THE RADIATION exposure of workers at nuclear facilities is low-dose and low-dose-rate exposure, and it is regarded as protracted or chronic exposure. To investigate the health effect associated with such radiation exposure, numerous epidemiological studies have been conducted on radiation workers in many nuclear facilities (UNSCEAR 2006). Recently, the International Nuclear Workers Study (INWORKS), in which pooled analysis was performed on radiation workers in France, the United Kingdom, and the United States, estimated the risk of not only cancer but also circulatory diseases (Leuraud et al. 2015; Richardson et al. 2015, 2018; Gillies et al. 2017). In the United States, the Million Worker Study has been continuing for nearly 25 y and includes radiation workers, atomic veterans, and radiologists (Boice 2017). In Japan, mortality risks among radiation workers, mainly workers at nuclear power plants, have been surveyed by the Radiation Effects Association as a government-commissioned project since 1990 (Kudo et al. 2017, 2018). Therefore, knowledge of the health effects of low-dose and low-dose-rate radiation exposure has been obtained both domestically and internationally.
On the other hand, the effects of dose rate are well known, especially in the field of radiation biology and from experimental results (Storer et al. 1979; NRC 2006). In light of the fact that the dose rate may affect the radiation effect, for example, there may be a difference in the risks between (a) exposure to radiation on the order of 100 mSv in a short period and (b) exposure to a small amount of radiation with a cumulative dose reaching on the order of 100 mSv over several decades. However, in epidemiological studies on radiation workers, the dose considered was mainly the cumulative dose, assuming a lag period.
Regarding this point, Metz-Flamant et al. (2012) estimated the leukemia risk associated with chronic radiation exposure in a cohort of French nuclear workers by considering a dose-rate window. The dose-rate window method involves assigning annual doses as either lower or higher than a specific annual dose rate (cut point of the dose rate) to calculate the cumulative dose. They found that the excess relative risk (ERR) of leukemia (excluding chronic lymphatic leukemia, CLL) appeared to be higher for doses received at exposure rates exceeding 20 mSv y−1 (Metz-Flamant et al. 2012) but noted that, other than for the 20 mSv y−1 cut point, the ERR did not converge under Poisson regression. For cancers excluding leukemia, Furuta et al. also applied this dose-rate window method to a cohort of Japanese radiation workers;3 nevertheless, such studies are still limited, and basic knowledge and information on the risks of chronic exposure considering the dose rate are still necessary to elucidate the radiation risks at the usual exposure level for workers and members of the public for radiological protection.
The International Commission on Radiological Protection (ICRP) established Task Group 91 to review the radiation risk coefficients based on the dose and dose-rate effectiveness factor (DDREF) (Rühm et al. 2015, 2016). Since the findings of epidemiological studies, including not only radiation workers but also residents of areas with high natural background radiation (Krestinina et al. 2005; Nair et al. 2009), were taken into consideration in the meta-analytic approach (Jacob et al. 2009; Shore et al. 2017), estimation and discussion of the effect of the dose rate from different aspects will be important in the development of the system of radiological protection.
In this study, we attempted the reanalysis of cancer mortality using publicly available epidemiological data from the viewpoint of the dose rate (dose-rate window). We used mortality data for Hanford site workers, which were used in the International Agency for Research on Cancer (IARC) cohort study of nuclear workers in the United States, the United Kingdom, and Canada (Three Countries Study) by Cardis et al. (1995), since the number of subjects is reasonable (Gilbert et al. 1993a and b), and the analytic data are publicly available online. First, by following the procedure of Cardis et al. (1995), we reproduced the ERR for cancers excluding leukemia and the ERR for leukemia excluding CLL to confirm the validity of the selected study population. Then, the ERR for cancers excluding leukemia was reanalyzed by varying the cut point of the dose rate from 2 to 40 mSv y−1. To verify the results, additional examinations were also carried out for different models, lag periods, and impacts of adjusting the monitoring period.
MATERIALS AND METHODS
The analytic data file sets of the Hanford site for the Three Countries Study, registered in the US Department of Energy (DOE) Comprehensive Epidemiological Data Resource (CEDR) database (US DOE 2018), were provided for this study. The Hanford site is a complex nuclear facility including reactors for plutonium production, reprocessing, and radioactive waste treatment, and workers employed during any period from 1944 to 1986 were included in the data sets. According to Gilbert et al. (1993a), who used almost the same subjects for risk analysis, the mean dose was 26 mSv. For age at exposure, 35% of workers were under 35 y old, 41% were 35–49 y old, and 24% were 50 y old or older from 1944 to 1986. It is likely that more than 77% of the subjects received doses from 0 to 10 mSv, according to the cumulative dose distributions in the paper by Cardis et al. (1995).
Definition of cohort
As mentioned before, the subjects for analysis were selected following the procedure and criteria in the Three Countries Study of Cardis et al. (1995). Specifically, among the 36,927 monitored workers, those whose length of employment was less than 6 mo were excluded; then the first year of monitoring was determined by considering the doses received by radiation workers not only at the main facility but also at other facilities (off-site radiation exposure). The date of entry was defined as the later of the date of starting employment plus 6 mo and the date of the first monitoring. Workers were excluded from the analysis when their year of entry and the year their last vital status was known were identical. Moreover, workers who may have received a higher dose rate (more than 250 mSv y−1) were excluded. Consequently, 32,988 subjects were selected in this study. Numbers of deaths due to cancers excluding leukemia and due to leukemia excluding CLL were counted using the international classification of disease, version 8 (ICD-8), code published by the World Health Organization (WHO 1966) (cancers excluding leukemia: ICD-8 codes 141–203; leukemia excluding CLL: ICD codes 204, 205–207), in the Hanford data.
In the Hanford data from the CEDR database, information on the dates of birth, entry, and death was available only for each calendar year; therefore we assume that all such dates were 1 July. Since the end of the study was 31 December 1986 for all workers, the exit date was defined as the date of the end of study or the date of the last vital status, whichever was earlier. The number of person-years at risk for the entry year and that for workers whose follow-up ended owing to death was set to be 0.5. The number of person-years in 1986 for workers who survived to the end of the study was set to 1.0.
For age, calendar year, sex, and socioeconomic status (SES), the following stratifications defined by levels of the confounding variables were adopted with reference to the method of Cardis et al. (1995): 5 y categories from ages 20 to 84 y and 85 y and over, 5 y categories from calendar years 1950 to 1979 and 1980 and over, two categories of sex (males and females), and four categories of SES. As discussed in detail by Gilbert et al. (1993b), SES was determined to be one of the following general socioeconomic categories: (a) professional and technical, (b) clerical, (c) skilled and semiskilled manual, and (d) unskilled manual.
The AMFIT program of Epicure software was used to calculate the ERRs (Preston et al. 1993). In order to reproduce the ERRs of the Three Countries Study, a lag period of 10 y was considered for cancers excluding leukemia, and that of 2 y was considered for leukemia excluding CLL. Including off-site radiation exposure, doses were accumulated from the first monitoring to the end of each worker’s follow-up. The ERR was estimated by Poisson regression analysis, where a linear model of the form λ = λ0(1 + βd) was assumed.
By contrast, to perform an analysis based on the dose rate obtained from the annual dose recorded using a personal dosimeter, the following model was considered:
Here, dL is the cumulative dose of the dose rate that is lower than the specified cut point of the dose-rate (DRcutpoint) windows, and dH is the cumulative dose of the dose rate that is higher than the specified cut point of the dose-rate windows. We also calculated 90% Wald-based confidence intervals, which are Gaussian approximations to the binomial distribution (Dunnigan 2008), since both Cardis et al. (1995) and Gilbert et al. (1993b) estimated 90% confidential intervals in their studies.
In the Life Span Study of atomic bomb survivors in Hiroshima and Nagasaki (Ozasa et al. 2012), a linear-quadratic model was also considered in the analysis of the dose response. Inspired by this, the models given by eqns (2) through (4) were adopted for comparison with the basic linear model given by eqn (1). Here, βL1, βL2,βH1, and βH2 were the parameters for linear and quadratic terms of dL and dH, respectively. The same stratifications and lag periods were adopted as before:
Moreover, different lag periods of 5, 10, 15, and 20 y were examined to assess the effect on the estimation of βL and βH for all cancers excluding leukemia. Adjustment for the monitoring period was also performed since Gilbert et al. (1993a) used the monitoring period (<5 y, 5+ y) as a stratification in the analysis of the Hanford data. In this study we also estimated the difference in risk estimates between short and long monitoring periods. These sensitivity analyses were performed with a cut-point dose rate of 20 mSv y−1 (DRcutpoint = 20 mSv y−1).
Recalculation of the ERR in the Three Countries Study
Table 1 summarizes the basic characteristics for the subjects, such as the number of workers, the numbers of deaths due to cancers excluding leukemia and due to leukemia excluding CLL, person-years at risk, and collective dose. The recalculated ERRs for the above-mentioned end points in the present study are also shown in the table, in addition to the ERRs from Cardis et al. (1995) and Gilbert et al. (1993a and b)
As shown in the Table 1, the ERR for all cancers excluding leukemia and its 90% confidential interval (CI), which was estimated by likelihood, were −0.22 and (−0.9, 0.6), respectively, which are compared with the results of Cardis et al. (1995). Although there were slight differences in the number of workers, number of deaths, and so forth, among the studies, the resultant ERR in the present study was in agreement with the previous studies. The ERR and CI for leukemia excluding CLL obtained in this study (−0.78 and (<0, 3.1)) were also similar to those by Cardis et al. (−0.9 and (<0, 2.9)) but not to those obtained by Gilbert et al. (−1.1 and (<0, 1.9)), although their CIs overlapped. This is considered to be due basically to the difference between the cohorts. The exclusion of the first 5 y of the follow-up by Gilbert et al. (1993b) may also have affected the ERR and CI.
Thus, we determined that the selection process of the subjects for the calculations in this study was appropriate.
In the following analysis, all cancers excluding leukemia was selected as the target cause of death since the number of deaths was larger than for leukemia excluding CLL, as shown in Table 1.
ERR considering the dose rate
Fig. 1 shows the ERRs (βL and βH) in eqn (1) for all cancers excluding leukemia, where the cut point of the dose rate was changed from 2 to 40 mSv y−1 with 2 mSv y−1 intervals. Stratifications for age, calendar year, sex, and SES were also adopted, and the lag period was 10 y.
In Fig. 1, the estimate of βL varies considerably when the cut point of the dose rate was lower than 6 mSv y−1 (βL = −8 and −4 when the cut points of the dose rate were 2 and 4 mSv y−1, respectively), although they did not differ from βH in a statistically significant manner. On the other hand, when the cut point of the dose rate was from 6 to 30 mSv y−1, the values of βL and βH were similar and appeared to be stable without any noticeable tendency. The estimates of βH (DRcutpoint = 26 mSv y−1 and 32 mSv y−1) were −1.8 (−3.4, −0.1) and −2.4 (−4.2, −0.6), which were the cases of a negative ERR with statistical significance in this analysis.
When the cut point of the dose rate exceeded 30 mSv y−1, the variation in βL became more stable while that in βH and its CI became larger and wider. This is due to the number of person-years corresponding to such higher dose rates decreasing suddenly above the annual dose limit for workers, increasing the statistical uncertainty. Since the ERRs for all cancers excluding leukemia at the cut point of a dose rate of 20 mSv y−1 appear to have a narrow CI, the cut point of 20 mSv y−1 was chosen in the sensitivity analysis.
As shown in Fig. 1, the values of βL for dose rates lower than 6 mSv y−1 have large CIs, where the maximum cumulative dose of such a case is less than approximately 100 mSv (see supplemental digital content 1 and 2; http://links.lww.com/HP/A159 and http://links.lww.com/HP/A160). This may increase the uncertainty in the estimation of the ERR in units of Sv−1.
A possible reason for large negative βL values may be the effect of healthier workers or other unadjusted factors (education, smoking, job category, etc.), which were identified as biases in a recent epidemiological study on Japanese radiation workers (Kudo et al. 2017, 2018). Information on such factors for Hanford workers is unfortunately not disclosed in the database, so further adjustment is not possible. Nevertheless, it is expected that the large negative βL values and the tendency for βL to decrease with decreasing cut point of the dose rate cannot be explained only by such unadjusted factors.
From the viewpoint of radiation biology, ICRP Publication 131 states that “chronic exposures at a dose rate of a few mGy per year mean that every cell in the body is hit by a track of radiation every few months … This elimination theory lowers the linear term” (ICRP 2015). There might be inconsistency between exposure conditions of radiation workers during normal operation and those of chronic environmental exposure, which might be within the scope of the above ICRP description. However, the difference between βL and βH might reflect the difference of these effects. Hence, it should be carefully examined whether biological phenomena eventually produce the large difference in mortality as shown in Fig. 1.
On the basis of the results, βL and βH for all cancers excluding leukemia did not differ significantly at the dose rate that radiation workers are normally exposed to under normal operational circumstances.
Table 2 shows the obtained estimates of βL1, βH1, βL2, and βH2 for all cancers excluding leukemia for models (1) through (4), as well as the deviance and the results of a likelihood test. As shown for models (2), (3), and (4), although the ERRs for the quadratic terms were estimated, their CIs were much wider than those of the linear terms.
As shown in Table 2, there was no statistically significant difference between βL1 and βΗ1 for models (1) through (4). Because there was no clear improvement of the deviance, no evidence to justify including the quadratic term was obtained.
In Table 3, the values of βL and βH for all cancers excluding leukemia for lag periods of 5, 10, 15, and 20 y are shown along with those of Gilbert et al. (1993a). The results of adjustment for the monitoring period as a stratification and a comparison between short (less than 5 y) and long (5 y and greater) monitoring periods are given in Table 4. When the cut point of the dose rate was not considered, the ERR was estimated to be −0.21 (−0.98, 0.56). This estimate is almost identical to that without adjustment by stratification, −0.22 (−0.99, 0.55), and the ERR for workers with a long monitoring period. On the other hand, the ERR for workers with a short monitoring period was estimated to be −32 with a large CI of (−220, 156).
In terms of the lag period without a dose rate cut point as shown in Table 3, the ERRs increase with the lag period. This tendency can also be observed in the results of Gilbert et al. (1993b); however, they estimated the ERR for all cancer deaths, and thus the ERR values are not commensurate. When the cut point of the dose rate is considered, βΗ shows an increasing tendency while βL does not. However, all the CIs in Table 3 also increase with the lag period. Overall, there was again no statistically significant difference between βL and βH, regardless of the lag period.
As shown in Table 4, when performing separate analysis by the monitoring period, clear differences in CIs were found regardless of the considered cut point of the dose rate. The ERRs for a short monitoring period have wider CIs than those for a long monitoring period, and βH for a short monitoring period cannot be obtained. It is expected that the maximum cumulative dose to workers who were monitored for a short period may not be sufficient to estimate the risk in units of Sv−1. Overall, adjustment of the monitoring period may change the ERR to some extent, although no statistically significant difference was observed for either βL or βH.
In this study, ERRs were estimated by taking the cut-point dose rate based on the annual dose into consideration, and sensitivity analyses were also conducted for different models and lag periods, with and without the monitoring period as a stratification. Many radiation biology results have indicated different risks between a high dose rate and low dose rate. However, the dose rates in Storer’s study (Storer et al. 1979) were considerably higher (80 mGy d−1 vs. 450 mGy min−1) than the radiation dose rate during normal operations at nuclear facilities. Concerning the dose rate in occupational radiation exposure, it will be two or three orders of magnitude smaller than the dose rate in the above-mentioned study on animals. Moreover, the United Nations Scientific Committee on the Effects of Atomic Radiation (UNSCEAR) defined low dose rates of low linear energy transfer (LET) radiation as those below 0.1 mGy min−1 averaged over 1 h (UNSCEAR 2010). It should also be emphasized that the cohort in the definition excludes individuals who might have had acute exposure to an annual dose greater than 250 mSv y−1.
Another possibility is the insufficient statistical power to distinguish between lower and higher cumulative dose indices due to the small number of person-years; a pooled analysis with appropriate compensation may improve the detectability of such differences.
Similarly to the present study, the dose rate at a workplace had to be inferred on the basis of recorded individual annual doses. However, actual doses at a workplace can differ for each location and period. Thus, there is a limitation on the resolution of the dose rate itself. If further information on the dose rate with appropriate time resolution (mSv h−1, mSv d−1) is available, the results may change, but this is unrealistic owing to the huge resources required to obtain such information for numerous radiation workers.
The present study reached the opposite conclusion to that of Metz-Flamant et al. (2012), although the target causes of death were different. When the mortality or morbidity of diseases, such as cancer or leukemia, is regarded as an end point in an epidemiological study, many possible modification factors among the subjects can be listed. Although smoking, alcohol consumption, and daily habits are expected to be modification factors (in particular, smoking is strongly correlated with radiation dose according to the latest study of Japanese radiation workers [Kudo et al. 2017]), there may be a limitation to the viability of covering a large number of subjects in epidemiological studies in the same manner, even if tremendous effort is made to reduce uncertainties. Therefore, to identify high-sensitivity biomarkers and to reveal their phenomena and response to radiation at lower dose rates, it will be helpful to interpret the results of epidemiological studies to improve radiological protection.
As mentioned before, few epidemiological studies have estimated ERRs from the viewpoint of the dose rate. We have assessed the effect of the radiation dose rate on mortality due to all cancers excluding leukemia among nuclear workers by reanalysis of the Hanford data. Fluctuations in the excess relative risk were seen upon varying the cut point of the dose rate, but there was no statistically significant difference between the cumulative doses obtained from lower and higher dose rates, even when the effects of different models, lag periods, and monitoring periods were also examined. One of the reasons for this is that even though we divided annual doses into high and low dose rates by considering the cut point, both dose rates in occupational exposure during normal operation at nuclear facilities are categorized as low dose rates according to the UNSCEAR definition.
The conclusive reason for differences between βL and βH, which are seen below 6 mSv y−1, is unclear at this time. To explore the present results in detail and to investigate the effect of the dose rate on other diseases, we will continue to analyze site-specific cancer as well as noncancer, such as circulatory disease, considering the possibility of pooled analysis by combining other available data sets.
We wish to thank the US Department of Energy (US DOE) for providing publicly available data sets at the Comprehensive Epidemiologic Data Resource. We also express our gratitude to Fumiyoshi Kasagi and Masami Ikai of the Radiation Effects Association (REA) and Toshiyasu Iwasaki of the Radiation Safety Research Center of the Central Research Institute of Electric Power Industry (CRIEPI) for their valuable comments.
This work was partly funded by the Nuclear Regulation Authority, Japan. The views of the authors do not necessarily reflect those of US DOE, REA, CRIEPI, and NRA.
Boice JD Jr. NCRP Vision for the future and program area committee activities. Health Phys 112:225–229; 2017. DOI 10.1097/HP.0000000000000638.
Cardis E, Gilbert ES, Carpenter L, Howe G, Kato I, Armstrong BK, Beral V, Cowper G, Douglas A, Fix J, Fry SA, Kaldor JM, Lave C, Salmon L, Smith PG, Voelz G, Wiggs L. Effects of low doses and low dose rates of external ionizing radiation: cancer
mortality among nuclear industry workers in three countries. Radiat Res 142:117–132; 1995.
Gilbert ES, Cragle DL, Wiggs LD. Updated analyses of combined mortality data for workers at the Hanford site, Oak Ridge National Laboratory, and Rocky Flats Weapons Plant. Radiat Res 136:408–421; 1993a.
Gilbert ES, Omohundro E, Buchanan JA, Holter NA. Mortality of workers at the Hanford site: 1945–1986. Health Phys 64:577–590; 1993b.
Gillies M, Richardson DB, Cardis E, Daniels RD, O’Hagan JA, Haylock R, Laurier D, Leuraud K, Moissonnier M, Schubauer-Berigan MK, Thierry-Chef I, Kesminiene A. Mortality from circulatory diseases and other non-cancer
outcomes among nuclear workers in France, the United Kingdom and the United States (INWORKS). Radiat Res 188:276–290; 2017. DOI 10.1667/RR14608.1.
International Commission on Radiological Protection. Stem cell biology with respect to carcinogenesis aspects of radiological protection. Oxford: ICRP; Publication 131; 2015.
Jacob P, Rühm W, Walsh L, Blettner M, Hammer G, Zeeb H. Is cancer
risk of radiation workers larger than expected? Occup Environ Med 66:789–796; 2009.
Krestinina LY, Preston DL, Ostroumova EV, Degteva MO, Ron E, Vyushkova OV, Startsev NV, Kossenko MM, Akleyev AV. Protracted radiation exposure and cancer
mortality in the Techa River Cohort. Radiat Res 164:602–611; 2005.
Kudo S, Ishida J, Yoshimoto K, Mizuno S, Ohshima S, Furuta H, Kasagi F. Direct adjustment for confounding by smoking reduces radiation-related cancer risk estimates
of mortality among male nuclear workers in Japan, 1999–2010. J Radiol Protect 38:357; 2018.
Kudo S, Ishida J, Yoshimoto K, Ohshima S, Furuta H, Kasagi F. The adjustment effects of confounding factors on radiation risk estimates
: findings from a Japanese epidemiological study on low-dose radiation effects (J-EPISODE). J Mol Genet Med 11:275; 2017. DOI 10.4172/1747-0862.1000275.
Leuraud K, Richardson DB, Cardis E, Daniels RD, Gillies M, O’Hagan JA, Hamra GB, Haylock R, Laurier D, Moissonnier M, Schubauer-Berigan MK, Thierry-Chef I, Kesminiene A. Ionising radiation and risk of death from leukaemia and lymphoma in radiation-monitored workers (INWORKS): an international cohort study. Lancet Haematol 2:e276–281; 2015.
Metz-Flamant C, Samson E, Caër-Lorho S, Acker A, Laurier D. Leukemia risk associated with chronic external exposure to ionizing radiation in a French cohort of nuclear workers. Radiat Res 178:489–498; 2012.
Nair RR, Rajan B, Akiba S, Jayalekshmi P, Nair MK, Gangadharan P, Koga T, Morishima H, Nakamura S, Sugahara T. Background radiation and cancer
incidence in Kerala, India-Karunagappally cohort study. Health Phys 96:55–66; 2009.
National Research Council. Health risks from exposure to low levels of ionizing radiation: BEIR VII—phase 2. Biological Effects of Ionizing Radiation (BEIR) Committee VII. Washington, DC: National Academy Press; 2006.
Ozasa K, Shimizu Y, Suyama A, Kasagi F, Soda M, Grant EJ, Sakata R, Sugiyama H, Kodama K. Studies of the mortality of atomic bomb survivors, Report 14, 1950–2003: an overview of cancer
and noncancer diseases. Radiat Res 177:229–243; 2012.
Preston DL, Lubin JH, Pierce DA, McConney ME. Epicure users guide. Seattle: Hirosoft International Corporation; 1993.
Richardson DB, Cardis E, Daniels RD, Gillies M, Haylock R, Leuraud K, Laurier D, Moissonnier M, Schubauer-Berigan MK, Thierry-Chef I, Kesminiene A. Site-specific solid cancer
mortality after eposure to ionizing radiation: a cohort study of workers (INWORKS). Epidemiol 29:31–40; 2018. DOI 10.1097/EDE.0000000000000761.
Richardson DB, Cardis E, Daniels RD, Gillies M, O’Hagan JA, Hamra GB, Haylock R, Laurier D, Leuraud K, Moissonnier M, Schubauer-Berigan MK, Thierry-Chef I, Kesminiene A. Risk of cancer
from occupational exposure to ionising radiation: retrospective cohort study of workers in France, the United Kingdom, and the United States (INWORKS). BMJ 351:h5359; 2015.
Rühm W, Azizova TV, Bouffler SD, Little MP, Shore RE, Walsh L, Woloschak GE. Dose-rate effects in radiation biology and radiation protection. In: Proceedings of the third international symposium on the system of radiological protection. Ann. ICRP 45(1S); 2016: 262–279.
Rühm W, Woloschak GE, Shore RE, Azizova TV, Grosche B, Niwa O, Ono T, Suzuki K, Iwasaki T, Ban N, Kai M, Clement CH, Bouffler S, Toma H, Hamada N. Dose and dose-rate effects of ionizing radiation: a discussion in the light of radiological protection. Radiat Environ Biophys 54:379–401; 2015.
Shore RE, Walsh L, Azizova TV, Rühm W. Risk of solid cancer
in low dose-rate radiation epidemiological studies and the dose-rate effectiveness factor. Int J Radiat Biol 93:1064–1076; 2017. DOI 10.1080/09553002.2017.1319090.
Storer JB, Serrano LJ, Darden EB Jr, Jernigan MC, Ullrich RL. Life shortening in RFM and BALB/c mice as a function of radiation quality, dose, and dose rate. Radiat Res 78:122–161; 1979.
US Department of Energy. The comprehensive epidemiologic data resource [online]. 2018. Available at https://apps.orau.gov/cedr/
. Accessed 15 August 2017.
World Health Organization. International classification of diseases, revision 8. Geneva: WHO; ICD-8; 1966.
3Furuta H, Kudo S, Ishida J, Yoshimoto K, Kasagi F. Dose-rate effects on cancer mortality risk estimates for Japanese nuclear workers. In: Proceedings of the 2nd European Radiological Protection Research Week, 10–12 October 2017, Paris, France.
cancer; epidemiology; exposure, occupational; risk estimates
Supplemental Digital Content
© 2019 by the Health Physics Society