriate approval, the inves-tigator can have access to the vital statis-tics schedule containing the cause ofdeath, usually at the health center ofjurisdiction. Mortality ascertainmenthas been continually monitored in avariety of ways and is known to be vir-tually complete.

Morbidity. Morbidity ascertainmentis, of course, much more complex, espe-

cially if the concept is broadened to in-clude anthropometric measurements andphysiologic changes that suggest no ill-ness but nevertheless are of interest inrelation to radiation dose. And at ABCC/RERF morbidity has been perhaps theleast successful of the observational sys-tems. Many approaches were tried or atleast given serious consideration: report-ing admissions to hospitals as they occur,funding a medical care plan that wouldreport each hospital admission of mem-bers of the fixed cohort, periodic homevisits, and interim mail questionnaires.In the end it was necessary to rely on thebiennial screening examination for vir-tually all the historical information onthe AHS sample, and to supplement theroutine inquiry at the time of the clinicvisit with ad hoc interviews and ques-tionnaires as required. Mail question-naires were used, however, to expandbase-line demographic and historical in-formation on the remainder of the LSSsample, and with some success, initiallyin connection with the needs of the Ni-Hon-San study.

The search for morbidity, and forphysiologic changes or other characteris-tics not pathognomic of disease, was thedriving force for the Adult Health Study.It was expected that a standard diagnos-tic examination and history, conducted attwo-year intervals, and supplementedby epidemiologic field surveys or hospitalreporting of interim illness and by anactive surgical pathology service, wouldbring to light the great bulk of the signi-ficant illness in the sample (6). It couldalso be strengthened by periodic exami-nation of all patients suspected of havinga particular disease, and by intensiveorgan-system reviews in support of whichspecial observational procedures wouldbe employed (6). There was, however, amuch greater emphasis on the detectionand diagnosis of illness than upon thesystematic recording of objective observa-tions. Thus, in time, the accumulateddata had their chief value as an index


to probable illness, the future study ofwhich then required calling in the pa-tients above a certain level of suspicionwith respect to a particular diagnosis sothat uniform, standardized studies couldbe performed.

As initially planned, clinical observa-tions were to be standardized by teaching,by supplementary written instructions,and by standard record forms, and checkedto the greatest extent possible by an inde-pendent observer. Anything unusual wasto be pursued in an appropriate way, sothat any existing disease or other condi-tion could be correctly diagnosed, andthe patient referred for any needed treat-ment. Despite considerable success inthe early years of the program, after thefirst few cycles it proved difficult to main-tain the high standards set, especiallywith respect to independent review andrecordkeeping.

Although the process of examining pa-tients proved very difficult to standardizeand control indefinitely, the AHS subjectsresponded well to the biennial solicita-tions, and their continued flow throughthe clinics, in 24 randomly selected sub-sets called up in turn throughout eachmonth of the cycle, provided an unusualopportunity for the special surveys thathave been the chief output of the AHSprogram (54). And, although it did notprove possible to improve their interimhistories by means of mail questionnairesor home visits, or by systematic tappinginto the medical care system, it was possi-ble to collect information on particulardiseases, especially individual cancers(37), from the local hospitals with whichto supplement the information obtainedin ABCC clinics. The cancer registriesoffer the chief device for supplementingthe observations made in the clinic, butit has become clear that the AHS sampleis too small and the entire LSS sample isnow the preferred sampling frame forcancer incidence as well as mortality.

Although the ABCC program hasexisted to detect and measure the late

effects of ionizing radiation, ABCC in-vestigators have taken an interest indisease whether or not radiation ap-peared to be involved. Both clinical andlaboratory interests have contributed tothe study of a wide variety of diseasesand potential disease-markers andgreatly enriched the observationalbase (55).

Genetics and cytogenetics. A remark-able aspect of the early genetic workwas the ascertainment of pregnancythrough the registration of pregnantwomen for supplementary rice rationsafter the fifth month of pregnancy, andof deliveries through the reports of at-tending midwives and, occasionally,physicians. Information on stillbirth,birthweight, sex, and any malformationwas obtained from the attending midwifeor physician. An ABCC physician thenexamined the child in the home, at once ifthere was a report of an abnormal termi-nation, otherwise on a more leisurelyschedule. By these methods it was pos-sible to study over 93 per cent of all thepregnancies that went to term in thestudy period, 1948-1953, over 76,000in all. In addition, a random 28 per cent ofthe births were selected for examinationat nine months of age to detect malforma-tions that might have been overlookedat birth, and to obtain anthropometricdata. Autopsies were also performed,especially in Hiroshima, where 62 percent of infants stillborn or dying in thefirst six days of life were examined atnecropsy. The indicators of possible radia-tion damage were sex ratio, birthweight,anthropometric measurements, still-birth, neonatal death, and gross mal-formations.

Cytogenetic methods first establishedby Bloom et al. (26) in 1965 play a largerole in the studies of A-bomb survivors(27). The cell preparations of survivors areexamined for interchange aberrationsand clones of cells with the same aberra-tion in single individuals. The study ofthe offspring of irradiated parents is


directed at genetic effects, particularlyas these may be manifested by cells con-taining abnormal numbers of sex chromo-some and structural chromosomal ab-normalities (28).

In the Biochemical Genetics Study,initially 25 biochemical markers wereselected for the search for variants amongthe Fi rare enough to justify examina-tion of parents to establish the variantas simply a rare, inherited form, or apossible mutant. Additional markershave since been added to give the inquirymore power.

Statistical methodsOf interest here is not the inventory of

specific statistical procedures employedby ABCC/RERF investigators, but theway in which they have been applied tothe problems of inference and estimationthat have arisen in the program. Statisti-cal methods are basic because, with thepossible exception of radiation cataracts,the lesions produced by ionizing radiationare not unique, or identifiable as to spe-cific etiology, so that statistical compari-son provides the only means of attribut-ing them to radiation.

Screening for effects. This was easyenough in the case of leukemia wherethe relative risk among those exposed tothe highest dose levels was on the order of50 to one. But for the general class ofcarcinogenic effects relative risks of 10or more are uncommon, and the average,on the order of two or three. The risk ofmissing real, but small, effects is everpresent (56). Most of the screening hasbeen done with dichotomies of high- vs.low-dose groups adjusted for any demo-graphic differences that might otherwisebias the comparison. In recent years thesehave been supplemented by regressionanalyses in an effort to achieve the ut-most power that the dose distributionswould permit.

In the screening efforts of ABCC inves-tigators, little attention has been paid tothe RBE for neutrons in analyses that

combine the experience of the two cities.Most such analyses have been, in fact,city-specific, and the more that is learnedabout the human effects of ionizing radia-tion the more necessary seems a city-specific approach to both screening andestimation. Although most dose-responsecurves for specific forms of cancer appearto be approximately linear for Hiroshima,dose-response curves for Nagasaki arequite uncertain. Further, not only do theNagasaki data look non-linear in someinstances but non-linearity cannot beexcluded on statistical grounds even forbreast cancer incidence, which appearslinear to the eye (37).

Dose-response. The linear hypothesisbecame rather firmly established inradiation protection work during themid-1950s on the basis of experimentalwork on the genetic effects of ionizingradiation. The force of numerous experi-mental demonstrations of non-lineardose-response functions for low-LETradiation, combined with the possiblycurvilinear dose-response function forleukemia among Nagasaki survivors(45), however, has injected a great dealof uncertainty into all discussions ofhuman data, especially as they are, em-ployed to estimate low-dose effects. Inthe NAS 1972 BEIR Committee reportABCC data were extensively relied uponto produce low-dose estimates on thebasis of the linear hypothesis (35).Manipulation of the Hiroshima andNagasaki leukemia data by severalinvestigators (45-48) has underscoredthe possibly linear-quadratic form of thedose-response relationship for gammaradiation. In none of these analyses,however, has it been possible to showthat the non-linear function fit the datasignificantly better than the linear func-tion. Moreover, the Nagasaki leukemiadata in the LSS sample, on which theseresults primarily depend, are very few,include 0 observations in one or two dose-intervals, depending on the grouping ofdose, and exclude 70 per cent of the infor-


mation in the Leukemia Registry of all.survivors diagnosed in the area in the1946—1974 period, whether or not mem-bers of the LSS sample. If the entireRegistry sample is examined in relationto the dose-distribution of all survivorsenumerated in Nagasaki at the time ofthe 1950 Census, a much less curvilineardose-response function is obtained (49). Itis by no means certain, therefore, that thetrue relationship for the leukemogenicresponse to low-LET radiation is notlinear.

More recent work (57) has made use of adose-response function for low-LETradiation suggested by Brown (58) onthe basis of experimental and some theo-retical considerations:

y =(aD + bD2)exp -*D ~ dD'2 (1)

where y denotes the excess above normalincidence, D is dose, and a, b, c and d arefitted constants constrained to be posi-tive. The exponential term imparts down-ward curvature in the region of cell-killing, but the number of constants tobe fitted is usually too large for humanseries. Hence,

y = aD + bD2 (2)

is the more practical form of expression1. When both Hiroshima and Nagasakidata are to be fitted, separate accountis taken of gamma and neutron dose inthe expression

y = oDy bDY + cDn (3)

where Dy is the gamma dose, Dn theneutron dose, and a, b, and c fitted con-stants. Depending on the data, the valuesof the fitted constants will emphasizeone or another of these terms more orless than others, so that the best fittingfunction may be linear in both gammaand neutron dose, quadratic in gammabut linear in neutron dose, or both linearand quadratic in gamma and linear inneutron dose. These functions have beenfitted to the death certificate data forthe period, 1950-1974, to breast cancer

incidence data, to the Leukemia Registrydata, and to head-size of the in uteroexposed (57, 59). In no case was it possi-ble to discriminate among the competingdose-response models for the effects ofgamma radiation. A corollary findingwas that RBE estimates had extremelylarge standard errors. Further, the esti-mates of a and b obtained in analyses ofboth cities jointly differed little fromthose obtained from fitting the Nagasakitotal dose, the neutron component ofwhich is only about 2 per cent.

RBE estimation. In their 1974 paper onleukemia among the A-bomb survivorsin the LSS sample, Rossi and Kellerer(45) suggested that the RBE for neutronsin the leukemogenic response was about45 Af* where Dn is the neutron dose.Although several other investigatorshave obtained similar results on the samedata (46—48), the Nagasaki leukemiacases in the LSS sample are too few toinspire great confidence in the result.The linear-in-gamma, linear-in-neutronmodel is equally consistent with the data,and implies a constant RBE of about 10(59).

Latent period, duration of effect. Whenmodels are created for radiation protec-tion purposes and the human experiencestudied for risk coefficients to be em-ployed with them, one must specify, ineffect, the natural history of whatevereffect is being modeled—how long afterexposure it is before the effect becomesapparent, when it peaks, and how longit lasts. For leukemia ABCC investiga-tors have shown that the effect probablybegan two to four years after the bombing,peaked six to seven years after the bomb-ing, and essentially disappeared about 30years after the bombing. These resultsare in accord with those obtained fromthe experience of the British patientswith ankylosing spondylitis treated byx-ray (60, 61). For solid tumors, however,beginnings and endings are much moredifficult to establish, although it is fairlycertain that none of the effects, except


perhaps bone cancer, is as short-lived asleukemia. The upswing marking the endof the latent period is a gradual process,and a purely empirical approach, withoutthe benefit of an a priori model, is rela-tively unsatisfactory (49). Present indica-tions are that latent periods for solidtumors probably vary, depending on thetumor, and that they are generally 10years or longer (57).

Patterning longitudinal observations.With a biennial examination cycle, astandard screening examination on about85 per cent of eligible subjects at eachcycle, and 10 cycles completed, ABCC/RERF investigators would seem to haveone of the greatest longitudinal data-banks in the world. Yet strictly longi-tudinal analyses, i.e., those in whichserial observations on each individualare assembled and patterned in variousways, as might be done with growth data,are notably few. In recent analyses ofblood pressures (62) and serum choles-terol (63), ABCC investigators haveexamined trends in the movement ofthese values over time or with age, butonly on a group basis. To predict hyper-tension and cardiovascular mortality,individuals were characterized on thebasis of a single early blood pressurereading; most readings were not used.In a growth and development study,however, Shohoji and Pasternack (64)were able to derive a function that couldbe fitted to the serial data of each indivi-dual to characterize his growth pattern,the parameters of which could then bestudied in relation to radiation dose orany other risk factor.

Competing risks. Although much ofthe early experimental work on the accel-erated aging effect of ionizing radiationdid not make use of life-table or othermethods to untangle competing risks (65),epidemiologists working with humandata are well aware of this problem. AtABCC life-table techniques have beenlittle used, but expectations have typi-cally been based on person-years or the

equivalent, i.e., the number of survivorsin successive short intervals. At the endof 1974, just 25 per cent of the A-bombsurvivors in the LSS sample had died, sothat there might be room for distortion ifthe problem of competing risks were notkept in mind. Groer (66) has recentlyquestioned the validity of ABCC/RERFanalyses on the grounds that competingrisks may have been ignored, but therecord suggests otherwise.

Other risk factors. Analyses of demo-graphic factors have shown an apprecia-ble sex difference in the risk of radia-tion-induced leukemia (67), and age dif-ferences in the risk of most radiation-induced solid tumors (49). The dose-groups are fairly well balanced as todemographic composition within city, butcustomarily age- and sex-adjustments aremade, either by means of the Mantel-Haenszel procedure (68) or indirectstandardization by means of age-, sex-,and calendar year-specific death ratesfor all Japan (34). There is more varia-tion among dose-groups with respect tosocioeconomic factors, however, since thedose in air is a function of distance fromthe hypocenter. The influence of socio-economic factors has been too littlestudied by ABCC/RERF investigators.Such studies as have been performedhave not uncovered evidence of con-founding sufficient to affect dose-responserelationships (49, 69, 70). Of greatercurrent interest is the possibility of inter-action between radiation and other riskfactors for cancer. Epidemiologic studiesof leukemia by ABCC investigators havefailed to identify other important riskfactors in the A-bomb experience (71), butfor both breast cancer (69) and lung can-cer (70), factors other than radiation havebeen shown to be important; in neithercase has interaction been demonstrated,but the studies are small. Finally, itmust always be borne in mind that mostof the energy released by the bombs wasin the form of heat and blast, and that allthree forms of energy-release are highly


correlated. ABCC investigators haverelied mainly on the correspondence oftheir results with experimental and otherhuman findings and have not generallyattempted to factor blast and heat intotheir analyses. An exception is the jointanalyses Jablon et al. (7, 72) made ofradiation and burns, which suggestedthat, for fixed distance, mortality wasunrelated to a history of burns.

CONTRIBUTIONS TO KNOWLEDGE

A full appreciation of the scientific con-tributions of the ABCC/RERF programis best gained from the reports of theUnited Nations Scientific Committee onthe Effects of Atomic Radiation (73),from the NAS BEIR Committee reports(35, 57), and from the 1975 supplementto the (Japanese) Journal of RadiationResearch (74). Here there is reason tomention merely the highlights.

Genetic effects. The genetic studies ofNeel and co-workers provide the only sub-stantial data on the probable geneticeffects of ionizing radiation in man. Theirlarger significance lies in their usefulnessin setting approximate limits on the fre-quency of radiation-induced mutationsin man and in suggesting that humangenes are not much more mutable thanthose of experimental organisms, notablymice (29). The pregnancy terminationstudy had approximately 90 per centpower to detect a doubling of the normalrates of stillbirth, neonatal death, andmalformation (2). In their second reporton the F! mortality study, Neel et al. (75)estimated the minimum genetic doublingdose for mutations resulting in deathduring the first 17 years of life to be 46rem for fathers, 125 rem for mothers.The promise of the present BGS is thatit may provide a more direct measure ofthe mutagenic effect of ionizing radiation.The studies of chromosomal aberrationsof leukocytes have been particularly use-ful in quantitating the dose-responserelationships for both gamma and neu-

tron radiation and in demonstrating theexistence of clones of aberrant cells (76).

Teratogenesis. The teratogenic effectof ionizing radiation was well establishedbefore the A-bomb (77), but it has beenboth better defined and more preciselymeasured by studies of the A-bomb sur-vivors than by studies of other series.Diminished stature, small head size, andmental retardation have all been shownto be dependent on the size of the dose tothe fetus, and probably also depend onthe quality of the radiation (4, 78). Neu-trons appear to be more effective thangamma radiation, and the dose-responsefunctions may also differ, at least formicrocephaly and mental retardation(78). Studies of fetal age ATB showedthat the central nervous system wasespecially sensitive to radiation duringthe seventh to the 15th weeks of gesta-tion (79).

Cataracts. The cataractogenic effectof ionizing radiation also was shownbefore 1950, when the first reports onA-bomb survivors began to appear (80).The ABCC/RERF studies have shownthat clinically significant cataracts wereprobably produced by ionizing radiationonly at high doses (81) and that the preva-lence of incomplete lesions is dose-dependent (82). Although the qualityfactor for neutrons is probably wellabove unity, this has not been definitelyshown by ABCC investigators (83).

Cancer. It is with respect to the car-cinogenic effect of ionizing radiation thatthe studies on A-bomb survivors havebeen most fruitful, contributing informa-tion on the specific tumors caused bywhole-body radiation (57, 73), on dose-response aspects of radiation carcino-genesis (34, 48, 49), on the quality factorfor neutrons (48, 59), on the significanceof other environmental factors and ofhost factors on the carcinogenic response,and on latency and duration of effect.

Life-shortening. The life-shorteningeffect of ionizing radiation has been


repeatedly demonstrated experimentally,and by implication also in man on thebasis of the carcinogenic effect. Thereare, however, few human studies thatprovide information on the basis for thelife-shortening effect. The ABCC/RERFstudies suggest that, in man, the life-shortening effect is probably dependentupon the carcinogenic effect alone, thatthere probably are not other effects thatcontribute materially to the mortalityburden of exposure to ionizing radiation(84). The data on patients with ankylosingspondylitis treated by x-ray suggest thesame conclusion (85), but the mortalityexperience of US radiologists, when con-trasted with that of medical specialistsless exposed to ionizing radiation, holdsopen some possibility that the life-short-ening effect may not depend entirely uponexcess cancer mortality (86).

Fertility. Investigations of fertility atABCC have been hampered by insufficientcontrol over the influence of contracep-tion and abortion. In failing to demon-strate a lower level of fertility (87) or ahigher prevalence of sterility (88) amongA-bomb survivors, however, ABCC in-vestigators have provided significantsupport for the view that human sterilityresults only when the dose level is high,perhaps in the lethal range for the male,and that any effect of lower doses onfecundity, e.g., through diminishedspermatogenesis, is short-lived. Studiesof the in utero exposed are in a prelimi-nary stage, but have not thus far pro-vided evidence that fetal exposure in-duced significant sterility in the A-bombsurvivors (89).

Non-radiation work. In addition totheir contributions to knowledge of theeffects of ionizing radiation on man,ABCC investigators have provided valu-able information on a number of diseasesin Japan, especially cardiovasculardisease, diabetes, rheumatoid arthritis,thyroid disorders, gastric cancer, tubercu-losis, pulmonary disease, urinary disease,

and diseases of the liver. This work hasrecently been reviewed by Hamilton andBrody (55). Of particular interest to epi-demiologists is the Ni-Hon-San study,the comparative clinical, pathologic, andepidemiologic study of cardiovascular andcerebrovascular disease in 45—69-year-old Japanese males living in Hiroshimaand Nagasaki, Honolulu, and California(19). The first genetic study led naturallyto one on the genetic effects of consan-guinity, as it enabled Neel to assemble asample of 5300 children of consanguineousparents. In their 1958-1960 study ofthese children and controls, Schull andNeel (90) found the effect of inbreeding onmortality and morbidity in the Japaneseto be well below estimates reported formost Caucasian populations.

SOCIOLOGIC PERSPECTIVES

US-Japanese relationsThe Atomic Bomb Casualty Commis-

sion was a unique institution, and itshistory, as well as its transition to aJapanese national foundation under theWelfare Ministry, deserves the criticalscrutiny of competent social scientists.Rooted in an act of war, nourished by anoccupying military force, and maintainedfor three decades by funds provided bythe US agency most responsible not onlyfor research in radiation biology, but alsofor the development of nuclear technology,it has nevertheless made major contri-butions to knowledge of the empiricalrisks of ionizing radiation on humanhealth. It is a tribute to the tolerance ofthe Japanese people, and to the under-standing attitude of the Japanese govern-ment that, despite some local hostility,despite some confusion in the press andlocal public opinion as to the mission ofABCC, despite the early ABCC decisionto engage in research without providingan extensive medical care service, anddespite the frequent portrayal of ABCCas a minimizer of radiation effects, theinstitution was permitted to continue


and to achieve an international reputa-tion for thorough scientific observationin a controversial field.

The transformation of the AmericanABCC to the Japanese RERF took place30 years after WW II, and was stimulatedmore by a fiscal crisis brought on byJapanese inflation and the decliningvalue of the dollar relative to the yenthan by considerations of the effective-ness of an American-controlled study inthe Japanese community. Darling,as director of ABCC since 1957, had thevision and the courage to call attention,in his 1967-1968 annual report, to theanachronism that was ABCC 23 yearsafter WW II. But it took the subsequentfiscal crisis, which carried the threat ofclosure, to bring the two governments tothe conference table.

Would ABCC have been more success-ful if the Japanese presence had becomedominant earlier in its history? Certainlyone can see in the contraction of ABCCin its last decade, in the difficulty ofrecruiting Japanese scientists for evenshort-term assignments, and in the in-creasing difficulty faced by ABCC inves-tigators in getting data from the medicalcare facilities in Hiroshima and Naga-saki, ample reason to believe that, underJapanese control, the institution wouldhave been even more successful. Andyet there is nothing in the four-yearhistory of RERF to support this. All theold difficulties face RERF as they didABCC. Japanese university investigatorsare active contributors to the radiationbiology journals, but have shown littleinterest in epidemiologic studies of thehuman effects of ionizing radiation. Epi-demiology, however, is not well developedin Japan and it lacks support from itssister discipline, statistics, which is littleapplied to medicine in Japan. The formalstudy of statistics is largely confined todepartments of mathematics in Japaneseuniversities, and no great interest hasbeen taken in medical applications. Thisis in sharp contrast to the statistical tech-

niques involved in quality control, carriedto Japan by W. Edwards Deming duringthe occupation, embraced by Japaneseindustry, and long taught mainly in in-dustrial seminars. There are no schoolsof public health in Japan, and the teach-ing program of the Institute of PublicHealth of the Welfare Ministry is aimedchiefly at training health officers.

Although for many years the onlysystematic access to medical care recordshas been through the registries for leu-kemia and other forms of cancer, and thearrangements for these activities have attimes been threatened with cancellation,under agreements with the nationalauthorities, access has been grantedto the family registers containing infor-mation on vital status, and to the causeof death information in the hands of thepublic health centers.

The survivorsIn response to an aggressive contacting

policy, participation in the biennialexamination program of the Adult HealthStudy has fallen off very little since theexaminations began in 1958, despite theABCC policy of gathering research in-formation without providing medical careand a generally unfavorable local press.Perhaps subjects have recognized thattheir examinations would be morethorough and competent than those theymight easily obtain at the hands of a busypractitioner seeing 50 to 100 patients aday. The press has often employed the"guinea-pig" model in inveighing againstthe failure to provide medical care. Insharp contrast to the stable participationrate in the clinical program, the autopsyprocurement rate rose rapidly after thestart of the new program in 1961, peakedin 1963-1964, and then declined to apoint where, in 1975, the Crow Com-mittee recommended discontinuationof the effort, a decision finally made in1979. Although the reasons for the in-ability of ABCC personnel for pathologycontacting to maintain the rate have been


actively debated at ABCC and RERF,factual data are few and no remedialstrategy was devised. There was, how-ever, considerable hostility to the pro-gram in press accounts of ABCC activi-ties, with ABCC pathologists caricaturedas vultures waiting for the A-bomb sur-vivors to die. And, although the autopsypercentage was generally higher if thedeceased had been a subject during life,it was discouraging to find that familieswho had assented to autopsy on a firstoccasion would more often refuse when asecond member died.

To a foreign observer it would appearthat there has been very little effort onthe part of Japanese authorities or themedical profession to reassure the A-bomb survivors about their health. Withso much public misunderstanding of itsmission it would have been folly for ABCCto have attempted in any systematic wayto alleviate the anxiety of the survivorsabout their future health. But one wouldhave thought that Japanese publicauthorities, or local physicians, mighthave paid more attention to this problem.Reading Lifton's accounts of interviewswith survivors (91), for example, one isoften struck by the discrepancy betweenthe survivor's account of his fears andhis stated distance from the hypocenterATB. And when a modified Cornell Medi-cal Health Questionnaire administeredto subjects of the Adult Health Studywas scored for anxiety about health,there seemed to be little relation betweenthis score and radiation dose, the ob-jective indicator of risk. In the 1950—1974 study of mortality among 82,000survivors it was estimated, on the as-sumption of linearity of dose-response,that all 285,000 A-bomb survivors regis-tered throughout Japan in 1950 had prob-ably sustained, in addition to 70,000deaths attributable to other causes in the1950-1974 interval, 415 deaths (90 percent confidence interval, 337-492) fromradiation-induced cancer, including leu-kemia (49). For the 214,000 exposed to

no more than one rad, the parallel fig-ures would be about 10 deaths fromradiation among about 52,500 deathsin all. But no systematic efforts appearto have been made to reassure survivors,especially those exposed to very low doses.For a long time, in fact, every death ofan A-bomb survivor was an "A-bombdeath" in the press, and it would appearthat much of the propaganda about A-bomb survivors was not in their bestinterest because it blurred the distinctionbetween deaths caused by radiation andthose that would have occurred in thenormal course of events. Whether the sur-vivors would, in fact, be able to respondto objective facts with a lessening ofanxiety remains unknown. The acuteexperience was emotionally so profoundthat factual material might have littlereassurance value. But the effort seemsnot to have been made. Instead, the ef-forts to impress the world with the veryreal horrors of nuclear war, and theJapanese government with their need forsupplemental income and medical assis-tance, may, in exaggerating the ill-health of the survivors, tend to perpetuateunnecessary fears among many survi-vors. Although we have learned moreabout the human effects of ionizing radia-tion from the experience of the A-bombsurvivors than from any other source,this knowledge may have benefited themvery little.

StaffingIn the early years of the post-war re-

construction, ABCC employment wasundoubtedly attractive to Japanese pro-fessional and scientific personnel, but asJapanese medical education advanced,and the economy prospered, much of theappeal was lost. A foreign institutionwith only minimal ties in Japan throughthe JNIH, which itself had little strengthto share, and lacking local institutionalroots such as might have been formedwith the local universities, ABCC haddifficulty building and maintaining a


strong Japanese professional staff. Itsgreatest successes have been in responseto unusual leadership and the stimula-tion of new scientific techniques and pro-cedures, as in Neel's genetic studies, inthe cytogenetic program developed byBloom, and in the early epidemiologicleadership of Keizo Nobechi. (Nobechi isbest known for his work on cholera.) Al-though it is difficult for a foreign observerto judge the strength of the various limit-ing factors, it does appear that the associa-tion of ABCC, and RERF, with the WelfareMinistry rather than with the EducationMinistry, and, in particular, the absenceof formal ties to medical schools under theaegis of the Education Ministry, has alsoinhibited Japanese participation in theprofessional and scientific work of theorganization. There was, however, reasonto fear that local control might makemore difficult the task of objective ob-servation and reporting of findings.

For US investigators a tour at ABCC/RERF is generally attractive only in thevery early stages of a career, before pat-terns are set and institutional tiesformed, or in the later stages, when tiesare loosening and interests changing.The opportunity to live and work inJapan is a great magnetic force for re-cruiting US personnel, but the best situa-tion is one in which one's career patternitself takes one to Japan with no disrup-tion of institutional ties. It was for thisreason that the Francis Committeerecommended that the NAS-NRC try toestablish firm links with US institutionscapable of providing staff under securearrangements. None of the three groupslinking themselves to ABCC, however,was both sufficiently dedicated to radia-tion biology, and sufficiently strong, tocontinue the relation indefinitely. Per-haps a consortium of universities wouldhave proved more effective. As long asthe doctor draft was in effect in the US,it was possible to recruit young physiciansfor two-year, alternative-service tours atABCC as officers of the Public Health

Service. This was a very important sourceof professional strength, particularly inclinical medicine and in pathology, butit provided men with little or no investi-gative experience and no training inradiation biology or nuclear medicine.The lack of trained epidemiologists canbe partly overcome by developing colla-borative working patterns between in-ternists, say, and statisticians, but thelack of training in radiation biology isnot so easily overcome.

Any discussion of staffing would beincomplete without some reference tothe fact that the ABCC experience hashad a major influence on the shaping ofresearch interests and professionalcareers for a good many, both Japaneseand Americans, and not only physiciansbut also statisticians, geneticists, andother scientists. This has occurred partly,of course, because so many younger menand women were exposed to ABCC oppor-tunities before their career patterns wereestablished, and partly because of theexceptional opportunities offered at ABCCto those able to perceive them.

Research and service

As the binational discussions leadingto RERF appeared on the horizon, therewas much taking-stock in the Japanesecommunity, especially in medical organi-zations and among local governmentofficials. Three themes dominated thesediscussions: 1) Japanese control; 2) localparticipation in management; and 3) thejoining of research with medical care ofthe survivors. This last theme repre-sented a general reaction against theABCC policy of not providing treatmentwhich was thought to reflect an attitudeof not caring about the welfare of thesurvivors. A research operation thatassumes responsibility for medical care,of course, thereby commits itself to anentirely different level of expenditureand if there was any single source offinancial pressure on ABCC over itsthree decades of operation it was the cost


of the clinical screening operation, evenwithout any assumption of responsibilityfor the general medical care of the clinical(AHS) sample. Further, there was nothingcallous or uncaring about either ABCCpolicy or practice. Not only were subjectsgiven written reports of their examina-tion findings but also their physicianswere informed. Where referral was indi-cated it was made. For many years theABCC clinic in Hiroshima maintained a"diagnostic" ward of 12-15 beds wheretreatment was provided to patients need-ing specialized care that ABCC physi-cians seemed best able to provide. Sometreatment was routinely provided inABCC outpatient clinics, e.g., to thosewith low hemoglobin levels. And amedical social work program assisted theindigent with funds and referrals forneeded medical care. Had the clinics notbeen run in a caring fashion, the sub-jects would probably not have been will-ing to participate in the biennial exami-nation program as fully as they actuallyhave for more than two decades, especial-ly in view of the great improvementsJapan has made in its medical care sys-tem. Nevertheless, the ABCC clinics weremerely on the periphery of that systemand their importance to the community,while great in the early years of recon-struction, gradually declined as Japanesemedical education and medical facilitiesimproved. Since about 85 per cent ofthe clinical subjects returned for theirbiennial examinations, and since the lateeffects of ionizing radiation are largelyconfined to cancer, it is by no means clearthat an integration of research and medi-cal care would have brought resultssooner, or that results would have beenmore definite or better understood. Butthe work of ABCC/RERF would have hadgreater acceptance in the community,and its foreign character might not haveseemed so blameworthy. Morbiditywould have been more fully ascertainedand the opportunity might have beencreated for systematic morbidity reports

akin to the mortality reports that havebeen so well accepted. And more infor-mation would have been available forexamining the role of ionizing radiationin the presence of other risk factors. Butit is hard to see how the general natureof the findings, their cogency, and theirsufficiency, could have been very dif-ferent even if full ascertainment of mor-bidity had been possible. If the late ef-fects of ionizing radiation were lessconfined to cancer, of course, a verydifferent evaluation might have beenmade.

SUMMARY AND CONCLUSIONS

Reflecting on the history of ABCC, andon the early years of RERF, the successororganization, one sees a very imperfectoperation that has, nevertheless, achieveda measure of success that warrants itsworldwide recognition in an area of in-creasing social importance. In its originsone discerns a clear vision of goals, butnot of research strategy; its early historywas characterized by some fumbling andmisdirection in the search for somaticeffects of ionizing radiation, while itssearch for genetic effects was marked byexcellence of both strategy and execution.Frequent changes in leadership and fiscalcrises brought it to the brink of disasteron more than one occasion, but the ABCChas been maintained by the vision andgood sense of some few individuals, bythe historic importance of its mission,and perhaps by some early positive find-ings. Had the military occupation beenless benign and the Japanese nationalcharacter less tolerant, ABCC mightwell have foundered in its difficultiesduring its first decade.

Of particular significance for epi-demiologists is the dynamic push thatwas given to the program by an epi-demiologic review and redirection underan overall research strategy that provedhighly effective for the ensuing twodecades. If there is any single, mostimportant lesson to be learned from


the ABCC experience, it is the impor-tance of a research strategy in any long-term, prospective, follow-up study. Thegeneral nature of the possible effectsmust be visualized, the observationsneeded to detect such effects specified,and sampling alternatives carefullyassessed. A particular merit of theFrancis proposals was that observationsof differing biologic depth became mutallyre-inforcing through the use of a commonfixed cohort of subjects. The strategyworked well not only during the imme-diate years after the Francis Report, butalso throughout the second decade whenthe goals became more specificallyfocused on cancer, and the emphasisbegan to shift from screening to themeasurement of effects. Even now it isoutmoded only in some of its particulars.

The lessons of ABCC apply ratherdirectly to the investigation of a disastersituation from which long-term healtheffects are feared, and with only some-what less force to any long-term, prospec-tive follow-up study. For me the mostimportant lessons are:

1) Devise an adequate overall strategyearly;

2) Register or identify the populationof interest as soon as possible, takingparticular care to include all the identi-fiers on which follow-up may depend;

3) Determine the parameters of theexposure in fine detail, with emphasison objective physical measurements;

4) Employ a cohort approach with aclear plan for making statistically power-ful comparisons to identify and to mea-sure effects, either in dose-specific fashionor in exposed vs. unexposed comparisons;

5) Evaluate carefully the potentialvalue and cost-effectiveness of alternativeend-points, particularly mortality vs.morbidity and defects;

6) Plan staffing patterns to provide notonly excellence of leadership and scientif-ic performance, but also continuity;

7) In a foreign operation of long dura-tion, endeavor to sink deep local roots.

Registration of the population anddetermination of its exposure need tobe done at the earliest possible moment,before migration or blurred memoriesmake the task difficult. A complete regis-tration permits flexibility in samplingand provides a source to which one canreturn if critical samples are too smallor need change. For the long term, goodidentification is essential.

Case-control methods have their place,even within cohort studies, but in a long-term, prospective study the investigatorneeds a solid grip on the population ofinterest that only the study of a fixedcohort can provide. The results of ABCCstudies based on all survivors in the city,or otherwise lacking a firm samplingbase, were always suspect, and it wasone of these that triggered the assign-ment to the Francis Committee. Butcohorts can be too small, and it is clearnow that a mistake was made in selectingthe Nagasaki sample, and that it shouldhave been made as large as possible. Itsimply was not appreciated, in the earlyyears, that the Nagasaki sample wouldbe more important than the Hiroshimasample because of the mixed gamma-neutron quality of the Hiroshima radia-tion. Crucial issues relating to the effectof low-dose, low-LET radiation and dose-response relationships perhaps remainmore uncertain today than they mighthave been had the importance of the Naga-saki experience been fully appreciatedwhen sampling plans were made.

An important lesson of the ABCCexperience is that serious considerationsshould always be given to mortality asan end-point, if this is consistent withthe expected nature of the effects beingsought. If the effects are being expressedin terms of both morbidity and mortality,then the cost-effectiveness of the mor-tality approach may outweigh that of themorbidity approach by a wide margin.Programs requiring clinical examina-tions can be prohibitively expensive ifthe effects being sought will be expressed


in increased mortality. There is a wide-spread tendency to deprecate the deathcertificate, but the experience of ABCCshows that this was the single mostimportant document pertaining to end-results. In another context, of course,this might not be the case, and in 1947it could not have been known that cancerwould be the dominant finding. But thereare ways of strengthening death certifi-cate information, as has been done atABCC, especially through the use ofautopsy information and disease regis-tries.

The experience of ABCC clearly showsthat a rotating staff can contribute toinstability and discontinuity of pro-gram, especially when top leadership isalso unstable. ABCC learned how to mini-mize this problem in time, but not untilthere was an overall research strategyto govern the establishment of prioritiesand the allocation of resources. As direc-tor, Darling insisted on research protocolsthat were binding on the institution andthe individual departments, for all toooften a man could not finish his projectin a two-year tour, and the larger pro-gram elements (LSS, AHS, Fi mortality,pathology, etc.) had to endure, whateverthe preferences of individual investiga-tors might be. Having a single, strongdirector for 15 years greatly facilitatedthe implementation of this policy andcontributed to the stability of the pro-gram in other ways. Finally, although itssignificance might not apply generally,in an overseas operation especially itis important that there be a home basefor personnel sent out on a short, e.g.,two-year, tour of service. The ABCC effortin this direction was helpful, but providedno ideal solution; the affiliated depart-ments were too small and too little com-mitted to radiation biology. The UCLADepartment of Pathology could send menfor only one year at a time and quicklyran out of staff to send. The arrangementwith Yale provided chiefs of medicineand supporting staff through 1974, and

Stuart C. Finch, who served as chiefin 1960-1962 while on leave from Yale,is the current Vice-Chairman and chiefof research at RERF. In its 18-year periodof responsibility for statistics and epi-demiology the NRC Medical Follow-upAgency provided from its own staff adepartment head for 12 years, and otherprofessional staff for an additional eightman-years. But a consortium of universi-ties might have had more power and pro-vided more effective input than thesethree separate departments.

Neel's continued interest in the geneticeffects of the nuclear radiation from thebombs has been very effectively imple-mented in an entirely different pattern.Without taking a staff position after 1952or stationing Michigan University staffat ABCC after 1956, Neel has neverthe-less managed to inspire and direct thework in genetics (apart from cytogenetics)over a subsequent period of more thantwo decades. He is now directing the Bio-chemical Grenetics Study on the basis ofvisits by himself and members of his staff,and is directing the assignment to theproject of RERF staff he has trained inAnn Arbor. This pattern is a very effec-tive one, provided that a viable core or-ganization has independent support.

Finally, although conditions will varyfrom country to country and time to time,the experience of ABCC suggests thatsurvival of a long-term research opera-tion in a foreign country may be difficultunless deep local roots are developed.ABCC and its staff played a large part inthe life of the medical community in theearly years but, in time, lacking any realresponsibility for the ABCC program, andhastening to achieve parity with medicalschools in the West, the local teachinginstitutions surpassed ABCC in all areasexcept the narrow field of its specificmission. These institutions providedoccasional junior staff, and cooperatedin various ABCC studies, but did so with-out committing any of their real strengthto the effort. Had the mission of ABCC


been as fascinating in 1965 as was themission of the Joint Commission in 1945,the situation might have been otherwise,even without shared responsibility.Perhaps seeing little opportunity in theABCC program for more than descriptivestudies, the local teaching institutionswent their own way and ABCC found itincreasingly difficult to maintain a strongJapanese staff or a vital place in themedical community.

REFERENCES

1. Oughterson AW, Warren S: Medical Effects ofthe Atomic Bomb in Japan. New York, McGraw-Hill, 1956

2. Neel JV, Schull WJ: The Effect of Exposure tothe Atomic Bombs on Pregnancy Terminationin Hiroshima and Nagasaki. Washington, DC,National Academy of Sciences, 1956

3. Folley JH, Borges W, Yamawaki T: Incidence ofleukemia in survivors of the atomic bomb inHiroshima and Nagasaki. Am J Med 13:311—321, 1952

4. Plummer GW: Anomalies occurring in childrenexposed in utero to the atomic bomb in Hiro-shima. Pediatrics 10:687-693, 1952

5. Francis T, Jr, Jablon S, Moore FE: Report ofAd Hoc Committee for Appraisal of ABCC Pro-gram, 1955. Hiroshima, Atomic Bomb CasualtyCommission TR 33-59, 1959

6. Hollingsworth JW, Beebe GW: ABCC-JNIHAdult Health Study. Provisional Research Plan.Hiroshima, Atomic Bomb Casualty CommissionTR 9-60, 1960

7. Jablon S, Ishida M, Yamasaki M: Studies of themortality of A-bomb survivors. 3. Description ofthe sample and mortality, 1950-1960. RadiatRes 25:25-52, 1965

8. Beebe GW, Kato H, Land CE: Studies of themortality of A-bomb survivors. IV. Mortalityand radiation dose, 1950-1966. Radiat Res48:613-649, 1971

9. Kato H: Mortality in children exposed to theA-bombs while in utero. Am J Epidemiol 93:435-442, 1971

10. Kato H, Schull WJ, Neel JV: Survival in chil-dren of parents exposed to the atomic bomb.A cohort-type study. Am J Hum Genet 18:339-373, 1966

11. Beebe GW, Yamamoto T, Matsumoto YS, et al:ABCC-JNIH Pathology Studies, Hiroshima-Nagasaki. Report 2, October 1950-December1965. Hiroshima, Atomic Bomb Casualty Com-mission TR 8-67, 1967

12. Ishida M, Beebe GW: Joint JNIH-ABCC Studyof Life-Span in Atomic Bomb Survivors. Re-search Plan. Hiroshima, Atomic Bomb CasualtyCommission TR 04-59, 1959

13. Auxier JA: Physical dose estimates for A-bombsurvivors—studies at Oak Ridge, USA. J RadiatRes 16 (Suppl):l-ll, 1975

14. Ishida M: Statistical aspects of tumor registry,

Hiroshima-Nagasaki. Bull Int Stat Inst 38:223-224, 1961

15. Steer A, Moriyama IM, Shimizu K: ABCC-JNIH Pathology Studies, Report 3, Hiroshimaand Nagasaki. The Autopsy Program and theLife Span Study. Hiroshima, Atomic BombCasualty Commission TR 16-73, 1973

16. Anderson RE: Aging in Hiroshima atomic bombsurvivors. Arch Pathol 79:1-6, 1965

17. Anderson RE, Yamamoto T, Ishida K, et al:Aging in Hiroshima and Nagasaki atomic bombsurvivors: Accumulation of cardiac lipofuscin.J Am Geriatr Soc 19:193-198, 1971

18. Anderson RE, Yamamoto T, Thorslund TW:Aging in Hiroshima Atomic Bomb Survivors:Soluble-Insoluble Collagen Ratio. Hiroshima,Atomic Bomb Casualty Commission TR 11-72,1972

19. Syme SL, Marmut MG, Kagan A, et al: Epi-demiologic studies of coronary heart diseaseand stroke in Japanese men living in Japan,Hawaii, and California: Introduction. Am JEpidemiol 102:477-480, 1975

20. Milton RC, Shohoji T: Tentative 1965 RadiationDose (T65D) Estimation for Atomic Bomb Sur-vivors, Hiroshima-Nagasaki. Hiroshima,Atomic Bomb Casualty Commission TR 1-68,1968

21. Kerr GD: Organ Dose Estimates for the JapaneseAtomic-Bomb Survivors. Oak Ridge, TN, OakRidge National Laboratory TR 5436, 1978

22. Hashizume T, Maruyama T: Physical dose esti-mates for A-bomb survivors—studies in Chiba,Japan. J Radiat Res 16 (Suppl):12-23, 1975

23. Ishimaru T, Hoshino T, Ichimaru M, et al:Leukemia in atomic bomb survivors, Hiro-shima-Nagasaki, 1 October 1950-30 Septem-ber 1966. Radiat Res 45:216-233, 1971

24. Beebe GW, Hamilton HB: Future research (onatomic bomb survivors). J Radiat Res 16(Suppl):149-164, 1975

25. Moorhead PS, Nowell PC, Mellman WJ, et al:Chromosome preparations of leukocytes cul-tured from human peripheral blood. Exp CellRes 20:613-616, 1960

26. Bloom AD, Neriishi S, Kamada N, et al: Cyto-genetic investigation of survivors of the atomicbombings of Hiroshima and Nagasaki. Lancet2:672-674, 1966

27. Awa AA: Cytogenetic and oncogenic effectsof the ionizing radiations of the atomic bombs.In Chromosomes and Cancer. Edited byJ German. New York, John Wiley and Sons,1974, pp 637-674

28. Awa AA: Genetic effects: Cytogenetic study (ofatomic bomb survivors). J Radiat Res 16(Suppl):75-81, 1975

29. Crow JF, Kaplan HS, Marks PA, et al: Reportof the Committee for Scientific Review ofABCC, February 1975. Hiroshima, AtomicBomb Casualty Commission TR 21-75, 1975

30. Brill AB, Tomonaga M, Heyssel RM: Leukemiain man following exposure to ionizing radiation.A summary of the findings in Hiroshima andNagasaki, and comparison with other humanexperience. Ann Intern Med 56:590-609, 1962

31. Socolow EL, Hashizume A, Neriishi S, et al:


Thyroid carcinoma in man after exposure toionizing radiation: A summary of the findingsin Hiroshima and Nagasaki. N Engl J Med268:406-410, 1963

32. Wanebo CK, Johnson KG, Sato K, et al: Breastcancer after exposure to the atomic bombingsof Hiroshima and Nagasaki. N Engl J Med279:667-671, 1968

33. Wanebo CK, Johnson KG, Sato K, et al: Lungcancer following atomic radiation. Am RevRespir Dis 98:778-787, 1968

34. Jablon S, Kato H: Studies of the mortality ofA-bomb survivors. 5. Radiation dose and mor-tality, 1950-1970. Radiat Res 50:649-698,1972

35. National Academy of Sciences Advisory Com-mittee on the Biological Effects of IonizingRadiation: The Effects on Populations of Expo-sure to Low Levels of Ionizing Radiation. Wash-ington, DC, National Academy of Sciences-National Research Council, 1972

36. Finch SC, Beebe GW: Aging (in atomic bombsurvivors). J Radiat Res 16 (Suppl):108-121,1975

37. Tokunaga M, Norman JE, Jr, Asano M, et al:Malignant breast tumors among atomic bombsurvivors, Hiroshima and Nagasaki, 1950-1974. J Natl Cancer Inst 62:1347-1359, 1979

38. Finch SC: Personal communication, 197839. Tachikawa K, Kato H: Mortality Among A-

Bomb Survivors, October 1945-September1964, Based on 1946 Hiroshima City CasualtySurvey. Hiroshima, Atomic Bomb CasualtyCommission TR 6-69, 1969

40. Beebe GW, Ishida M, Jablon S: Studies of themortality of A-bomb survivors. 1. Plan of studyand mortality in the medical subsample (Selec-tion 1), 1950-1958. Radiat Res 16:253-280,1962

41. Watanabe S: Cancer and leukemia developingamong atom-bomb survivors. In Handbuch derallgemeinen Pathologie, Geschwulste. Mor-phologie, Epidemiologie, Immunologie. Editedby E Grundmann. Berlin, Springer-Verlag,1974, pp 461-577

42. Arakawa ET: Residual Radiation in Hiroshimaand Nagasaki. Hiroshima, Atomic BombCasualty Commission TR 2-62, 1962

43. Moriyama EM, Otake M: Projections of AtomicBomb Survivor Populations in the Major ABCCSamples: 1970 to 1995. Hiroshima, AtomicBomb Casualty Commission TR 1-73, 1973

44. Auxier JA, Cheka JS, Haywood FF, et al: Free-field radiation-dose distributions from theHiroshima and Nagasaki bombings. HealthPhys 12:425-429, 1966

45. Rossi HH, Kellerer AM: The validity of riskestimates of leukemia incidence based onJapanese data. Radiat Res 58:131-190, 1974

46. Jablon S: Environmental factors in cancerinduction: Appraisal of epidemiologie evidence.Leukemia, lymphoma, and radiation. InCancer Epidemiology, Environmental Factors.Proceedings of the Eleventh InternationalCancer Congress, Florence, Italy, 1974. Vol 3.Amsterdam, Excerpta Medica ICS 351, 1975,pp 239-243

47. Kerr GD, Jones TD, Hwang JML, et al: AnAnalysis of Leukemia Data from Studies ofAtomic-Bomb Survivors Based on Estimatesof Absorbed Dose to Active Bone Marrow.Proceedings of the Fourth International Con-gress of the International Radiation ProtectionAssociation. Vol 3. Paris, France, April 24-30,1977, pp 714-718

48. Ishimaru T, Otake M, Ichimaru M: Incidenceof Leukemia Among Atomic Bomb Survivorsin Relation to Neutron and Gamma Dose, Hiro-shima and Nagasaki, 1950-1971. Hiroshima,Radiation Effects Research Foundation TR14-77, 1978

49. Beebe GW, Kato H, Land CE: Studies of themortality of A-bomb survivors. 6. Mortality andradiation dose, 1950-1974. Radiat Res 75:138-201, 1978

50. Hubbell HH, Jones TD, Cheka JS: The Epi-center of the Atomic Bombs. 2. Reevaluationof All Available Physical Data with Recom-mended Values. Hiroshima, Atomic BombCasualty Commission TR 3-69, 1969

51. Arakawa ET: Radiation dosimetry in Hiroshimaand Nagasaki atomic-bomb survivors. N EnglJ Med 263:488-493, 1960

52. Cheka JS, Sanders FW, Jones TD, et al: Distri-bution of Weapons Radiation in Japanese Resi-dential Structures. US Atomic Energy Com-mission Report CEX 62-11, 1965

53. Jablon S: Atomic Bomb Radiation Dose Esti-mation at ABCC. Hiroshima, Atomic BombCasualty Commission TR 23-71, 1971

54. Belsky JL, Tachikawa K, Jablon S: ABCC-JNIH Adult Health Study. Report 5, Resultsof the First Five Examination Cycles, Hiro-shima-Nagasaki 1958-68. Hiroshima, AtomicBomb Casualty Commission TR 9-71, 1971

55. Hamilton HB, Brody JA: Health surveillancestudies (on atomic bomb survivors). J RadiatRes 16 (Suppl):138-148, 1975

56. Beebe GW: Statistical aspects of the programof the Atomic Bomb Casualty Commission. BullInt Stat Inst 38:203-215, 1961

57. National Academy Of Sciences, Committee onthe Biological Effects of Ionizing Radiation: TheEffects on Populations of Exposure to LowLevels of Ionizing Radiation. Washington, DC,National Academy of Sciences (in preparation)

58. Brown JM: The shape of the dose-responsecurve for radiation carcinogenesis: Extrapola-tions to low dose. Radiat Res 71:34-50, 1977

59. Land CE, Beebe GW, Jablon S: Role of neutronsin late effects of radiation among A-bomb sur-vivors. Presented at the Meeting of the HealthPhysics Society, Minneapolis, MN, June, 1978

60. Court Brown WM, Doll R: Mortality from can-cer and other causes after radiotherapy forankylosing spondylitis. Br Med J 2:1327-1332, 1965

61. Smith PG, Doll R: Age and Time DependentChanges in the Rates of Radiation InducedCancers in Patients with Ankylosing Spondy-litis Following a Single Course of X-ray Treat-ment. Vienna, Proceedings of the InternationalSymposium on the Late Effects of Ionizing


Radiation, International Atomic EnergyAgency, March, 1978

62. Dock DS, Fukushima K: A longitudinal studyof blood pressure in the Japanese, 1958-72.J Chronic Dis 31:669-689, 1978

63. Danzig MD, Lee KK: Time trends in serumcholesterol levels in Hiroshima and Nagasaki(in preparation)

64. Shohoji T, Pasternack B: Adolescent growthpatterns in survivors exposed prenatally tothe A-bombs in Hiroshima and Nagasaki.Health Phys 25:17-27, 1973

65. Walburg HE, Jr: Radiation-induced life short-ening and premature aging. In Advances inRadiation Biology. Vol 5. Edited by JT Lett,H Adler. New York, Academic Press, 1975,pp 145-179

66. Groer PG: Dose-response curves and competingrisks. Proc Natl Acad Sci USA 75:4087-4091,1978

67. Bizzozero OJ, Jr, Johnson KG, Ciocco A, et al:Observations on Type-Specific Leukemia, Sur-vivorship and Clinical Behavior of Radiation-Related Leukemia, Hiroshima-Nagasaki,1946—1964. Hiroshima, Atomic Bomb CasualtyCommission TR 6-66, 1966

68. Mantel N, Haenszel W: Statistical aspects ofthe analysis of data from retrospective studiesof disease. J Natl Cancer Inst 22:719-748,1959

69. Nakamura K, McGregor DH, Kato H, et al:Epidemiologic Study of Breast Cancer in A-Bomb Survivors. Hiroshima, Radiation EffectsResearch Foundation TR 9-77, 1978

70. Ishimaru T, Chihak RW, Land CE, et al: Lungcancer at autopsy in A-bomb survivors and con-trols, Hiroshima and Nagasaki, 1961-1970.II. Smoking, occupation, and A-bomb exposure.Cancer 36:1723-1728, 1975

71. Ishimaru T, Okada H, Tomiyasu T, et al: Occu-pational factors in the epidemiology of leukemiain Hiroshima and Nagasaki. Am J Epidemiol93:157-165, 1971

72. Ishida M, Jablon S: JNIH-ABCC Life SpanStudy, Hiroshima and Nagasaki. Report 4,Mortality in A-Bomb Survivors by Age Co-horts 1950-1959. Atomic Bomb CasualtyCommission TR 14-64, 1964

73. United Nations Scientific Committee on Ef-fects of Atomic Radiation: Sources and Effectsof Ionizing Radiation, 1977 Report to theGeneral Assembly. New York, United Nations,1977

74. Okada S, Hamilton HB, Egami N, et al (eds):A Review of Thirty Years Study of Hiroshimaand Nagasaki Atomic Bomb Survivors. J RadiatRes 16 (Suppl): 1-164, 1975

75. Neel JV, Kato H, Schull WJ: Mortality inchildren of atomic bomb survivors and con-trols. Genetics 76:311-326, 1974

76. Awa AA: Chromosome aberrations in somaticcells (in atomic bomb survivors). J Radiat Res16(Suppl):122-131,1975

77. Goldstein L, Murphy DP: Etiology of the ill-health in children born after maternal pelvicirradiation. Part II. Defective children bornafter postconception pelvic irradiation. Am JRoentgenol 22:322-331, 1929

78. Blot WJ: Growth and development followingprenatal and childhood exposure (to atomicradiation). J Radiat Res 16 (Suppl):82-88,1975

79. Miller RW: Delayed effects occurring withinthe first decade after exposure of young indi-viduals to the Hiroshima atomic bomb.Pediatrics 18:1-18, 1956

80. Cogan DG, Martin SF, Kimura SJ: Atomicbomb cataracts. Science 110:654-655, 1949

81. Cogan DG, Donaldson DD, Reese AB: Clinicaland pathological characteristics of radiationcataract. Arch Ophthalmol 47:55-70, 1952

82. Sinskey RM: The status of lenticular opacitiescaused by atomic radiation. Am J Ophthalmol39:285-293, 1955

83. Jablon S, Fujita S, Fukushima K, et al: RBE ofNeutrons in Atomic Bomb Survivors, Hiro-shima-Nagasaki, Hiroshima, Atomic BombCasualty Commission TR 12-70, 1970

84. Beebe GW, Land CE, Kato H: The hypothesisof radiation-accelerated aging and the mor-tality of Japanese A-bomb victims. In LateBiological Effects of Ionizing Radiation. Vol 1.Vienna, International Atomic Energy Agency,1978, pp 3-27

85. Smith PG, Doll R, Radford EP: Cancer mor-tality among patients with ankylosing spondy-litis not given x-ray therapy. Br J Radiol 50:728-734, 1977

86. Matanoski GM, Seltser R, Sartwell PE, et al:The current mortality rates of radiologistsand other physician specialists: Deaths fromall causes and from cancer. Am J Epidemiol101:188-198, 1975

87. Seigel DG: Frequency of live births among sur-vivors of Hiroshima and Nagasaki atomicbombs. Radiat Res 28:278-288, 1966

88. Blot WJ, Sawada H: Fertility among femalesurvivors of the atomic bomb, Hiroshima-Nagasaki. Am J Hum Genet 24:613-622,1972

89. Blot WJ, Shimizu Y, Kato H, et al: Frequencyof marriage and live birth among survivorsprenatally exposed to the atomic bomb. Am JEpidemiol 102:128-136, 1975

90. Schull WJ, Neel JV: The Effects of Inbreedingon Japanese Children. New York, Harper andRow, 1965

91. Lifton RJ: Death in Life: Survivors of Hiro-shima. New York, Simon and Schuster, 1967

REFLECTIONS ON THE WORK OF THE ATOMIC BOMB … bg/Beebe GW 1979... · The atomic bombs dropped on Hiro- ... Japanese National Institute of Health; NAS, National Academy of Sciences;

Documents