A.1 RADIATION AS A CAUSE OF CANCER
At low doses of radiation, cells may be damaged. The main initiating event by which radiation damages the cells in the long term is damage to DNA in the cell nucleus. With well-orchestrated and efficient mechanisms, cells respond to the induced damage and attempt to repair it, but sometimes the damage cannot be repaired or is misrepaired, which may lead to mutations. The modifications induced by low levels of radiation dose may be transmitted to daughter cells and may lead to uncontrolled cell growth and consequently cancer, the health effect of primary concern in the context of radiation. Exposure to radiation is not the only way in which the DNA within a cell can be damaged and become cancerous. In fact, DNA damage can occur spontaneously or due to a number of other stressors such as chemical exposure (for example, smoking and lung cancer) and infectious agents (for example, hepatitis B virus and liver cancer). In other words, as ionizing radiation exposure induces DNA damage to the tissue, that tissue will already carry some damaged cells from other stressors.
Although small increases in the chance of developing cancer is the main health effect of low levels of radiation, such effects in individuals are probabilistic and known as stochastic effects. In other words, there appears to be no threshold below which effects do not occur, but the greater the exposure, the higher the probability that they will occur. Severity of the effects does not depend on dose. This is in contrast to the “deterministic” or “nonstochastic” radiation effects of high doses of radiation, that is, doses of several sieverts that can kill enough cells to cause injury such as skin reddening,
burns, organ damage, radiation sickness, and even death. Patients receiving radiation treatment for cancer often experience controlled acute radiation sickness because they receive relatively high levels of radiation. Infertility and cataract are two other examples of nonstochastic effects of radiation; cataract may not occur until several years after exposure. Doses to people near nuclear facilities are far below levels that would cause deterministic effects.
In the case of the effects of exposure to low levels of radiation (less than 0.1 Gy, or 100 mSv effective dose), the scientific uncertainty of radiation-induced cancer is considerable as there is little or no empirical knowledge. Despite the uncertainty, decisions are needed for use in setting standards for protection of individuals against the side effects of low-level radiation. Based on current scientific knowledge (or lack thereof), regulatory agencies in the United States currently use a model that describes radiation injury as a linear function of radiation dose that has no threshold; this is called the linear no-threshold (LNT) model. According to LNT, if a dose equal to 1 Gy gives a cancer risk X, the risk from a dose of 0.01 Gy is X/100, the risk from 0.00001 Gy is X/100,000, and so on. Thus, the risk of health effects including cancer risk is not zero regardless of how small the dose is.
In the LNT model, data from high levels of exposure where radiogenic cancers have been observed are used to extrapolate risks at lower doses where cancers have not been observed, and if they exist they are beyond the current science to observe and measure. One result of following the LNT model is that a very small estimated risk, when multiplied by a large number such as the population of the United States, results in an estimate of a substantial number of cases or deaths, which in reality may not exist.
Scientific groups such as the International Commission on Radiological Protection (ICRP), the National Council on Radiation Protection and Measurements (NCRP), and the National Research Council Committee on the Biological Effects of Ionizing Radiation (BEIR), repeatedly review and endorse the use of this model for assessing risk, which is used to set radiation protection standards and operating policies, such as the “as low as reasonably achievable” (ALARA) policy. This approach is often considered to be conservative and gives emphasis to public health. Data provided by the updated report of the atomic bombing survivors in Japan continue to be in support of the LNT model across the entire dose range. However, a concave curve was the best fit for data restricted to doses of 0-2 Gy. This resulted because risk estimates for exposure to 0.3-0.7 Gy were lower than those in the linear model (Ozasa et al., 2012). The finding was not explained.
Not all countries support the LNT model at this time, but in general it is perceived that with so much uncertainty about the effects at low doses, it is appropriate to continue with the LNT model that has been in place for several decades for purposes of radiation protection.
A.2 BIOLOGICAL RESPONSES AT LOW DOSES
A variety of different biological responses have been identified at low doses of radiation, although it is difficult to identify effects at doses that are close to those encountered from natural background radiation. It is highly unlikely that epidemiologic studies of populations around nuclear facilities will contribute toward knowledge of the effects of radiation at very low doses. Because of the epidemiologic limitations, efforts are directed toward improving understanding of the effects, response, and defense mechanisms to low-dose radiation at the cellular and molecular levels. The Department of Energy’s Low Dose Radiation Program is focused on understanding the effects of doses of radiation under 100 mSv by supporting research of the molecular and cellular responses to very-low-dose exposures. Some scientists have argued that DNA repair capabilities are effective at low doses, preventing the accumulation of DNA damage and mutations following low-dose exposures, while others have argued that low doses may be even more damaging per unit dose than high doses.
Major discussion on the biological consequences of low-dose radiation despite being controversial has also led to the identification of pathways of radiation damage that are evident at low doses but difficult to measure at high doses in light of overwhelming DNA damage. Among these is the adaptive response, which would tend to dampen the potential adverse effects and perhaps even provide a beneficial (or hormetic) effect of radiation exposure at low doses. In most studies of adaptive responses, cells in vitro are given a “tickle” low dose of radiation (for example 20 cGy or 0.2 Gy) followed by a high dose of radiation (1 Gy). The administration of the “tickle” dose prevents some of the damaging effects of the high dose, including cell killing and chromosomal injury. In animal models a variety of investigators have documented that low doses of radiation can enhance immune responses (Cheng et al., 2010).
There are also several damaging responses observed at low doses, including the bystander effect and delayed genomic instability. The bystander effect is defined as genetic changes (chromosome damage, mutations) induced in cells that are not directly hit by the radiation beam. The exact mechanism by which the bystander effect occurs is unclear, although data support both transmission of a factor either in conditioned medium (Sowa Resat and Morgan, 2004) or through gap junctions (Gaillard et al., 2009). Recent studies have documented that such bystander effects may occur in vivo as well (Singh et al., 2011). Delayed genomic instability has also been identified in irradiated cell populations where mutations do not occur in the irradiated cells themselves but rather in the progeny of these irradiated cells sometimes up to 13 generations later (Little et al., 1997; Morgan, 2003).
Another detrimental effect of low-dose exposures (mostly in the cGy range) is low-dose hypersensitivity in which some cells in culture show an
enhanced response to the killing effects of x-rays at the very low doses (10-60 cGy) than they do to slightly higher doses (1 Gy, for example). Whether this is really a low-dose hypersensitivity or an induced radiation resistance at the slightly higher doses (1 Gy) is not clear, and the mechanism for it has not been defined, although some attribute it to the need for a threshold number of double-strand breaks to induce cell-cycle arrest (Marples et al., 2004).
Dose-rate factors are also important in considering the effects of low-dose radiation. Most studies have documented that low-dose-rate exposure is less damaging than similar doses administered at high rates, although these studies are limited, difficult to conduct, and predominantly in animal populations (Brooks, 2011; Vares et al., 2011). In long-term animal studies carried out at Argonne National Laboratory in 1960-1990, dogs and mice were exposed to doses of radiation daily with very low doses per day and equal doses given in a single exposure; these studies revealed that life shortening and cancer incidence was significantly higher for animals given the high-dose-rate compared to the low-dose-rate exposures (Carnes and Fritz, 1991; Carnes et al., 1998). In other mouse strains (AKR), a lower incidence of cancer-induced thymic lymphoma was also found in mice exposed to low-dose-rate compared to high-dose-rate radiation (Shin et al., 2011), suggesting that there are significant differences in biological consequences (Uehara et al., 2010).
Radiobiological data, some based on animal experiments, have been the basis of the dose and dose-rate effectiveness factors (DDREFs), that is, factors used to convert risk estimates from populations exposed in larger acute doses such as the atomic bombing survivors to populations who are exposed to lower low-rate doses. The ICRP derived estimates of the excess cancer risk after low-dose exposures and after exposures with higher doses but low-dose rates by reducing the corresponding risk value for the atomic bombing survivors by a DDREF of 2.0 (ICRP, 2007). The BEIR VII Committee used a DDREF of 1.5 (National Research Council, 2005). It has been speculated that these DDREFs underestimate the risks from low-dose-rate exposures. For example, in a recent paper by Jacob et al. (2009), comparisons of risks of radiation workers who receive chronic exposures with those of the atomic bombing survivors who received acute exposures indicated that risks among workers tended to be higher, contrary to expectations.
Most individuals exposed to radiation do not wear physical dosimeters such as film badges or thermoluminescent dosimeters; therefore, reconstructing their exposure requires collecting information through interviews and available models and thus estimated exposures often contain a high
level of uncertainty. In an attempt to overcome this problem, biological markers are being developed as a useful tool for estimating the exposure and the effects of, or the response to, radiation. A biomarker is in general an end point that is objectively measured and can be used as an indicator of a biological state. Studies have highlighted the importance of biomarker research in radiation epidemiology specifically in assessing occupational exposure (Schneider et al., 1999), exposure following industrial accidents (Menz et al., 1997), as well as response to radiation therapy (Wickremesekera et al., 2001). Two types of purpose-oriented categorizations of irradiation biomarkers have been proposed. Brooks segregates them into markers of exposure, sensitivity, and disease (Brooks, 1999), while others mention predictive, prognostic, diagnostic, and dosimetric markers (Okunieff et al., 2008). A single biomarker can often fit into several of these categories which serve different purposes. For example, biomarkers of effect measure the biological responses in individuals who have been exposed to an agent (and also include elements of individual sensitivity to that agent); markers of exposure, on the other hand, do not necessarily indicate effects. A methodology-focused categorization of radiation biomarkers would separate them into cytological and molecular markers, both with numerous subcategories. In addition, while cytological markers in radiation research are often very specific, molecular-based radiation biomarkers are often compendia of molecules rather than isolated molecular species. Today, the use of biomarkers in epidemiologic studies of low doses is unlikely to help with dose reconstruction, as the variability of the assays within a person and between persons is a major problem. However, the rapid advances in the research on biomarkers may in the future provide more sensitive tools that may also prove useful for epidemiologic purposes and significantly reduce the uncertainties related with current dose reconstruction models.
A.4 EPIDEMIOLOGIC STUDIES OF IONIZING RADIATION
A.4.1 Studies of Residents near Nuclear Facilities
A British television program in 1983 reported a cluster of childhood leukemia in Seascale, a village 3 km from the nuclear fuel reprocessing facility Sellafield on the Cumbrian coast, then known as Windscale. The television team discovered seven childhood leukemia cases over the previous 30 years, while less than one case was expected (Urquhart et al., 1984). Given the proximity of the village to the nuclear reprocessing plant, and in the absence of any other obvious causative agent, a direct effect of environmental pollution with radioactive waste was hypothesized. The British government appointed an independent advisory group to investigate the claims. The group produced its report within seven months (Black, 1984), confirming
the TV broadcast, but could not explain the finding in terms of radioactive discharges. In response, a governmental Committee on Medical Aspects of Radiation in the Environment (COMARE) was set up in 1985 and over the past 25 years has published several reports using data from the national registry of children’s tumors. The reports include an extensive investigation of the Sellafield area (COMARE, 1996) and the sites of Dounreay in Scotland (COMARE, 1988), Aldermaston in Berkshire, and Burghfield in North Hampshire (COMARE, 1989). Reviews by COMARE of the discharges from the nuclear installations showed that the doses that the general public residing in the area were likely to have received were far too small to have caused increases in childhood leukemia (COMARE, 1988, 1989, 1996). In 2011, COMARE published an update on the issue as its fourteenth report (COMARE, 2011), undertaking a further review of the issues addressed in the tenth report that covered the years 1969-1993 (COMARE, 2005). The latest report covered the period 1969-2004 and found no significant evidence of an association between risk of childhood leukemia and living in proximity to a nuclear power plant (COMARE, 2011).
The sequence of cluster or ecologic studies finding excess cancers around a nuclear site and more detailed examination following to confirm the findings and research the associations has been a common approach for many years. Studies from Great Britain, Germany, France, and the United States contribute the most to the literature. Childhood leukemia is primarily investigated as it is recognized to be a “sentinel indicator” for radiation effects occurring with a shorter time latency following exposure and with a stronger dose-risk relationship. Although initially mortality data were used to evaluate the possible impact of living near nuclear facilities under normal operating conditions, it was soon realized that, given the advances in cancer treatment and consequent improvements in survival, incidence data (the number of newly diagnosed cases in a given period of time) could provide more relevant estimates.
Studies on the cancer risks associated with living near nuclear facilities have come to different conclusions, with some suggesting a positive association between living in proximity to a nuclear facility and cancer risk and others suggesting that there is not a risk, or that the risk is too small to be detected with the methodology used. The power of a study to detect an effect, if there is one, depends highly on the hypothesized strength of the association to be detected and the sample size. Neither of these variables is likely to be high in an epidemiologic study of cancer risks in populations near nuclear facilities:
a. The size of the estimated risks from reported radioactive effluent releases from nuclear facilities is likely to be small. Consequently,
epidemiologic studies have a limited ability to discern associations between radiation exposure and cancer risk in these populations.
b. The size of the populations most likely to be exposed (that is, those living in very close proximity to a nuclear facility, for example within a 5-10-km radius) is relatively small. This limits the expected number of informative (exposed) incident cases or deaths that will be available for study, especially for rare cancers such as those of childhood.
Study conclusions are based on a very small local population size, which makes the risk estimations statistically unstable because a single additional case, or one less case, can change the rate estimate dramatically. For example, in the study in Germany with 23 years of follow-up, out of the 593 leukemia cases in children under 5 years old diagnosed in the study area, only 37 cases (6 percent) were observed in the risk zone ( 5 km from a facility) (Kaatsch et al., 2008). Similarly, in the recent COMARE report (2011) with 35 years of follow-up, out of the 430 leukemia cases in children under 5 diagnosed in an area up to 25 km from the nuclear power plants in Britain, only 20 (5 percent) were in the risk zone (Table A.1). It is expected that a study in the United States would contain a larger number of exposed individuals than those in the European studies because the number of nuclear power plants in the United States is larger than that in any of the European countries.
For this and other reasons related to differences in study design or analysis stages (results may be influenced, for example, by unrecognized bias in the data, the effect of other relevant factors, or by chance variation; these need to be discussed by the investigators even if they cannot be quantified), interpretation of epidemiologic findings is not always easy and there are often subjective elements to their interpretation that experts may disagree upon. Evaluating well-designed studies that do not suggest the existence of an association between a factor and a disease is equally important to evaluating
TABLE A.1 Number of Cases in the At-Risk Zone ( 5 km from a facility) in European Studies of Pediatric Cancers (children 5 years old)
|Country||Reference||Study Years||End Point||Cases ( 5 km)|
|Germany||Spix et al., 2008
Kaatsch et al., 2008
|France||Sermage-Faure et al., 2012||17||leukemia||24|
|Switzerland||Spycher et al., 2011||24||all cancers
studies that show an association. However, it is often harder to convince stakeholders of the validity of the so-called “negative” studies especially if there are flaws or inefficiencies in their design, methods, or analysis. A better term for flawed studies would be “uninformative.”
In absence of biological plausibility, a positive or somewhat positive association may be underinterpreted. In studies that assess cancer risks associated with releases from nuclear facilities, there are examples where investigators are hesitant to conclude that evidence supported the hypothesis when they find a positive association between risk and exposure associated with nuclear facilities (Baker and Hoel, 2007; Hatch et al., 1990; Kaatsch et al., 2008; Nuclear Safety Council and the Carlos III Institute of Health, 2009), even though direct radiation measurements were not made. This phenomenon has led a researcher to emphasize the importance of having explicit study hypotheses (Wing et al., 2011) and to the question, “Why conduct a study if the results cannot be interpreted as providing evidence in support of the hypothesis?” (Wing, 2010). Of course, there is the opposite error, too—that of overinterpretation. A balanced “weight-of-evidence” approach is the most appropriate.
It is important to be open to new information or novel interpretation and alternative hypotheses that can impact assumptions about exposure effects. A recent study from France demonstrated that children living in very close proximity to nuclear power plants are twice as likely to develop leukemia compared to those living farther away from the plants. However, analysis of the same population of children using a dose-based geographic zoning approach instead of distance, did not support the findings. The absence of any association with the dose-based geographic zoning approach may indicate that the observed association with distance may be due to factors other than the releases from the nuclear power plants (Sermage-Faure et al., 2012). Among such potential factors are population mixing (Kinlen, 2011a), a hypothesis that could not be evaluated in this study, and exposures to agents including natural or manmade exposures to radiation not modeled in the study.
From the reports published the past 4 years alone from Germany (Kaatsch et al., 2008), Finland (Heinavaara et al., 2010), Great Britain (COMARE, 2011), Switzerland (Spycher et al., 2011), and France (Sermage-Faure et al., 2012), it is obvious that additional scientific resolution to the question of whether living near a nuclear facility increases one’s risk of developing cancer remains. Authors have called for collaborative analysis of multisite studies conducted in various countries (Sermage-Faure et al., 2012). Similarly, the need for a well-conducted meta-analysis that would provide a more precise estimate of the risk remains.
Two meta-analyses were conducted recently in an effort to provide more precise estimates of the possible risks associated with living near a
nuclear facility (Baker and Hoel, 2007; Greiser, 2009). Baker and Hoel combined and statistically analyzed studies of childhood leukemia around nuclear facilities published until 1999, but only included studies that calculated standardized incidence ratios (SIRs) or standardized mortality ratios (SMRs) (see Sidebar A.1 for risk measures) for individual facilities. Studies that calculated rates for multiple sites or those that did not distinguish leukemia from lymphoma were excluded. Seventeen published studies (out of 37 individual studies published at the time) addressing 136 nuclear sites in 7 countries (Great Britain, Canada, France, United States, Germany, Japan, and Spain) met the criteria. Due to variability between study designs, eight separate analyses were performed stratified by age and zone. Meta-SMRs and meta-SIRs were all greater than the reference group, implying an increase in risk. More specifically, the overall estimated relative risk was 1.22 (95% CI=1.05-1.41) and the 0-9 age group accounted for the majority of the excess cases and deaths. Excluding the Aldermaston nuclear weapons plant and Amersham plant that produces radioisotopes (both in Britain) reduced the overall estimate to a nonsignificant 14 percent increase in risk (RR=1.14, 95% CI=0.98-1.33). The authors discuss that although the meta-analysis showed an increase in childhood leukemia near nuclear facilities, it “does not support a hypothesis to explain the excess” (Baker and Hoel, 2007).
The meta-analysis by Baker and Hoel was criticized by authors of the German Kinderkrebs in der Umgebung von Kernkraftwerken (KiKK) study (Spix and Blettner 2009). The first issue they identified with the meta-analysis was the general problem of combining heterogeneous data such as different age groups (0-9 years or 0-25 years), the different types of nuclear facilities (nuclear power plants and other facilities), and the different exposure zone definitions (10 km or county). Beyond that, there was criticism over the completeness of the publication search and lack of justification for excluding the 20 studies which were identified but did not fit the criteria for inclusion; possible selection bias resulting from the exclusion of sites with zero observed leukemia cases or deaths from leukemia; and a methodological problem with the confidence intervals presented in the forest plots which should be symmetric on a logarithmic scale but, contrary to expectation, were skewed (Spix and Blettner, 2009).
The meta-analysis by Greiser included data from 80 nuclear power plants in five countries (Germany, France, Great Britain, United States, and Canada). Data were retrieved in the literature but also from cancer registries. (Rather than relying on the data used in the Jablon et al. 1991 analysis of risks in nuclear facilities in the United States, the author retrieved cancer incidence data from cancer registries of Illinois, Pennsylvania, and Florida.) The incidence of leukemia was estimated to increase by 13 percent (95% CI = 10%-17%) relative to the corresponding average national or regional rate (Greiser, 2009). The latest COMARE report (2011) discusses the key
Risk Measures, P Values, and Confidence Intervals
Several types of estimates of relative risk (RR) are used in epidemiologic studies. RR is generically defined as the ratio of the risk of developing the disease or of dying of the disease among an exposed population compared to an unexposed population. A simple type of estimate of the RR is the standardized incidence ratio (SIR) or standardized mortality ratio (SMR) for the exposed group. An SIR is the ratio of the number of cases observed in the exposed group in some time period to the number of cases expected if the group had the same disease occurrence rates as a standard population. The standard population is often the general population or a large reference population with characteristics similar to the study group except for the exposure of interest, and comparisons typically are based on cancer rates from population cancer registries. The ratio of observed to expected cases is often multiplied by 100 to yield results without decimals. Thus, an SIR of 100 indicates that the observed number of cases is the same as that expected in the standard population. Thus, an SIR of 140 indicates that incidence is 40 percent higher than expected, while an SIR of 80 indicates 20 percent fewer cases than expected.
SIRs should be interpreted with caution as their significance partially depends on the number of cancer cases in the exposed group. Imagine a situation where 5 cases were expected and 6 were observed and a second situation where 500 cases were expected and 600 were observed. In both instances the SIR is 120; however, because in the second scenario the SIR is based on a greater number of cases, the estimate is more precise, and hence more meaningful. In other words, although the one excess case could have occurred due to chance alone, it is highly unlikely that an excess of 100 incident cases has occurred by chance. This is a common issue in interpreting studies of risks in populations near nuclear facilities where the number of excess cancers in the exposed region is particularly small when rare diseases such as childhood leukemia are examined (see Table A.1).
The SMR is similar to the SIR, except it is based on deaths due to some cause rather than cancer occurrences to draw conclusions regarding whether there is excess mortality. As age is one of the main determinants of mortality, and other factors such as gender and racial composition may influence the mortality or tumor rates, SMRs and SIRs are usually calculated by summing the observed and expected numbers of deaths or cancers across categories of gender, age, and sometimes race with the expected numbers calculated separately for each category.
Results from cohort and ecologic studies are sometimes described in terms of SMRs or SIRs, but other techniques are often preferred which permit comparisons of disease rates (often called rate ratios) between exposed and unexposed study groups, usually with adjustment for gender, age, and perhaps other factors. More advanced techniques use some type of “regression analysis” to estimate exposure-effect associations, with study subgroups or individuals defined according to graded amounts of exposure.
Case-control studies (which compare exposures observed in cases to those observed in control subjects) are typically unable to calculate actual disease rates since they lack appropriate population denominators, which means that SIRs, SMRs, and rate ratios cannot be used. However, for case-control studies the odds ratio (OR) can be calculated. The OR and relative risk are closely related (and are nearly identical for “rare” diseases). The OR indicates the ratio of the probability of exposure to the probability
of nonexposure among those with the disease of interest divided by the similar ratio of probabilities among those without the disease. A value greater than 1 means that the odds of disease are greater among the exposed than the unexposed. A value less than 1 means that the odds are higher in the unexposed than in the exposed. Similar to all the other statistics mentioned, the number of disease cases with exposure has a major influence on the precision and statistical significance of the OR.
A useful measure of risk in epidemiologic studies is that of “excess” risk associated with an exposure. Excess risk can be expressed as excess relative risk (ERR) or excess absolute risk (EAR). The ERR and EAR in principle are estimates of the amount of risk due to the exposure of interest when the effects of other risk factors are removed. Statistically, ERR = RR-1 and EAR = RE - RU, where RE is the rate of occurrence of disease or death in the exposed group in a specified period, and RU is the corresponding rate of occurrence in the unexposed group, which is the baseline rate. In contrast to ERR, which represents the ratio of the excess rate associated with exposure to the baseline rate, the EAR represents the additional rate of a disease due to the exposure of interest over a given period of time. As baseline disease rates depend on a number of factors, excess risks can vary not only with radiation dose but also with age at exposure, time after exposure, age at risk (attained age), gender, and other factors such as smoking. Therefore, risk estimates are usually reported for a specified combination of these factors. ERR and EAR estimates can best be calculated in a cohort study although ecologic studies can sometimes permit such estimates to be made. A statistic analogous to the ERR estimate can be calculated as OR-1 for case-control studies, but usually EAR estimates cannot be obtained from a case-control study due to the lack of population denominators.
By describing the excess number of people affected by the disease of interest, EAR is a better descriptor than the ERR of the public health impact that an exposure may have in a population. For example, in the Life Span Study (LSS) follow-up of the Japanese atomic bombing survivors the ERR for leukemia is the highest among the various cancer effects of radiation exposure (RR approximately 5 for a dose of 1 Gy, which translates into an ERR of about 4), and the total number of radiation-related cases of leukemia among the LSS survivors is estimated to be about 90-100. In contrast, the ERR for solid cancers is much smaller (RR approximately 1.5 at 1 Gy, or an ERR of about 0.5), yet the total number of LSS survivors who have developed solid cancers due to the bombing is estimated to be about 850. This is because of the relative rarity of leukemias compared to the group of cancers described as solid cancers. Common cancers may appear to have a low ERR in an epidemiologic study, but the risk may translate to a large number of cases, or a large EAR. One can say that the ERR is an appropriate measure to assess disease etiology, whereas the EAR is useful for estimating the extent of a health problem.
Applying ERR or EAR estimates derived from individuals in one population to those in another population sometimes has substantial uncertainties. Since most types of cancer vary substantially in their baseline frequency according to age, both ERR and EAR estimates can be affected by differences in the age distributions of populations being compared. For instance, it would be inappropriate to compare radiation-related leukemia risk of children in one population with adult leukemia risk in another population. Sometimes there also are differences in the baseline rates of disease in different populations even with the same age distributions. For example, the Japanese have historically had much higher rates of stomach and liver cancers than in the United States. It is therefore uncertain as to how to extrapolate stomach or liver cancer ERR
or EAR risk estimates from the Japanese atomic bomb survivors to the U.S. population. Careful analysis and interpretation is required in making projections of risk across populations.
By itself a point estimate whether it is an SMR, SIR, OR, or RR is difficult to interpret because it does not indicate the extent to which chance may have played a role. This additional information regarding the reliability of an estimate is provided by calculating the confidence interval. A confidence interval with a particular confidence level, commonly set up at 95 percent, is intended to give the assurance that, if the statistical model is correct, the true value of the parameter is within the range indicated. If the 95 percent CI range does not include 1, then the estimated risk is significantly different from that of a comparison group. For example, if the risk ratio of a smoker being diagnosed with lung cancer is estimated to be 10 when compared to the risk of a nonsmoker and the 95 percent confidence interval (CI) is 8.6-12.7, then the investigator can conclude that the risk ratio is significantly higher than 1 as there is less than 5 percent chance that the observed difference is the result of random fluctuation. The width of the CI is also very important as it indicates the precision with which the risk is estimated. Narrow estimation indicates a fair level of certainty that the calculated estimate falls within a narrow range. A wide interval makes the estimation “imprecise” and leaves considerable doubt as to the accuracy of the estimate. However, confidence intervals do not account for the uncertainty resulting from bias in exposure estimates,
problems with the analysis, which are both methodological and also relate to lack of justification for excluding studies from the meta-analysis (for example, data from Japan).
The limitations of the two meta-analyses discussed here defeat their purpose, which is to estimate the effect size with higher precision than the single studies which are often underpowered. In addition, the selection of data to be included or excluded from the meta-analysis can influence the results. Although meta-analyses often suffer from the general problem of summarizing heterogeneous data and the possibility of “publication bias”1 (studies that find a positive association are more likely to be published compared to studies that find no association), they ought to be based on a thorough literature search that identifies relevant studies and to clearly state the criteria and justify excluding studies from the analysis.
A review of the literature that includes all cancer types and all ages is presented here. Table A.2 summarizes information from selected multisite leukemia studies of children that investigated place of residence at time of
1 Negative studies often do not interest the publishers, who may be biased in favor of positive or promising results (Simes, 1986), or the researchers themselves fail to write them up and submit them for publication (Angell, 1989). The results from the meta-analysis would then be skewed toward a positive association.
or from confounders that investigators were not able to fully adjust for, or confounders that were unidentified.
The P value is a statement of the probability that the association observed could have occurred by chance under the assumption that the null hypothesis is true. Traditionally, a P value 0.05 is considered as sufficiently unlikely for the association to have occurred by chance and justifies the designation “statistically significant.” The smaller the P value, the less likely the observed association could have occurred by chance under the null hypothesis. P values can be either two-tailed (also called two-sided) or one-tailed (or else one-sided) depending on the alternative hypothesis tested. The one-tailed test provides more power to detect an effect in the direction tested and should be used only after considering the consequences of missing an effect in the untested direction. The KiKK study, for example, used a one-tailed test and limited attention to identification of increases associated with living near a nuclear facility (Kaatsch et al., 2008).
Inferences about an association between a disease and an exposure are considerably strengthened if information is available to support a dose response in the relationship between the degree of exposure and the disease. In that case, risks are estimated for every category of exposure and a P for trend is estimated (that is, the alternative hypothesis reflects a trend of effect across exposure values rather than an increase or decrease for particular ranges of exposure).
diagnosis or death, or place of birth in relation to nuclear facilities as a risk factor for the disease.
A.4.1.1 Great Britain
In Great Britain the first multisite study came immediately as a response to the reported cluster in Sellafield. In 1984, Baron examined cancer mortality trends for the small areas around 14 nuclear installations in England and Wales using census and survey data for the years 1974-1979 (Baron, 1984). In the short period of observation, the data did not indicate any increase in mortality in areas around the major nuclear facilities examined. A year later, a preliminary report on the incidence of leukemia for the years 1972-1984 in children with age equal to or less than 9 years living near two nuclear establishments, the Atomic Weapons Research Establishment at Aldermaston and the Royal Ordnance Factory at Burghfield in the West Berkshire District Health Authority, showed that the incidence among those 0-4 years of age increased 60 percent (Barton et al., 1985). The study did not include children residents of the West Berkshire District Health Authority who were referred elsewhere for diagnosis and treatment. An updated and extended study that included incident cases diagnosed in 1985, those aged 10-14 years and residents in the above-mentioned district and neighboring
TABLE A.2 Selected Multisite Studies of Leukemia among Young People Living near Nuclear Facilities
|A. Ecologic studies|
|Country||Reference||No. of Sites||Study Period||Age||I/M||Exposed Areas||Comparison Areas||No. Cases||SIR or SMR|
|Britain||Forman et al., 1987||14||1959-1980||0-24||M||10 km||Control local authority||44||2|
|Britain||Cook-Mozaffari et al., 1989a||15 (+8 possible)||1969-1978||0-24||M||16 km||Other districts||635||1.15|
|United States||Jablon et al., 1991||62||1950-1984||0-9||M + I||107 counties||292 counties||1,390||1.03|
|Canada||McLaughlin et al., 1993a||5||1950-1987||0-14||M + I||25 km||Province rates||54||1.17|
|Germany||Michaelis et al., 1992||20 (+6 possible)||1980-1990||0-14||I||15 km||30-100 km||274 13||1.06
7.09 (children <5 yr)
|Britain||Bithell et al., 1994||23 (+6 possible)||1966-1987||0-14||I||25 km||National rates||4,100||0.98 (for NPPs)
1.02 (possible sites)
|Scotland||Sharp et al., 1996||7||1968-1993||0-14||I||25 km||National rates||399||1.99 (reprocessing plant)
0.90 (for NPPs)
|Germany||Kaatsch et al., 1998||20||1991-1995||0-14||I||15 km||30-100 km||550||1.05|
|France||White-Komng et al., 2004||29||1990-1998||0-14||I||20 km||National rates||670||0.92|
|Japan||Yoshimoto et al., 2004||44||1972-1997||0-14||M||10 km||10-80 km||473||1.01|
|France||Evrard et al., 2006||23||1990-2001||0-14||I||40 km2||national||750||0.94|
|Britain||Bithell et al., 2008||13||1969-1993||0-4||5,10,
|National rates||409||1.36 (<5 km)
0.90 (<10 km)
0.97 (<25 km)
|Finland||Heinavaara et al., 2010||2||1975-2004 0-20 I||15 km||Stratum-s incidence rates||16||1.01|
|Britain||COMARE, 2011||13||1969-2004 0-4 I||25 km||National rates||511||1.01 (<5 km)
|NOTE: I, incidence; M, mortality; SIR, standardized incidence ratio; SMR, standardized mortality ratio; OR, odds ratio; RR, relative risk; NPP, nuclear power plant.|
|B. Case-control studies|
|Country||Reference||No. Sites||Period of Diagnosis||Age||Area Examined||No. Cases||OR|
|Germany||Kaatsch et al., 2008||16||1980-2003||0-4||≤5 km
|Finland||Heinavaara et al., 2010||2||1975-2004||0-14||5-10 km vs ≥30 km||16||0.7|
|France||Sermage-Faure et al., 2012||19||2002-2007||0-14||≤5 km vs ≥20 km||2,753||1.9 with distance
1.0 with dose-based zoning
|C. Cohort studies|
|Country||Reference||No. Sites||Period Examined||Age||Area Examined||No. Cases||RR|
|Finland||Heinavaara et al., 2010||2||1975-2004||0-14||<15 km vs 15-50 km||16||1.0|
|Switzerland||Spycher et al., 2011||5||1985-2009||0-14||≤5 km vs >15||953||1.24|
districts that may have been referred elsewhere for diagnosis, was conducted (Roman et al., 1987). Among the 60,000 children residents within a 10-km radius of a nuclear establishment, the recorded incidence was three cases per year while two cases per year were expected.
In 1986 a cluster of leukemia among children was reported around the area of the Dounreay nuclear reprocessing plant in Scotland (Heasman et al., 1986). In 1987 and 1989 two reports were published of an increased rate of leukemia in children under 15 years of age that reside within a 16km (10-mile) radius of the nuclear weapons plants in Aldermaston and Burghfield (Forman et al., 1987; Roman et al., 1987) and the Hinkley Point nuclear power station in Somerset, England (Cook-Mozaffari et al., 1989a; Ewings et al., 1989). This later cluster was not confirmed by follow-up studies (Bithell et al., 1994). In 1992, a fifth cluster was reported in Britain among children under 10 years of age near the Amersham plant that produces radioisotopes (Goldsmith, 1992). Again the increased incidence was not confirmed by others (Bithell et al., 1994).
Using more comprehensive data sets and analyses, Draper and colleagues (1993) aimed first to reappraise the original report of possible excess of childhood leukemia incidence and non-Hodgkin’s lymphoma in areas around the Sellafield nuclear installation and second to determine whether the excess incidence persisted in the years following the original report. All ages and other cancers were included. The authors confirmed an increased incidence in cancer, especially leukemia in young people. Cook-Mozaffari et al. (1989b) analyzed data on mortality for 400 districts of England and Wales where there was an existing nuclear installation or the construction of nuclear installations had been considered or occurred at a later date. The authors report an excess mortality due to leukemia in young people who lived near potential sites similar to that in young people who lived near existing sites, implying the presence of unidentified risk factors associated with the sites where nuclear stations reside or are selected to reside but not associated with the nuclear installations themselves.
A study aiming to examine the contribution of potential risk factors to the observed excess of childhood leukemia ( 25 years of age) and lymphoma near the Sellafield nuclear plant in Cumbria, England, was conducted, this time using a case-control design (Gardner et al., 1990). Fifty-two cases of leukemia, 22 cases of non-Hodgkin’s lymphoma, and 23 Hodgkin’s disease patients diagnosed in the period 1950-1985 and 1001 controls matched on sex and date of birth were compared. Antenatal abdominal x-ray examinations, viral infections, behavioral data, lifestyle factors, and parental employment at Sellafield were examined as potential risk factors. The authors concluded that there is an association between childhood leukemia and paternal exposure before conception to relatively high doses of radiation. More specifically, the relative risk for paternal
estimated dose of 100 mSv before the child’s conception was 8.4 (95% CI: 1.4-52.0 based on 4 exposed cases). However, the relative risk for the next-highest paternal preconception dose category of 50-99 mSv was only 0.78 (CI: 0.1-7.8 based on 1 exposed case), which was not very supportive of a dose-related risk. When doses received 0-6 months before conception were examined, the relative risks for the highest (10 mSv) category was 8.2 (CI: 1.6-42 based on 4 exposed cases) and for 5-9 mSv was 3.0 (CI: 0.3-33, 1 exposed case). The authors speculate that radiation exposure during work may have an effect on the father’s germ cells, producing genetic changes in sperm that may be leukemogenic in the offspring. The evidence, however, seems mixed and subsequent independent investigations in England, France, Scotland, and Canada did not support this association (Draper and Vincent, 1997; Draper et al., 1997; Kinlen et al., 1993; McLaughlin et al., 1993b; Pobel and Viel, 1997).
Bithell and colleagues (1994) performed the largest (at the time) incidence study for all of England and Wales and examined the relationship between the risk of childhood leukemia (15 years of age) and non-Hodgkin’s lymphoma and proximity of residence to 23 nuclear installations for the period 1966-1987. The authors investigated regions of 25-km radius and six control sites that had been considered for generating stations but were never used. Observed and expected numbers of cases were calculated and analyzed by standard methods based on ratios and by linear rank score test. Overall, there was no evidence of an increase of childhood leukemia or of non-Hodgkin’s lymphoma around nuclear installations. The only significant results for the linear rank score test were for Sellafield and a weaker but significant association for Burghfield. The authors noted that a more appropriate analysis would be one based on place of residence at birth as an analysis based on place of diagnosis may fail to detect the effect of prenatal or preconception factors. A year later, a mortality study investigated seven districts near the sites of Harwell, Aldermaston, and Burghfield for the period 1981-1995, among children younger than 15 years. Excess leukemia deaths were reported in two districts (Newbury, 11 deaths observed, 5.7 expected; South Oxfordshire, 12 deaths observed, 4.9 expected) (Busby and Cato, 1997). However, the ranking of the seven districts by incidence rates for the period 1969-1993 did not agree with that for mortality and no excess of leukemia cases existed (Draper and Vincent, 1997). In Scotland, Sharp and colleagues carried out a similar study of the seven nuclear sites for the period 1968-1993. The only significant observation was the reported excess around the Dounreay reprocessing plant (Sharp et al., 1996).
The reported cluster around Dounreay, Scotland, was referred to COMARE for consideration and the committee recommended further epidemiologic investigations, including a cohort study of the incidence of leukemia among children born locally and those who attended school in the
area but were born elsewhere (Black et al., 1992) and a case-control study to examine possible risk factors for leukemia (Urquhart et al., 1991). The aim of the cohort study was twofold: (a) to determine whether the excess of leukemia and other cancer cases occurred in children born to mothers that were residents in the Dounreay area or in children who moved to the area after birth and (b) to determine whether any leukemia cases occurred in children born near Dounreay who may have moved elsewhere. The cohort included 4,144 children born in the area in the period 1969-1988 and 1,641 children who attended local schools in the same period who had been born elsewhere. Cancer registration records were linked to birth and school records and observed rates were compared to national rates. The authors showed that the incidence of leukemia and non-Hodgkin’s lymphoma was raised in both the birth and school cohorts with observed-to-expected ratios of 2.3 and 6.7, respectively, suggesting that the place of birth was not a more important factor than place of residence in the series of cases observed near the Dounreay area. No cases were found in children who were born in Dounreay and moved elsewhere (Black et al., 1992).
The excess incidence of leukemia and non-Hodgkin’s lymphoma in children and young adults in the area less than 25 km from the Dounreay nuclear installation was later reexamined for the period 1968-1991 and was found to continue to be a matter of concern (Black et al., 1994). In the case-control study, the study participants were 14 cases of leukemia and non-Hodgkin’s lymphoma occurring in children aged less than 15 years diagnosed in the area between 1970 and 1986 and 55 matched controls. Antenatal abdominal x-ray examination, drugs taken, and viral infections during pregnancy were examined as potential risk factors by interviews and structured questionnaires. Given the findings of Gardner et al. (1990), who reported a possible association between paternal employment and development of leukemia by the offspring, detailed information on father’s occupation, father’s employment at Dounreay, and radiation dose preconception exposure to nonionizing radiation of the father was collected. The study in Dounreay did not provide any evidence of father’s employment as a risk factor for childhood leukemia. (However, a possible but weak association between the children’s use of local beaches and risk of leukemia was identified.) The paternal preconception exposure theory of genetically transmitted disease was also rejected by Doll, who published a commentary entitled “Paternal exposure not to blame,” emphasizing the fact that the hypothesis that irradiation of the testes causes any detectable risk of leukemia in the offspring does not agree with what is known of radiation genetics or of the heritability of childhood leukemia (Doll et al., 1994).
A year earlier, Kinlen et al. (1993) also argued that paternal exposure as a risk factor for childhood leukemia would not explain the excess. Kinlen speculated that nuclear plants that were built in unusually isolated places,
for example, Dounreay and Sellafield in Britain, led to large influxes of people such as construction workers, scientists, and “nuclear” employees in the 1950s to those areas. Indeed, the development of the Dounreay plant, which started its operations in 1958, raised the population in the area of nearby Thurso almost 150 percent between 1951 and 1961. This or similar situations (irrelevant to the radiation industry) may result in bringing into contact susceptible and infected individuals for some unidentifiable transmissible agent whose route and nature of the infection remain unknown. Infected individuals could have been present in any of the groups and given a sufficient population density could have caused outbreaks (Kinlen, 2011a; Kinlen et al., 1995). The theory of population mixing was originally applied on the North Sea oil industry in Scotland (Kinlen et al., 1993) and was also tested later on the Nord Cotentin region in France, which shares some characteristics with the Sellafield and Dounreay regions in terms of population influx between the years 1978 and 1992 with the construction of the La Hague nuclear waste reprocessing site and the Flamanville nuclear power station (Boutou et al., 2002). Although the hypothesis of an infectious agent has some plausibility, the studies assessing the hypothesis are ecologic and have inherent limitations that would not allow them to prove a causal relationship between the unknown infectious agent and the disease. Still, the Kinlen hypothesis of population mixing is well perceived today and, although it has not been explicitly examined, it is part of the discussion of the studies on cancer risks in populations around nuclear facilities published the past 2 years (COMARE, 2011; Sermage-Faure et al., 2012; Spycher et al., 2011).
Following the publication of the results from the KiKK study showing an increased risk among children 5 years of age or younger that live within the 5-km radius from German nuclear power plants (Kaatsch et al., 2008; Spix et al., 2008), Bithell et al. (2008) conducted a study to reexamine the incidence of childhood leukemia around nuclear power plants in Britain. The main reason was that results from Germany did not support those of COMARE published in 2005, and this discrepancy could be accounted for by methodological differences, especially those related to the distances from the power stations and the ages of the children investigated. Bithell and colleagues used the same data as considered by COMARE’s tenth report and modified the methodology to apply as similar of an approach as possible to that of the KiKK study. The incidence of childhood leukemia observed (18 cases against 14.58 expected within the 5-km zone) was not significantly raised. The original paper (Bithell et al., 2008) made no adjustments for demographic characteristics to resemble the methodology of the KiKK study. Follow-up analysis (Bithell et al., 2010) adjusted for population density at the ward level without altering the overall conclusions.
The latest report from Britain and the fourteenth in series by COMARE
presented a new geographic data analysis on the incidence of leukemia in children under 5 years of age, living in the vicinity of 13 nuclear power plants (COMARE, 2011). The investigators used cancer registration data for the period 1969-2004 extending the previous analysis presented in COMARE’s tenth report for 1969-1993. The report concluded that there is no evidence to support an increased risk of childhood leukemia and other cancers in the vicinity of nuclear power plants due to radiation effects. COMARE recommended that monitoring of liquid carbon-14 discharges from the plants continues, as this radioactive isotope of carbon is a major contributor to the radiation doses which the public receive from discharges. Moreover, the report recommends that research continues for all possible causative mechanisms of leukemia, including the role of infectious agents. An extensive review of the KiKK study as well as useful unpublished analyses of the data are presented in the report.
An excess of childhood leukemia cases in the small rural community of Elbmarsch in Northern Germany, close to the Krümmel nuclear power plant, was first reported in the early 1990s (Schmitz-Feuerhake et al., 1993). Between 1990 and 1995, six cases of childhood leukemia were diagnosed, five of whom resided within a 5-km radius from the plant (Hoffmann et al., 1997). The cluster persisted until at least 2005 (Grosche et al., 1999; Hoffmann et al., 1997, 2007), and together with that of Sellafield and Dounreay (both fuel reprocessing plants) was a confirmed cluster of childhood leukemia near nuclear facilities (Laurier et al., 2008b). The modestly elevated levels of cesium detected in rainwater and air samples led to postulations that there was an accidental release of radionuclides from the nuclear research facility near the community (Schmitz-Feuerhake et al., 1997).
An ecologic study that compared disease rates within 15 km of German nuclear plants with those in control areas was designed following an approach almost identical to the British studies (Michaelis et al., 1992). The German study was based on 1,610 childhood malignancies identified from the country’s childhood cancer registry including leukemia cases that were diagnosed before the child’s fifteenth birthday from 1980 to 1990. An increased risk of all cancers or leukemia within the 15-km zone was not confirmed. However, exploratory analysis indicated that in children younger than 5 years old living within the 5-km zone, the increase in leukemia risk was statistically significant. A second study was undertaken to validate the results of the previous exploratory analysis and include independent data for the period 1991-1995 (Kaatsch et al., 1998). Results did not support the original hypothesis or the exploratory findings from the 1980-1990
period, although a tendency toward an increased risk estimation for leukemia to occur in children younger than 5 years within the 5-km vicinity persisted. Although the authors concluded that at that point no further investigations were necessary in Germany, discussions on the potential elevated risk of cancer in populations living near nuclear facilities under routine operation did not cease. This led the German federal government to start a case-control study, the third one in a series of corresponding investigations which differs from the previous ecologic studies that were based on aggregate data. The case-control study investigated exact information on distance of the family’s place of residence at the time of diagnosis to the chimney of the nearest nuclear power plant with a precision of 25 m (Kaatsch et al., 2008).
The study is known as the KiKK study and was carried out by researchers from the German Childhood Cancer Registry in Mainz, on behalf of the Federal Office of Radiation Protection. Control subjects were randomly selected from the records of the appropriate registrar’s office and matched to cases for the date of birth, age, gender, and nuclear power plant area. Five hundred and ninety-three leukemia cases and 1,766 matched controls were included in the study; however, only 37 cases lived within the 5-km zone, the most important number to assess the meaningfulness and strength of the observed association. Analysis indicated a statistically significant odds ratio (OR) of 2.19 [lower limit of the 95% confidence interval (CI) = 1.51] for residential proximity within 5 km of one or more of the 16 nuclear power plants compared to residence outside these areas for children aged less than 5 years. No effect was seen for the distance 5-10 km from a plant (OR = 1.09, based on 58 cases). A negative trend for distance was identified; the farther the residence was from the nuclear power plant, the lower the risk. No association between distance to the nuclear power plants and risk of developing leukemia was observed when children aged 0-15 years were examined together. The investigators attempted to collect data on exposures, residential history, and other potential confounders such as socioeconomic characteristics, pesticides, and immunological factors by administering questionnaires to a subset of the study participants. Because the response rates varied remarkably with distance to the plants (total response was 78 percent for cases, 61 percent for controls; response in the inner 5-km zone was 63 percent for cases, 45 percent for controls), the results were not summarized due to the high risk of selection bias. In the absence of a questionnaire survey, potential confounders could not be investigated; therefore, the study overall did not differ substantially from ecologic studies. Still, the study was associated with wide publicity (http://www.bfs.de/en/bfs/presse/pr07/pr0712, http://www.bfs.de/en/kerntechnik/kinderkrebs/statement_kikk_en.pdf) and some have argued that the sponsoring body made extravagant claims of its importance (Kinlen, 2011b).
The study has been criticized for potential defective control selection (COMARE, 2011; Little et al., 2008a), but also for the misleading presentation of study findings by zone, time period, and malignancy subtype (Kinlen, 2011b). As discussed in a recent critical review (Kinlen, 2011b), some 10 percent of community registrars tasked with control selection declined to cooperate, the proportion being higher within the 5-km zone (16 percent). Moreover, some registrars did not follow instructions regarding matching criteria of cases and controls, selecting potential control children for an inappropriate calendar year, that is, not for the year the matched case was diagnosed. Moreover, the increased risk was driven by risks associated with early operational years: The data from the most recent 8 years (1996-2003) were suggestive of a trend, though the association was not as strong as the earlier period (1980-1995, OR = 1.8, 95% lower bound of the CI: 0.99). Additionally, results seemed to be driven by the notable excess of cases of childhood leukemia around the Krümmel plant in northern Germany, an analysis that was not undertaken by the original authors but by COMARE (COMARE, 2011).
The same group published results from a larger population (1,592 cases and 4,735 controls) that included all other childhood cancers and concluded that leukemia was driving the positive association of cancer risk and living near the installations (Spix et al., 2008).
The Northern Germany Leukemia and Lymphoma (NLL) study is a population-based case-control study that preceded the KiKK and was designed to address the risk associated with three environmental exposures simultaneously: ionizing radiation released from nuclear power plants, electromagnetic fields, and pesticides (Hoffmann et al., 2008). In contrast to the KiKK study, which relied on distance to the residence as a surrogate of exposure, the NLL study reconstructed radiation doses arising from routine discharges of radioactive material from four nuclear power plants by extracting relevant information obtained from questionnaires. Exposure to ionizing radiation due to medical diagnostic or therapeutic radiation was also assessed. The NLL study did not find an elevated risk with the radiation doses assessed to have been received as a result of routine discharges from the nuclear power plants.
Following the cancer mortality study around nuclear installations in Great Britain (Forman et al., 1987), Hill and Laplanche (1990) reported the results of a similar study for the population residing around six nuclear installations in France, four of which were nuclear power plants. In the period 1968-1987, the number of leukemia deaths among children and young adults aged 0-24 was 58, compared to 62 in control areas. In the
same period, two studies examined mortality from leukemia among those aged 0-24 near the La Hague reprocessing plant in Nord Cotentin, a region with particularly high density of nuclear installations. No findings of excess mortality were reported (Dousset, 1989; Viel and Richardson, 1990,1993). An extended multisite study that included observed leukemia deaths for the years 1988 and 1989 around 13 nuclear installations, of which 11 were nuclear power plants, also showed no excess in mortality (Hattchouel et al., 1995).
In 1993, Viel et al. published the results of a study of the incidence of leukemia among persons up to 24 years of age living within 35 km of the La Hague nuclear reprocessing plant in the region of Nord-Cotentin in France and diagnosed between 1978 and 1990 (Viel et al., 1993). Twenty-three cases were diagnosed, giving an incidence rate of 2.99 per 100,000, which is close to the expected rate. Two years later, the same group continued their initial survey by including data through 1992 (Viel et al., 1995). Although the study did not show excess of leukemia for the zone as a whole, a non-statistically significant increased risk was observed if analysis was restricted to an administrative unit in the 10-km zone around the plant (four cases observed over 15 years while 1.4 were expected). These studies together with a third study on cancer incidence that covered the period 1978-1996 (Guizard et al., 1997) led to the conclusion that the potential elevated risk associated with living near the La Hague site should be kept under review. A follow-up ecologic study of incidence using zones defined according to their distance from the La Hague site (0-10, 10-20, and 20-35 km) was conducted to describe the occurrence of leukemia for each age group and cytological type for the period 1978-1998. The highest SIR was observed in the 5-9-year-old group (SIR = 6.38, 95% CI = 1.32-18.65) within the 10-km zone from the plant (Guizard et al., 2001).
Pobel and Viel (1997) reported the first case-control study in France. The study was undertaken within a 35-km radius of the nuclear waste reprocessing plant of La Hague. The aim was to investigate the association between childhood leukemia (25 years of age) and established risk factors or other factors related to the plant. Twenty-seven cases of leukemia diagnosed during the period 1978-1993 and 192 matched controls were investigated, and information on antenatal and postnatal exposure to x-rays and viral infections, occupational exposure of parents, and lifestyle of parents and children was extracted through administered questionnaires and face-to-face interviews. A threefold increased risk of developing leukemia and frequent use of local beaches was found. Consumption of local fish and shellfish also showed an increased trend with risk. No association with occupational radiation was observed. The authors suggest an environmental route of exposure of children to radioactive material associated with certain lifestyle risk factors. These findings have been debated especially concerning
control selection, possible recall bias, multiple comparisons, and biological plausibility of the causal associations inferred (Clavel and Hemon, 1997; Law and Roman, 1997; Wakeford, 1997).
To respond to public concerns, the French government commissioned complementary epidemiologic investigations and also requested an analysis to be carried out by the Nord-Cotentin radioecology group to estimate the local population’s exposure to radiation. No risk associated with radiation-induced leukemia was found (Rommens et al., 2000). More recent multisite studies in France like the one by (White-Koning et al., 2004) examined childhood leukemia (15 years of age) incidence rates within 20 km of the 29 nuclear sites in the period 1990-1998. Comparison of the observed rates in areas surrounding the sites to expected rates based on national registry data did not provide any evidence of an excess leukemia in those areas. Results from intermediate analyses performed at the time of the White-Koning study that focused on leukemia incidence among children less than 5 years of age—to resemble the KiKK study in Germany (Kaatsch et al., 2008)—did not show an association (Laurier et al., 2008). However, the number of cases within the 5-km zone was small (5 observed cases compared to 5.2 expected from national rates).
The above-mentioned studies, as the majority of studies of incidence of leukemia around nuclear facilities, use distance to the site as a surrogate for radiation dose exposure, assuming an isotropic distribution of discharges. Evrard et al. (2006) investigated for the first time the incidence of childhood leukemia (15 years of age) around 23 French nuclear installations (18 nuclear power plants, 2 nuclear fuel-cycle plants, 1 nuclear fuel reprocessing plant, 2 research centers) using a geographic zoning based on estimated doses to the bone marrow due to gaseous radioactive discharges. Direct radiation and liquid discharges were not considered. Compared to the study period of the previous report (White-Koning et al., 2004), this one included 3 additional years of observation (study period was 1990-2001). Risk was estimated for each of the five zones defined on the basis of estimated exposure levels, and trends of increasing risk with increasing exposure were recorded. Analysis showed no evidence of general increase of risk or trend in the incidence of childhood leukemia according to the zoning method developed in the study. More specifically, for the nuclear power plants, 242 cases were observed over the study period against 253 expected (SIR = 0.96), with no observed trend with dose. When the other nuclear facilities were included, the SIR was 0.94. Further analyses for the individual diagnosis age groups, 0-4-, 5-9-, and 10-14-year-olds, also did not show any significant trends by estimated exposure categories. Specifically, for the ages 0-4 years, the SIRs for the two highest exposure categories were 0.92 (based on 19 cases) and 0.93 (based on 5 cases), compared to the total SIR of 0.95 (based on 395 cases) for that age group. This study is notable in
that it was the first multisite study to conduct analyses based on estimates of exposure levels, although those estimates did not consider liquid discharges. An updated study with an additional 5 years of observations (2002-2007) that used both a case-control and an ecologic approach showed that for the recent years, children living within 5 km of nuclear power plants (14 cases) are twice as likely to develop leukemia compared to those living 20 km or farther away from the plants. However, analysis of the same population of children using a dose-based geographic zoning approach, instead of distance, did not support the findings. The authors discuss that the absence of any association with the dose-based geographic zoning approach may indicate that the observed association of distance and cancer risk may be due to factors other than the releases from the nuclear power plants (Sermage-Faure et al., 2012). Among the potential factors are population mixing (Kinlen, 2011a) (a hypothesis that could not be evaluated in this study) and exposures to agents including natural or manmade exposures to radiation not modeled in the study. At least two additional aspects of this study are worth emphasizing: (a) While the KiKK study showed a doubling of risk in childhood leukemia only in children less than 5 years of age that live close to a nuclear power plant in Germany, the observed increase in leukemia incidence in this study was not restricted to the very young children but to all children ages 0-14. (b) The risk estimations from the case-control and ecologic approaches were in high concordance (OR = 1.9, 95% CI = 1.0-3.3 and SIR = 1.9, 95% CI = 1.0-3.2, respectively).
A.4.1.4 United States
In 1990, a national study by the National Cancer Institute (NCI), and the broadest of its kind ever conducted, investigated the potential excess of cancer deaths in 107 counties containing or closely adjacent to 62 nuclear facilities (Jablon et al., 1990, 1991). Three comparison counties were selected for each study county matched to study counties by the percent of persons in the population over 25, race, household income, and population size among other characteristics. The facilities included in the study were 52 nuclear power plants, nine Department of Energy (DOE) research and weapon plants and one commercial fuel reprocessing plant; all had begun operation before 1982 (Jablon et al., 1991). The survey examined 16 types of cancer that included those of the stomach, colorectal, primary liver, lung, female breast, and especially focused on leukemia. SMRs were calculated within “exposed” counties before and after the plant started operation and between “exposed” and “unexposed” counties both before and after plant startup. Over 900,000 cancer deaths occurred from 1950 through 1984 in the counties examined. The study results were essentially negative. No general increase in cancer mortality was found in counties with or near nuclear power plants and, unlike some reports in Britain (Black, 1984; COMARE,
1988, 1989; Heasman et al., 1986), no excess incidence of leukemia was found in children who lived near reprocessing and weapons plants. At the time the study was designed, county was the smallest geographic unit for which nationwide data on mortality could be quickly evaluated. However, it is well recognized that this was a limitation of the study because a county may be too large to detect risks present only in limited areas, which results in a dilution of any dose-associated effect. The limited incidence data available from two states (Iowa and Connecticut) provided inconclusive results.
Boice and colleagues (2005, 2006, 2007a, b) extended by 16-17 years the 1990 NCI study results at St. Lucie nuclear power plant in Florida, the Department of Energy’s Hanford nuclear facility in Washington, and the uranium mining and milling facilities in Montrose County, Colorado. The team investigated cancer mortality rates among residents of counties near the facilities and found no evidence for increased risk compared to control counties that could be attributed to radiation exposures. Cancer mortality and incidence were also investigated in counties near the Apollo-Parks former nuclear materials processing facilities in Pennsylvania (Boice et al., 2003a, b). Although there was no observed increase in risk as measured by either mortality or incidence rates, the authors emphasize that mailing addresses in small rural areas may not always reflect actual residences, and validation by contacting area postmasters and using Census Bureau geocoding information may be necessary to prevent misleading conclusions. An update of the study showed consistent findings of lack of evidence for increased incidence near the former Apollo-Parks nuclear facilities (Boice et al., 2009).
Cancer risks were also investigated among residents in relation with the uranium milling and mining operations in Grants, located in Cibola County, New Mexico. Cancer mortality data were analyzed for the period 1950-2004 and cancer incidence data for the period 1982-2004 (Boice et al., 2010). Lung cancer mortality and incidence were significantly increased among men (SMR = 1.11, 95% CI = 1.02-1.21; SIR = 1.40, 95% CI = 1.18-1.64) but not women. Analysis among the population of the three census tracts near the Grants Uranium Mill revealed a higher risk for lung cancer among men (SMR = 1.57; 95% CI = 1.21-1.99). The authors discuss that etiologic inferences are not possible because of the ecologic study design. However, the excess in lung cancer among men is likely to be due to previously reported risks among underground miners from exposure to radon and its decay products, coupled with heavy smoking and possibly other factors.
Mortality among residents of Uravan, a company town built around the uranium mill in Montrose County, Colorado, was investigated in more detail using a retrospective cohort study design (Boice et al., 2007b). The study population was originally identified from worker and community
records (Austin, 1986). Workers at the Uravan mill and nearby uranium mines, their spouses and children, and other workers in the town such as teachers and postal clerks were included in this study. Approximately 1,900 men and women who lived in Uravan for at least 6 months within the period 1936-1984 and were alive after 1978 were included in the study. Results showed that among the approximately 450 residents who had worked in underground uranium mines, a significant twofold increase in lung cancer was found. No significant elevation in lung cancer was seen among the female residents of Uravan or the uranium mill workers. The excess of cancer among uranium miners was attributed to the historically high levels of radon in uranium mines of the Colorado Plateau, and heavy smoking among the workers (Boice et al., 2007).
Previous smaller studies of mortality or incidence in the United States, such as that around the San Onofre power plant in California (Enstrom et al., 1983), the Rocky Flats nuclear weapon production facility in Colorado (Crump et al., 1987), and Hanford and Oak Ridge in Washington State and Tennessee, respectively (Goldsmith, 1989), showed no evidence of increased risk. Mangano (1994) concluded that between 1950-1952 and 1987-1989, cancer risk from all types of cancer and all age groups increased significantly around the Oak Ridge site; however, a radius of 160 km was analyzed as a whole. An excess of incident leukemia across all age groups reported by Clapp and colleagues (1987) for the period 1982-1984 in Massachusetts seemed to be counterbalanced by a lower-than-expected incidence of cases the 2 following years (Poole et al., 1988; Wilson, 1991).
State health departments have also specifically addressed concerns of their communities on increased cancer rates around nuclear facilities. Such an example is the recent publication from the Illinois Department of Public Health, which analyzed childhood cancer rates in the vicinity of the plants in the state (Ma et al., 2011).
One of the largest and most comprehensive studies conducted in the United States regarding the risk of cancer near a nuclear facility, in this case thyroid disease, is the Hanford Thyroid Disease study. The Hanford Nuclear Site in southeastern Washington State was established in 1943 to produce plutonium for atomic weapons. In the mid 1980s it was revealed that during the 1940s and 1950s of plutonium production at Hanford, large amounts of gaseous and vaporized radionuclides were released into the atmosphere including about 740,000 Ci of 131I resulting in estimated mean dose to the thyroid of 174 mGy. In response, the U.S. congress mandated the Hanford Thyroid Disease study in 1988 to investigate the widespread concerns among people living near the site that such releases may have increased their risk of developing thyroid disease. The primary analyses focused on living participants who received medical examinations to detect thyroid disease, and for whom thyroid radiation doses were estimated using
the dosimetry system developed by the investigators; dose reconstructions were based on environmental measurements and personal interviews (Davis et al., 2004). The investigators concluded that there was no evidence of a relationship between Hanford radiation dose and thyroid cancer incidence or other thyroid diseases. In an attempt to reconcile the study results with the evidence for thyroid disease that has been reported for the Chernobyl accident (see Section A.4.3), which also includes exposures primarily to 131I, the investigators suggest that differences in the dose and dose rates delivered may account for the differences in observed risks. Other investigators recommend that the results are interpreted as inconclusive (rather than negative) because of possible inadequate power to detect an effect due to uncertainties associated with the models and assumptions used for individual dose reconstruction (Hoffman et al., 2007).
Potential health effects associated with the 1979 accidental releases of the Three Mile Island nuclear plant in Pennsylvania have been examined and have been a subject of controversy. Immediately after the accident, a presidential commission expressed confidence that the maximum external radiation dose to a person in the general population was less than the average background (~ 1 mSv) and that no health effects would be detectable and that the sole health consequence for the population in close proximity to the installation was mental distress (Kemenu et al., 1979). Karl Morgan, one of the founders of the field of radiation health physics, estimated that there would be 50 excess cancer cases in the area surrounding the plant, a presumptive risk characterized as “exaggerated” based on current knowledge of radiation effects at the doses surrounding populations would be exposed (Upton, 1980).
The initial cancer risk survey was conducted by Columbia University for the period 1975-1985 and was supported by the Three Mile Island Health Fund, which was created and governed by a court order (Hatch et al., 1990, 1991). Estimates of the emissions delivered to small geographic study zones were derived from mathematical dispersion models. Although the data provided hints of increased risk of leukemia and lung cancer in the surrounding areas, they were interpreted as not convincing based on the assumption that the doses were too low to produce a measurable effect (Hatch et al., 1990). Given the “mental distress” health consequence that the government reported for populations that lived near the facility when the accident happened, a study was conducted to test whether mental distress could be linked with the somewhat elevated cancer incidence in the area (Hatch et al., 1991). Stress following local community disasters has been linked with increased cancer in early studies (Bennet, 1970; Janerich et al., 1981); however, studies on the relationship between psychological stress and cancer have revealed conflicting results (Garssen, 2004), although it is known that stress can affect the immune system (Segerstrom and Miller, 2004). In the absence of individual and direct measures of stress, residential
proximity to the site was used as a surrogate (Hatch et al., 1991). Using this crude test of an accident-stress hypothesis, a 40 percent increased risk between postaccident cancer rates and proximity was estimated. The authors state that radiation emissions as modeled mathematically did not account for the observed increase (Hatch et al., 1991).
The topic of health effects related to the Three Mile Island accident reappeared in 1997 when attorneys representing more than 2,000 area residents asked epidemiologist Stephen Wing from the University of North Carolina to examine the original work. The examination, with severe criticism on the study approach followed by Hatch and colleagues, reanalyzed and reinterpreted exactly the same data. The claim was that the original study may have been biased, as analysis was driven by the belief that no association could exist at low exposures. The new analysis showed that incidence of leukemia and lung cancer following the accident increased more in areas estimated to have been in the pathway of radioactive plumes compared to areas outside the pathway (Wing et al., 1997a). An exchange of published responses between the Columbia team and Wing followed (Hatch et al., 1997; Susser, 1997; Wing et al., 1997b). To this day, Wing’s article remains the only one to present original health data supporting an association between releases from the Three Mile Island accident and cancer.
A case-control study by McLaughlin and colleagues (1993b) of workers at nuclear facilities in Ontario, Canada, can possibly be directly compared with that of Gardner et al. (1990) because it tested the hypothesis of an association between childhood leukemia and the occupational exposure of fathers to ionizing radiation before a child’s conception. In this study, cases (n = 112) were children (15 years of age) who died or were diagnosed with leukemia in the period 1950-1988 and were born to mothers living near one of the five operating facilities under investigation (one research development facility, a uranium refinery, a uranium mining and milling facility, and two nuclear power plants). No association with paternal occupational exposure was found in the analysis (McLaughlin et al., 1993b). Also, an ecologic study examined the mortality and incidence of childhood leukemia for the period 1950-1987 among children less than 15 years of age living in the vicinity of the Ontario nuclear facilities (McLaughlin et al., 1993a). Overall, the observed number of leukemia deaths (O = 54) was slightly greater than expected (E = 46.1) during the period when the facilities operated, but the difference was not statistically significant (O/E = 1.17, 95% CI = 0.88-1.53).
Lopez-Abente and colleagues (1999) studied the mortality due to hematological tumors in towns lying within 30 km of seven nuclear power plants and five nuclear fuel facilities during the period 1975-1993. No study area yielded evidence of a raised risk of leukemia mortality among persons under the age of 25. A recent updated ecologic study that included all nuclear power plants and other nuclear fuel facilities in the country, regardless of whether they are in operation, studied mortality due to different types of cancer including leukemia in municipal areas within a radius of 30 km around the facilities and in control counties (50-100 km). The study period was 1975-2003. The main original contribution of the study was the reconstruction of the exposure of the population in each municipality accounting for both liquid and gaseous discharges from the facilities, described as means of effective dose (Nuclear Safety Council and the Carlos III Institute of Health, 2009). The spatial distribution of the data by the different dose categories differs from the radius pattern produced by distances used in most previous studies, since specific characteristics of each site, including land and water use, have been incorporated in the models. The dose estimates are conservative, constituting the upper limit for the exposures actually received by the populations.
Risk estimates were adjusted for natural radiation and other covariates. The investigators interpret their findings as there being overall no association of living near the nuclear facilities and cancer mortality. Increases in risk such as those observed for lung and bone cancer around specific nuclear fuel-cycle facilities were interpreted as inconsistent, as they were not replicated across the facilities examined and cannot be attributed to the effect of the doses generated as a result of their operation, primarily because the releases are too low to have an impact.
The existence of leukemia clusters among those less than 15 years of age living near four nuclear facilities was examined for the period 1980-1990. No consistent evidence was found for childhood leukemia clusters associated with living in the proximity of nuclear power plants (Waller et al., 1995).
A recent multiapproach investigation in Finland (ecologic, case-control, and cohort studies) suggests no association of leukemia and vicinity to the two nuclear power plants (Heinavaara et al., 2010). However, the 5-km zone around the nuclear plants was not investigated.
The results of the Childhood Cancer and Nuclear Power Plants in Switzerland (CANUPIS) study were recently published (Spycher et al., 2011). CANUPIS was a large census-based cohort study that analyzed distance of residence at birth as well as distance of residence at diagnosis to determine if children who grew up near the country’s five nuclear power plants had an increased risk of developing childhood cancer. Children aged 0-15 years born in Switzerland from 1985 to 2009 based on the 1990 and 2000 Swiss censuses and identified cancer cases from the Swiss Childhood Cancer Registry were included in the study. Completeness of registration was greater than 90 percent. In the study period, 2,925 children were diagnosed with cancer, 953 of whom had leukemia. The number of diagnosed children that lived within the 5-km zone was small: 18 and 31 children at ages 0-4 and 0-15 years, respectively, were diagnosed with cancer overall, while 8 and 12 children in the above-mentioned age groups were diagnosed with leukemia. Compared with children born at a distance greater than 15 km from the plant, the RRs (95% CIs) for leukemia in the 0-4 and 0-15 age groups were 1.20 (0.60-2.41) and 1.05 (0.60-1.86), respectively.
Results presented little evidence for an association between residence at birth or diagnosis near nuclear power plants and risk of leukemia or other childhood cancers. Potential confounders that were considered included background ionizing radiation, electromagnetic radiation from power lines and other sources, carcinogens related to traffic, pesticide exposure, socioeconomic status, and proxies of population mixing and exposure to childhood infection (average number of children per household in the community and degree of urbanization) (Law, 2008). Although no data on radiation releases from the nuclear plants were available, additional analysis was performed where main dispersal directions of airborne emissions were accounted in the model. Results were consistent with the main results. Among the limitations of this study were the small sample size, particularly of 0-4-year-olds living close to the nuclear power plants, and lack of coverage of the earlier time periods when higher dose exposures may have occurred.
In Israel a study of the population near the Dimona nuclear plant (Sofer et al., 1991) examined new leukemia cases among those under 25 years of age who lived within 45 km of the station. The authors concluded that there was no excess incidence near the power plant.
A study by Yoshimoto et al. (2004) that covered the period 1972-1997 in 20 municipalities in Japan, containing 16 nuclear power plants showed no evidence of increased risk compared to control municipalities among the young residents. However, rates of mortality due to leukemia for the population overall were higher among those populations living in proximity to nuclear power plants in Japan.
A.4.2 Atomic Bombing Survivor Studies
The atomic bombs that exploded over the city of Hiroshima and three days later over Nagasaki, Japan, in August 1945 exposed the people of each city to whole-body doses of penetrating ionizing radiation. The number of deaths before the end of 1945 were estimated to be between 90,000 and 120,000 in Hiroshima (population at the time was 330,000) and between 60,000 and 80,000 in Nagasaki (with a population of about 250,000) and were attributed to traumatic blast injuries, burns, bone marrow depletion, and other physical consequences associated with the exposure. The information available on atomic bombing survivors and their children is highly relevant to the radiation protection policy of the general public (National Research Council, 2005; NCRP, 2009; UNSCEAR, 2006a, b).
The Radiation Effects Research Foundation and its predecessor, the Atomic Bomb Casualty Commission, track the mortality and cancer incidence—among other health effects—of the survivors of the bombings. The LSS cohort consists primarily of about 94,000 survivors of the atomic bombings of Hiroshima and Nagasaki. The cohort includes both a large proportion of survivors who were within 2.5 km of the hypocenters at the time of the bombings and a similar sized sample of survivors who were between 3 and 10 km from the hypocenters and whose radiation doses were almost negligible. Periodic analyses of the LSS mortality data have resulted in a series of reports; the fourteenth report (Ozasa et al., 2012), which covers the period 1950-2003 and includes an additional 6 years of follow-up since the last report of the series (Preston et al., 2003), was recently published.
Although the follow-up of the atomic bombing survivors is often perceived as a high-dose study (exposures 0.5-3 Sv range), about 86 percent of the survivors with estimated doses (i.e., 74,000 persons presenting 11,000 cancer cases) had colon doses under 0.2 Sv (Preston et al., 2007). Demo-graphically, the population is large, and individuals were unselected with respect to sociodemographic or health-related status at the time of the bombings, but in order to be included they must have survived for at least 5 years after the bombings. All ages and both genders of individuals exposed to a wide range of radiation exposure levels are included, permitting a
dose-response analysis. Importantly, estimates of these individual doses are reasonably precise. Additionally, the population has a high rate of mortality and cancer-incidence follow-up. These strengths of the LSS study provide a high-quality, informative epidemiologic study. However, the radiation exposures were acute, received in a matter of seconds, and the population was exposed to a small amount of neutrons and not just gamma rays. Moreover, the fact that the population had to live in a war-torn country where there was malnutrition, poor sanitary conditions, and other severe difficulties makes generalizability of the findings to other populations an issue (Ozasa et al., 2012).
Subcohorts of LSS include the in utero cohort where persons born to mothers pregnant at the time of the bombing and controls are being followed, and the F1 cohort, where children of the exposed and unexposed parents are being followed for disease occurrence. While radiation doses were not directly measured at the time of the bombings, information on survivor locations and shielding were obtained in the early years, which combined with extensive physics calculations of the radiation source and transport have been used to retrospectively estimate the doses received by individual survivors (Cullings et al., 2006).
By the late 1940s, there were suggestions of an increased risk of leukemia among the atomic bombing survivors; the earliest evidence of an increased leukemia was reported in 1952 (Folley et al., 1952). The latest published LSS mortality data for leukemia are through 2000 and a 46 percent excess (93 excess deaths) are attributable to radiation exposure among the survivors to 0.005 Gy (Preston et al., 2004; Richardson et al., 2009). A clear dose-response relationship exists, with 90 percent of the leukemia deaths among those exposed to doses 1 Gy being excess deaths. Separate analyses also indicated strong dose responses for most subtypes of leukemia except chronic lymphocytic leukemia (Preston et al., 1994).
Because the atomic bombing survivors received whole-body exposure from penetrating radiation, a large number of organ sites were affected. An analysis by Preston et al. (2007) on solid cancer incidence in atomic bombing survivors for the period 1958-1998 showed that an excess of 11 percent of solid cancers are attributed to exposures 0.005 Gy (mean 0.23 Gy). The attributable proportion increases with increasing dose and reaches 48 percent among those who received at least 1 Gy. In ranking the sites based on excess cancers observed because of the exposure, the highest relative excess was found for bladder, female breast, and lung cancers, followed by cancers of the central nervous system, ovary, thyroid, colon, and esophagus (Preston et al., 2007). Overall, estimates for solid cancers were 50 percent higher among women, but if female cancers are excluded from the analysis, the estimates by gender are more comparable. Examination of the excess absolute risks (EARs) shows that the number of excess radiationrelated
cancers occurring among males per 10,000 persons per year per Gy is about the same as among females. Excess risks are highly dependent on age at exposure and attained age. The excess relative risk (ERR) for persons exposed to the bombs at a younger age is higher than those exposed to the bombs when they were older, but it declines over time with increasing attained age (or time since exposure). However, the number of excess cancers occurring among 10,000 persons per year per Gy increases with attained age and indicates that radiation risk persists throughout the remaining lifetime. Both the in utero and early childhood groups exhibited statistically significant dose-related increases in incidence rates of solid cancer. At present, not only is there no evidence to support the hypothesis that in utero exposure confers greater adult-cancer risk than childhood exposure, but the risk might be lower (Preston et al., 2008).
Of particular pertinence to this document are the considerations related to risks among the low-dose part of the study population. In the most recent update of cancer incidence there was a statistically significant dose response within the range 0-150 mSv (Preston et al., 2007), suggesting there is dose-related risk even at relatively low dose levels. For cancer mortality, statistically significant upward curvature has been seen, but this is associated primarily with a sublinear degree of risk in the dose range of about 300-800 mSv and not sublinearity at low doses. However, other uncertainties need to be kept in mind in evaluating the low-dose data. First is the fact that some were exposed to residual radiation from neutron activation of soil elements which may have affected those who entered the high-exposure areas in the first few days after the bombings (e.g., in search of missing relatives). Certain areas also received “black rain,” fallout which sometimes may have contained a degree of radioactive elements. There is very little information about who among the atomic bombing survivors may have received such exposures. In addition, the risk estimates may be affected by sociodemographic factors such as rural and urban differences and by selection effects having to do with the hardiness of the survivors of acute radiation effects. (However, the selection effects would more likely apply to high- and moderate-dose survivors than to low-dose survivors.) Because of these uncertainties, plus the other issues of generalizing to protracted exposures and to Western populations, corroborating evidence is needed from other studies to increase certainty in projecting low-dose risks.
A.4.3 Studies of Accidental Releases to Populations
The Chernobyl nuclear power station accident in 1986 in northern Ukraine resulted in the largest accidental release of radionuclides (principally
131I and 137Cs) into the environment in history. Although there was a wide geographic dispersion of radionuclides, the accident had the greatest impact in Belarus, Ukraine, and the Russian Federation. A number of epidemiologic studies have investigated the impact of the Chernobyl accident and cancer risk, and most of the studies have been ecologic, where information on dose and health outcomes is available only at the population level. The radiation effects from the Chernobyl accident are comprehensively summarized in a recent report (UNSCEAR, 2008b). The most notable health consequence of the accident has been the large increase in thyroid cancer among those exposed as children or teenagers. The latency period for thyroid cancer was estimated to be 4-5 years after exposure (Ivanov et al., 2006; Kazakov et al., 1992). The increase in incidence of thyroid cancer was first observed in the early 1990s in Belarus. It is estimated that the thyroids of several thousand children received 131I doses of at least 2 Gy. By 1995, the incidence of childhood thyroid cancer had increased to 4 per 100,000 per year compared to less than 0.05 cases per 100,000 per year prior to the accident (Stsjazhko et al., 1995). For the three most affected countries combined, the increase in incidence rate translated to 5,000 excess thyroid cancer cases in the first 16 years following the accident (Cardis et al., 2005a). A recent study—an update of an earlier report (Tronko et al., 2006)—evaluated the dose-response relationship for incident thyroid cancers using measurement-based individual 131I thyroid dose estimates taken within 2 months after the accident. The 12,000 individuals who were part of the prospective cohort study were 18 years of age at the time of the accident and resided in three contaminated regions of Ukraine. Results suggested that thyroid cancers attributable to 131I exposure continued to occur two decades after the exposure; the estimated ERR for incident thyroid cancer per gray was 1.91 (95% CI = 0.43-6.34) (Brenner et al., 2011). There is some indication that iodine deficiency at the time of exposure to 131I may have increased the risk of developing thyroid cancer; conversely, prolonged iodine dietary supplementation may be protective for the disease (Cardis et al., 2005a).
Data on solid cancers other than thyroid among residents of the affected areas are limited. Among residents of the contaminated region of Kaluga in Russia, no indication of increased incidence or mortality of solid cancers was observed (Ivanov et al., 1997a). Exposure to ionizing radiation is a known risk factor for breast cancer. Pukkala et al. (2006) conducted an ecologic study to describe the trends in breast cancer incidence in Belarus and Ukraine. Despite the evident trends of increased breast cancer incidence due to improvements of diagnosis and registration, the authors showed that during the period 1997-2001, there was a twofold increase in risk in the highly contaminated (average accumulative dose 40 mSv or more) compared to the least contaminated areas.
Whether there is leukemia excess following the accident is much less clear, although several ecologic studies have examined the association between leukemia risk and exposure to radiation from Chernobyl in childhood. For example, the International Program on the Health Effects of the Chernobyl Accident pilot projects study aimed to examine leukemia and lymphoma incidence among populations residing in selected radioactively contaminated areas of the Ukraine, Russia, and Belarus during 1980-1992. Incidence was estimated before and after the Chernobyl accident and a statistically significant increase was observed following the accident (WHO, 1996). However, application of better screening systems and diagnostic procedures could account for the reported increase in incidence. The European Childhood Leukemia-Lymphoma Incidence Study examined trends in leukemia based on cancer registration data from 23 countries among children aged 0-14 years (Parkin et al., 1996). No significant associations with exposure to radiation from Chernobyl were identified. Other studies have not provided consistent evidence for an association (Ivanov et al., 1993, 1996; Noshchenko et al., 2001; Prisyazhiuk et al., 1991) but are limited by dependence on historical and current registration data of varying quality and lack of reliable dosimetry.
A case-control study was conducted to estimate the radiation-induced acute leukemia risk among those aged 0-20 at the time of the Chernobyl accident in Ukraine. Individual estimations of accumulated absorbed radiation dose to the bone marrow were assessed. The period of investigation was 1987-1997. Ninety-eight verified cases were compared to 151 randomly selected controls, matched for age, gender, and administrative region. The mean value of the estimated accumulated equivalent dose to the bone marrow was 4.5 mSv and the maximum was 101 mSv. Analysis showed that males whose estimated radiation exposure was higher than 10 mSv had a threefold higher risk of developing leukemia compared to those exposed to 1.9 mSv or less (Noshchenko et al., 2002). Many of the youngest subjects of the above-mentioned study were also participants of a larger multinational population-based case-control study of acute leukemia diagnosed among children who were in utero or less than 6 years of age at the time of the accident. Confirmed cases of leukemia diagnosed between 1986 and 2000 in Belarus, Russia, and Ukraine were included and compared to the same age, gender, and residence controls. The major findings of the study were that the median radiation doses received by the participants were low (10 mGy), and there was an overall significant increase of leukemia risk with increasing dose, an association that was most evident in Ukraine, apparent in Belarus, and not evident in Russia (Parkin et al., 1996).
A.4.3.2 The Techa River Study
The Techa River cohort of an unselected population of men and women of all ages provides a unique opportunity to evaluate long-term human health risks from low-dose radiation exposures. Between 1949 and 1956, radioactive materials were released into the Techa River as a result of technological processes at the Mayak complex that produced plutonium for the Soviet nuclear weapons program. At the time of the Mayak releases, there were about 30,000 people living in 41 rural villages downstream on the river. This population received both external exposure primarily due to gamma exposure due to proximity to sediments and shoreline, and internal low-dose-rate radiation exposures, the more significant included drinking of water from the river (Degteva et al., 2000; Krestinina et al., 2005, 2007). Enhanced dose reconstruction efforts for individuals of the Techa River cohort were performed. Dosimetry information derived from annual village mean dose estimates that allowed for dose rate in air at the river bank and in residence areas, representative behavior patterns, intake of radionuclides with river water and food, and other factors (Degteva et al., 2000). Results provided clear evidence for radiation-associated increases in cancer mortality risks of the cohort. More specifically, the excess relative risk per gray for deaths from leukemia was 4.2 (95% CI = 1.2, 13). It was estimated that 2.5 percent of the solid cancer deaths and 63 percent of the leukemia deaths were associated with the radiation exposure (Krestinina et al., 2005). Studies on incidence of solid cancers (Krestinina et al., 2007) and leukemia (Ostroumova et al., 2006) in the cohort confirmed the association. More specifically, analysis of 83 leukemia cases diagnosed within the period 1950-1997 and 415 matched controls showed that the ORs per gray of total, external, and internal doses were 4.6, 7.2, and 5.4, respectively.
A.4.4 Studies of Nuclear Workers
Extrapolating results from databases such as that of the LSS to residential settings is problematic due to major differences in magnitude of dose and exposure periods (high-dose acute exposures versus low-dose protracted or fractionated exposures), study group demographics, and health of exposed populations. Studies of cancer risk assessment among workers in the nuclear industry could provide more relevant estimates of the effects of protracted, low-level ionizing radiation exposure. The great advantage of this approach is the availability of well-standardized and generally computerized individual whole-body dosimetry records that provide reliable information as the basis for epidemiologic estimates of radiation-induced cancer risk. The major limitation, however, is the “healthy worker effect,” a concern in occupational epidemiology when health risk factors associated with workers (such as intended selection of more healthy persons for
employment, work-related medical care, higher socioeconomic status) are compared to those of the general population from which the workers are drawn.
The “healthy worker effect” reflects that an individual must be relatively healthy to be employable in a workforce; therefore, both disease and mortality rates are usually lower among workers than in the general population. Moreover, within the workforce studies, healthier workers are more likely to stay employed for longer periods of time than less healthy workers. This may give rise to a healthier occupational cohort (Li and Sung, 1999). There are several comprehensive reviews of the biases related to the comparison of workers and general population that includes selection bias, information bias, and confounding (Li and Sung, 1999; Pearce et al., 2007). An example of the latter is that some health-related behaviors such as smoking are not permitted during the hours of work, and certain personal traits such as obesity may be thought unfit for particular labor forces by industry (Wilcosky and Wing, 1987). Therefore, in view of the deficiency of background risk factors, the possibility of differential effects of ionizing radiation cannot be excluded. Although direct comparisons between the workforce and the general population in relation to the effects of ionizing radiation may be somewhat deceptive, examining the variation of the health outcome across a gradient of increasing exposure within the nuclear industry is very informative. It is worth noting that the healthy worker effect has often been found to be smaller for cancer than for other disease categories.
Workers in the nuclear plants are at potential risk of exposure to ionizing radiation both externally from radioactivity in the working environment and internally from radionuclides which enter the body by inhalation, ingestion, or through accidents that result in percutaneous wounds. The exposures may accumulate over a lifetime to doses of the order of 100 mGy. The possible carcinogenic effects of exposure to external sources of radiation among nuclear workers have been the subject of numerous investigations over the past 20 years. Estimates from these analyses are of limited precision because the sample sizes are small and the follow-up time not sufficiently long (Shore, 1990, 2009). Among white male employees of the Oak Ridge National Laboratory, leukemia mortality rates were 60 percent higher than national rates; however, there was no evidence of a dose-response relationship (Wing et al., 1991). Mortality data among 5,413 workers at the Rocky Flats plutonium weapons facility, although with limited precision, suggested an elevated risk for esophageal, stomach, colon, and prostate cancers among individuals with plutonium body burdens of 2 nCi or greater. No excess risk was reported for cancers of the bone, liver, and lung, the cancer sites most likely to be associated with plutonium exposure (Wilkinson et al., 1987). Combined analyses of mortality workers at
the Hanford Site, Oak Ridge National Laboratory, and Rocky Flats nuclear weapons plants provided no evidence of an association between radiation exposure and mortality from all cancers or from leukemia (Gilbert et al., 1989). The exception was multiple myeloma, which was found to exhibit a statistically significant correlation with radiation exposure. However, the observed association could be due to chance alone.
More recently, Schubauer-Berigan et al. (2007) combined the data from five nuclear facilities in the United States to evaluate leukemia mortality risk from ionizing radiation using a nested case-control study design. The authors reported an adjusted ERR per 10 mSv of 1.44 percent (95% CI = –1.03% - 7.59%). In both reports, the results suggest that risks among nuclear workers are comparable to those observed in populations exposed acutely to high doses. An analysis of observed versus expected mortality of more than 29,000 nuclear workers in France, employed between 1950 and 1994 at two nuclear installations, showed a strong healthy worker effect with an observed 40 percent lower mortality rate among workers than expected from national mortality statistics (Telle-Lamberton et al., 2007). Of the 21 cancer sites examined, a statistically significant excess was observed only for skin melanoma. A significant dose-effect relationship was observed for leukemia after exclusion of chronic lymphoid leukemia (CLL). A larger study of 75,000 employees of the United Kingdom atomic energy authority, the atomic weapons establishment, and the Sellafield plant of British nuclear fuels demonstrated an approximately 20 percent lower all-cause mortality and 4 percent lower cancer associated mortality among workers compared to national rates. A positive association was observed for leukemia risk and exposure to radiation and weaker associations for melanoma and other skin cancers (Carpenter et al., 1994).
A.4.4.1 The Three-Country Study and the 15-Country Study of Nuclear Workers
The three-country study was coordinated by the International Agency for Research on Cancer (IARC) of the World Health Organization (WHO). In the analysis, Cardis and colleagues (1995) found a statistically significant correlation between mortality from leukemia (excluding CLL) and the cumulative individual dose of external radiation. The ERR coefficient was 2.18 (90% CI = 0.13, 5.7) per sievert. Cardis et al. (2005b) extended the IARC study to include countries with nuclear programs such as France and Japan to produce what is probably the largest study to date of cancer in the nuclear workforce. The investigation assessed mortality among workers in 155 nuclear facilities in 15 countries and was conducted to improve the precision of direct estimates of cancer risk following protracted low
doses of ionizing radiation and to advance the scientific basis for radiation protection standards. Analysis included more than 400,000 nuclear workers monitored individually for external radiation and followed up for an average of 12.7 years. The number of workers included in the study is approximately four times greater than in the three-country study. However, as discussed in a recent review, the increase in statistical power is not as great as the number of workers in the cohort may imply, primarily because of the inclusion of workers with low average doses and short periods of follow-up (Wakeford, 2005). About 10 percent of the cohort of workers received external doses exceeding 50 mGy, while 0.1 percent received doses exceeding 500 mGy. Additional problems of the 15-country study include the fact that the results are driven by the contribution of only one country, Canada (Ashmore et al., 2010). The Canadian data are being reexamined for their quality and validity of results. Areas of uncertainty in the 15-country study related to dosimetry, analytical methods, smoking data, and others have been described (Boice, 2010).
Thirty-one cancer types were examined in the 15-country study. A significant association was seen between radiation dose and all-cause mortality (ERR = 0.42 per Sv, 90% CI = 0.07, 0.79); 18,993 deaths were attributed to mortality from all-cancer types (ERR/Sv = 0.97, 90% CI = 0.28, 1.77; 5,233 deaths). Lung cancer was the only cancer to show a statistically significant rise in the risk estimate; however, the association should be interpreted with caution as data on individual smoking characteristics were missing from the analysis. A borderline significant association was found for multiple myeloma. Stratified analysis by duration of employment had a large effect on the ERR/Sv, reflecting a strong healthy worker survivor effect in these cohorts.
A.4.4.2 The British National Registry of Radiation Workers
Perhaps the most precise estimates to date of mortality and cancer risks following occupational radiation exposure come from the third analysis of the British National Registry of Radiation Workers (Muirhead et al., 2009). Two earlier analyses that only looked at mortality data found a strong healthy worker effect and some evidence of an increasing trend in cancer risk (particularly leukemia) with increasing external dose; however, the confidence intervals for the observed trends were wide (Kendall et al., 1992; Muirhead et al., 1999). The third analysis of the series looked at an enlarged cohort of 175,000 workers, adding 9 years of follow-up (87,000 of these workers also were in the 15-country study described above). Due to the higher dose distribution and the larger number of cancers, this study had a greater statistical power than the 15-country study.
Within the cohort, there was evidence of an increasing trend in cancer
mortality with increasing external radiation dose. The trend with dose in the risk of all cancers other than leukemia was maintained when lung cancer was excluded from the analysis, supporting that the trend is not an artifact due to smoking. The cancer risk estimates obtained were consistent with values used to set radiation protection standards.
A.4.4.3 Emergency Chernobyl Workers
Cancer incidence (as opposed to mortality) data among nuclear workers is less available. An analysis has been published of solid cancer incidence rates during an 11-year follow-up (1991-2001) of emergency and cleanup workers after the Chernobyl accident in Russia. These persons worked in the 30-km zone in 1986-1987 and received on average higher doses than those involved in recovery operations in 1988-1990 and have been subject to annual medical checkups (Ivanov et al., 2004). Two control groups were selected for comparison: an “external control” representing age-adjusted incidence rates in Russia and an “internal control” representing emergency workers who were not exposed. The SIR and its 95% CI are similar to that obtained from the Russian population. The values of excess relative risk per unit dose (ERR/Gy) was estimated to be 0.33 (95% CI = –0.39, 1.22) for the follow-up period 1991-2001 and 0.19 (95% CI = –0.66, 1.27) for 1996-2001 compared to the internal control. The authors translate their findings as positive yet statistically insignificant excess of radiogenic solid cancers in the cohort of emergency workers (Ivanov et al., 2004).
Chernobyl recovery operation workers also have theoretically a high risk of developing cancer as a consequence of radioactivity from the accident. However, a number of investigations conducted among recovery workers have not found associations between leukemia incidence and exposure (Ivanov et al., 1997b, 2004). Risk factor analysis among 55 cases of leukemia among Chernobyl emergency workers reported between 1986 and 1995 showed that the risk of developing leukemia was not associated with radiation dose, effective exposure dose rate, or duration of stay in the zone (Konogorov et al., 2000).
A.4.4.4 The Mayak Workers Study
A cohort of about 25,000 Russian nuclear workers who worked at the Mayak plutonium production complex in the period 1948-1972 provides a great opportunity to evaluate cancer risks from exposure to plutonium. These workers were exposed to chronic low-dose-rate external gamma radiation as well as internal (inhaled) plutonium at levels much higher than workers in other countries. For example, for the nearly 11,000 monitored workers hired before 1959, the mean cumulative external dose was 1.2 Gy,
more than an order of magnitude higher than any of the nuclear cohorts described. Leukemia death rates increased significantly with increasing gamma-ray dose (Shilnikova et al., 2003). Excess cancers of the lung, liver, and bone, the organs that receive the largest doses of plutonium, have been described (Gilbert et al., 2000; Koshurnikova et al., 2000). Recent analysis with improved plutonium and external dose estimates verified the increase (Sokolnikov et al., 2008).
A.4.5 Studies of Medical Exposures to Radiation
Diagnostic and therapeutic radiation has been used in medicine for over a century. The continuing improvements in diagnostic imaging and radiotherapy as well as the aging of the population have led to greater use of medical radiation (Ron, 2003). Epidemiologic studies of persons exposed to radiation for medical reasons have provided unique opportunities in understanding the risks associated with fractionated radiation exposure. Additionally, medical records often contain information on a patient’s personal past medical history as well as on demographic data and information on personal habits such as smoking, alcohol drinking, and medications. On the negative side, because of their possible underlying disease, patients may have different sensitivity to the radiogenic effects compared to a somewhat healthy population. Other concurrent treatments can affect radiation risks and it can prove difficult to untangle the impact of those different factors. Also, because patients come back for follow-up, other diseases are more likely to be detected and reported, leading to overrepresentation of diseases on this group compared to the general population (Ron, 2002).
A recent report from the NCRP entitled “Ionizing Radiation Exposure of the Population of the United States” indicated that in 2006, people in the United States were exposed to more than seven times as much ionizing radiation from medical diagnostic procedures than in 1980; the increase is fueled largely by the use of CT scans (NCRP, 2009). In 2006, over 67 million scans were performed, 4 to 7 million in children, and many patients receive multiple scans.
Diagnostic exposures are typically characterized by fairly low doses to individual patients (effective doses are typically in the range 0.1-10 mSv), sufficient to provide the required medical information. Because doses are typically low, their effects are difficult to study unless multiple examinations are performed. For example, an excess risk of breast cancer has been reported among women with tuberculosis who had multiple chest fluoroscopies (Delarue et al., 1975; Miller et al.,1989), women treated for benign breast disease (Mattsson et al., 1993), as well as among scoliosis patients who had frequent diagnostic x-rays during their late childhood and adolescence
(Doody et al., 2000). The potential risk attributed to mammography screening programs and understanding the balance between the number of breast cancer deaths induced and breast cancer deaths prevented continues to be an issue of debate especially when extended to women under the age of 50 (de Gelder et al., 2011; Hellquist et al., 2011). Exposure to diagnostic radiography in utero has been associated with increased risk of childhood cancer, particularly leukemia (Linet et al., 2009; Rajaraman et al., 2011; Wakeford, 2008).
In contrast to diagnostic radiation doses, therapeutic doses are much higher and precisely delivered to the targeted area such as the tumor (doses can be as high as 40 Gy or more) (Gilbert, 2009; UNSCEAR, 2008a) aiming to produce cell killing. Physicians need to consider the risks of the treatment against the potential benefits. Overall more than 100 studies of patients receiving diagnostic or therapeutic radiation have evaluated the potential risks and have been comprehensively reviewed elsewhere (Gilbert, 2009; NRC, 2005). Briefly, an association between leukemia and medical radiation exposure was first identified in a study of ankylosing spondylitis patients more than 50 years ago. Since then, leukemia has been linked with many medically exposed persons primarily adults (UNSCEAR, 2008a).
A.4.6 Exposure of the Offspring
Radiation could increase cancer risk of the offspring through parental preconception exposures that potentially cause germline mutations, or by in utero exposure of the fetus to radiation, which may cause somatic mutations.
A.4.6.1 Parental Preconception Exposure
Heritable mutations are particularly concerning, especially among women, as their oocytes are fixed at birth. A study in Sweden investigated, among other outcomes, risk of childhood malignancies in the offspring of women exposed to therapeutic radiation for treatment of skin hemangioma, when 18 months or less (Kallen et al., 1998). The mean ovarian dose was 6 cGy and the maximum was 8.6 Gy. No increase in childhood malignancies was detected. Similar results were obtained from a collaborative study from five countries: Denmark, Finland, Iceland, Norway, and Sweden, which included cancer survivors diagnosed when they were less than 20 years old (Sankila et al., 1998). Results from maternal or paternal radiation exposure from medical diagnostic procedures before conception were not associated with childhood cancer in some (Patton et al., 2004) but were in other studies (Graham et al., 1966; Shu et al., 1994a, b). Comprehensive
studies of the children of cancer survivors exposed to high-dose radiotherapy and chemotherapy provide no evidence for heritable diseases (Signorello et al., 2012; Winther et al., 2012).
In Section A.4.1 we discussed the rejection of the hypothesis—known as the Gardner hypothesis, named after the investigator (Gardner et al., 1990)—that nuclear radiation exposure during work may have an effect on a father’s germ cells, producing genetic changes in sperm that may be leukemogenic in the offspring (Draper et al., 1997; Kinlen et al., 1993; McLaughlin et al., 1993b; Pobel and Viel, 1997). Even in the offspring of male atomic bombing survivors in Hiroshima and Nagasaki, no increase in childhood cancer risk was observed (Izumi et al., 2003; Schull and Neel, 1959). A study examined the childhood cancer in the offspring of radiologic technologists in the United States, born in 1921-1984. Testis or ovary doses were estimated by undertaking a comprehensive dose reconstruction using work history data, badge dose data, and literature doses. No convincing evidence of an increased risk of childhood cancer in the offspring of ra-diologic technologists in association with parental occupational radiation exposure either preconception or in utero was found.
A.4.6.2 In Utero Exposure
A historic study, now known as the Oxford Survey of Childhood Cancers, was the first large study of in utero exposure to low doses of ionizing radiation (1-10 cGy) from diagnostic radiography and risk of childhood cancer. The study examined more than 15,000 case-control pairs and showed an approximately 50 percent increase in the frequency of childhood cancer among the exposed (Stewart et al., 1956). A consistent association has been found in many case-control studies; however, it is not universally accepted that the relationship is causal and not the effect of bias or confounding. Many people think that the observed association is the result of recall bias; mothers of the children who died of the disease would be more motivated to recall in detail the number of medical examinations they undertook during pregnancy, compared to the mothers of healthy children. It was not until later that a study in the United States that relied on hospital records rather than on mother’s memory reported similar findings (MacMahon, 1962) that the results were taken seriously. Others believed that the relationship is due to confounding with some aspect of pregnancy that had given rise to the need for radiographic examinations itself. However, the theory was rejected when reanalysis of published data from the Oxford Childhood Cancer Survey showed that the frequency of leukemia and of solid cancers in childhood is greater following antenatal x-radiography, not only in singleton births but also in twins. The radiography rate for singletons and twins differed and was 10 and 55 percent, respectively, as
mothers of twins are x-rayed to determine fetal position before delivery, and not necessarily because of any illness or condition. A similar excess of leukemia and of solid cancers in the x-rayed with such different rates of radiography was strong evidence for irradiation as the cause (Mole, 1974). In support of a causal relationship is the demonstrated increase in risk with the increase in number of x-ray films used during the examination (Bithell and Stewart, 1975); the reduction in risk over time with reduction in fetal dose (Bithell and Stiller, 1988); and animal experiments that show the fetus to be susceptible to the induction of cancer by radiation. Based on the review of the evidence, it was concluded that “radiation doses of the order of 10 mGy received by the fetus in utero produce a consequent increase in the risk of childhood cancer. The excess absolute risk coefficient at this level of exposure is approximately 6% per gray” (Doll and Wakeford, 1997). Under the assumption that the relationship between in utero exposure to medical imaging and cancer is causal, the medical profession has in large part replaced x-rays by ultrasounds.
A reason for doubt of a causal relationship between cancer risk in childhood following prenatal exposure to ionizing radiation is the lack of evidence of a corresponding increased risk in cohort studies, most notably the atomic bombing survivors. Observations of those exposed in utero following the atomic bombings have been published since 1970. Possibly due to the small number of observed cancers, a dose-related increase in cancer mortality before age 15 could not be demonstrated (Jablon and Kato, 1970; Kato, 1971). More specifically, during the period 1950-1984, among atomic bombing survivors exposed in utero, there were only 18 cancer cases; 5 of them were in the “zero-dose” group. Two of these subjects developed childhood cancer and all the others developed cancer in adulthood. At present, there is no evidence to support the hypothesis that in utero exposure confers greater adult-cancer risk than childhood exposure (Preston et al., 2008).
An additional reason for doubt of a causal relationship is the unusual homogeneity of the relative risk of all childhood cancers in the Oxford Survey of Childhood Cancers. Regardless of the type of malignancy (i.e., childhood brain cancer, leukemia, neuroblastoma, Wilms tumor), the relative risks were consistent to a 40 to 50 percent increase in risk (Boice and Miller, 1999). Furthermore, in questioning the biological plausibility of increased cancer risk in childhood following prenatal exposure to ionizing radiation is whether embryonic tumors such as Wilms tumor and neuroblastoma could be induced by exposures that occurred primarily just before birth during pelvimetry in the measurement of the birth canal. These issues are sufficiently important to raise doubts as to the causal nature of the association and the ICRP in their most recent review concluded that the evidence for solid tumors, and in particular childhood brain cancer, was not strong (ICRP, 2003).
A.4.7 Noncancer Diseases and Radiation
The atomic bombing survivor studies and specifically the Adult Health Study is the principle source for information on diseases other than cancer related to radiation exposure. This is particularly true as there are no population-based disease incidence registries other than cancer.
A.4.7.1 Cardiovascular Diseases
The issue of radiation-induced cardiac damage has been demonstrated in studies of breast cancer and Hodgkin’s lymphoma patients that received high-dose therapeutic radiation (>30-40 Gy) (Adams et al., 2003; Senkus-Konefka and Jassem, 2007). These patients have a life-long increased risk of fatal cardiovascular events. Data from the Japanese survivors demonstrated for the first time that subtherapeutic doses (5 Gy) can also be associated with cardiovascular disease (Preston et al., 2003; Shimizu et al., 1992). A recent report indicated an excess relative risk of 14 percent per Sv (95% CI = 6%-23%) with an essentially linear dose response (Shimizu et al., 2010). However, there was substantial uncertainty in the amount of cardiovascular disease risk at doses under 0.5 Sv. Outside the atomic bombing studies, there is mixed epidemiologic evidence to support the notion that exposure to low doses of ionizing radiation increases risk of cardiovascular diseases (Little et al., 2008b, 2010; McGale and Darby; 2005; UNSCEAR, 2006b).
Posterior subcapsular or cortical cataracts are characteristic of radiation exposure. Cataracts were observed in survivors that received high doses of radiation within 3-4 years after the bombings in Hiroshima and Nagasaki (Cogan et al., 1949). More recent studies have shown an excess of opacities and cataracts at lower doses to the lens, both in the atomic bombing study (Nakashima et al., 2006; Neriishi et al., 2007) and in Chernobyl cleanup workers who received protracted radiation exposures (Worgul et al., 2007). Those studies suggest there may be a threshold for opacity effects at approximately 0.5 Sv.
A.4.7.3 Thyroid Diseases and Hyperparathyroidism
Nonmalignant thyroid diseases have been examined among those exposed as children or young adults as a result of fallout from the Chernobyl nuclear power plant accident in Ukraine (Zablotska et al., 2002). A significant
but small association between 131I thyroid dose estimates and prevalent subclinical hypothyroidism with an excess estimated odds ratio per Gray of 0.10 (95% CI = 0.03-0.21) was observed in this cohort.
Together with thyroid cancer, the Hanford Thyroid Disease study examined risks associated with nonmalignant thyroid diseases such as benign thyroid nodules, thyroid nodules, autoimmune thyroiditis, and hypothyroidism. The study provided no evidence of an increase in any of the outcomes measured (Davis et al., 2004).
A study evaluated the prevalence of thyroid diseases and their radiation dose responses in atomic bombing survivors, some 55 years after the bombings. A significant linear radiation dose response for thyroid nodules (malignant and benign) was observed with an excess relative risk of 2.01 per Gray (Imaizumi et al., 2006). The prevalence of hyperparathyroidism was found to increase with an estimated excess relative risk of 3.1 at 1 Gy in the atomic bombing study (Fujiwara et al., 1992) and an excess relative risk of 1.1 at 1 Gy in a follow-up of those with medical irradiation in Chicago (Schneider et al., 1995); however, it was not clear whether there is an effect at low doses.
A.4.7.4 Neurological Effects
High doses of radiation to those with prenatal exposure to the atomic bombing were shown to increase the risk of mental retardation and decrements in intelligence (IQ) more generally (ICRP, 2003; Otake et al., 1996), but were limited to those exposed between 8 and 25 weeks of gestation. A review of the data by the ICRP concluded that there were dose thresholds for these effects of 300 mSv or greater for mental retardation and 100 mSv or greater for IQ (ICRP, 2003). Other related effects seen among those exposed during 8-25 weeks of gestation were diminished school performance and increased episodes of neurological seizures (Dunn et al., 1990; ICRP, 2003).
A.4.7.5 Life-Span Shortening
Life-span shortening provides an index that integrates a variety of possible adverse effects of ionizing radiation and has been seen in animal-model studies at high doses of several sieverts. A study of atomic bombing survivors indicated small amounts of life-span shortening at doses below 1 Sv, but proportionately more at higher doses. About 70 percent of the lifespan shortening was due to excess cancer risk (Cologne and Preston, 2000).
Adams, M. J., P. H. Hardenbergh, et al. (2003). Radiation-associated cardiovascular disease. Crit Rev Oncol Hematol 45(1):55-75.
Angell, M. (1989). Negative studies. N Engl J Med 321(7):464-466.
Ashmore, J. P., N. E. Gentner, and R. V. Osborne (2010). Incomplete data on the Canadian cohort may have affected the results of the study by the International Agency for Research on Cancer on the radiogenic cancer risk among nuclear industry workers in 15 countries. J Radiol Prot 30:121-129.
Austin, S. G. (1986). A Study of the Health Experience of Residents of Uravan, Colorado. Final Report. Fort Collins, CO: Austin Health Consultants, Inc.
Baker, P. J., and D. G. Hoel (2007). Meta-analysis of standardized incidence and mortality rates of childhood leukaemia in proximity to nuclear facilities. Eur J Cancer Care (Engl) 16(4):355-363.
Baron, J. A. (1984). Cancer mortality in small areas around nuclear facilities in England and Wales. Br J Cancer 50(6):815-824.
Barton, C. J., E. Roman, et al. (1985). Childhood leukaemia in West Berkshire. Lancet 2(8466):1248-1249.
Bennet, G. (1970). Bristol floods 1968. Controlled survey of effects on health of local community disaster. Br Med J 3(5720):454-458.
Bithell, J. F., and A. M. Stewart (1975). Pre-natal irradiation and childhood malignancy: A review of British data from the Oxford Survey. Br J Cancer 31(3):271-287.
Bithell, J. F., and C. A. Stiller (1988). A new calculation of the carcinogenic risk of obstetric X-raying. Stat Med 7(8):857-864.
Bithell, J. F., S. J. Dutton, et al. (1994). Distribution of childhood leukaemias and non-Hodgkin’s lymphomas near nuclear installations in England and Wales. BMJ 309(6953): 501-505.
Bithell, J. F., T. J. Keegan, et al. (2008). Childhood leukaemia near British nuclear installations: Methodological issues and recent results. Radiat Prot Dosimetry 132(2):191-197.
Bithell, J. F., T. J. Keegan, M. E. Kroll, M. F. Murphu, and T. J. Vincent (2010). Response to letter to the editor. Radiat Prot Dosimetry 138:89-91.
Black, D. (1984). Investigation of the possible increased incidences of cancer in West Cumbria. London, United Kingdom, Her Majesty’s Stationary office.
Black, R. J., J. D. Urquhart, et al. (1992). Incidence of leukaemia and other cancers in birth and schools cohorts in the Dounreay area. BMJ 304(6839):1401-1405.
Black, R. J., L. Sharp, et al. (1994). Leukaemia and non-Hodgkin’s lymphoma: Incidence in children and young adults resident in the Dounreay area of Caithness, Scotland in 1968-91. J Epidemiol Community Health 48(3):232-236.
Boice, J. D., Jr. (2010). Uncertainties in studies of low statistical power (Editorial). J Radiol Prot 30:115-120.
Boice, J. D., Jr, and R. W. Miller (1999). Childhood and adult cancer after intrauterine exposure to ionizing radiation. Teratology 59:227-233.
Boice, J. D., Jr., W. L. Bigbee, et al. (2003a). Cancer incidence in municipalities near two former nuclear materials processing facilities in Pennsylvania. Health Phys 85(6):678-690.
Boice, J. D., Jr., W. L. Bigbee, et al. (2003b). Cancer mortality in counties near two former nuclear materials processing facilities in Pennsylvania, 1950-1995. Health Phys 85(6):691-700.
Boice, J. D., Jr., M. T. Mumma, et al. (2005). Childhood cancer mortality in relation to the St Lucie nuclear power station. J Radiol Prot 25(3):229-240.
Boice, J. D., Jr., M. T. Mumma, et al. (2006). Cancer mortality among populations residing in counties near the Hanford site, 1950-2000. Health Phys 90(5):431-445.
Boice, J. D., Jr., M. T. Mumma, et al. (2007a). Cancer and noncancer mortality in populations living near uranium and vanadium mining and milling operations in Montrose County, Colorado, 1950-2000. Radiat Res 167(6):711-726.
Boice, J. D. Jr., S. S. Cohen, M. T. Mumma, B. Chadda, and W. J. Blot (2007b). Mortality among residents of Uravan, Colorado who lived near a uranium mill, 1936-1984. J Radiol Prot 27:299-319.
Boice, J. D., Jr., W. L. Bigbee, et al. (2009). Cancer incidence in municipalities near two former nuclear materials processing facilities in Pennsylvania—an update. Health Phys 96(2):118-127.
Boice, J. D. Jr., M. T. Mumma, and W. J. Blot (2010). Cancer incidence and mortality in populations living near uranium milling and mining operations in Grants, New Mexico, 1950-2004. Radiat Res 174:624-636.
Boutou, O., A. V. Guizard, et al. (2002). Population mixing and leukaemia in young people around the La Hague nuclear waste reprocessing plant. Br J Cancer 87(7):740-745.
Brenner, A. V., M. D. Tronko, M. Hatch, T. I. Bogdanova, V. A. Oliynik, J. H. Lubin, L. B. Zablotska, V. P. Tereschenko, R. J. McConnell, G. A. Zamotaeva, P. O’Kane, A. C. Bouville, L. V. Chaykovskaya, E. Greenebaum, I. P. Paster, V. M. Shpak, and E. Ron (2011). I-131 dose response for incident thyroid cancers in Ukraine related to the Chornobyl accident. Environ Health Perspect 119(7):933-939.
Brooks, A. L. (1999). Biomarkers of exposure, sensitivity and disease. Int J Radiat Biol 75(12):1481-1503.
Brooks, A. L. (2011). Is a dose dose-rate effectiveness factor (DDREF) needed following exposure to low total radiation doses delivered at low dose-rates? Health Phys 100(3):262.
Busby, C., and M. S. Cato (1997). Death rates from leukaemia are higher than expected in areas around nuclear sites in Berkshire and Oxfordshire. BMJ 315(7103):309.
Cardis, E., et al. (1995). 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
Cardis, E., A. Kesminiene, et al. (2005a). Risk of thyroid cancer after exposure to 131I in childhood. J Natl Cancer Inst 97(10):724-732.
Cardis, E., M. Vrijheid, et al. (2005b). Risk of cancer after low doses of ionising radiation: Retrospective cohort study in 15 countries. BMJ 331(7508):77.
Carnes, B. A., and T. E. Fritz (1991). Responses of the beagle to protracted irradiation. I. Effect of total dose and dose rate. Radiat Res 128(2):125-132.
Carnes, B. A., S. J. Olshansky, et al. (1998). An interspecies prediction of the risk of radiation-induced mortality. Radiat Res 149(5):487-492.
Carpenter, L., C. Higgins, et al. (1994). Combined analysis of mortality in three United Kingdom nuclear industry workforces, 1946-1988. Radiat Res 138(2):224-238.
Cheng, G. H., N. Wu, et al. (2010). Increased levels of p53 and PARP-1 in EL-4 cells probably related with the immune adaptive response induced by low dose ionizing radiation in vitro. Biomed Environ Sci 23(6):487-495.
Clapp, R. W., S. Cobb, et al. (1987). Leukaemia near Massachusetts nuclear power plant. Lancet 2(8571):1324-1325.
Clavel, J., and D. Hemon (1997). Leukaemia near La Hague nuclear plant. Bias could have been introduced into study. BMJ 314(7093):1553; author reply 1555.
Cogan, D. G., S. F. Martin, et al. (1949). Atom bomb cataracts. Science 110(2868):654.
Cologne, J. B., and D. L. Preston (2000). Longevity of atomic-bomb survivors. Lancet 356(9226):303-307.
COMARE (Committee on Medical Aspects of Radiation in the Environment) (1988). Second Report. Investigation of the Possible Increased Incidence of Leukaemia in Young People near the Dounreay Nuclear Establishment Caithness, Scotland. London: HMSO.
COMARE (1989). Third Report. Report on the Incidence of Childhood Cancer in the West Berkshire and North Hampshire area, in Which Are Situated the Atomic Weapons Research Establishment, Aldermaston and the Royal Ordance Factory, Burghfield. London: HMSO.
COMARE (1996). Fourth Report. The Incidence of Cancer and Leukaemia in Young People in the Vicinity of the Sellafield Site, West Cumbria; Further Studies and an Update of the Situation Since the Publication of the Report of the Black Advisory Group in 1984. London: Department of Health.
COMARE (2005). Tenth Report: The Incidence of Childhood Cancer Around Nuclear Installations in Great Britain. London: Department of Health.
COMARE (2011). Fourtheenth report: Further Consideration of the Incidence of Childhood Leukemia Around Nuclear Power Plants in Great Britain. London: Department of Health.
Cook-Mozaffari, P. J., S. C. Darby, et al. (1989a). Geographical variation in mortality from leukaemia and other cancers in England and Wales in relation to proximity to nuclear installations, 1969-78. Br J Cancer 59(3):476-485.
Cook-Mozaffari, P., S. Darby, et al. (1989b). Cancer near potential sites of nuclear installations. Lancet 2(8672):1145-1147.
Crump, K. S., T. H. Ng, et al. (1987). Cancer incidence patterns in the Denver metropolitan area in relation to the Rocky Flats plant. Am J Epidemiol 126(1):127-135.
Cullings, H. M., S. Fujita, et al. (2006). Dose estimation for atomic bomb survivor studies: its evolution and present status. Radiat Res 166(1 Pt 2):219-254.
Davis, S., K. J. Kopecky, T. E. Hamilton, and L. Onstad (Hanford Thyroid Disease Study Team) (2004). Thyroid neoplasia, autoimmune thyroiditis, and hypothyroidism in persons exposed to iodine 131 from the hanford nuclear site. JAMA 292:2600-2613.
de Gelder, R., G. Draisma, et al. (2011). Population-based mammography screening below age 50: balancing radiation-induced vs prevented breast cancer deaths. Br J Cancer 104(7):1214-20
Degteva, M. O., M. I. Vorobiova, et al. (2000). Dose reconstruction system for the exposed population living along the Techa River. Health Phys 78(5):542-554.
Delarue, N. C., G. Gale, et al. (1975). Multiple fluoroscopy of the chest: Carcinogenicity for the female breast and implications for breast cancer screening programs. Can Med Assoc J 112(12):1405-1413.
Doll, R., and R. Wakeford (1997). Risk of childhood cancer from fetal irradiation. Br J Radiol 70:130-139.
Doll, R., H. J. Evans, et al. (1994). Paternal exposure not to blame. Nature 367(6465): 678-680.
Doody, M. M., J. E. Lonstein, et al. (2000). Breast cancer mortality after diagnostic radiography: Findings from the U.S. Scoliosis Cohort Study. Spine (Phila Pa 1976) 25(16): 2052-2063.
Dousset, M. (1989). Cancer mortality around La Hague nuclear facilities. Health Phys 56(6): 875-884.
Draper, G. J., and T. J. Vincent (1997). Death rates from childhood leukaemia near nuclear sites. Findings were probably due to chance fluctuations in small numbers of deaths. BMJ 315(7117):1233; author reply 1234.
Draper, G. J., C. A. Stiller, et al. (1993). Cancer in Cumbria and in the vicinity of the Sellafield nuclear installation, 1963-90. BMJ 306(6870):89-94.
Draper, G. J., M. P. Little, et al. (1997). Cancer in the offspring of radiation workers: A record linkage study. BMJ 315(7117):1181-1188.
Dunn, K., H. Yoshimaru, et al. (1990). Prenatal exposure to ionizing radiation and subsequent development of seizures. Am J Epidemiol 131(1):114-123.
Enstrom, J. E. (1983). Cancer mortality patterns around the San Onofre nuclear power plant, 1960-1978. Am J Public Health 73(1):83-92.
Evrard, A. S., D. Hemon, et al. (2006). Childhood leukaemia incidence around French nuclear installations using geographic zoning based on gaseous discharge dose estimates. Br J Cancer 94(9):1342-1347.
Ewings, P. D., C. Bowie, et al. (1989). Incidence of leukaemia in young people in the vicinity of Hinkley Point nuclear power station, 1959-86. BMJ 299(6694):289-293.
Folley, J. H., W. Borges, et al. (1952). Incidence of leukemia in survivors of the atomic bomb in Hiroshima and Nagasaki, Japan. Am J Med 13(3):311-321.
Forman, D., P. Cook-Mozaffari, et al. (1987). Cancer near nuclear installations. Nature 329(6139):499-505.
Fujiwara, S., R. Sposto, et al. (1992). Hyperparathyroidism among atomic bomb survivors in Hiroshima. Radiat Res 130(3):372-378.
Gaillard, S., D. Pusset, et al. (2009). Propagation distance of the alpha-particle-induced bystander effect: The role of nuclear traversal and gap junction communication. Radiat Res 171(5):513-520.
Gardner, M. J., M. P. Snee, et al. (1990). Results of case-control study of leukaemia and lymphoma among young people near Sellafield nuclear plant in West Cumbria. BMJ 300(6722):423-429.
Garssen, B. (2004). Psychological factors and cancer development: Evidence after 30 years of research. Clin Psychol Rev 24(3):315-338.
Gilbert, E. S. (2009). Ionising radiation and cancer risks: What have we learned from epidemiology? Int J Radiat Biol 85(6):467-482.
Gilbert, E. S., S. A. Fry, et al. (1989). Analyses of combined mortality data on workers at the Hanford Site, Oak Ridge National Laboratory, and Rocky Flats Nuclear Weapons Plant. Radiat Res 120(1):19-35.
Gilbert, E. S., N. A. Koshurnikova, et al. (2000). Liver cancers in Mayak workers. Radiat Res 154(3):246-252.
Goldsmith, J. R. (1989). Childhood leukaemia mortality before 1970 among populations near two US nuclear installations. Lancet 1(8641):793.
Goldsmith, J. R. (1992). Nuclear installations and childhood cancer in the UK: Mortality and incidence for 0-9-year-old children, 1971-1980. Sci Total Environ 127(1-2):13-35; discussion 43-55.
Graham, S., M. L. Levin, et al. (1966). Preconception, intrauterine, and postnatal irradiation as related to leukemia. Natl Cancer Inst Monogr 19:347-371.
Greiser, E. (2009). Leukämie-Erkrankungen bei Kindern und Jugendlichen in der Umgebung von Kernkraftwerken in fünf Ländern Meta-Analyse und Analyse [Leukaemia in children and young people in the vicinity of nuclear power stations in five countries. Meta-analyses and analyses.] Commissioned by the Bundesfraktion B’90/The Greens: MUSAweiler. Available at http://www.ippnw.de/commonFiles/pdfs/Atomenergie/090904-Metanalyse-Greiser.pdf.
Grosche, B., D. Lackland, et al. (1999). Leukaemia in the vicinity of two tritium-releasing nuclear facilities: a comparison of the Kruemmel Site, Germany, and the Savannah River Site, South Carolina, USA. J Radiol Prot 19(3):243-252.
Guizard, A. V., A. Spira, et al. (1997). [Incidence of leukemias in people aged 0 to 24 in north Cotentin]. Rev Epidemiol Sante Publique 45(6):530-535.
Guizard, A. V., O. Boutou, et al. (2001). The incidence of childhood leukaemia around the La Hague nuclear waste reprocessing plant (France): A survey for the years 1978-1998. J Epidemiol Community Health 55(7):469-474.
Hatch, M., M. Susser, et al. (1997). Comments on A reevaluation of cancer incidence near the Three Mile Island nuclear plant. Environ Health Perspect 105(1):12.
Hatch, M. C., J. Beyea, et al. (1990). Cancer near the Three Mile Island nuclear plant: Radiation emissions. Am J Epidemiol 132(3):397-412; discussion 413-397.
Hatch, M. C., S. Wallenstein, et al. (1991). Cancer rates after the Three Mile Island nuclear accident and proximity of residence to the plant. Am J Public Health 81(6):719-724.
Hattchouel, J. M., A. Laplanche, et al. (1995). Leukaemia mortality around French nuclear sites. Br J Cancer 71(3): 651-653.
Heasman, M. A., I. W. Kemp, et al. (1986). Childhood leukaemia in northern Scotland. Lancet 1(8475):266.
Heinavaara, S., S. Toikkanen, et al. (2010). Cancer incidence in the vicinity of Finnish nuclear power plants: an emphasis on childhood leukemia. Cancer Causes Control 21(4):587-595.
Hellquist, B. N., S. W. Duffy, et al. (2011). Effectiveness of population-based service screening with mammography for women ages 40 to 49 years: evaluation of the Swedish Mam-mography Screening in Young Women (SCRY) cohort. Cancer 117(4):714-722.
Hill, C., and A. Laplanche (1990). Overall mortality and cancer mortality around French nuclear sites. Nature 347(6295):755-757.
Hoffman, F. O., A. J. Ruttenber, A. I. Apostoaei, R. J. Carroll, and S. Greenland (2007). The Hanford Thyroid Disease Study: An alternative view of the findings. Health Phys 92(2):99-111.
Hoffmann, W., H. Dieckmann, et al. (1997). A cluster of childhood leukemia near a nuclear reactor in northern Germany. Arch Environ Health 52(4):275-280.
Hoffmann, W., C. Terschueren, et al. (2007). Childhood leukemia in the vicinity of the Geesthacht nuclear establishments near Hamburg, Germany. Environ Health Perspect 115(6):947-952.
Hoffmann, W., C. Terschueren, et al. (2008). Population-based research on occupational and environmental factors for leukemia and non-Hodgkin’s lymphoma: The Northern Germany Leukemia and Lymphoma Study (NLL). Am J Ind Med 51(4):246-257.
ICRP (International Commission on Radiological Protection) (2003). Biological Effects after Prenatal Irradiation (Embryo and Fetus). ICRP Publication 90. Ann. ICRP 33(1-2).
ICRP (2007). The 2007 Recommendations of the International Commission on Radiological Protection. ICRP Publication 103. Ann. ICRP 37(2-4).
Imaizumi, M., T. Usa, et al. (2006). Radiation dose-response relationships for thyroid nodules and autoimmune thyroid diseases in Hiroshima and Nagasaki atomic bomb survivors 55-58 years after radiation exposure. JAMA 295(9):1011-1022.
Ivanov, E. P., G. Tolochko, et al. (1993). Child leukaemia after Chernobyl. Nature 365(6448): 702.
Ivanov, E. P., G. V. Tolochko, et al. (1996). Childhood leukemia in Belarus before and after the Chernobyl accident. Radiat Environ Biophys 35(2):75-80.
Ivanov, V. K., A. F. Tsyb, et al. (1997a). Cancer risks in the Kaluga oblast of the Russian Federation 10 years after the Chernobyl accident. Radiat Environ Biophys 36(3):161-167.
Ivanov, V. K., A. F. Tsyb, et al. (1997b). Leukaemia and thyroid cancer in emergency workers of the Chernobyl accident: estimation of radiation risks (1986-1995). Radiat Environ Biophys 36(1):9-16.
Ivanov, V. K., A. I. Gorski, et al. (2004). Solid cancer incidence among the Chernobyl emergency workers residing in Russia: Estimation of radiation risks. Radiat Environ Biophys 43(1):35-42.
Ivanov, V. K., A. I. Gorski, et al. (2006). Radiation-epidemiological studies of thyroid cancer incidence among children and adolescents in the Bryansk oblast of Russia after the Chernobyl accident (1991-2001 follow-up period). Radiat Environ Biophys 45(1):9-16.
Izumi, S., K. Koyama, et al. (2003). Cancer incidence in children and young adults did not increase relative to parental exposure to atomic bombs. Br J Cancer 89(9):1709-1713.
Jablon, S., and H. Kato (1970). Childhood cancer in relation to prenatal exposure to atomic-bomb radiation. Lancet 2(7681):1000-1003.
Jablon, S., Z. Hrubec, J. D. Boice Jr., and B. J. Stone (1990), Cancer in Populations Living near Nuclear Facilities, Vols. 1-3. NIH Publication No. 90-874.
Jablon, S., Z. Hrubec, et al. (1991). Cancer in populations living near nuclear facilities. A survey of mortality nationwide and incidence in two states. JAMA 265(11):1403-1408.
Jacob, P., W. Rühm, L. Walsh, M. Blettner, G. Hammer, and H. Zeeb (2009). Is cancer risk of radiation workers larger than expected?, Occup Environ Med 66(12):789-796.
Janerich, D. T., A. D. Stark, et al. (1981). Increased leukemia, lymphoma, and spontaneous abortion in Western New York following a flood disaster. Public Health Rep 96(4):350-356.
Kaatsch, P., U. Kaletsch, et al. (1998). An extended study on childhood malignancies in the vicinity of German nuclear power plants. Cancer Causes Control 9(5):529-533.
Kaatsch, P., C. Spix, et al. (2008). Leukaemia in young children living in the vicinity of German nuclear power plants. Int J Cancer 122(4):721-726.
Kallen, B., P. Karlsson, et al. (1998). Outcome of reproduction in women irradiated for skin hemangioma in infancy. Radiat Res 149(2):202-208.
Kato, H. (1971). Mortality in children exposed to the A-bombs while in utero, 1945-1969. Am J Epidemiol 93(6):435-442.
Kazakov, V. S., E. P. Demidchik, et al. (1992). Thyroid cancer after Chernobyl. Nature 359(6390):21.
Kemenu, J. G., B. Babbitt, et al. (1979). Report of the President’s commission on the accident at three mile island—the need for change: The legacy at TMI. Washington, DC: U.S. Government Printing Office.
Kendall, G. M., C. R. Muirhead, et al. (1992). Mortality and occupational exposure to radiation: First analysis of the National Registry for Radiation Workers. BMJ 304(6821): 220-225.
Kinlen, L. (2011a). Childhood leukaemia, nuclear sites, and population mixing. Br J Cancer 104(1):12-18.
Kinlen, L. (2011b). A German storm affecting Britain: Childhood leukaemia and nuclear power plants. J Radiol Prot 31(3):279-284.
Kinlen, L. J., F. O’Brien, et al. (1993). Rural population mixing and childhood leukaemia: Effects of the North Sea oil industry in Scotland, including the area near Dounreay nuclear site. BMJ 306(6880):743-748.
Kinlen, L. J., M. Dickson, et al. (1995). Childhood leukaemia and non-Hodgkin’s lymphoma near large rural construction sites, with a comparison with Sellafield nuclear site. BMJ 310(6982):763-768.
Konogorov, A. P., V. K. Ivanov, et al. (2000). A case-control analysis of leukemia in accident emergency workers of Chernobyl. J Environ Pathol Toxicol Oncol 19(1-2):143-151.
Koshurnikova, N. A., E. S. Gilbert, et al. (2000). Bone cancers in Mayak workers. Radiat Res 154(3):237-245.
Krestinina, L. Y., D. L. Preston, et al. (2005). Protracted radiation exposure and cancer mortality in the Techa River Cohort. Radiat Res 164(5):602-611.
Krestinina, L. Y., F. Davis, et al. (2007). Solid cancer incidence and low-dose-rate radiation exposures in the Techa River cohort: 1956-2002. Int J Epidemiol 36(5):1038-1046.
Laurier, D., D. Hemon, et al. (2008a). Childhood leukaemia incidence below the age of 5 years near French nuclear power plants. J Radiol Prot 28(3):401-403.
Laurier, D., S. Jacob, et al. (2008b). Epidemiological studies of leukaemia in children and young adults around nuclear facilities: A critical review. Radiat Prot Dosimetry 132(2): 182-190.
Law, G., and E. Roman (1997). Leukaemia near La Hague nuclear plant. Study design is questionable. BMJ 314(7093):1553; author reply 1555.
Law, G. R. (2008). Host, family and community proxies for infections potentially associated with leukaemia. Radiat Prot Dosimetry 132(2):267-272.
Li, C. Y., and F. C. Sung (1999). A review of the healthy worker effect in occupational epidemiology. Occup Med (Lond) 49(4):225-229.
Linet, M. S., K. P. Kim, et al. (2009). Children’s exposure to diagnostic medical radiation and cancer risk: Epidemiologic and dosimetric considerations. Pediatr Radiol 39(Suppl 1):S4-S26.
Little, J., J. McLaughlin, et al. (2008a). Leukaemia in young children living in the vicinity of nuclear power plants. Int J Cancer 122(4):x-xi.
Little, J. B., H. Nagasawa, et al. (1997). Radiation-induced genomic instability: Delayed mu-tagenic and cytogenetic effects of X rays and alpha particles. Radiat Res 148(4):299-307.
Little, M. P., E. J. Tawn, et al. (2008b). A systematic review of epidemiological associations between low and moderate doses of ionizing radiation and late cardiovascular effects, and their possible mechanisms. Radiat Res 169(1):99-109.
Little, M. P., E. J. Tawn, et al. (2010). Review and meta-analysis of epidemiological associations between low/moderate doses of ionizing radiation and circulatory disease risks, and their possible mechanisms. Radiat Environ Biophys 49(2):139-153.
Lopez-Abente, G., N. Aragones, et al. (1999). Leukemia, lymphomas, and myeloma mortality in the vicinity of nuclear power plants and nuclear fuel facilities in Spain. Cancer Epidemiol Biomarkers Prev 8(10):925-934.
Ma, F., M. Lehnherr, J. Fornoff, and T. Shen (2011). Childhood cancer incidence in proximity to nuclear power plants in Illinois. Arch Environ Occup Health, 66(2):87-94.
MacMahon, B. (1962). Prenatal x-ray exposure and childhood cancer. J Natl Cancer Inst 28:1173-1191.
Mangano, J. J. (1994). Cancer mortality near Oak Ridge, Tennessee. Int J Health Serv 24(3):521-533.
Marples, B., B. G. Wouters, et al. (2004). Low-dose hyper-radiosensitivity: A consequence of ineffective cell cycle arrest of radiation-damaged G2-phase cells. Radiat Res 161(3): 247-255.
Mattsson, A., B. I. Ruden, et al. (1993). Radiation-induced breast cancer: long-term follow-up of radiation therapy for benign breast disease. J Natl Cancer Inst 85(20):1679-1685.
McGale, P., and S. C. Darby (2005). Low doses of ionizing radiation and circulatory diseases: A systematic review of the published epidemiological evidence. Radiat Res 163(3): 247-257.
McLaughlin, J. R., E. A. Clarke, et al. (1993a). Childhood leukemia in the vicinity of Canadian nuclear facilities. Cancer Causes Control 4(1):51-58.
McLaughlin, J. R., W. D. King, et al. (1993b). Paternal radiation exposure and leukaemia in offspring: The Ontario case-control study. BMJ 307(6910):959-966.
Menz, R., R. Andres, et al. (1997). Biological dosimetry: the potential use of radiation-induced apoptosis in human T-lymphocytes. Radiat Environ Biophys 36(3):175-181.
Michaelis, J., B. Keller, et al. (1992). Incidence of childhood malignancies in the vicinity of west German nuclear power plants. Cancer Causes Control 3(3):255-263.
Miller, A. B., G. R. Howe, et al. (1989). Mortality from breast cancer after irradiation during fluoroscopic examinations in patients being treated for tuberculosis. N Engl J Med 321(19):1285-1289.
Mole, R. H. (1974). Antenatal irradiation and childhood cancer: causation or coincidence? Br J Cancer 30(3):199-208.
Morgan, W. F. (2003). Non-targeted and delayed effects of exposure to ionizing radiation: II. Radiation-induced genomic instability and bystander effects in vivo, clastogenic factors and transgenerational effects. Radiat Res 159(5):581-596.
Muirhead, C. R., A. A. Goodill, et al. (1999). Occupational radiation exposure and mortality: second analysis of the National Registry for Radiation Workers. J Radiol Prot 19(1): 3-26.
Muirhead, C. R., J. A. O’Hagan, et al. (2009). Mortality and cancer incidence following occupational radiation exposure: Third analysis of the National Registry for Radiation Workers. Br J Cancer 100(1):206-212.
Nakashima, E., K. Neriishi, et al. (2006). A reanalysis of atomic-bomb cataract data, 2000-2002: A threshold analysis. Health Phys 90(2):154-160.
NCRP (National Council on Radiation Protection and Measurements) (2009). Ionizing Radiation Exposure of the Populations of the United States. Report 160.
Neriishi, K., E. Nakashima, et al. (2007). Postoperative cataract cases among atomic bomb survivors: Radiation dose response and threshold. Radiat Res 168(4):404-408.
Noshchenko, A. G., K. B. Moysich, et al. (2001). Patterns of acute leukaemia occurrence among children in the Chernobyl region. Int J Epidemiol 30(1):125-129.
Noshchenko, A. G., P. V. Zamostyan, et al. (2002). Radiation-induced leukemia risk among those aged 0-20 at the time of the Chernobyl accident: A case-control study in the Ukraine. Int J Cancer 99(4):609-618.
NRC (National Research Council) (2005). Committee to Assess Health Risks from Exposure to Low Levels of Ionizing Radiation, Health Risks From Exposure to Low Levels, of Ionizing Radiation: BEIR VII—Phase 2. Washington, DC: The National Academies Press.
Nuclear Safety Council and the Carlos III Institute of Health (2009). Epidemiological study of the possible effect of ionizing radiations deriving from the operation of Spanish nuclear fuel cycle facilities on the health of the population living in their vicinity, Spain.
Okunieff, P., Y. Chen, et al. (2008). Molecular markers of radiation-related normal tissue toxicity. Cancer Metastasis Rev 27(3):363-374.
Ostroumova. E., B. Gagnière, D. Laurier, N. Gudkova, L. Krestinina, P. Verger, P. Hubert, D. Bard, A. Akleyev, M. Tirmarche, and M. Kossenko (2006). Risk analysis of leukaemia incidence among people living along the Techa River: A nested case-control study. J Radiol Prot 26(1):17-32.
Otake, M., W. J. Schull, et al. (1996). Threshold for radiation-related severe mental retardation in prenatally exposed A-bomb survivors: A re-analysis. Int J Radiat Biol 70(6):755-763.
Ozasa, K., Y. Shimizu, A. Suyama, F. Kasagi, M. Soda, E. J. Grant, R. Sakata, H. Sugiyama, and K. Kodama (2012). Studies of the Mortality of Atomic Bomb Survivors, Report 14, 1950-2003: An Overview of Cancer and Noncancer Diseases. Radiat Res. 177(3):229-243.
Parkin, D. M., D. Clayton, et al. (1996). Childhood leukaemia in Europe after Chernobyl: 5 year follow-up. Br J Cancer 73(8):1006-1012.
Patton, T., A. F. Olshan, et al. (2004). Parental exposure to medical radiation and neuroblas-toma in offspring. Paediatr Perinat Epidemiol 18(3):178-185.
Pearce, N., H. Checkoway, et al. (2007). Bias in occupational epidemiology studies. Occup Environ Med 64(8):562-568.
Pobel, D., and J. F. Viel (1997). Case-control study of leukaemia among young people near La Hague nuclear reprocessing plant: The environmental hypothesis revisited. BMJ 314(7074):101-106.
Poole, C., K. J. Rothman, et al. (1988). Leukaemia near Pilgrim nuclear power plant, Massachusetts. Lancet 2(8623):1308.
Preston, D. L., S. Kusumi, et al. (1994). Cancer incidence in atomic bomb survivors. Part III. Leukemia, lymphoma and multiple myeloma, 1950-1987. Radiat Res 137(2 Suppl): S68-S97.
Preston, D. L., Y. Shimizu, et al. (2003). Studies of mortality of atomic bomb survivors. Report 13: Solid cancer and noncancer disease mortality: 1950-1997. Radiat Res 160(4): 381-407.
Preston, D. L., D. A. Pierce, et al. (2004). Effect of recent changes in atomic bomb survivor dosimetry on cancer mortality risk estimates. Radiat Res 162(4):377-389.
Preston, D. L., E. Ron, et al. (2007). Solid cancer incidence in atomic bomb survivors: 1958-1998. Radiat Res 168(1):1-64.
Preston, D. L., H. Cullings, et al. (2008). Solid cancer incidence in atomic bomb survivors exposed in utero or as young children. J Natl Cancer Inst 100(6):428-436.
Prisyazhiuk, A., O. A. Pjatak, et al. (1991). Cancer in the Ukraine, post-Chernobyl. Lancet 338(8778):1334-1335.
Pukkala, E., A. Kesminiene, et al. (2006). Breast cancer in Belarus and Ukraine after the Chernobyl accident. Int J Cancer 119(3):651-658.
Rajaraman, P., J. Simpson, et al. (2011). Early life exposure to diagnostic radiation and ultrasound scans and risk of childhood cancer: case-control study. BMJ 342:d472.
Richardson, D., H. Sugiyama, et al. (2009). Ionizing radiation and leukemia mortality among Japanese atomic bomb survivors, 1950-2000. Radiat Res 172(3):368-382.
Roman, E., V. Beral, et al. (1987). Childhood leukaemia in the West Berkshire and Basingstoke and North Hampshire District Health Authorities in relation to nuclear establishments in the vicinity. Br Med J (Clin Res Ed) 294(6572):597-602.
Rommens, C., D. Laurier, et al. (2000). Methodology and results of the Nord-Cotentin radio-ecological study. J Radiol Prot 20(4):361-380.
Ron, E. (2002). Ionizing radiation and cancer risk: Evidence from epidemiology. Pediatr Radiol 32(4):232-237; discussion 242-234.
Ron, E. (2003). Cancer risks from medical radiation. Health Phys 85(1):47-59.
Sankila, R., J. H. Olsen, et al. (1998). Risk of cancer among offspring of childhood-cancer survivors. Association of the Nordic Cancer Registries and the Nordic Society of Paediatric Haematology and Oncology. N Engl J Med 338(19):1339-1344.
Schmitz-Feuerhake, I., H. Schroder, et al. (1993). Leukaemia near water nuclear reactor. Lancet 342(8885):1484.
Schmitz-Feuerhake, I., B. Dannheim, et al. (1997). Leukemia in the proximity of a German boiling-water nuclear reactor: Evidence of population exposure by chromosome studies and environmental radioactivity. Environ Health Perspect 105(Suppl 6):1499-1504.
Schneider, A. B., T. C. Gierlowski, et al. (1995). Dose-response relationships for radiation-induced hyperparathyroidism. J Clin Endocrinol Metab 80(1):254-257.
Schneider, J., P. Presek, et al. (1999). Serum levels of pantropic p53 protein and EGF-receptor, and detection of anti-p53 antibodies in former uranium miners (SDAG Wismut). Am J Ind Med 36(6):602-609.
Schubauer-Berigan, M. K., R. D. Daniels, et al. (2007). Risk of chronic myeloid and acute leukemia mortality after exposure to ionizing radiation among workers at four U.S. nuclear weapons facilities and a nuclear naval shipyard. Radiat Res 167(2):222-232.
Schull, W. J., and J. V. Neel (1959). Atomic bomb exposure and the pregnancies of biologically related parents. A prospective study of the genetic effects of ionizing radiation in man. Am J Public Health Nations Health 49:1621-1629.
Segerstrom, S. C, and G. E. Miller (2004). Psychological stress and the human immune system: A meta-analytic study of 30 years of inquiry. Psychol Bull 130(4):601-630.
Senkus-Konefka, E., and J. Jassem (2007). Cardiovascular effects of breast cancer radiotherapy. Cancer Treat Rev 33(6):578-593.
Sermage-Faure, C., D. Laurier, S. Goujon-Bellec, M. Chartier, A. Guyot-Goubin, J. Rudant, D. Hémon, and J. Clavel (2012). Childhood leukemia around French nuclear power plants—the Geocap study, 2002-2007. Int J Cancer, [Epub ahead of print], Sharp, L., R. J. Black, et al. (1996). Incidence of childhood leukaemia and non-Hodgkin’s lymphoma in the vicinity of nuclear sites in Scotland, 1968-93. Occup Environ Med 53(12):823-831.
Shilnikova, N. S., D. L. Preston, et al. (2003). Cancer mortality risk among workers at the Mayak nuclear complex. Radiat Res 159(6):787-798.
Shimizu, Y., H. Kato, et al. (1992). Studies of the mortality of A-bomb survivors. 9. Mortality, 1950-1985: Part 3. Noncancer mortality based on the revised doses (DS86). Radiat Res 130(2):249-266.
Shimizu, Y., K. Kodama, et al. (2010). Radiation exposure and circulatory disease risk: Hiroshima and Nagasaki atomic bomb survivor data, 1950-2003. BMJ 340:b5349.
Shin, S. C., K. M. Lee, et al. (2011). Differential expression of immune-associated cancer regulatory genes in low- versus high-dose-rate irradiated AKR/J mice. Genomics 97(6):358-363.
Shore, R. E. (1990). Occupational radiation studies: status, problems, and prospects. Health Phys 59(1):63-68.
Shore, R. E. (2009). Low-dose radiation epidemiology studies: status and issues. Health Phys 97(5):481-486.
Shu, X. O., F. Jin, et al. (1994a). Diagnostic x-ray and ultrasound exposure and risk of childhood cancer. Br J Cancer 70(3):531-536.
Shu, X. O., G. H. Reaman, et al. (1994b). Association of paternal diagnostic x-ray exposure with risk of infant leukemia. Investigators of the Childrens Cancer Group. Cancer Epidemiol Biomarkers Prev 3(8):645-653.
Signorello, L. B., J. J. Mulvihill, D. M. Green, H. M. Munro, M. Stovall, E. J. Tawn, R. E. Weathers, A. C. Mertens, J. A. Whitton, L. L. Robison, and J. D. Boice Jr. (2012). Congenital anomalies in the children of cancer survivors: A report from the Childhood Cancer Survivor Study. J Clin Oncol 30:239-245.
Simes, R. J. (1986). Publication bias: The case for an international registry of clinical trials. J Clin Oncol 4(10):1529-1541.
Singh, H., R. Saroya, et al. (2011). Radiation induced bystander effects in mice given low doses of radiation in vivo. Dose Response 9(2):225-242.
Sofer, T., J. R. Goldsmith, et al. (1991). Geographical and temporal trends of childhood leukemia in relation to the nuclear plant in the Negev, Israel, 1960-1985. Public Health Rev 19(1-4):191-198.
Sokolnikov, M. E., E. S. Gilbert, et al. (2008). Lung, liver and bone cancer mortality in Mayak workers. Int J Cancer 123(4):905-911.
Sowa Resat, M. B., and W. F. Morgan (2004). Radiation-induced genomic instability: a role for secreted soluble factors in communicating the radiation response to non-irradiated cells. J Cell Biochem 92(5):1013-1019.
Spix, C., and M. Blettner (2009). Re: BAKER P.J. & HOEL D.G. (2007) European Journal of Cancer Care16, 355-363. Meta-analysis of standardized incidence and mortality rates of childhood leukaemia in proximity to nuclear facilities. Eur J Cancer Care (Engl) 18(4):429-430.
Spix, C., S. Schmiedel, et al. (2008). Case-control study on childhood cancer in the vicinity of nuclear power plants in Germany 1980-2003. Eur J Cancer 44(2):275-284.
Spycher, B. D., M. Feller, et al. (2011). Childhood cancer and nuclear power plants in Switzerland: A census-based cohort study. Int J Epidemiol 40(5):1247-60.
Stewart, A. M., J. Webb, B. D. Giles, and D. Hewitt (1956). Malignant disease in childhood and diagnostic irradiation in utero. Lancet 2:447.
Stsjazhko, V. A., A. F. Tsyb, et al. (1995). Childhood thyroid cancer since accident at Chernobyl. BMJ 310(6982):801.
Susser, M. (1997). Consequences of the 1979 Three Mile Island accident continued: Further comment. Environ Health Perspect 105(6):566-570.
Telle-Lamberton, M., E. Samson, et al. (2007). External radiation exposure and mortality in a cohort of French nuclear workers. Occup Environ Med 64(10):694-700.
Tronko, M. D., G. R. Howe, T. I. Bogdanova, A. C. Bouville, O. V. Epstein, A. B. Brill, I. A. Likhtarev, D. J. Fink, V. V. Markov, E. Greenebaum, V. A. Olijnyk, I. J. Masnyk, V. M. Shpak, R. J. McConnell, V. P. Tereshchenko, J. Robbins, O. V. Zvinchuk, L. B. Zablotska, M. Hatch, N. K. Luckyanov, E. Ron, T. L. Thomas, P. G. Voillequé, and G. W. Beebe (2006). A cohort study of thyroid cancer and other thyroid diseases after the chornobyl accident: Thyroid cancer in Ukraine detected during first screening. J Natl Cancer Inst 98(13):897-903.
Uehara, Y., Y. Ito, et al. (2010). Gene expression profiles in mouse liver after long-term low-dose-rate irradiation with gamma rays. Radiat Res 174(5):611-617.
UNSCEAR (United Nations Scientific Committee on the Effects of Atomic Radiation). (2006a). Sources and Effects of Ionizing Radiation, Volume I, Annex A: Epidemiological Studies of Radiation and Cancer.
UNSCEAR (2006b). Sources and Effects of Ionizing Radiation, Volume I, Annex B: Epidemio-logical Evaluation of Cardiovascular Disease and Other Non-cancer Disease Following Radiation Exposure.
UNSCEAR (2008a). Effects of Ionizing Radiation, Volume I, Annex A: Medical Radiation Exposures.
UNSCEAR (2008b). Effects of Ionizing Radiation, Volume II—Annex D: Health Effects Due to Radiation from the Chernobyl Accident.
Upton, A. C. (1980). Radiation risks from nuclear power exaggerated. N Engl J Med 302(21): 1205.
Urquhart, J., M. Palmer, et al. (1984). Cancer in Cumbria: The Windscale connection. Lancet 1(8370):217-218.
Urquhart, J. D., R. J. Black, et al. (1991). Case-control study of leukaemia and non-Hodgkin’s lymphoma in children in Caithness near the Dounreay nuclear installation. BMJ 302(6778): 687-692.
Vares, G., Y. Uehara, et al. (2011). Transcription factor-recognition sequences potentially involved in modulation of gene expression after exposure to low-dose-rate gamma-rays in the mouse liver. J Radiat Res (Tokyo) 52(2):249-256.
Viel, J. F., and S. T. Richardson (1990). Childhood leukaemia around the La Hague nuclear waste reprocessing plant. BMJ 300(6724):580-581.
Viel, J. F., S. Richardson, et al. (1993). Childhood leukemia incidence in the vicinity of La Hague nuclear-waste reprocessing facility (France). Cancer Causes Control 4(4):341-343.
Viel, J. F., D. Pobel, et al. (1995). Incidence of leukaemia in young people around the La Hague nuclear waste reprocessing plant: a sensitivity analysis. Stat Med 14(21-22):2459-2472.
Wakeford, R. (1997). Leukaemia near La Hague nuclear plant. Scientific context is needed. BMJ 314(7093):1553-1554; author reply 1555.
Wakeford, R. (2005). Cancer risk among nuclear workers. J Radiol Prot 25(3):225-228.
Wakeford, R. (2008). Childhood leukaemia following medical diagnostic exposure to ionizing radiation in utero or after birth. Radiat Prot Dosimetry 132(2):166-174.
Waller, L. A., B. W. Turnbull, et al. (1995). Detection and assessment of clusters of disease: an application to nuclear power plant facilities and childhood leukaemia in Sweden. Stat Med 14(1):3-16.
White-Koning, M. L., D. Hemon, et al. (2004). Incidence of childhood leukaemia in the vicinity of nuclear sites in France, 1990-1998. Br J Cancer 91(5):916-922.
WHO (World Health Organization) (1996). Health Consequences of the Chernobyl Accident.
Results of the IPHECA Pilot Projects and Related National Programs. Geneva: WHO.
Wickremesekera, J. K., W. Chen, et al. (2001). Serum proinflammatory cytokine response in patients with advanced liver tumors following selective internal radiation therapy (SIRT) with (90)Yttrium microspheres. Int J Radiat Oncol Biol Phys 49(4):1015-1021.
Wilcosky, T., and S. Wing (1987). The healthy worker effect. Selection of workers and work forces. Scand J Work Environ Health 13(1):70-72.
Wilkinson, G. S., G. L. Tietjen, et al. (1987). Mortality among plutonium and other radiation workers at a plutonium weapons facility. Am J Epidemiol 125(2):231-250.
Wilson, R. (1991). Leukemias in Plymouth county, Massachusetts. Health Phys 61(2):279.
Wing, S. (2010). Testable hypotheses for cancer risks near nuclear facilities. Statement to the Nuclear and Radiation Studies Board of the National Academies.
Wing, S., C. M. Shy, et al. (1991). Mortality among workers at Oak Ridge National Laboratory. Evidence of radiation effects in follow-up through 1984. JAMA 265(11):1397-1402.
Wing, S., D. Richardson, et al. (1997a). A reevaluation of cancer incidence near the Three Mile Island nuclear plant: the collision of evidence and assumptions. Environ Health Perspect 105(1):52-57.
Wing, S., D. Richardson, et al. (1997b). Reply to comments on A reevaluation of cancer incidence near the Three Mile Island. Environ Health Perspect 105(3):266-268.
Wing, S., D. B. Richardson, and W. Hoffmann (2011). Cancer risks near nuclear facilities: The importance of research design and explicit study hypotheses. Environ Health Perspect 119(4):417-421.
Winther, J. F., J. H. Olsen, H. Wu, Y. Shyr, J. J. Mulvihill, M. Stovall, A. Nielse, M. Schmiegelow, J. D. Boice Jr. (2012). Genetic disease in the children of Danish survivors of childhood and adolescent cancer. J Clin Oncol 30:27-33.
Worgul, B. V., Y. I. Kundiyev, et al. (2007). Cataracts among Chernobyl clean-up workers: Implications regarding permissible eye exposures. Radiat Res 167(2):233-243.
Yoshimoto, Y., S. Yoshinaga, et al. (2004). Research on potential radiation risks in areas with nuclear power plants in Japan: Leukaemia and malignant lymphoma mortality between 1972 and 1997 in 100 selected municipalities. J Radiol Prot 24(4):343-368.
Zablotska, L. B., T. I. Bogdanova, E. Ron, O. V. Epstein, J. Robbins, I. A. Likhtarev, M. Hatch, V. V. Markov, A. C. Bouville, V. A. Olijnyk, R. J. McConnell, V. M. Shpak, A. Brenner, G. N. Terekhova, E. Greenebaum, V. P. Tereshchenko, D. J. Fink, A. B. Brill, G. A. Zamotayeva, I. J. Masnyk, G. R. Howe, and M. D. Tronko (2008). A cohort study of thyroid cancer and other thyroid diseases after the Chornobyl accident: Dose-response analysis of thyroid follicular adenomas detected during first screening in Ukraine (1998-2000). Am J Epidemiol 167(3):305-312.