Articles

The Effects of Internal Radiation Exposure on Cancer Mortality in NuclearWorkers at Rocketdyne/Atomics InternationalBeate Ritz,1'2 Hal Morgenstern,1'2 Douglas Crawford-Brown,3 and Bambi Youngl1Department of Epidemiology, School of Public Health, University of California-Los Angeles, Los Angeles, California, USA; 2Center forOccupational and Environmental Health, School of Public Health, University of California-Los Angeles, Los Angeles, California, USA;3institute for Environmental Studies and Department of Environmental Sciences and Engineering, University of North Carolina at ChapelHill, Chapel Hill, North Carolina, USA

We examined the effects of chronic exposure to radionuclides, primarily uranium and mixed-fis-sion products, on cancer mortality in a retrospective cohort study ofworkers enrolled in the radia-tion-monitoring program of a nuclear research and development facility. Between 1950 and1994, 2,297 workers were monitored for intenal radiation exposures, and 441 workers died, 134(30.4%) of them from cancer as the underlying cause. We calculated internal lung-dose estimatesbased on urinalysis and whole-body and lung counts reported for individual workers. We exam-ined cancer mortality of workers exposed at diffirent cumulative lung-dose levels using completerisk-set analysis for cohort data, adjusting for age, pay type, time since first radiation monitored,and external radiation. In addition, we examined the potential for confounding due to chemicalexposures and smoking, explored whether external radiation exposure modifies the effects ofinternal exposure, and estimated effects after excduding exposures likely to have been unrelated todisease onset. Dose-response relations were observed for death from hemato- and lymphopoieticcancers and from upper aerodigestive tract cancers, adjusting for age, time since first monitored,pay type, and external (gamma) radiation dose. No association as found for other cancers,including cancers of the lung. Despite the small number-of exposed deaths from specific cancertypes and possible bias due to measurement error and confounding, the positive findings andstrong dose-response gradients observed suggest carcinogenic effects of internal radiation to theupper aerodigestive tract and the blood and lymph system in this occupational cohort. However,causal inferences require replication of our results in other populations or confirmation with anextended follow-up of this cohort. Key words cancer mortality, hematopoietic cancers, internal(alpha) radiation, lymphopoietic.cancers, occupational cohort study, upperaerodigestive tract can-cers. Environ Healtb Perpct 108:743-751 (2000). (Online 28 June 20001kttp://ehpnetl.nie/s.ni/.gov/docs/2000/108p743-75li*s/abstraa.iltml

Compared to a wealth of information abouteffects of low-dose external radiation expo-sures (gamma and X rays), considerably fewerdata are available for quantifying humanhealth risks associated with chronic internalexposure to radionuclides. In animal experi-ments, high internal doses from alpha- andbeta/gamma-emitting radionuclides haveresulted in immunosuppressive and carcino-genic effects in organs where these radionu-clides concentrate (1). The carcinogenicpotential of such radionuclides has been con-firmed in a few human populations exposedto high doses, including uranium miners andmillers, radium dial painters, and patientstreated with Thorotrast and 224Ra (2,3). Thesites of cancer have coincided with distribu-tion patterns for the radionuclides within thebody, with increases in the incidence of lung,liver, and head-sinus carcinomas, as well asleukemias and bone sarcomas.

Studies published to date examininghealth effects in workers in the nuclearindustry who were exposed internally toradionuclides have yielded inconsistent find-ings at dose levels less than 1 Sv (100 rem)(Table 1). The lack of consistency may bepartly a function of differences in the types

of alpha radiation-emitting particles towhich workers have been exposed at differ-ent nuclear facilities; for example, someworkers were primarily exposed to 239Pu and238Pu, others to uranium dusts, a mixture oftritium, plutonium, and other radionuclides,and others to 222Rn or 210po (Table 1). Afteringestion or inhalation, radioactive particles,depending on their size, solubility, andchemical structure, differ in their distribu-tion through the body, their organ residencetime, and the transfer, dissolution, andabsorption of the radioactivity associatedwith the particles (3), and hence might beexpected to vary in their effects across organsystems. Moreover, there has been consider-able variation from study to study in themethods used to estimate internal dose lev-els. Some studies simply used monitoringstatus and/or duration as a crude proxy mea-sure of internal exposure, whereas othersrelied on environmental monitoring of air-borne dust concentrations to approximatepersonal exposures. Several studies usedmore extensive dose-modeling approachesbased on variable combinations of urinalysis,fecal analysis, and in vivo organ or whole-body count data, sometimes in association

with environmental measures, to calculatewhole-body burden (a measure that appliesan equal dose to all organs) or organ-specificdoses such as to the lung, kidney, or spleen(Table 1). Because of large differences inexposure assessment and the lack of power insmaller cohorts with the most in-depth expo-sure characterization, comparisons of internaldose levels and of results across studies areproblematic and the generalizability of find-ings may be limited. However, although thisheterogeneity across studies may prohibit usfrom calculating a common effect estimate orvalidly comparing results across studies, eachstudy contributes information about thepotential carcinogenicity of specific radionu-clides prevalent in the work environment of anuclear facility.

In our study we calculated lung dosesusing several kinds of individual-level moni-toring data provided by the facility to exam-ine the cancer mortality risk associatedprimarily with exposures to uranium andmixed-fission products. Most of the employ-ees included in the analyses were also moni-tored for external (gamma) radiation.

Materials and MethodsStudy Design and Subject Selection

We carried out a retrospective cohort mortal-ity study of workers employed since 1950 atRocketdyne/Atomics International (RAI), ofwhom 4,607 were enrolled in the company'shealth physics radiation monitoring programbetween 1 January 1950 and 31 December1993. The analyses were restricted to those

Address correspondence to B. Ritz, Department ofEpidemiology, School of Public Health, UCLA,Box 951772, Los Angeles, CA 90095-1772 USA.Telephone: (310) 206-7458. Fax: (310) 206-7371.E-mail: britz@oucla.eduiWe gratefully acknowledge the contributions of

G. Wilkinson, F. Yu, Y. Wang, J. Moncau, and T.Riggs. We especially thank R. Harrison, L. Bilick,and miembers of the study's advisory panel.

Vhis sttudy was supported by a grant from theDepartment of Eniergy (subcontract 324A-8701-S0163), which was administered by the PublicHealth Institute in association with the CaliforniaDepartment of Health Services. B. Ritz was sup-ported in part by a fellowship of the GermanAcademic Exchange Program.Received 24 January 2000; accepted 11 April

2000.

Environmental Health Perspectives * VOLUME 108 1 NUMBER 8 1 August 2000 743

Articles * Ritz et al.

ToNal. *1 .7 r TflJ~~~~~~~-i ~1 TI 2r$fl:* t.fNo. ofworkersexpd bdimslide Eposw Mufmnosli

Studysitefrefer T monitored monodfor sures montredworkers Ma findisRockzi V.tal. N -;:e:iess .4 - 13'Wu' l by ..cancers.Inrae jiskieported for al he

fabrication .{ab;4nand.... . . -.burden;(ur.ine (1952-1979, Mand.Jq ppo qncor mortaltyassembly bioassays) 14.5 years) for wtb ptniumbody

,;- - s-wIn. :-- Pu.

,: .. ..%

1. ..7-11l9 .)

ahdMlr^incfr&-fj-(- wo--ns. radim .uniosasa) TEC: 587 ungcncea aporraduumrIning (1vsance)

;:and irocss'mg Femald: 51liung cancers.(1951-19893L,Mallinckrodt: 27 lungcanbcers(194.2-l983)1.(1c:1matching of

:,.: c,on..;t,z.ro,.liaiif.^7̂,s.etocase)

*jIUa' Il-1UAJfR--

.. ..:

plant (13) thoron (1940-1976)Increased ri'sk r6portedfor lung andpancreaticcnrmotlitepoure status-

VOLUME 1081 NUMBER 8 1 August 2000 * Environmental Health Perspectives744

Articles * Internal radiation and cancer mortality

2,297 workers involved in nuclear fuelassembly and disassembly operations whowere monitored for internal radionuclideexposure. We chose to exclude radiationworkers who were not monitored for internalexposures for two reasons: to minimize expo-sure misclassification, since some unmoni-tored workers probably were exposed toradionuclides, especially before 1963; and tominimize possible selection bias resultingfrom differences in unmeasured risk factorsbetween monitored and unmonitored work-ers, a phenomenon demonstrated in theRocky Flats cohort by Wilkinson andMorgenstern (23).

It was necessary to exclude 39 otherwiseeligible workers for whom the records lackedenough information to determine vital sta-tus. We did not restrict the cohort on thebasis of employment duration, race, or gen-der. All but 44 of the workers included inthe internal radiation assessments had alsobeen monitored for external radiation expo-sure. Follow-up for each subject began at thestart of internal monitoring or on 1 January1950, whichever date was later. Follow-upended on the date of death of a cohort mem-ber or on 31 December 1994, whicheverdate came earlier.

Ascertainment of DeathsVital status determinations identified 441subjects who died between 1959 and 1994.We received death certificates of vestedcohort members from the company. If twoindependent company data sources identi-fied an employee as active, and thus alive, atthe end of follow-up, we counted him or heras living. About 10% of the cohort memberswere identified as living.

Employees not identified as alive or deadby company records were checked againstthree different record systems: the SocialSecurity Administration (SSA) beneficiary-records files (period covered, 1935-1994),the vital statistics files for the State ofCalifornia (period covered, 1960-1994),and the U.S. National Death Index (NDI)(period covered, 1979-1994). Matches wereverified from a review of information ondeath certificates. We were able to obtain allbut 12 death certificates for deceased sub-jects. Among these 12 deaths, 7 (58%)workers were unexposed, 4 (33%) belongedto the 1-< 5 mSv group, and 1 (8%) to the5-30 mSv group. Because this exposure dis-tribution is similar to the one observed forall workers (Table 2), we concluded that wedid not differentially lose exposed or unex-posed workers, and thus did not expect thislack of information to bias our results.

The underlying and contributing causesof death recorded on the certificates werecoded using the International Classification

of Diseases, Ninth Revision (ICD-9) (24,25)by a licensed nosologist. The accuracy of thecoding was verified by members of the studyteam. For some analyses the ICD-9 codeswere translated into Eighth Revision of theInternational Classification of Diseases (ICD-8) codes (Table 3). All results presented inthis paper are based on cancers as the under-lying cause of death.

Radiation Measurements

Throughout the study period, RAI conduct-ed periodic bioassays of urine or feces, aswell as in vivo whole-body counts and lungcounts, to estimate internal doses for work-ers assigned to areas potentially contaminat-ed by radioactive materials. These dosesresulted from inhalation and, to a lesser

Table 2. Characteristics of 2,297 cohort members monitored for internal radiation, by sex.

Number of employeesAverage follow-up time (years)Average age at entry into cohort (years)Number of person-yearsNumber of deathsTotal mortality rate (per 1 05/year)Total cancer mortality rate (per 1 05/year)Pay typeSalaried managerial/professionalSalaried technical/administrativeHourly/unionUnknown

Internal radiation dose (mSv)00-

Articles * Ritz et al.

degree, ingestion and skin absorption ofradionuclides. Most of the available internaldose records were for the period 1963-1983.Before 1963, few measures of internal expo-sures were taken, and by 1983, all majoroperations involving radionuclides had beendiscontinued. Company policy during theearly years was to monitor only those indi-viduals with a significant possibility ofreceiving annual lung-dose equivalents inexcess of 150 mSv.

We calculated an estimated internalcumulative dose to the lung for eachemployee. The primary radionuclidesincluded in the dose estimates were a) urani-um, with a range of degrees of enrichmentfor 235U; b) mixed-fission products (unspeci-fied as to radionuclide); c) 90Sr; a) 137Cs;and e) small amounts of plutonium. In addi-tion, measurements of gross beta and grossalpha radiation in samples were available forsome individuals. Methods for convertingbioassay results to annual dose (in units ofmillisieverts) were based on the biokineticmodels of International Commission onRadiological Protection publications (26,27)and on the mathematical techniquesdescribed in a report by Crawford-Brownand co-workers (28,29). This approachyielded the following conversion factors forthe primary radionuclides of interest:* For uranium urinalyses, each 15 dpmexcreted per day translates to an averagevalue of 5 niSv exposure to the lung peryear. This conversion factor represents amean lung removal half-time of 120 daysand a urine excretion fraction of 0.8.

* For in vivo uranium lung counts, the con-version factor is obtained directly from theRAI estimate of the percent maximum per-missible lung burden. In each case, thetime-averaged percent maximum permissi-ble lung burden (%MPLB ) for an individ-ual is multiplied by 1.5 mSv. Theconversion factor for the dose to the bonemarrow is approximately 0.2 mSv/%MPLB.

* For mixed-fission products, the conversiondepends on the availability of informationon the radionudide involved. In cases wherethe radionuclide was specified in the records(e.g., 90Sr or 137Cs), committed effectivedose equivalents had already been calculatedby the facility health physics staff. These cal-culations were checked and, if confirmed,used as the dose for an individual. Where theradionuclide was not specified, a representa-tive conversion factor based on an assump-tion of 90Sr intakes and a class Y retentionhalf-time in the lung was used. The resultingconversion factor is 5 mSv/year to the lungper 250 dpm excreted per day.

* For plutonium, the conversion factor usedwas 10 mSv/year exposure to the lung perdisintegrations per minute per day. This

factor is appropriate for a class Y plutoni-um compound.For urinalysis measurements of uranium,

plutonium, and mixed-fission products, weused radiometric or fluorometric techniques.The more reliable radiometric method wasthe primary basis of dose estimation for ura-nium intakes in this study. When recordslisted "mixed-fission products," it was possi-ble only in a few cases to determine theradionuclide present in the sample. In addi-tion, there were limited in vivo lung countingresults for 235U. For every worker, we exam-ined records for each of the radionuclidesseparately and sorted them by calendar timewithin each year. A time-weighted averagemeasurement for an individual was thenobtained for each year by weighting eachreading in that year by the fraction of theyear until the next reading in the temporalsequence. For example, if XI were a readingobtained on 1 January of a year and if X2were a reading obtained on July 1 of thatsame year, then the average for the yearwould be 0.5Xj + 0.5X2, since each readingwould represent the exposure measure forapproximately 50% of that year. The excep-tion was at the end of the monitoring period(indicated by the end of monitoring recordsfor an individual), in which case the radionu-clide was assumed to be removed with a half-life depending on the particular radionudide,and the resulting integral of activity versustime was calculated. We obtained a time-weighted average lung burden for eachworker. More than 95% of the reported orcalculated doses from internal exposures werefrom uranium and mixed-fission products.

At RAI, the health physics team identifiedworkers potentially exposed to significantinternal radiation doses from airborne conta-minants for inclusion in a routine quarterlymonitoring program. Some workers weremonitored only in the event of accidentsinvolving radioactive material spills. Thus,there might be no measurements available foran individual during a certain period, eventhough exposure may have occurred. Mostrecords did not distinguish between routineand accident-driven monitoring, and weassumed that the record represented a routinemeasurement. Consequently, the assumptionof time weighting used in this study overesti-mates doses for instances in which the mea-surement was due to an accident, but was notdesignated as such.

Fortunately, it was possible to separateroutine and accident-related measurementsfor individuals with large annual doses (> 10mSv in a year). For other measurements thatwere due to an accident, however, we overes-timated the true average annual dose bycounting the measurement as an average dose,instead of a one-time peak dose. On the other

hand, a potential for underestimation of thetrue average annual dose existed due to theminimum detection limits (MDLs) of theassay methods in use (the MDL was 2 mSvfor uranium and plutonium and 0.5 mSv formixed-fission products).

We used RAI records of external radia-tion monitoring, including whole-body dosemeasurements for gamma rays and X rays, tocalculate cumulative dose from externalexposure (30). For those 44 workers nevermonitored for external radiation exposure,we assumed an external dose of 0 mSv.

Treatment of Potential ConfoundersWe used personnel and medical records toexplore such potential confounders as occu-pational/socioeconomic status, race, work-place exposure to carcinogenic chemicals,and smoking history. Based on personnelrecords, workers were assigned to one ofthree pay-type categories (hourly, salariedtechnical/administrative, or managerial/pro-fessional); this variable was used as a proxyfor occupational/socioeconomic status.Employees who changed titles or pay typewere categorized according to the titles andpay types held longest at RAI. The 78 sub-jects lacking job titles and pay type wereassigned to the hourly category. Because RAIdid not systematically collect data on therace of its employees before 1972, we wereunable to control for the influence of thisfactor in our analyses. According to theinformation on death certificates, however,96% of all deceased workers were white.

Job titles, employment periods, and,when available, job locations were used tocreate proxy measures for chemical exposuresduring the study period. We determinedthat hydrazine, asbestos, beryllium, andmany solvents had been used extensively atRocketdyne/AI. We categorized workers ashighly, moderately, potentially, or not likelyto be exposed to asbestos and hydrazine.

Information about tobacco smoking wassystematically recorded for two subgroups ofsubjects in routinely administered medicalquestionnaires from different periods.Questionnaires from 1961 to 1969 indicatedonly whether the worker was a smoker(yes/no); after 1980, the level ofsmoking anddates of starting and quitting were specified.Because information on smoking was notavailable for most of the study cohort, weexamined the association between smokingstatus and cumulative radiation dose in thoseworkers for whom information on smokingwas available (658 subjects) to assess potentialconfounding in the larger cohort.

Statistical MethodsWe used two different analytic approaches:external comparisons of our monitored

VOLUME 1081 NUMBER 8 1 August 2000 * Environmental Health Perspectives746

Articles . Internal radiation and cancer mortality

workers with the general U.S. white malepopulation; and internal comparisonsamong monitored workers according tomeasured dose levels of radiation exposure(dose-response analyses). In external com-parisons, the Monson program (31) was usedto estimate standardized mortality ratios(SMRs; = observed/expected deaths) for themonitored study population. We estimatedexpected numbers of deaths from the mortal-ity rates of the U.S. white male population,stratified by age (5-year categories) and calen-dar year (5-year intervals). Estimation of95% confidence limits for the SMRs wasbased on a formula derived by Byar and rec-ommended by Breslow and Day (32).

Because our study population yielded 10or fewer deaths for many types of cancer, itwas not possible to perform informativedose-response analyses; thus, it was neces-sary to combine deaths from selected can-cers. The choice ofwhich cancers and cancergroups to evaluate was made a priori on thebasis of the distribution within the body ofthe radionuclides of major concern. Theseradionuclides emit densely ionizing alpharadiation that usually reaches and damagesonly the tissues in its immediate vicinity-within micrometers of the particle (1).Exceptions are the air-filled spaces in thelung, which allow alpha particles to reachgreater distances, such that almost any tissueconstituent of the lung may receive a consid-erable dose of radiation, and radionuclidesthat dissolve from particles into systemic cir-culation from which they deposit in othertissues. Cells located at bifurcations, whereremoval is significantly slower than in thetubular airways, will experience significantlyhigher doses than those lining the tubularairways. In addition, for alpha emitters suchas those considered here, microdosimetricconsiderations show that most cells will havea dose of zero, with a small fraction of cellshaving doses on the order of tens of rads dueto the passage of one or a few alpha particlesthrough the nucleus. Because risk coeffi-cients generally are developed using meantissue doses, however, we chose to use meandose in the present study rather than themore detailed microscopic dose distribution.

Relatively insoluble radioactive particlesthat reach the alveoli are gradually translo-cated to tracheobronchial and other thoraciclymph nodes, which may accumulate con-centrations of inhaled material several hun-dred times greater than in the regions of thelung (1). Larger particles (2 10 pm) rarelyreach the lower respiratory tract or, if theydo, are cleared rapidly and completely. Suchparticles can deliver intense doses of concen-trated alpha radiation to regions of the naso-and oropharyngeal systems and the uppergastrointestinal tract.

Thus, any effects of internally depositedradionuclides are most likely to be evident inthose tissues receiving the highest dose. Ingeneral, these will be the tissues of the portalorgans (lungs for inhalation and gastro-intestinal tract for ingestion) for the highlyinsoluble compounds, or the bone for themore soluble compounds (for translocationof uranium and strontium). Because solubili-ty is unknown for this population, it was notpossible to estimate doses to tissues otherthan the lung tissue, and even for the lungwe obtained only a relative measure of dose,as the absolute value of the dose depends onsolubility. Accordingly, we conducteddose-response analyses for a) lung cancer(ICD-9 162); b) upper aerodigestive tractcancers encompassing the naso-oropharyn-geal regions, esophagus, and stomach (ICD-9 140-151); c) hemato- and lymphopoieticcancers (ICD-9 200-208, excluding chroniclymphatic leukemias); a) urinary-tract can-cers (ICD-9 188-189); and e) prostate can-cer (ICD-9 185). Other organs to whichsome radionuclides are translocated andstored are the liver (Thorotrast), bones (plu-tonium), and the thyroid (iodine). We didnot, however, observe any bone, thyroid, orprimary liver cancers among workers moni-tored for internal radiation.

To estimate effects in the dose-responseanalyses, we used the risk-set approach forthe analysis of cohort study data, which wasrecommended by Breslow and Day (32),using the full cohort information. In thisapproach, conditional logistic regression isused to compare individuals who have diedof cancer (outcome events) with all individu-als still at risk of dying from cancer (sur-vivors). We constructed risk sets of deathsand survivors matched on calendar time foruse in the analysis by matching to each can-cer death all cohort members who were stillalive at the time of the index subject's death.This approach allowed us to treat cumulativedose and all other time-varying variables,such as time since first monitoring, as timedependent, (i.e., values for these factors weredetermined for all risk-set members at thetime of each index death).

We modeled cumulative internal radia-tion dose both as a set of binary variables andas a continuous variable (in 1 0-mSv incre-ments). Based on the dose distribution in ourcohort, we categorized dose equivalents into 4levels: 0 mSv, > 0-5 mSv, > 5-< 30 mSv, and> 30 mSv. To allow for a period of induc-tion/latency between radiation exposure andcancer death and to reduce possible selectionbias (33), we lagged cumulative doses by 0, 2,and 10 years. Lagging entailed limiting thelevel of cumulative dose for each individual ina risk set to the dose level achieved 0, 2, and10 years before the index death occurred. As

recommended, we adjusted in all models fortime since first monitored to avoid the possi-ble selection bias inherent in the analyses ofcumulative exposures (34).

We used results of the conditional logisticregression analyses to estimate rate ratios (RR)and 95% confidence intervals (CI) for inter-nal radiation and other covariates in themodel. To test for a monotonic trend in theassociation between cumulative dose and can-cer mortality, the mean of the four dose cate-gories were used as exposure scores. Weexplored a variety of potential confounders,but retained in the final models only thosecovariates that changed the estimated RR forradiation exposure by > 10% for any outcome(35). Accordingly, pay status, time since firstmonitored, and age at risk (continuous) butnot exposure to chemicals were included in allmodels presented in this paper. BecauseCheckoway et al. (6) reported a positive asso-ciation between internal and external radia-tion dose in Oak Ridge workers, all analysesof internal radiation effects were also adjustedfor the effect of external radiation dose (treat-ed as continuous in 10-mSv increments).

ResultsThe Rocketdyne/AI cohort monitored forinternal radiation exposure was characterizedby a long follow-up period (average 25.4years), a high percentage of salaried employees(40.1%), and few women (Table 2). Only0.7% of these workers received estimatedinternal radiation doses to the lung > 30 mSv,and slighdy more than half of the workers hadrecorded doses of0 mSv.

During the study period, 19.2% of thecohort members died (441 total deaths). Weobserved 133 deaths from cancer as the under-lying cause among males and one such cancerdeath among females, yielding a total cancermortlity rate of235 per 105/ year (Table 1).

Comparing the mortality experience ofmale RAI workers monitored for internalradiation with the white male U.S. popula-tion resulted in SMRs of 0.72 (95% CI,0.66-0.80) for all causes, 0.87 (95% CI,0.73-1.03) for all cancers, and 0.68 (95%CI, 0.58-0.78) for all circulatory system dis-eases (Table 3). These results indicate thatmembers of the RAI cohort are healthierthan the general population [i.e., a stronghealthy worker (selection) effect exists], aneffect we would expect to observe in a cohortwith a large proportion of higher socioeco-nomic status employees and extensive healthinsurance coverage (see also "Discussion").For specific cancer sites, we did not observeSMRs for which the 95% CIs excluded thenull value of 1.

In dose-response analyses, monotonicassociations were observed between cumu-lative internal dose and mortality from

Environmental Health Perspectives * VOLUME 108 1 NUMBER 8 1 August 2000 747

Articles * Ritz et al.

hemato- and lymphopoietic cancers andfrom cancers of the upper aerodigestive tract(Tables 4 and 5). The rate ratios for hemato-and lymphopoietic cancers, comparing acumulative dose . 30 mSv with 0 mSv, was44.6 (95% CI, 5.64-353), and the corre-sponding rate ratio for upper-aerodigestivetract cancers was 57.2 (95% CI, 8.17-401).Total cancer mortality was also elevatedsomewhat for cumulative doses > 30 mSv(RR = 2.56; 95% CI, 0.93-7.09). We foundno effects of internal radiation on mortalityfrom lung cancer, urinary tract cancers, orprostate cancer (Tables 4 and 5). Laggingdoses by 2 and 10 years did not change theresults of the analyses (Table 5), nor didadding to the cancers specified as underlyingcauses of death all cancers listed as contribut-ing causes (results not shown).

For the 2,253 workers monitored forboth external and internal radiation, we esti-mated the combined effects on total cancermortality of both types of radiation, cross-classified into nine dose categories. Althoughthere were no cancer deaths in the highestcombined dose category (> 200 mSv exter-nal and > 30 mSv internal radiation), the

cancer mortality RRs were elevated apprecia-bly (RR > 5) for monitored workers in thenext highest combined dose categories(Table 6). However, the 95% confidenceintervals are quite wide for these estimates,indicating low precision of these estimatesbased on small numbers.

We did not find an association betweensmoking and cumulative internal radiationdose during the 1960s (Table 7). On theother hand, exposed workers who were stillemployed in the 1980s were more likely thanunexposed workers to have quit smoking, andthe fraction of unexposed workers who con-tinued smoking remained disproportionatelyhigh relative to both their exposed co-workersand California males in general (36).

DiscussionWe observed a strong healthy worker (selec-tion) effect in our cohort: compared to theU.S. population, monitored RAI nuclearworkers experienced lower rates of deathfrom all causes, from all cancers, and partic-ularly from all circulatory system diseases.This phenomenon is characteristic of occu-pational cohorts in general, but is especially

strong in the nuclear industries for whichmean all-cause SMRs have been reported tobe even lower (0.79) than the correspondingmean SMRs (0.83) reported for a large num-ber of other industries (37). The all-causeSMR in our cohort is low (0.72) mainlybecause RAI employees exhibit a large deficitin cardiovascular disease mortality (SMR =0.62) which may be due to differences inlifestyle factors (diet, smoking, physicalactivity) when comparing these Californianworkers to the rest of the United States ormay be related to the extensive health insur-ance coverage these nuclear workers enjoyedthroughout their employment. Greaterhealth insurance coverage of workers mayalso be responsible for reducing fatality ratesof many common cancers such as those ofthe colon, prostate, and bladder; for theseorgans, fatality depends on early detectionand medical treatment of the cancer (38).

Exposure levels in the cohort studied wererelatively low; the mean lung dose from inter-nal radionuclide exposure for 2,297 moni-tored workers was estimated to be 2.1 mSv, adose much lower, for example, than the aver-age lung dose of 82.1 mSv reported for 3,491

Table 4. Adjusted rate ratio (RR) estimates (and 95% confidence intervals) for the effect of cumulative internal radiation dose and other factors on cancer mortali-ty, by cancer type, assuming zero lag for exposure: results of conditional logistic regression analyses.

Predictors All cancersAge at riskb 1.10

(1.08-1.12)Time since first 0.99monitoredb.c (0.97-1.01)

Pay typeSalaried managerial/ 0.75professional vs. other (0.51-1.10)

External radiation dose 1.02(10 mSv)cd (0.98-1.06)

Internal radiation dose (mSv)c0 1.00

(n = 79)e> 0-< 5 0.86 (0.58-1.27)

(n= 36)2 5-< 30 0.87 (0.45-1.67)

(n= 15). 30 2.56 (0.93-7.09)

(n=4)p for trendf 0.087

Hemato- and lympho-poietic cancers(ICD-9 200-208)a

1.10(1.03-1.18)

0.99(0.89-1.09)

1.05(0.26-4.27)

1.06(1.00-1.13)

1.00In= 2)

2.31 (0.37-14.2)(n= 3)

6.10 (0.89-41.7)(n= 3)

44.6 (5.64-353)(n= 2)0.0001

Lung cancers(ICD-9 162)

1.10(1.07-1.13)

0.97(0.93-1.02)

0.49(0.21-0.97)

1.06(1.01-1.11)

1.00In= 30)

0.58 (0.28-1.21))n=9)

0.45 (0.12-1.67)(n=5)0.00(n=0)0.20

Upper aerodigestivetract cancers(ICD-9 140-151)

1.09(1.04-1.15)

0.94(0.85-1.03)

0.64(0.17-2.35)

0.92(0.76-1.12)

1.00In= 3)

4.75 (1.12-20.2)In= 6)

10.56 (1.91-58.4)(n= 3)

57.2 (8.17-401)(n=2)0.0001

Bladder andkidney cancers(ICD-9 188,189)

1.18(1.09-1.27)

0.95(0.84-1.07)

0.79(0.16-4.06)

1.05(0.91-1.21)

1.00(n=5)

1.07 (0.23-5.02)In= 3)0.00(n= 0)0.00(n=0)0.43

Prostate cancers(ICD-9 185)

1.20(1.10-1.31)

0.98(0.92-1.04)

1.23(0.22-6.95)

0.19(0.03-1.32)

1.00In=5)

1.59 (0.28-9.06)In= 2)0.00(n= 0)0.00(n=0)0.65

&Excluding chronic lymphatic leukemias. bMeasured in one year increments. cTreated as time-dependent. dAssumes dose due to radionuclides equal to zero for employees not moni-tored for external radiation. Measured in 10-mSv increments. 'Number of cancer deaths shown in parentheses. frhe test for trend was performed by entering an interval variable withthe category means as the score values into the logistic regression model.

Table 5. Adjusted rate ratio (RR) estimate (and 95% confidence interval) and two-tailed p-value for the effect of cumulative internal radiation dose in 10-mSVincrements, by cancer type and lag for exposure: results of conditional logistic regression analyses.

Internal All cancersradiation dose (n = 134)(per 10 mSv) RR p0-year lag 1.03 (0.88-1.20) 0.702-year lag 1.03 (0.89-1.21) 0.661 0-year lag 1.04 (0.88-1.22) 0.68

&Excluding chronic lymphatic leukemias.

Hematopoietic andlymphopoietic cancersa

(ICD-9 200-208)n= 10

RR p1.23 (0.97-1.55) 0.081.23 (0.97-1.55) 0.081.24 (0.98-1.55) 0.07

Lung cancers(ICD-9 162)

n = 44RR p

0.75 (0.32-1.76) 0.500.76 (0.33-1.76) 0.520.74 (0.29-1.92) 0.54

Upper aerodigestivetract cancers(ICD-9 140-151)

n= 14RR p

1.25 (1.05-1.48) 0.011.25 (1.05-1.49) 0.011.23 (1.01-1.50) 0.04

Bladder andkidney cancers(ICD-9 188, 189)

n=8RR p

0.13 (0.00-18.5) 0.420.13 (0.00-18.8) 0.430.19 (0.00-20.8) 0.49

Prostate cancers(ICD-9 185)

n=7RR p

0.08 (0.00-375) 0.560.08 (0.00-374) 0.560.09 (0.00-371) 0.57

VOLUME 108 1 NUMBER 8 1 August 2000 * Environmental Health Perspectives748

Articles * Internal radiation and cancer mortality

workers monitored for uranium exposure atthe Oak Ridge Y-12 facility (6). Moreover,because the quality of our internal radiationdata did not allow us to calculate specificorgan doses other than to the lung, all inter-nal exposure risk estimates were calculated onthe basis of expected doses to the lung. Thus,we relied on lung doses to approximate doselevels to a range of organs involved inradionuclide passage through the body (seeMethods), some ofwhich may have been sub-jected to very different levels of exposure,depending on the radioactive decay processand the retention function of the radionudidefor different organs. Although the computedlung doses can serve as crude indicators of themagnitude of doses delivered to other organs,our dose estimates should be interpreted inrelative rather than in absolute terms. In gen-eral, dose comparisons with other studies maynot be appropriate even for those also relyingon lung doses because we lacked informationon solubility, on which accurate estimates oflung dose depend.

Despite these limitations, we detectedincreases in mortality from hemato- and lym-phopoietic cancers with increasing internalradiation dose among RAI employees, a find-ing also reported by two previous studies ofnuclear workers exposed to plutonium(4,5,18). Wilkinson et al. (4) reported thatRocky Flats employees with positive plutoni-um body burdens experienced elevatedmortality from blood and lymphatic systemcancers. These results were more pronouncedand showed a dose-response gradient whenthe follow-up of the Rocky Flats cohort wasextended (5). Omar et al. (18) observed anincrease in the incidence of these cancers withincreasing cumulative plutonium plus exter-nal radiation doses among Sellafield workers.

Elevated rates of hemato- and lymphopoieticcancers have also been observed in groups ofmedical patients treated with high doses ofThorotrast (3). Archer et al. (39) reported anSMR of 4 for these cancers among uraniumminers and millers (based on four cases), andWaxweiler et al. (4Q) found a small increasefor lymphatic cancers among uranium millers(also based on small numbers).

The dose-response relationship that weobserved between internal radiation exposureand death from cancers of the upper aerodi-gestive tract has not previously been describedin occupational cohorts exposed to low doses.At high levels of exposure, radium dialpainters have suffered an excess of head-sinuscancers (3). Furthermore, the effect estimatesbased on the continuous dose (Table 5) didnot change when we considered only the 11esophageal and gastric cancers out of thegroup of 14 upper aerodigestive tract cancers.Wilkinson (41) reported a strong ecologicassociation between uranium deposits andgastric cancer mortality among counties inNew Mexico. These results should be inter-preted with caution, however, because of pos-sible ecologic bias (42) and confounding dueto the effects of other environmental carcino-gens such as arsenic and cadmium.

Other studies of nuclear workers havenot reported increases in cancers of the upperaerodigestive tract at (lung) doses apparentlyhigher than those in our cohort, yet it is notclear whether other researchers ever examinedeffects on these organs in a dose-responseanalysis. Our external comparisons did notalert us to an excess mortality for cancers ofthese organs compared to the general U.S.population, possibly because RAI workersdrank less alcohol (the observed number ofliver cirrhosis deaths was about half that

expected; Table 3), while dose-responseanalyses showed strong effects with increasingradionuclide exposure. Thus, researchersusing external comparisons to guide theirchoice of organ sites for dose-response analy-ses may have been misled. Our positive find-ings may be due, in part, to the long follow-upperiod in our study and the properties of theradioactive particles to which workers at RAIwere exposed. Moreover, the true dose deliv-ered to the upper aerodigestive tract may behigher than indicated by our exposure mea-sures, which were calculated as doses to thelung and derived mainly from urinalysis andlung-count data. Although most internalexposures are likely to involve inhalation,some radionuclide particles, depending onsize, will not reach the lower respiratory tractor will be cleared by the ciliary system andswallowed. Because such particles have littleor no residence time in the lungs, they areunlikely to be detected in a lung count.Because they are excreted through feces, theywould also be missed by urinalysis; however,fecal analyses were rarely performed in ourcohort. Nevertheless, these particles candeliver intense doses of concentrated alpharadiation to regions of the naso- and oropha-ryngeal and upper gastrointestinal system (1).Thus, our dose categories based on lung-doseestimates should be interpreted in a qualita-tive rather than quantitative fashion for thegastro-intestinal tract and other organs. It isreasonable, however, to assume that workerswith higher lung doses were at greater risk forexposure to non-respirable radioactive parti-cles, although the ratio of respirable to non-respirable particles may have varied.

The observed excess rate of total cancermortality in workers in the highest dose cate-gory (> 30 mSv) (Table 4) is due entirely to

Table 6. Adjusted rate ratio (RR) estimates (and 95% CIs) for the combined effects of cumulative internal and external radiation dose on total cancer mortalityamong all 2,253 subjects monitored for both internal and external radiation, by dose level assuming a zero year lag for both exposures: results from a conditionallogistic regression analyses.,

Internal dose

Articles * Ritz et al.

deaths from cancers of the hemato- and lym-phopoietic system and upper aerodigestivetract. We did not observe an effect of internalradiation on cancers of the urinary tract orprostate among RAI workers. British studiesfound increased incidence rates of prostateand renal cancers in nuclear workers whowere either exposed primarily to tritium or toa variety of different radionuclides (14-16).Our negative results for urinary tract andprostate cancers might be attributable to theabsence of tritium exposures in our cohort,the lower power of our study, lower radionu-clide doses delivered to the urinary system(perhaps with a greater degree of partition-ing to the gastrointestinal tract), or the useof mortality rather than incidence data.Furthermore, the absence of bone, liver, orthyroid cancers may be due to the fact thatRAI workers were primarily exposed to ura-nium compounds and not other radionu-clides favorably deposited in the latter twoorgans and/or the small number of such can-cers expected in our cohort.

We also did not detect a positive associa-tion between internal radiation dose and lungcancer mortality in our cohort. British studiesdemonstrated a trend of increasing mortalityfrom lung cancers with increasing externalradiation doses only among those workerswho were also monitored for internal expo-sure to radionuclides, and an overall increasein respiratory tract cancers among plutoniumworkers could not be explained by externalradiation doses (14-16). Similarly, at OakRidge Y-12, Checkoway et al. (6) found thestrongest gradient for the effect of cumulativegamma radiation dose (external) on lung can-cer mortality in a subgroup of workersexposed to > 50 mSv of internal alpha radia-tion, primarily from uranium. Dupree et al.(7) were unable to confirm these results whenextending the Oak Ridge follow-up by 3years. However, the later analysis differedfrom the original in one important aspect:nonmonitored workers were included in theunexposed group. Wiggs et al. (10) reporteda slightly elevated lung cancer mortalityamong plutonium-exposed workers at theLos Alamos National Laboratories. Fernalduranium processing workers exposed to alpharadiation at levels > 200 mSv also experi-enced an increased risk of dying from lungcancers (9). Several Russian studies of pluto-nium workers employed at the Mayaknuclear enterprise recently also reported anincreased risk of lung cancers among workersexposed to high levels of plutonium (19-22).

The lack of a positive associationbetween lung cancer mortality and radionu-clide dose in our cohort may be due to RAIworkers having actually received relativelylow doses to the lung tissue, very few work-ers having been exposed to plutonium, or

incomplete control for confounding by otherrisk factors. The most likely potential con-founders are smoking and exposure to chem-ical carcinogens such as asbestos, hydrazine,and beryllium. We did not have the infor-mation necessary to adjust for berylliumexposures, and our measures of asbestos andhydrazine exposure, based almost entirely onjob titles, are likely to suffer from misclassifi-cation. Although we could not adjust forsmoking in the analyses, we were able toevaluate smoking data in a subgroup ofinternally monitored workers. We foundthat among those still employed in the1980s, the proportion of current smokerswas substantially higher for unexposed thanfor exposed workers. This disparity suggeststhat negative confounding due to smokingmight be occurring in our cohort, potentiallyobscuring the effect of radiation exposure onlung cancer mortality.

In summary, despite the small samplesize and relatively low lung doses recordedfor workers at RAI, this study has demon-strated a dose-response association betweencumulative internal radiation dose and mor-tality from hemato- and lymphopoietic can-cers. In addition, we have seen evidence for adose-response association with upper aerodi-gestive tract cancers that may have resultedfrom the ingestion of nonrespirable particlesthat were cleared from the upper and lowerrespiratory tract. Our latter finding is basedon a pooling of specific cancer types thathave not been examined as a group in previ-ous radiation studies. Although we foundstrong dose-response gradients for these twotypes of cancers, our estimates are imprecisedue to the small number of cases in eachgroup and should be confirmed by furtherfollow-up of the present cohort.

REFERENCES AND NOTES

1. International Commission on Radiological Protection.Biological Effects of Inhaled Radionuclides. Publicationno. 31. Oxford:Pergamon Press, 1980.

2. Mays CW. Alpha-particle-induced cancer in humans.Health Phys 55(4):637-652 (1988).

3 National Research Council. Health Effects of Exposure toLow Levels of Ionizing Radiation. Washington, DC:NationalAcademy Press, 1988.

4. Wilkinson GS, Tietjen GL, Wiggs LD, Galke WA, AcquavellaJF, Reyes M, Voeiz GL, Waxweiler RJ. Mortality among plu-tonium and other radiation workers at a plutoniumweapons facility. Am J Epidemiol 125:231-250 (1987).

5. Wilkinson GS, Baillargeon J, Ray L, Baillargeon G, Trief N.Cancer mortality among plutonium and radiation workers[Abstract]. Am J Epidemiol 145(suppl):S40 (1997).

6. Checkoway H, Pearce N, Crawford-Brown J, Cragle DL.Radiation doses and cause-specific mortality amongworkers at a nuclear materials fabrication plant. Am JEpidemiol 127:255-266 (1988).

7. Dupree EA, Watkins JP, Ingle JN, Wallace PW, WestCM, Tankersley WG. Uranium dust exposure and lungcancer risk in four uranium processing operations.Epidemiology 6:370-375 (1995).

8. Loomis DP, Wolf S. Mortality of workers as a nuclearmaterial production plant at Oak Ridge, Tennessee,1947-1990. Am J Ind Med 29:131-141 (1996).

9. Ritz B. Cancer mortality and radiation exposures in urani-um processing workers. Epidemiology 10:531-538 (1999).

10. Wiggs LD, Johnson ER, Cox-De-Vore CA, Voeiz GL.Mortality through 1990 among white male workers at theLos Alamos National Laboratory: considering exposuresto plutonium and external ionizing radiation. Health Phys67(6):577-588 (1994).

11. Wiggs LD, Cox-Devore, Voeiz GL. Mortality among acohort of workers monitored for Po-210 exposure:1944-1972. Health Phys 61:71-76 (1991).

12. Dupree EA, Cragle DL, McLain RW, Crawford-Brown DJ,Teta MJ. Mortality among workers at a uranium processingfacility, the Linde Air Products Company Ceramics Plant,1943-1949. Scand J Work Environ Health 13:100-107 (1987).

13. Polednak AP, Stehney AF, Lucas HF. Mortality amongmale workers at a thorium-processing plant. Health Phys44(suppl 1):239-251 (1983).

14. Beral V, Inskip H, Fraser P, Booth M, Coleman D, Rose G.Mortality of employees of the United Kingdom AtomicEnergy Authority, 1946-1979. Br Med J 291:440-447 (1985).

15. Rooney C, BerI V, Macononhie N, Fraser P, Davies G.Case-control study of prostatic cancer in employees ofthe United Kingdom Atomic Energy Authority. Br Med J307:1391-1397 (1993).

16. Beral V, Fraser P, Carpenter L, Booth M, Brown A, RoseG. Mortality of employees of the Atomic WeaponsEstablishment, 1951-82. Br Med J 297:757-770 (1988).

17. Carpenter LM, Higgins, Douglas AJ, Maconochie NES,Omar RZ, Fraser P, Beral V, Smith PG. Cancer mortality inrelation to monitoring for radionuclide exposure in threeUK nuclear industry workforces. Br J Cancer 78(9):1224-1232 (1998).

18. Omar RZ, Barber JA, Smith PG. Cancer mortality and mor-bidity among plutonium workers at the Sellafield plant ofBritish Nuclear Fuels. Br J Cancer 79(7-8):1288-1301(1999).

19. Tokarskaya ZB, Okladnikova ND, Belyaeve ZD, Drozhko EG.The influence of radiation and nonradiation factors on thelung cancer incidence among the workers of the nuclearenterprise Mayak. Health Phys 69(3):356-366 (1995).

20. Khokhriakov VF, Romanov SA. Estimation of the temporaldistribution and dose dependency of lung cancer amongworkers of nuclear fuel reprocessing plants. Health Phys71(1):83-85 (1996)

21. Koshurnikova NA, Bolotnikova MG, llyin LA, Keirim-MarkusIB, Menshikh ZS, Okatenko PV, Romanov SA, Tsvetkov VI,Shilnikova NS. Lung cancer risk due to exposure to incor-porated plutonium. Radiat Res 149(4(:366-371 (1998).

22. Koshurnikova NA, Bysogolov GD, Bolotnikova MG,Khokhryakov VF, Kreslov W, Okatenko PV, Romanov SA,Shilnikova NS. Mortality among personnel who workedat the Mayak complex in the first years of its operation.Health Phys 71(1l:90-93 (1996).

23. Wilkinson GS, Morgenstern H. Internal comparisons donot always control for the healthy worker effect[Abstract]. Am J Epidemiol 141(suppl(:S58 (1995).

24. U.S. Department of Health and Human Services.International Classification of Diseases: 9th revision,Clinical Modification. 3rd ed. (PHS) 89-1260. Los Angeles,CA:Professional Medical Management Corporation, 1989.

25. Karaffa MC, ed. The International Classification ofDiseases, 9th revision: Clinical Modification, Vol 1, 2. 4thed. (PHS) 91-260. Washington DC:United StatesGovernment Printing Offices, 1993.

26. International Commission on Radiological Protection.Limits for Intakes of Radionuclides by Workers.Publication 30. New York:Pergamon Press, 1978.

27. International Commission on Radiological Protection.Design and Interpretation: Individual Monitoring forIntakes of Radionuclides by Workers. Publication 54.New York:Pergamon Press, 1987.

28. Crawford-Brown DJ, Watson J, Strom J, Tankersley W.Procedures for Assessing Occupational RadiationMonitoring Data for Use in Epidemiologic Studies, 1989.Oak Ridge Associated Universities, Rpt No ORAU 89/A127. Oak Ridge, TN: Oak Ridge National Laboratory, 1989.

29. Checkoway H, Crawford-Brown D. Metabolic modelingof organ-specific doses to carcinogens as illustratedwith alpha-radiation emitting radionuclides. J Chron Dis40(suppl 2): 191S-200S (1987).

30. Ritz B, Morgenstern H, Froines J, Young BB. Effects ofexposure to external ionizing radiation on cancer mortal-ity in nuclear workers monitored for radiation at

750 VOLUME 108 1 NUMBER 8 August 2000 * Environmental Health Perspectives

Articles * Internal radiation and cancer mortality

Rocketdyne/Atomics International. Am J Ind Med35:21-31 (1999).

31. Monson RR. Documentation Accompanying the MonsonProgram. Boston, MA:Harvard School of Public Health,1994.

32. Breslow NE, Day NE. Statistical Methods in CancerResearch. Vol Il: The Design and Analysis of CohortStudies. IARC Sci Pub] 82 (1987).

33. Arrighi HM, Hertz-Picciotto I. The evolving concept ofthe healthy worker survivor effect. Epidemiology5:189-196 (1994).

34. Arrighi HM, Hertz-Picciotto I. Controlling for time-since-hire in occupational studies using internal comparisonsand cumulative exposure. Epidemiology 6:415-418 (1995).

35. Greenland S. Modeling and variable selection in epidemi-ologic analysis. Am J Public Health 79:340-349(1989).

36. Burns D, Pierce JP. Tobacco Use in California 1990-1991.Sacramento, CA:California Department of HealthServices, 1992.

37. Park RM, Maizlish NA, Punnett L, Moure-Eraso R,Silverstein MA. A comparison of PMRs and SMRs as esti-mators of occupational mortality. Epidemiology 2:49-59(1991).

38. Demers PA, Vaughan TL, Checkoway H, Weiss NS,Heyer NJ, Rosenstock L. Cancer identification using atumor registry versus death certificates in occupationalcohort studies in the United States. Am J Epidemiol136(10):1232-1240 (1992).

39. Archer VE, Brinton HP, Wagoner JK. Cancer mortalityamong uranium mill workers. J Occup Med 15:11-14 (1973).

40. Waxweiler RJ, Archer VE, Roscoe A, Watanabe A, ThunJJ. Mortality patterns among a retrospective cohort ofuranium mill workers. In: Proceedings of the HealthPhysics Society, 428-435 (1983).

41. Wilkinson GS. Gastric cancer in New Mexico countieswith significant deposits of uranium. Arch Environ Health40:307-31211985).

42. Morgenstem H. Ecologic studies. In: Modern Epidemiology,2nd ed (Rothman K, Greenland S, eds). Philadelphia,PA:Lippincott-Raven, 1998;459-480.

* C l ~ Excellence in basic research at theNational Institute of Environmental Health Sciences

NIEHS scientists and grantees are performing basic studies of our susceptibility to environment-related disease:demonstrating that a carcinogen in cigarette smoke (benzo(a)pyrene) alters part of a gene to cause lung cancer ...showing the effects of fetal exposure to PCBs .. . developing a strain of mouse that lacks functional estrogen receptorsand that helps evaluate how some pesticides and other estrogen-like compounds might affect development andreproduction . . . discovering the genes for breast, ovarian, and prostate cancers ... identifying women's optimaldays of fertility ... seeking to reverse the damage from lead exposure ... finding alternatives to traditional amaltests ....pinpointingthefunctions ofspecificgenes byeliminatingthemfrom specially bred'ing :a way, using ordinary yeast cells, to isolate and clone genes and other fragments of genetiJ.shoing the effects of urban air on lung function...

&x Yj> .. .S. S. .... .... ^ . -; S . . S:. . ~. a .. -.

_^ ~~~~~~~~~~a_S~~~~~~~~~~~~~~ l

Address Chan e?To be sure to continue receiving your subscription,

Lcall 1-800-315-3010 or e-mail ehis@niehs.nih.govInclude your old address as it appears on the mailing label, as well as your new address

Environmental Health Perspectives * VOLUME 1081 NUMBER 81 August 2000 751