Epidemiologic Evaluation of Leukemia Incidence in Children ... · Epidemiologic Evaluation of Leukemia Incidence in Children and Adults in the Vicinity of the Nuclear Power Plant

Epidemiologic Evaluation of Leukemia Incidence in Children and Adults in theVicinity of the Nuclear Power Plant Krummel (KKK)

Epidemiologic Evaluation of Leukemia Incidence inChildren and Adults in the Vicinity of the Nuclear PowerPlant Krummel (KKK)

Wolfgang Hoffmann, Eberhard Greiser, Bremen

AbstractBetween February 1990 to the end of 1995six cases of childhood leukemia were dia-gnosed among residents of the small ruralcommunity Elbmarsch in Northern Ger-many. Five of these cases were diagnosedin only ]6 months between Feb. ]990 andMay 1991. The cases all lived in closeproximity (500-4500 m) to Germany's larg-est capacity nuclear boiling water reactor(KKK).Standardized incidence ratios (SIR) and ex-act 95% confidence intervals were calcula-ted for a circular area with a 5 km radiusaround the plant. For the time period 1990to 1995 the SIR was 464 (208-1034). If theanalysis is restricted to the years ]990 and199], the SIR increases to 1175 (489-2827).This cluster of childhood leukemia casescannot presently be explained in terms ofestablished and putative risk factors forchildhood leukemia, including radiation.from medical sources.After this unprecedented childhood leu-kemia cluster had been detected, it was hy-pothesized that leukemia incidence couldalso be increased in adults. To test this hy-pothesis, a retrospective incidence studywas conducted which covered 10 years(1984-]993) and three adjacent countieswith a total population of about 470,000.Case ascertainment was done perusingdocuments of multiple primary data sourcesincluding all hospitals, institutes of patho-logy, oncology centers, as well aspracticing physicians and three Depart-ments of Health within the study area and

in adjacent counties, respectively. Afterlinking of documents from all sources a to-tal of ]985 verified cases of leukemias,malignant lymphomas and multiple mye-loma remained. Incidence rates were stan-dardized to the 1987 West Germany censuspopulation structure. Proximity to KKKwas operationalized by constructing con-centric circles with radii 5 km (10, 15, 20km) around the site of the plant.When overall incidence was used as a refe-rence, SIR for leukemia in the 5 km-regionwas elevated for both genders (observed51, expected 40; SIR 128; p=0.046),whereas no such elevation was observed inany of the other circles. The increased inci-dence was attributable to males (SIR ]56;p=O.Ol) and was most pronounced, but notconfined to, chronic myeloid leuke-mia.When the analysis was restricted topersons below age 65, SIR were signifi-cantly increased for both males and fe-males. The excess risk observed over the 10year study period was exclusively due to anincreased incidence from 1989 through1993.A subsequent case control study includingall cases identified in the incidence study isindicated to identify possible causes for theexcess leukemia risk for children and adultsin the vicinity of KKK.

IntroductionThe rural community Elbmarsch is situatedon the southern bank of the river Elbe,some 30 km south-east of the City of Ham-burg in Lower Saxony. In a population ofless than 2000 children under 15 years 6 ca-

237


ses of leukemia were diagnosed since 1990. Established and suspected risk factors forFive of these cases had occurred in the time leukemia include low dose ionizing radia-period between February 1990 and May tion from natural [11, 43, 57, 59] and man-1991. The sixth case was diagnosed made sources [54, 60] at all agesrecently in 1995. This cluster is hitherto (preconceptional [26, 66, 80], prenatal [18,unprecedented in Germany with respect to 30, 61, 64], and postnatal [28, 56]), expo-its temporal and spatial dimension [19]. sure to benzene [1, 6, 50-52], low-frequen-Though we cannot exclude the existence of cy electromagnetic fields [23, 76, 91] (seea comparable childhood leukemia cluster [74, 75, 90] for recent reviews), and pesti-elsewhere, we are presently unaware of any cides [9, 21, 34, 58, 68]. Despite detailedmentioning in the international literature. investigations into the medical, residential,The cases were all living in close vicinity to family, social and occupational histories ofthe nuclear power plant Kernkraftwerk the afflicted families which were supple-Krummel (KKK) which is situated at the mented with extensive physical, chemical,northern bank of the river Elbe, just oppo- and toxicological analyses, no unusual ex-site of the Elbmarsch. KKK is Germany's posure to any of these factors was detected.largest capacity nuclear boiling water reac- The expert committee therefore concluded,tor (1300 MWel). It was put into operation that only a detailed epidemiologic studyin 1984. Adjacent to KKK the nuclear re- could shed further light on the etiology ofsearch institute GKSS (GeseJlschaft zur the cluster. In order for any epidemiologi-Kernenergieverwertung in Schiffbau und cal approach to be sensible, however, bothSchiffahrt) operates two research reactors the age range and the area that is beingof 5 and 15 MW, respectively, since the studied needed to be extended. As a firstlate 1950s. step the expert committee suggested aThis epidemiological situation generated retrospective incidence study, coveringconsiderable public concern and media children and adults in the three adjacentcoverage. The local authorities reacted rural districts of the Elbmarsch (total popu-promptly: In order to investigate possible lation 480,000). The project was assignedcauses of the increased incidence of leu- to the Bremen Institute for Prevention Re-kemias, the Federal State of Lower Saxony search and Social Medicine, where con-in Feb. 1992 established an interdiscipli- siderable experience was available fromnary scientific committee, which initially earlier studies with similiar questions [33,consisted of 25 experts. 35, 36,45-47].Tab. 1 The Retrospective Incidence Study Elbmarsch: Design elements

Study area

Target diagnoses

Inclusion criteriaAge groupsStudy periodMethods

Rural districts Liineburg, Harburg (Lower Saxony); Rural districtHerzogtum Lauenburg (Schleswig-Holstein)Total population 480.000

Leukemia (all forms), malignant lymphoma, multiple myeloma,related diseases and syndromes

Residence in study area at time of I st diagnosis, German citizenship

all ages (children, adults)

]984 - ]993

Active case ascertainment, compilation of multiple primary datasources

238


Materials and MethodsDesignPrincpal design of the study presented here-in is a retrospective geographical incidenceanalysis. The most important design featu-res are summarized in Tab. 1.

Case ascertainment and data sourcesFundamental requirement for any geogra-phical incidence study is the completeascertainment of all incident cases alikeover all single years of the study period andin all regions of the study area.Whereas this task is trivial in countries inwhich central cancer registries exist, it be-comes particularly tedious in countrieswithout established registration. In Ger-many only childhood cancers are routinelyregistered in a central cancer registry [55].Hence case ascertainment needed to resortto the vast array of primary data sources.Among the sources analysed were the pa-tient documentation and medical historyfile archives of all hospitals in the studyarea, the pathologic and cytologic instituteswhich cover the study area with respect totissue samples and biopsies, and the privatepractices of all General Practitioners and anumber of other specialties in the studyarea. Since patients with target diseases areoften referred to University Hospitals, spe-cializeddinics, or treatment facilities, se-veral sources outside the study area were

- included. Finally all death certificates forthe study period and area were abstracted inthe respective Departments of Health (Tab.2). Wherever available, secondary sources(discharge books, laboratory protocols, re-ferral books, nurses' notes etc.) were alsocompletely screened for candidate cases.Any hint was validated thoroughly, how-ever cases were only accepted if the secon-dary data could be confirmed in the medi-cal history or any equivalent written docu-ment.Altogether some 2.5 million files wereanalysed for this study.

Tab. 2 Primary data sources

Hospitals in study area 8Hospitals in adjacent rural 14 *)districts and regionalcenters

- Departments 46Hematology/Oncology 4

Internal Medicine 15

Pediatrics 3Pathology 8

Radiation therapy 6Surgery 7

Computer tomography 2Ear, nose, throat 1

Private practices 314- Specialities -.

General Practitioners 223Internal Medicine/Oncology 63

Pediatrics 28Regional oncological clinics 3

Dept. of Health 3

*) Including Medical Universities: Kiel,Lubeck, Hamburg (UKE)

Record linkageAll cases were subsequently compared byvarious dimensions (Date of birth, sex, dia-gnosis, date of first diagnosis, rural com-mune of residence) to link the datasets ofthose patients who had been abstractedfrom more than one data source. As a con-sequence of the complete screening of allavailable data sources some 60% of the pa-tients had been documented in at least twoindependent sources (not counting differentdata sources within the same department orhospital).

239


The first stages of the record linkage werealways performed while the screening invarious primary and secondary data sourcesand departments of the same hospital wasstill ongoing. Most discrepancies could thusbe resolved and many incomplete datasetscould still be supplemented by specific in-vestigations with full access to the originaldata.Remaining inconsistencies and unconfir-med cases could in some cases be solvedand completed in the final step of the re-cord linkage process, in which the datafrom all sources were compiled.

Standardized incidence ratiosAs a measure for the geographical inci-dence pattern standardized incidence ratios(SIR) were calculated. SIR reflect the ratiobetween an observed (age- )standardizedincidence and a reference incidence. AnSIR of 100 indicates that the standardizedincidences in the region of interest is thesame as in the reference population. AnSIR below 100 means a lower standardizedincidence in the region of interest, an SIRabove 100 corresponds to an increasedstandardized incidence in the region of in-terest.P-values were calculated for all SIR underthe assumption of a Poisson distribution [7]to quantify the probability that an observedincrease in incidence is actually due tochance. In line with the general convention,statistical significance is assumed if thisprobability is less than 0.05. In the resultssection of this paper approximated confi-dence intervals are given in addition to thep-values. Unlike the p-values, these confi-dence intervals are based on rounded SIR.The purpose of these intervals is visuali-zation of the differential precision of singlevalues in the figures. They should not beinterpreted in terms of statistical signifi-cance.

Reference populationThe most desirable reference population ina geographical incidence study would bethe general population of the Federal Re-public of Germany as a whole. However,incidence data on the national level do notexist in Germany. In Federal State of Saar-land incidences have routinely been publi-shed for each year after 1972, but due toconsiderable differences with respect topopulation density, industrialization andother factors we assume that the figures donot readily apply to the population in Nor-thern Germany.To avoid systematic under- or overestima-tion of the expected incidence in the studyarea, we selected as the reference theoverall incidence in the pooled populationof the study region. Indirect age standardi-zation for both the incidences in the regionsof interest and the reference was performedapplying the age-specific incidences to thepopulation of the Federal Republic of Ger-many, as determined in the national censusof 1987.

HypothesisIn principle, an incidence study can be donefor purely descriptive purposes and hence-does not necessarily require an a priori hy-pothesis. However, the circumstances· underwhich the Retrospective Incidence StudyElbmarsch was commissioned clearly im-plied a geographical analysis with respectto the nuclear power plant Krummel andthe GKSS. It was to be determined,whether or not an increase in leukemia in-cidence, which had been observed in child-ren in the direct vicinity of the nuclearfacilities would also be present in adults.To allow for analysis of a trend with dis-tance, concentric regions of 0-5, 5-10, lO-IS, 15-20 and more than 20 km around thefacilities were constructed.The second purpose of the study was toderive SIR for each of the 35 administrative

240


Chronic myeloid leukemiaChronic myeloid leukemia (CML) accoun-ted for about half of the excess cases in the0-5 km region. In a subgroup analysis theSIR for CML was 194 for both sexes com-bined (p = 0.049), the excess again being

Leukemias attributable to excess cases in males (obs.Main result of the analysis was a statisti- 10, expo 5.4; SIR = 275; p=O.013; Fig. 2).Tab. 3 Distribution of cases over diagnostic groups and subgroups

units of the study area ("incidence map-ping"; results not presented).

ResultsFor the years 1984-1993 4600 cases withtarget diseases were abstracted from theprimary data sources. After extensive re-cord linkage 2253 distinct cases remained.In 261 of these information on the year ofprimary diagnosis was insufficient orcompletely absent. The majority of thesecases had been abstracted from death certi-ficates which in Germany do not routinelycontain this information. To avoid anypossible overestimation of the SIR, all 261were excluded from the analysis, althougha major proportion of these cases had likelybeen diagnosed in the study period.Another seven cases were documented withtypical monoclonal gammopathy, but thediagnosis of multiple myeloma could not besubstantiated before the study was closed.These cases were also excluded.Consequently, the analysis was based on1985 cases (Table 3).

cally significant increase in the incidence ofleukemia (all forms) for both sexes com-bined in the 0-5 km-region around theplant. Between 1984 and 1992 51 case.swere observed, whereas only 40 caseswould have been expected. The (age-)standardized incidence exceeded theoverall age-standardized incidence in thestudy area by 28 % (SIR=128; p=0,046,Fig. 1).No significant deviations from the expectedvalues were observed in any of the remai-ning concentric regions for both sexescombined and in separate analyses for ma-les and females. The gender specific analy-ses, however, revealed that the excess inci-dence in the vicinity of the plant was con-fined to males (32 cases observed, 21 ex-pected; SIR = 156; p=0.0099; Fig. 1).

Diagnoses Males Females TotalLeukemias (all forms) 338 295 633Major subgroups:

- acute leukemia 133 99 232- other leukemia 205 196 401

Subtypes:- CML 48 37 85- CLL 137 135 272

NHL 321 274 595M.Hodgkin 91 66 157Mult. myeloma 142 175 317Myeloproliferative syndrome/ 115 139 254myelodysplastic syndromeAplastic anemia 8 21 29Total 1015 970 1985CML = chronic myloid leukemia; CLL = chronic lymphatic leukemia; NHL = Non Hodgkin'slymphoma; Mult myeloma = multiple myeloma

241


Age distribution of the SIRIn another subgroup analysis, cases weredichotomized according to their ages at thetime of first diagnosis. When age 65 is usedas a cutpoint roughly half of the patientsend up in the younger age group.The results of the cases below 65 yearsshow a different pattern as was observedfor all age groups combined. Again, inci-dence of all leukemias was found signifi-cantly increased in the central region (obs.31, expo 16.8; SIR=I78; p=0.0023). How-ever, unlike in the previous analysis for allages combined, amog the younger casesboth males and females contributed to thisincrease to a comparable extent. About halfof the excess cases under 65 years of agewere males (obs. 17, expo 9.8; SIR=65.6; P= 0.041) and females, respectively (obs. 14,expo 7; SIR=194.4; p=O.013; Fig. 3). Thispattern was similiar in separate analyses forthe agegroups 0-<15 and 15-64 years(results not shown).

Comparison of SIR in 1984-1988 vs.1989-1993To evaluate any temporal trend in leukemiaincidence in the vicinity of the nuclearpower plant the study period was dividedinto two consecutive 5 year periods, 1984-1988 and 1989-1993, respectively. Theexcess in leukemia incidence in the 0-5 kmregion for both genders combined wasconfined almost exclusively to the morerecent 5 year period (obs. 30, expo 19.3;SIR=155; p=0.015), whereas between1984-1988 no deviation from the expectedvalue was observed (21 obs., 21.1 exp.).The increased incidence in the 0-5 km re-gion in 1989-1992 again was mostly attri-butable to excess cases in males (obs. 18,expo 10.1; SIR=180; p=0.016), but SIRwere also elevated in females (obs. 8, expo5.5; SIR=127.3), though this elevation wasnot statistically significant (p>0.2).

Discussion"Cluster" analysis has become a widelyused term for the study of spatial diseasepatterns in small geographical areas [15].Though by definition ecological at the de-sign level, many small area studies havebeen supplemented with investigations atthe individual level. Cluster analysis thendevelopes into "micro-epidemiology" [71].While micro-epidemiology has been criti-cized for its conceptual and methodologicallimitations [72, 73], some well-known ex-amples nevertheless prove its principal po-tential as a tool in environmental epidemio-logy [2, 13,44,67, 78].Results of cluster analysis and micro-epi-demiology have become key arguments in adebate on childhood and juvenile leukemiarisks in the vicinity of nuclear facilities (see[25, 62, 81, 89] for reviews). Increasedrisks were observed in geographical studiesin England and Wales [3-5, 12,22,24,27,70, 84, 85], Scotland [41, 42, 48], the US[10, 14,31, 32,40,53], Germany [17, 37,38, 46, 63, 69] and recently in France [86-88].Hardly ever, however, have detailed geo-graphical leukemia incidence studies inclu-ded adults of all ages. This is likely due toboth methodological difficulties and thetremendous endeavour of a complete caseascertainment in terms of time and cost.In the Southeastern Massachusetts HealthStudy, all leukemias except CLL wereascertained among adults over a period ofthree years in five towns adjacent to thenuclear boiling water reactor Pilgrim nearPortsmouth, Massachusetts (USA). Datasources were vital statistics records and theMassachusetts cancer registry. Usingstatewide incidences as a reference, the SIRfor all leukemias (except CLL) was 160 (27obs., 16.9 exp.; 95% confidence interval108-237) [10]. In a subgroup analysis, alsothe SIR for myelogenous leukemias wassignificantly increased (19 obs., 9.9 exp.;SIR=191; 95%CI 120-304). As in our

242


study, the increase was confined to males(13 obs., 5 exp.; SIR 252).In a subsequent case control study inclu-ding I 15 leukemia cases and 230 healthycontrols, information on individual riskfactors like smoking, socioeconomicfactors, and occupational exposures wereobtained by means of standardized tele-phone interviews and the analyses wereadjusted for each of these factors simultan-eously. A composite exposure index wasthen calculated using distance of residenceand place of work to the plant, weightedwith the time spent at each location and thetotal time the respective locations weredownwind of the plant. When levels of thisexposure index were trichotomized, and thelowest third was used as a reference, resul-ting odds ratios increased from 2.2 (95%CI0.9-5.2) for males and 1.5 (0.4-5.6) for fe-males in the medium third to 3.12 (1.2-8.5)and 6.5 (1.53-27.2) for males and females,respectively in the highest exposure catego-ry [65].

Radioactive exposureThe childhood leukemia cluster in the 0-5km region immediately suggests a potentialrole of radioactive releases of KKK and/orGKSS. However, careful examination ofprotocols of the routine environmental ra-diation surveillance around the nuclearfacilities yielded no indication for anymajor accidental release. Radioactive con-tamination of lifestock, soil, and vegetationwith long-lived radionuclides was in therange which was usually observed in Nor-thern Germany after the Chernobyl cata-strophy. Whole-body counting of Elb-marsch inhabitants in 1991 likewise yieldedmoderately elevated specific activities ofincorporated Caesium-137 which could beattributed to the Chernobyl fallout.Many authors have stressed that exposuredoses in the vicinity of nuclear powerplants in normal operation are several or-

ders of magnitude too smaIl to cause anysomatic effect [16, 20, 39, 65].Discussing a potential contribution of ra-dioactive releases of the plants in quantita-tive terms, however, it should be taken intoaccount that• the susceptibility for radiation-induced

leukemia increases with younger ageby a factor of 5-10 (adults vs. childrenbelow age 5 [79]). When prenatal ex-posure is considered this ratio comesclose to 100 [82].

• preconceptional exposure adds a com-paratively new exposure pathwaywhich is still not accounted for in con-ventional risk coefficients [66, 80].The magnitude of this effect is present-ly unknown.

• effects of complex combinations ofenvironmental exposures are stilI verypoorly understood. Nevertheless,synergistic effects between radiationand other, e.g. chemical exposurescould well enhance the leukemogenicpotential of either of the factors [29,92].

Finally, exposure assessment in such stu-dies is usuaIly considerably less than per-fect. Routine environmental radiation sur-veillance can miss exposures due to *- andB-emitters or extremely inhomogenousspatial/temporal distributions of raadio-nuclides. Indeed, we havespeculated earlier[77] that elevated rates of structural chro-mosomal aberrations in a casual sample ofElbmarsch residents would be compatiblewith past releases of short-lived fissionproducts which could have been missed byroutine surveillance.Some of the subgroup results are well inline with the radiation hypothesis. For allleukemias, the increase in SIR was highestin children, but, to a lesser degree was alsopresent in adults. CML, which w"as alsosignificantly increased in both sexes and inmales, is a sentinel radiation-induced leu-kemia. The fact that the increase in inci-

243


dence began only after 1989, whereas nosuch elevation was observed for the years1984-1988 could correspond to the com-missioning of the nuclear power plant in1984. A minimum latency period of 5 yearsfor radiation induced leukemia is compa-tible to the findings in the survivors ofHiroshima and Nagasaki [8] as well as inthe Oxford Survey of Childhood Cancer[83]. The maximum of the latency distribu-tion for adults, however, is considerably la-ter than 5 years [49]. Hence, if the increasein leukemia incidence in the 0-5 km regionwould indeed be radiation-induced thiswould imply that the bulk of excess adultcases is still to come, even if the radioac-tive exposure was confined to the earlyyears of operation of the plant.At this point geographical epidemiologycannot provide any further insight. Theabundance of investigations and measure-ments which have so far been performedcan be condensed into just a few causalhypotheses, which now must be tested inanalytical studies. For this purpose a largecase-control study is presently being desi-gnedand will probably be commissionedby the Federal Ministry for the Environ-ment, Natural Protection and ReactorSafety in early 1996.

ConclusionAn increased leukemia incidence in adultswas observed in the 0-5 km region aroundthe nuclear boiling water reactor Krummeland the nuclear research facility GKSS, anarea where a hitherto unprecedented child-hood leukemia cluster had been confirmedearlier.It cannot be stressed enough, though, thatdue to the ecologic character of the Retro-spective Incidence Study Elbmarsch thefindings must not be interpreted in terms ofcausality. So far, the association betweenan increased incidence of leukemia and theimmediate vicinity of the nuclear reactorKrummel is merely geographical.

However, main results of this study as wellas findings in subgroup analyses are com-patible with a contribution of the nuclearfacilities to the excess cases:• a significant increase in leukemia inci-

dence in the 0-5 km region around thenuclear power plant, for both genderscombined and males only (1984-1993),together with expected incidences inall other geographical regions.

• a significant increase in chronic mye-loid leukemia in both genders and ma-les only (1984-1993)

• significantly elevated SIRs for leuke-mia in males and females below 65years of age (1984-1992)

• appearance of excess cases only after1989 whereas incidences were as ex-pected in the first 5 years of the opera-tion of the plant.

These results clearly require further inve-stigations into the possible causes for thisunique cluster of childhood and adult leu-kemia. Most appropriately this could bedone by means of an epidemiologic case-control study.

References1. Aksoy M, Erdem S. Follow-up study on

the mortality and the development of leukemiain 44 pancytopenic patients with chronic expo-sure to benzene. Blood 1978;52:285-292.

2. Baris YI, Simonato L, Artvinli M, et al.Epidemiology and environmental evidence ofhealth effects of exposure to erionite: A fouryear study. International Journal of Cancer1987;39: 10-17.3. Baron JA. Cancer mortality in small areas

around nuclear facilities in England and Wales.Br J Cancer 1984;50:815-824.

4. Barton CJ, Roman E, Ryder HM, WatsonA. Childhood leukemia in West -Berkshire[letter]. Lancet 1985:1248-1249.5. Black D. Investigation of the possible in-

creased incidence of cancer in West Cumbria:Report of the Independent Advisory Group.London Her Majesty's Stationary Office, 1984.

244


6. Brandt L, Nilsson PG, Mite]man F. Occu-pationa] exposure to petroleum products in menwith acute non-]ymphatocytic leukaemia. BMJ]978;4:553.7. Bres]ow NE, Day NE. Statistical methods

in cancer research. Vol. II - The design andanalysis of cohort studies. Lyon IARC, ] 987.

8. Brill AB, Tomonaga M, Heysse] RM.Leukemia in man following exposure to ioni-zing radiation. Ann Intern Med 1962;56(4):590-609.

9. Brown LM, Blair A, Gibson R, et al. Pe-sticide exposures and other agricultural riskfactors for leukemia among men in Iowa andMinnesota. Cancer Res 1990;50:6585-6591.10. Clapp R, Cobb S, Chan C, Walker B.

Leukemia near Massachusetts nuclear powerplant [letter]. Lancet ]987;2:1324-1325.11. Collmann GW, Loomis DP, Sandler DP.

Childhood cancer mortality and radon concen-tration in drinking water in North Carolina. Br JCancer 1991;63:626-629.12. Cook-Mozaffari PJ, Darby SC, Doll R, et

al. Geographica] variation in mortality fromleukemia and other cancers in England andWales in relation to proximity to nuclear instal-lations, 1969-78. Br J Cancer 1989;59:476-485.13. Creech J, Johnson MN. Angiosarcoma of

the liver in the manufacture of po]yvinyl ch]ori-de. J Occup Med 1974;16:150-151.14. Crump KS, Ng T-H, Cuddihy RG. Cancer

incidence patterns in the Denver metropolitanarea in relation to the Rocky Flats Plant. Am JEpidemioI1987;126:127-135.15. Cuzick J, Elliott P. Small-area studies:

Purpose and methods. In: Elliott P, Cuzick J,Eng]ish D, Stem R, editors. Geographica] andEnvironmenta] Epidemio]ogy. Oxford, NewYork, Tokyo: Oxford University Press,]992:14.16. Darby SC, Doll R. Fallout, radiation doses

near Dounreay, and childhood leukaemia. BMJ1987;294:603-607.17. Demuth M. Leukamieerkrankungen bei

Kindem in der Umgebung von Atomanlagen.In: Kohnlein W, Kuni H, Schmitz-Feuerhake I,editors. Niedrigdosisstrah]ung und Gesundheit.Berlin: Springer Verlag, 1990: ]27-35.18. Diamond EL, Schmerler H, Li]ienfe]d

AM. The relationship of intra-uterine radiationto subsequent mortality and development of

leukemia in children. Am J Epidemiol1973;97:283-313.19., Dieckmann H. Haufung von Leukamieer-

krankungen in der Elbmarsch. Gesundheitswe-sen 1992; 10:592-596.20. Dionian J, Wan SL, Wrixon AD. Radia-

tion doses to members of the public aroundAWRE, Aldermaston, ROF, Burghfield andAERE, Harwell. Chilton, Didcot, Oxon Natio-nal Radiation Protection Board, 1987.2]. Epstein S, Ozonoff D. Leukemias and

blood dyscrasias following exposure to Chlor-dane and Heptach]or. Teratogenesis CarcinogMutagen 1987;7:527-540.22. Ewings PD, Bowies C, Phillips MJ, John-

son SAN. Incidence of leukemia in young peo-ple in the vicinity of Hink]ey Point nuclearpower station, 1959-1986. BMJ 1989;299:289-293.23. Feychting M, Ah]born A. Magnetic fieldsand cancer in people residing near Swedishhigh-voltage power lines. Am J Epidemio]1993;138:467-481.24. Forman D, Cook-Mozaffari P, Darby S, eta!. Cancer near nuclear installations. Nature1987;329:499-505.25. Gardner MJ. Review of reported increases

of childhood cancer rates in the vicinity ofnuclear installations in the UK. Journal of theRoyal Statistical Society A 1989; 152:307-325.26. Gardner MJ, Snee MP, Hall AJ, Powell

CA, Downes S, Terrell JD. Results of a case-control study of leukemia and lymphoma amongyoung people near Sellafied nuclear plant inWest Cumbria. BMJ 1990;300:423-442.27. Gardner MJ, Winter PD. Mortality in

Cumberland during 1959- I978 with reference tocancer in young people around Windsca]e[letter]. Lancet 1984:217-2] 8.28. Gibson RS, Graham S, Li]ienfe]d A,

Schuman L, Dowd IE, Levin ML. Irradiation inthe epidemiology of leukemia among adults.JNCII972;48:30]-311.29. Gibson RW, Bross IDJ, Graham S, et al.

Leukemia in children exposed to multiple riskfactors. N Eng] J Med 1968;279(17):906-909 .. 30. Gi]man EA, Kneale GW, Knox EG, Ste-wart _A. Pregnancy and childhood cancers: Ef-fects of exposure age and radiation dose. J Ra-dio] Prot 1988;8(1):3-8.

245


31. Goldsmith JR. Childhood leukemia mor-tality before 1970 among populations near twoUS nuclear installations [letter]. Lancet1989; 1:793.32. Goldsmith JR. Childhood leukemia mor-

tality before 1970 among populations near twoUS nuclear installations. Lancet 1989;2:1443-1444.33. Greiser E, Lotz 1. Leukamie-Inzidenzinder Nahe einer SondermUl1deponie. BremenBremer Institut fUr Praventionsforschung, 1990.34. Greiser E, Lotz I, Hoffmann W, Schill W,

Hilbig K. Nahe zu einer SondermUlldeponie undandere Risikofaktoren fUr die Entstehung vonLeukamien, malignen Lymphomen und multi-plen Myelomen. Bremen, Germany Bremer In-stitut fUr Praventionsforschung (BIPS), 1995.35. Greiser E, Lotz I, Hoffmann W, Wosniok

W. Leukamien, maligne Lymphome und Mul-tiple Myelome im Umdkreis Holzminden.Orientierende Einschatzung der epidemiologi-schen Situation unter besonderer BerUcksichti-gung der Gemeinde Bevern. Bremen Bremer In-stitut fUr Praventionsforschung und Sozialme-dizin (BIPS), 1994.36. Greiser E, Lotz I, Steinkohl M, Brand H.

Leukamien und maligne Lymphome in den nie-dersachsischen Landkreisen Nienburg undSchaumburg - Regionale Inzidenzen in 24 Ge-meinden. Bremen Bremer Institut fUr Praven-tionsforschung, 1991.37. Grosche B, Hinz G, Tsavachidis C, Kaul

A. Analyse der Leukamiemorbiditat in Bayernin den Jahren 1976-1981. Teil I Grundlagen,Methodik und medizinische Aspekte. Neuher-berg Bayerisches Staatsministerium fUr Landes-entwicklung und Umweltfragen (Hrsg.), InstitutfUr Strahlenhygiene des Bundesgesundheitsam-tes, 1987.38. Grosche B, Hinz G, Tsavachidis C, Kaul

A. Analyse der Leukamiemorbiditat in Bayernin den Jahren 1976-1981. Teil II. Risikofakto-ren, regionaJe Verteilung und epidemiologischeAspekte. Neuherberg Bayerisches Staatsmini-sterium fUr Landesentwicklung und Umweltfra-gen (Hrsg.), Institut fUr Strahlenhygiene desBundesgesundheitsamtes, 1987.39. Hatch MC, Beyea J, Nieves JW, SusserM. Cancer near the Three Mile Is]and nuclearplant: Radiation emissions. Am J Epidemiol1990; 132(3):397-4] 2.

40. Hatch MC, Wallenstein S, Beyea J, Nie-ves JW, Susser M. Cancer rates after the ThreeMile Island nuclear accident and proximity ofresidence to the plant. Am J Public Health1991 ;81 (6):719-724.41. Heasman MA, Kemp IW, MacLaren A,

Trotter P, Gillis CR, Hole AJ. Incidence of leu-kemia in young persons in the West of Scotland[letter]. Lancet 1984;1:1188-1189.42. Heasman MA, Kemp IW, Urquhart JD,

Black R. Childhood leukemia in northern Scot-land [Jetter]. Lancet 1986; I :266.43. Henshaw DL, Eatough JP, RichardsonRB. Radon as a causative factor in induction ofmy]oid leukemia and other cancers. Lancet1990: I 008-10 12.44. Herbst AL, Ulfelder H, Poskanzer DC.Adenocarcinoma of the vagina: Association ofmaternal stilbestrol therapy with tumor appear-ance in young women. N Engl J Med1971;284:878-881.45. Hoffmann W, Kranefeld A, Schmitz-Feu-erhake I. Radium-226-contaminated drinkingwater: hypothesis on an exposure pathway in apopulation with elevated childhood leukemia.Environ Health Perspect 1993; 10 I Suppl 3: 113-115.46. Hoffmann W, Lotz I, Greiser E. Leuka-miefalle im Landkreis Steinburg, orientierendeEinschatzung der epidemiologischen Situation.Bremen Bremer Institut fUr Praventionsfor-schung und Sozialmedizin (BIPS), 1994.47. Hole DJ, Gillis CR. Childhood leukemiain the west of Scotland [letter]. Lancet1986;2:524-525.48. Hostrup 0, Witte I, Hoffmann W, GreiserE, Butte W, Walker G. Biozidanwendungen imHaushalt als moglicher Risikofaktor fUr die Ge-sundheit der Raumnutzer. AbschluBbericht zurerweiterten Nachuntersuchung del' Fall-Kon-troll-StudieMUnchehagen ("Nahe zu einer Son-dermUIldeponie und andere Risikofaktoren fUrdie Entstehung von Leukamien, malignen Lym-phomen und multiplen.Myelomen"). Praxis derBiozidanwendung, cemisch-analytische Expo-sitionserfassung uns Leukamierisiko durch An-wednung von Bioziden in Innenraumen. Bre-men Bremer Institut fUr Praventionsforschung(BIPS), 1996.49. Ichimaru M, Ishimaru T, Belsky JL, et al.Incidence of leukemia in atomic bomb survi-


vors, Hiroshima and Nagasaki 1950-71, by ra-diation dose years after exposure, age, and typeof leukemia. Hiroshima Radiation Research andEffects Foundation, Technical Report RERF 10-76, 1977.50. Infante PF, Rinsky RA, Wagoner JK,

Young RJ. Leukaemia in benzene workers.Lancet 1977;2:76-78.51. International Agency for Research on

Cancer (IARC). IARC Monographs on theevaluation of carcinogenic risks of chemicals tohumans. Some industrial chemicals anddyestuffs. v. 29. Lyon IARC, 1982:93-148.52. Jakobsson R, Ahlborn A, Bellander T,

Lundberg 1. Acute myloid leukemia among pe-trol station attendants. Arch Environ Health1993;48(4):255-259.53. Johnson CJ. Cancer incidence in an area

contaminated with radionuclides near a nuclearinstallation. Colorado Medicine 1981 ;78:385-392.54. Johnson CJ. An increase in leukemia

deaths in South Dakota associated with thirty-two radioactive fallout plumes that traversed theState from nuclear bomb tests at the NevadaTest Site. International Perspectives in PublicHealth 1989;5:3-6.55. Kaatsch P, Haaf HG, Michaelis J. Jahres-

bericht 1993 des Deutschen Kinderkrebsregi-sters. Mainz Institut fUr Medizinische Statistikund Dokumentation der Universitat Mainz,1994.56. Kitabake TT, Watanabe T, Koga S. Ra-

diation cancer in japanese radiological workers.Strahlentherapie 1973; 146:599-606.57. Kneale GW. Geographic and sociodemo-

graphic factors in the etiology of juvenile neo-plasms: Implications for leukemia clusters ofseveral factors including density, backgroundradiation, infections and immunisations. In:Lengfelder E, Wendhausen H, editors. NeueBewertung des Strahlenrisikos. Niedrigdosis-Strahlung und Gesundheit. Miinchen: MMVMedizin Verlag, 1992:141-57.58. Leiss JK, Savitz DA. Home pesticide useand childhood cancer: A case-control study. AmJ Public Health 1995;85:249-252.59. Lyman GH, Lyman CG, Johnson W. As-

sociation of leukemia with radium groundwatercontamination. JAMA 1985;254:621-626.

60. Lyon JL, Klauber MR, Gardner.JW, UdallKS. Childhood leukemia associated with falloutfrom nuclear testing. N Engl J Med1979;300:397-402.61. MacMahon B. Prenatal X-ray exposure

and childhood cancer. JNCI 1962;28: 1173-1191.62. MacMahon B. Leukemia clusters around

nuclear facilities in Britain. Cancer Causes Con-troI1992;3:283-288.63. Michaelis J, Keller B, Haaf G, Kaatsch P.

Incidence of childhood malignancies in thevicinity of West German nuclear power plants.Cancer Causes Control 1992;3:255-263.64. Monson RR, MacMahon B. Prenatal X-

ray exposure and cancer in children. In: BoiceJD, Fraumeni JF, editors. Radiation Carcinoge-nesis: Epidemiology and biological signifi-cance. New York: Raven Press, 1984:97-105.65. Morris M, Knorr RS. Southeastern Mas-sachusetts Health Study 1978-1986. Boston,Mass. Division of Environmental Health as-sessment, Massachusetts Dept. of Public Health,1990.66. Natarajan N, Bross IDJ. Preconceptionradiation and leukemia. J Med 1973;4:276-281.67. Neutra RR. Counterpoint from a cluster

buster. Am J Epidemiol 1990;132(1):1-8.68. Pearce NE, Sheppard RA, Howard JK,Fraser J, Lilley BM. Leukemia among NewZealand agricultural workers. A cancer registry-based study. Am J Epidemiol 1986;124:402-409.69. Prindull G, Demuth M, Wehinger H. Can-

cer morbidity rates of children from the vicinityof the nuclear power plant of Wurgassen (FRG).Acta HaematoI 1993;90:2.0-93.70. Roman E, Beral V, Carpenter L, et al.Childhood leukemia in the West Berkshire andBasingstoke and North Hampshire DistrictHealth Authorities in relation to nuclear estab-lishments in the vicinity. BMJ 1987;294:597-602.71. Rose G. [Preface]. In: Elliott P, Cuzick J,

English D, Stern R, editors. Geographical andEnvironmental Epidemiology. Oxford, NewYork, Tokyo: Oxford University Press,1992: VII.72. Rothman KJ. Clustering of disease. Am J

Public Health 1987;77: 13-15.

247


73. Rothman KJ. A sobering start for the clu-ster busters' conference. Am J Epidemiol1990; 132(\ ):S6-S 13.74. Savitz DA. Epidemiologic studies of

electric and magnetic fields and cancer: Strate-gies for extending knowledge. Environ HealthPerspect 1993;IOI(SuppI4):83-91.75. Savitz DA, Ahlborn A. Epidemiologicevidence on cancer in relation to residential andoccupational exposures. In: Carpenter DO,Ayrapetyan S, editors. Biological effects ofelectric and magnetic fields. Beneficial andharmful effects. v. 2. San Diego, New York,Boston, London, Sydney, Tokyo, Toronto:Academic Press, 1994:233-62.76. Savitz DA, Wachtel H, Barnes FA, John-

son EM, Tvrdik JG. Case-control study ofchildhood cancer and exposure to 60-Hz ma-gnetic fields. Am J Epidemiol 1988; I 28:21-38.77. Schmitz-Feuerhake I, Schroeder H, Dann-

heim B, et al. Leukaemia near water nuclear re-actor [letter]. Lancet 1993;342:1484.78. Shands KN, Schmid GP, Dan BB, et al.

Toxic-shock syndrome in menstruating women.N Engl J Med 1980;303:1436-1442.79. Shimizu Y, Kato H, Schull WJ. Studies ofthe mortality of A-bomb survivors. Mortality1950-1985: Part 2. Cancer mortality based onthe recently revised doses (DS86). Radiat Res1990; 121(2): 120- 141.80. Shiono PH, Chung CS, Myrianthopoulos

NC. Preconception radiation, intrauterine dia-gnostic radiation, and childhood neoplasia.JNCII980;65(4):681-686.81. Shleien B, Ruttenber AJ, Sage M. Epi-

demiologic studies of cancer in populations nearnuclearfacilities. Health Phys 1991 ;61 :699-713.82. Stewart AM, Gilman EA, Kneale GW.

Radiation dose effects in relation to obstetric X-ray and childhood cancer. Lancet 1970;2: I 185-1188.83. Stewart AM, Kneale GW. Age-distribu-

tion of cancers caused by obstetric X-rays andtheir relevance to cancer latent periods. Lancet1970;1:4-8.84. Urquhart J, Cutler J, Burke M. Leukemia

and lymphatic cancer in young people nearnuclear instaJlations [Jetter]. Lancet 1986:384.

85. Urquhart J, Palmer M, Cutler J. Cancer inCumbria: The Windscale connection. Lancet1984:217-218.86. Viel Jean-F, Pobel D, CamS A. Incidence

of leukemia in young people around the La Ha-gue nuclear waste reprocessing plant: A sensi-tivity analysis. Statistics in Medicine1995;14:2459-2472.87. Viel JF, Richardson ST. Childhood leu-

kaemia around the LaHague nuclear waste re-processing plant. BMJ 1990;300:580-58 1.88. Viel Jean-F, Richardson S, Danel P, et al.

Childhood leukemia incidence in the vicinity ofLa Hague nuclear-waste reprocessing facility(France). Cancer Causes Control 1993;4:341-343.89. Wakeford R, Binlas K. Childhood leu-

kemia and nuclear instaJlations. Journal of theRoyal Statistical Society A 1989;152 (part1):61-86.90. Washburn EP, Orza MJ, Berlin JA, et al.Residential proximity to electricity transmissionand distribution equipment and risk of child-hood leukemia, childhood lymphoma, andchildhood nervous system tumours: Systematicreview, evaluation, and meta analysis. CancerCauses Control 1994;5:299-309.91. Wertheimer NW, Leeper E. Electrical wi-ring configurations and childhood cancer. Am JEpidemiol 1979; 109:237-284.92. Wheldon TE, Mairs R, Barrett A. GermceJl injury and childhood leukemia clusters.Lancet 1989:792-793.

(This text is a reprint of a chapter publishedpreviously by the authors. Hoffmann W, GreiserE: Inceased Incidence of leukaemias in thevicinity of the Kriimmel nuclear power plant inNorthern Germany. In: Environmental Epide-miology in Europe 1995. Proceedings of an In-ternational Symposium. Frentzel-Beyme R,Ackermann-Liebrich U, Bertazzi PA, Greiser E,Hoffmann W, Olsen J (eds). Bremen Institutefor Prevention Research and Social Medicine(BIPS) for the European Commission, Di-rectorate General V, Bremen 1996, 185-206)

248


Fig. 1 Age-standardized incidence ratios (SIR) for all leukemias

Total<5km

5 -< 10 km10 -< 15 km15 -< 20 km

> 20 kmMales<5km

5 -< 10 km10 -< 15 km15 -< 20 km

> 20 kmFemales<5km

5 -< 10 km10 -< 15 km15 -< 20 km

> 20 km

OlE 51/40I

I

-l ~OlE 32/21

,I

f-- --j

I

I I

I

p (Poisson)

0.04570.81860.18200.88930.6181

0.00990.93470.58680.85320.5085

0.56340.59270.08980.76490.6700

o 50 100 150 200 250ole = observed/expected number of cases

Fig. 2 Age-standardized incidence ratios (SIR) for chronic myeloid leukemiap (Poisson)

0.04940.94560.85960.18570.7246

Total<5 km

5 -< 10 km10 -< 15 km15 -< 20 km

> 20 kmMales<5km

5 -< 10 km10 -< 15 km15 -< 20 km

> 20 kmFemales<5km

5 -< 10 km10 -< 15 km15 -< 20 km

>20km

(OlE 10/5.4)

Bt-t(OlE 8/3)

I

HI

t-+

I

,-'- J

I

I

0.01290.77920.27980.7912

0.68490.72040.81680.28500.5278

o 100 200 300 400 500 600ole = observed/expected number of cases

249


p (Poisson)

0.00230.94590.91200.74500.3635

Fig. 3 Leukemia incidence in agegroups below 65 years(age at time of first diagnosis)

Total<5 km

5 -< 10 km10 -< 15 km15 -< 20 km

> 20 kmManner< 5 km

5 -< 10 km10 -< 15 km15 -< 20 km

> 20 kmFrauen<5 km

5 -< 10 km10 -< 15 km15 -< 20 km

> 20 km

I,,

1-HI ,

,,

,,

l ,I

0.04070.99470.81800.64660.3528

0.01330.52270.86520.74490.4798

SIRFig. 4

200

150

100

50

o

o 50 100 150 200 250 300 350

Comparison of SIR in two consecutive 5 year periods (1984-'88 vs. 1989-'93)1~1984-1988 D1989-1993 I

OlE 30/19.3 * (OlE 815.5 ns)

* OlE 18/10.1

1I

Ji J

Total Males Femalesole = observed/expected number of cases

250

Documents

Epidemiologic Evaluation of Leukemia Incidence in Children ... · Epidemiologic Evaluation of Leukemia Incidence in Children and Adults in the Vicinity of the Nuclear Power Plant