2
THE COMMITTEE’S REVIEW

The National Research Council committee conducted a general review of the CDC-NCI draft report, evaluating its merits and limitations and addressing the particular questions directed to it by CDC. The committee believes that the CDC-NCI working group performed a very competent scoping assessment of the geographic distribution of probable doses to the population, the projected risks associated with those doses, and potential communication plans. However, the committee has a number of suggestions for improvements and has identified some weaknesses of the feasibility study and the draft report that are detailed below.

TECHNICAL APPROACH AND CONTENTS OF THE DRAFT REPORT

The draft report of the CDC-NCI feasibility study consists of two volumes; the first presents the study’s general findings and suggestions regarding further research, and the second sets out in a series of eight appendixes the technical details that support the main text. Collectively, the two volumes represent a substantial amount of work, including the development and application of methods that have produced a credible dose reconstruction from a variety of data that were, in general, not originally collected for such a purpose. However, the draft should make clear, perhaps in the introduction, that this type of analysis cannot be used to identify either individual doses or individual risks and that collective dose calculations are not meaningful in the context of the very low exposures involved (NCRP, 1995). That caveat could be important in later communications with the public. Moreover, without a context, presenting dose information as absolute values has little meaning to the general public or even to the general scientific community. Some method of presentation that shows relative values would be useful. Comparisons with natural background and its geographic variations and with discretionary exposures, such as flying or the use of nonemergency medical radiation, might be helpful. (For example, the French have recently opened a Web site—http://www.sievert-system.org/ (the site was accessed on January 2, 2003)—where anyone can estimate the amount of radiation received during any flight of the French airlines.)



The National Academies | 500 Fifth St. N.W. | Washington, D.C. 20001
Copyright © National Academy of Sciences. All rights reserved.
Terms of Use and Privacy Statement



Below are the first 10 and last 10 pages of uncorrected machine-read text (when available) of this chapter, followed by the top 30 algorithmically extracted key phrases from the chapter as a whole.
Intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text on the opening pages of each chapter. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

Do not use for reproduction, copying, pasting, or reading; exclusively for search engines.

OCR for page 15
Exposure of the American Population to Radioactive Fallout from Nuclear Weapons Tests 2 THE COMMITTEE’S REVIEW The National Research Council committee conducted a general review of the CDC-NCI draft report, evaluating its merits and limitations and addressing the particular questions directed to it by CDC. The committee believes that the CDC-NCI working group performed a very competent scoping assessment of the geographic distribution of probable doses to the population, the projected risks associated with those doses, and potential communication plans. However, the committee has a number of suggestions for improvements and has identified some weaknesses of the feasibility study and the draft report that are detailed below. TECHNICAL APPROACH AND CONTENTS OF THE DRAFT REPORT The draft report of the CDC-NCI feasibility study consists of two volumes; the first presents the study’s general findings and suggestions regarding further research, and the second sets out in a series of eight appendixes the technical details that support the main text. Collectively, the two volumes represent a substantial amount of work, including the development and application of methods that have produced a credible dose reconstruction from a variety of data that were, in general, not originally collected for such a purpose. However, the draft should make clear, perhaps in the introduction, that this type of analysis cannot be used to identify either individual doses or individual risks and that collective dose calculations are not meaningful in the context of the very low exposures involved (NCRP, 1995). That caveat could be important in later communications with the public. Moreover, without a context, presenting dose information as absolute values has little meaning to the general public or even to the general scientific community. Some method of presentation that shows relative values would be useful. Comparisons with natural background and its geographic variations and with discretionary exposures, such as flying or the use of nonemergency medical radiation, might be helpful. (For example, the French have recently opened a Web site—http://www.sievert-system.org/ (the site was accessed on January 2, 2003)—where anyone can estimate the amount of radiation received during any flight of the French airlines.)

OCR for page 15
Exposure of the American Population to Radioactive Fallout from Nuclear Weapons Tests The maps showing county-by-county distribution of radioactive dose is an important aid to public understanding of exposure distributions but could be improved in two ways. First, the display of doses for each county projects a false precision in that distinct boundaries (county boundaries) may be identified as separating various doses. Sophisticated viewers of such a map understand that there is considerable overlap in potential exposure between adjacent but differently coded counties, but this may not be recognized by the general public. A smoothed color-graded contour plot with state or county borders overlaid would be useful in representing visually the true state of limited knowledge about small-scale local differences in exposure. Second, the maps lack a contextual framework. It would be useful to compare graphically, using histograms or other visual presentation tools, the distribution of fallout exposures relative to estimates of natural background radiation either over the country as a whole or by region (East, Midwest, West, and South). That should be done for selected birth cohorts and for selected organ doses to illustrate, for example, that thyroid doses from fallout to children born in 1950 are estimated to be higher than thyroid doses from background for many regions of the country. In contrast, for almost all other organ doses or age cohorts, the estimated doses from fallout are smaller by an order of magnitude or so than background throughout the country, and this should be depicted graphically as well. The committee notes that the main text of the draft report does not always appear to reflect what was done as stated in the appendixes. For example, the committee finds it disturbing that a footnote on page E-36 alludes to a probable error in the 131I deposition values associated with four or more weapons tests that is not mentioned in the main text; nor is the impact of this error on estimated doses described. This error, although it would affect the estimated doses, would not alter the draft’s basic conclusion about the feasibility of the study of the health consequences of exposure to the fallout from weapons testing in the United States or elsewhere. It would have been helpful if, in the presentation of their conclusions, the authors had set out a summary of the possible changes that might occur with further refinement and had essayed an evaluation of how much better the results might be if a major effort were made. ASSESSMENT OF THE DOSE RECONSTRUCTION To estimate radiation doses to the US public due to fallout from nuclear-weapons tests conducted at the NTS and global fallout from nuclear-weapons tests conducted at other sites around the world, it is necessary to estimate the deposition density of fallout radionuclides across the United States as a function of time during and after the period of testing. With this information in hand, calculations of external and internal doses can be based on a number of assumptions such as route of exposure, lifestyle, age, and diet. Recognizing that they were preparing a feasibility study, not a complete detailed analysis of exposure to all radionuclides under all conditions, the authors of the draft report note that the dose estimates given are relatively crude values based on approximate evaluations. The calculations, of moderate complexity, are based on data obtained from readily available publications in the open literature. Some of the data sources and analytic approaches used are discussed below as are some of the key results. Calculations related to the NTS are discussed first, and then those related to global fallout.

OCR for page 15
Exposure of the American Population to Radioactive Fallout from Nuclear Weapons Tests Deposition Density of Nevada Test Site Fallout Radionuclides In the draft report, estimates of deposition of fallout radionuclides from NTS tests are based on county-by-county estimates of the deposition of 131I released from NTS weapons tests that were published as part of the earlier NCI study on the exposure of the American public to 131I from nuclear-weapons tests (NCI, 1997). Because of the importance of these results to the calculations and conclusions in the draft report, the committee briefly describes here the nature of the estimation procedure used in the earlier NCI study. Iodine-131 The development of the 131I estimates from the source data (which consisted of gummed-film data supplemented by cloud-tracking information and rainfall estimates) posed a complex statistical data-processing problem (NCI, 1997). Much of the deposition estimate was based on a general-purpose statistical interpolation method known as kriging. The kriging method requires specification of user-defined parameters that determine specifics of the statistical model used for the interpolation and also required covariate data (in this case, rainfall). It may be of more than just academic interest to determine the degree to which the deposition calculations depend on the choice of input parameters by the NCI investigators. However, the calculations were done with software written specifically for this task rather than with commercially available programs. The publication and archiving of the computer codes and the source data used in the kriging calculations is strongly recommended.6 Regardless of the statistical analyses used, it is important to remember that the underlying data on which average county deposition values are based for the 3,000-odd counties in the United States were obtained from 100 or fewer fallout-collector (gummed-paper) stations throughout the country. That is an inherent limitation in any further refinements of any future analysis. Given initial deposition, the data used in the 1997 NCI study to estimate thyroid exposure to 131I involved a reconstruction of the milk and food pathway, of which a key element was the development of a county-by-county estimate of milk production and a distribution matrix for flows of milk between counties (NCI, 1997). The development of the distribution matrix involved considerable detective work. The data from which the matrix was generated need to be published and archived. The feasibility study used a simplified approach that assumed that the milk and food pathways throughout the United States are similar to those near the NTS. 6   The kriging program used in the NCI study is not included in the published NCI report, for the stated reason that it was not developed by NCI. The committee finds such a reason for not publishing the code inadequate. The meteorologic programs used for the tests for which no gummed films were available were also not published, for the same stated reason. They, too, should be published and archived.

OCR for page 15
Exposure of the American Population to Radioactive Fallout from Nuclear Weapons Tests Other Radionuclides With the 131I deposition data in hand for each county, deposition densities for an additional 42 fallout radionuclides were estimated by using calculated ratios of these radionuclides to 131I as a function of time after each test was conducted, as provided by Hicks (1981). The deposition densities were then used to calculate external and internal radiation doses received by members of the public. The most-detailed calculations were for persons who were adults in 1951. Some specific calculations were also made for persons who were born on January 1, 1951, to compare with the adult values. This birth cohort was estimated in the 1997 report to be one of the more highly exposed groups to NTS-related 131I, although other birth cohorts (1952–57) may have had even higher exposure to this radionuclide. External Doses from Nevada Test Site Fallout The calculations of external dose were based on radionuclides deposited on the ground. It was assumed that persons spent 20% of their time outdoors and the other 80% in structures that reduced their radiation exposure to 30% of the outdoor value. In this way, the daily external dose was calculated as 44% of what it would have been for a person remaining outdoors 100% of the time. For fallout that arrived soon after a test, the short-lived iodine radionuclides contributed substantially to the external dose. A few days later, the main contributors were 132I, 140Ba, 95Zr-95Nb, and 103Ru. Because of the short half-lives of those radionuclides, most of the external dose from NTS fallout was received during the year after a given test. Of the six test series covered in the feasibility study, two, UPSHOT-KNOTHOLE in 1953 and PLUMBOB in 1957, accounted for about half the total population-weighted external dose from NTS fallout.

OCR for page 15
Exposure of the American Population to Radioactive Fallout from Nuclear Weapons Tests Internal Doses from Nevada Test Site Fallout During the timeframe of about 1979–1987, a major effort was conducted to calculate external and internal doses from the NTS tests. That effort, the Off-Site Radiation Exposure and Review Project (ORERP), covered the population living in Nevada, Utah, Arizona, New Mexico, and portions of several other states (Church et al., 1990). The following relationship was used in the draft report to calculate absorbed dose or equivalent/effective dose (see Volume 2, Appendix E, page 7; for further references): D=P×I×Fg, where D=absorbed dose, Gy, or equivalent/effective dose, Sv; P=deposition density of radionuclide of interest at time of fallout arrival, Bq m−2; I=integrated intake by ingestion of radionuclide per unit deposition, Bq per Bq m−2; and Fg=ingestion-dose coefficient for radionuclide, GyBq−1 or SvBq−1. The factors I were estimated in the ORERP study for the integrated intakes in states near the NTS and were used in the feasibility study for all other counties in the coterminous United States. In essence, the parameters for all of the United States were assumed to be those found earlier to be appropriate for Nevada. Those assumptions were made to facilitate calculations in the feasibility study. As noted in the draft report, the method led to a calculated collective thyroid dose of 2,000,000 person-Gy. Calculations in the earlier NCI (1997) study, using a geographic area-specific method, led to a collective thyroid dose of 4,000,000 person-Gy. The difference, which apparently arises from the assumptions listed above,7 may indicate an uncertainty of similar size for the values of all the other radionuclide doses computed in the feasibility study because they are all based on the use of ratios of each radionuclide’s deposition to that of 131I, values for which were available from the NCI (1997) report. A factor of 2 in a study with wide geographic variation is reasonable agreement, particularly for a feasibility study with simplifying assumptions. Refinement of the approach is not likely to add much value to the dose information obtained, particularly because of the generally low doses that are calculated. Even if the estimated doses in the draft report proved to be low by a factor of 2, the corrected doses for radionuclides other than 131I would still be too small to be of significance for health. Deposition Density of Global Fallout Radionuclides The characteristics of fallout from weapons tests at other sites around the world were different, in large part because of the nature of the tests. In general, they were tests of weapons with much larger energy yields from both fission and fusion components. The high-yield explosions injected much of the radioactive material into the stratosphere, where it traveled 7   Appendix E of the draft report speculates that a major contributor to the difference was the implicit assumption of a constant initial retention (of 0.39 m2/kg) of fallout radionuclides on vegetation, whereas this measure is known to vary with distance from the NTS (because particles of different sizes fall out at different distances) and with season (for unknown reasons).

OCR for page 15
Exposure of the American Population to Radioactive Fallout from Nuclear Weapons Tests around the globe largely staying within latitudinal bands; that delayed its fallout for months or years. During the prolonged stratospheric retention, the short-lived radionuclides decayed substantially before the fallout occurred. Thus, for the most part, the deposition of global fallout involved longer-lived fission products, such as 90Sr and 137Cs, and radionuclides produced by other processes in the explosions, 3H and 14C. Beyond that general picture, there were sporadic observations of localized deposition of the short-lived radionuclide 131I (and other short-lived radionuclides presumably were also deposited), apparently because of local meteorologic conditions. Many of the determinations of deposition density of global fallout radionuclides in the United States are based on soil samples collected periodically from a network of about 30 stations and analyzed for 90Sr and perhaps 89Sr. The extension of the results to other years has been based heavily on a model developed by Bennett (1978). The United Nations Scientific Committee on the Effects of Atomic Radiation (UNSCEAR) has been active in assessing global fallout for many years. The authors of the draft report used information provided in UNSCEAR (1993) to establish a list of important radionuclides to study. The deposition density of 90Sr in each of the 3000-odd counties was estimated as being proportional to average rainfall. Deposition densities for other radionuclides were calculated from the ratios of the amounts of each of these radionuclides present relative to the amount of 90Sr present according to Bennett’s model. External Doses from Global Fallout Given the fallout deposition densities calculated above, external doses were computed for each county for each year from 1953 to 1972, together with a single calculation for external dose for 1973–2000. The calculations showed that 137Cs and 95Zr-95Nb accounted for 70% of the total external dose from global fallout. The resulting population-weighted external doses over the years 1953–2000 totaled 0.74 mGy/person compared with the population-weighted external dose of 0.5 mGy/person from NTS fallout. Thus, fallout of longer-lived radionuclides resulted in protraction of the period over which the external dose was received, and the total external dose was about 50% higher than the external dose received from NTS fallout. Internal Doses from Global Fallout As was true for the NTS fallout, ingestion of contaminated foodstuffs was the major contributor to the calculated internal doses from global fallout. Previous reports showed that there were five major contributors to the internal dose: 3H, 14C, 90Sr, 131I, and 137Cs. The ORERP approach described above was used to calculate the doses for 90Sr, 131I, and 137Cs. Special models were used to describe the disposition of 3H and 14C throughout the environment.

OCR for page 15
Exposure of the American Population to Radioactive Fallout from Nuclear Weapons Tests DOCUMENT LOCATION AND RETRIEVAL The draft report identifies many sources of information that were used in the feasibility study (or in the studies on which the feasibility study relied) in the calculation of doses from weapons testing and the methods used. It also identifies other sources of potentially informative data that were unavailable when the feasibility study was performed. Some of those data are in well-archived repositories, but many are poorly stored and cataloged. The committee recognizes the potential value of such data to fallout research that may be undertaken in the future and recommends that CDC extend programs to identify and preserve records and materials. Such a program would have several elements as listed below: Finding relevant data A number of sources of information should be considered when searching for relevant data on fallout from nuclear-weapons testing. The national laboratories under the jurisdiction of the Department of Energy (DOE) have all monitored fallout in their areas, and this information and that from the laboratories’ special studies need to be archived. Special attention is needed to identify and catalog unique data stored at laboratories that have been heavily involved in fallout research, such as the Environmental Measurements Laboratory (EML) and the Lamont Geophysical Laboratory. In 1982, the US Public Health Service collected its documents on nuclear fallout, and they contain material of potential importance, such as data on fallout radionuclides in milk. The US Navy and US Air Force also have extensive documents of likely importance. The extensive measurements made of 131I in animal thyroid glands should be compiled. There are other sources of information of probable utility for future studies of fallout. For example, during the years of nuclear-weapons testing, Congress held many hearings on the health effects of, and need for, further nuclear-weapons testing. The published hearings are out of print, but CDC has found extensive collections of them in several locations such as the Coordination and Information Center in Las Vegas, university libraries, and libraries at DOE national laboratories, Atomic Energy Commission repository libraries, and EML. The records of hearings are valuable in two ways: they contain useful information themselves, and they point to where more information can be found. In addition, attention should be given to the availability of fallout-related information available from UNSCEAR and foreign laboratories that have had active programs in fallout measurement and analysis. Safeguarding and storing data Data searches and cataloging will not be possible if the underlying records and related materials are destroyed. Recognizing that, DOE has placed a moratorium on the destruction of possibly relevant records. At present, there is no such moratorium on the destruction of DOD

OCR for page 15
Exposure of the American Population to Radioactive Fallout from Nuclear Weapons Tests fallout-related records. We recommend that CDC urge Congress to prohibit the destruction of relevant records held by federal agencies and to permit appropriate access to them. The implementation of the modern-day dosimetry system used to estimate radiation doses for the atomic-bomb survivor study rests on data on individual survivors that were collected in the 1950s (Roesch, 1987). That speaks to the importance of archiving historical data for future dose-reconstruction purposes. How such data are stored is a key to ensuring their availability to future investigators. Conscious of changes in computer storage media and the importance of archival material, CDC has embarked on the laudable project of transferring data stored on magnetic tape, punched paper tape, and punched cards to modern computer-readable media. That should be done at all sites where archival fallout information is stored. Ensuring continued accessibility to the data When relevant data and analytic reports have been found and cataloged, it will be important for their contents and locations to be made known to interested potential users. The establishment and maintenance of public reading rooms would be highly desirable to provide continued public access to copes of reports and catalogs of sample data and other resources (like original samples). ASSESSMENT OF THE ESTIMATES OF CANCER RISK The CDC-NCI investigators estimated risks posed by the external exposure generated by the fallout radionuclides and by the internal doses from 131I. As the basis of their risk estimates, they used the estimate of five excess cancer deaths per 100 persons per sievert of dose to the whole body, which was derived by the International Commission on Radiological Protection (ICRP, 1991) and takes into account that the exposures were protracted rather than acute. A linear non-threshold (LNT) model was assumed for projecting possible cancer risks in the feasibility study. While this seems to be a reasonable assumption based on available epidemiologic data from higher doses and high dose-rates, such as in the Japanese atomic-bomb study, no data are available to definitively substantiate or contradict the LNT model at the low levels of dose and low dose-rates contributed by NTS and global fallout. Experimental data are also not definitive: some suggest linearity, but others suggest no effect or even the possibility of a protective effect, at low doses of radiation. The available data generally do not suggest that the risk at low doses is greater than that predicted by the LNT model. The LNT model is widely used and is generally appropriate in this case because it represents a prudent approach to health protection of the public in the absence of definitive information on radiation risks at low exposure levels. One limitation of the findings presented in the draft report is that to calculate the plausible range of risk estimates, the authors relied on a general estimate of the “credibility interval” of risk estimates (draft report, p. 155) per unit of dose—namely, a factor of 3 less or

OCR for page 15
Exposure of the American Population to Radioactive Fallout from Nuclear Weapons Tests more than the central estimate—and on the assumption of a similar factor of 3 of a “credibility interval” for the dose estimates (draft report, p. 156). The basis of the factor of 3 for the dose estimates was not explained; it should be, because it is a key factor in generating the ranges of putative risks that are given in the draft report. The feasibility study could not conduct a detailed uncertainty analysis based on the specific uncertainties that go into dose estimates. Such an approach is acceptable as a rough guide, but if a more comprehensive study is performed, a thorough uncertainty analysis of doses and risks would be needed. In a number of places in Chapter 4 of the draft report there is considerably more emphasis on population risk (posed by collective dose) than on individual risk. While collective dose and resulting collective risk may be of interest from a historical, academic, or political standpoint, the critical information for exposed people is their degree of risk, which has more important implications for individual health than does how many people in the population are expected to have adverse effects. It is therefore more useful to put the primary emphasis on individual risk than on population risk, when this is possible. Page 130 of the draft report indicates that “population risk” will be useful for developing public-health policy, although it does not state how. Arguably, the individual-risk information is more useful because it helps to clarify whether preventive actions (such as cancer screening because of high risk) might be useful in a way that collective dose or risk cannot be. That said, we recognize that there are even greater complexities in estimating individual risk than collective risk since apportioning the collective risk to individuals is dependent upon dose reconstruction techniques with many uncertainties, and the individual increments in risks, so derived, are small. Explaining the consequence of a small and uncertain increment in individual risk to an individual is inherently problematic. More perspective is needed on the NTS and global fallout-dose effects. We recommend two ways to provide such perspective. First, present the estimated incremental cancer risk posed by fallout as a percentage of the current observed cancer risk. For example the current lifetime cancer-death risk is about 20%, and the risk posed by fallout is about 0.03% (with a claimed credibility interval of about 0.01–0.09%), so the lifetime cancer risk would be raised from 20% to 20.03% (with an estimated range of 20.01–20.09%). Second, compare the estimated risk posed by fallout with that posed by lifetime exposure to natural background radiation, and its variations from place to place. It will then be possible for the reader to understand that the fallout risk is small compared with the risk we all presumably incur from exposure to natural background radiation. A comparison of the fallout risk with putative background radiation risks for the residents of the eastern seaboard and residents of Rocky Mountain states would also be illuminating, given the variation in background radiation for these two populations. Expanding on this theme, for the 3.8 million children born in 1951 (representing the most heavily exposed Americans), the authors of the draft report estimate that fallout will lead to about 2,000 excess cancer cases and about 1,000 excess cancer deaths (which represents an excess risk of death of one in 3,800) on the basis of extrapolation of existing epidemiologic data with a linear “non-threshold” (LNT) model. The draft report compares that number of excess cases with the total lifetime risk in this cohort dying of cancer, (20%, or 760,000 total cancer deaths). An additional comparison of the excess-cancer calculations for fallout with the total number of deaths due to background radiation in this model would highlight the dependence of the calculation on the use of an LNT model. Not only would the comparison help further to put fallout-related risk into perspective by comparison with the “natural” radiation risks that people

OCR for page 15
Exposure of the American Population to Radioactive Fallout from Nuclear Weapons Tests must accept as a part of everyday life, but it would help to distinguish between cancers whose risk from fallout will potentially increase compared with background risk (specifically thyroid cancer) and the remaining cancers, whose risk it will not increase. For example, a simplified calculation based on an LNT model leads to an estimate that about 2% of all cancer cases (or 28,120 excess cases in the 1951 birth cohort on the basis of an estimated “spontaneous” lifetime cancer risk of 37%8 without any added radiation exposure) might be due to background radiation. That estimate is based on the assumptions that exposure to external background radiation amounts to an average of 1 mSv/year in the United States and that there is an excess lifetime cancer risk of 10% per sievert of background exposure. The excess risk per unit dose is calculated as half the excess risk per sievert seen in the youngest atomic-bomb survivors who experienced acute rather than chronic exposures (see the following site that was accessed on December 10, 2002; http://www.rerf.or.jp/eigo/radefx/late/cancrisk.htm). From this example calculation, NCI’s estimate of 2,000 excess cancer cases due to fallout is seen to amount to about 7% of the roughly 28,100 excess cancer cases that are due to natural background radiation. The corresponding fractions—cases caused by fallout divided by cases due to background radiation—are even smaller when computed for other age cohorts, the members of which were either exposed to less radiation from fallout or exposed at ages at which they were less sensitive. One implication of the example calculation is that even in the most heavily exposed age cohort it will be impossible to detect differences in total cancer incidence (or mortality) due to fallout with current methods of observational epidemiology. Epidemiologic analysis of the US population for increases in overall cancer rates by predicted fallout exposure is extremely unlikely to add scientific information about the existence of excess cancer risks or about the validity of the LNT model used to predict them. In fact, comparing the cancer rates seen in the naturally higher versus naturally lower background-radiation regions in the United States yields little evidence of the predicted 2% excess cancer rate due to background radiation (Eckhoff et al., 1974, Jacobson et al., 1976). Attempting to detect fallout risks that are more than an order of magnitude smaller than background-radiation risks in a smaller cohort restricted to the most-exposed people is impractical with any epidemiologic method currently available. A useful illustration to show that very-low-dose radiation exposure is not an overwhelming factor in cancer causation would be a comparison of the rates of cancer observed in residents of the eastern seaboard with those in residents of the Rocky Mountain states, where background radiation is higher. The failure of this and other observational analyses to detect the cancer increases due to background radiation that are predicted by the LNT model in the Rocky Mountain region, although not a disproof of the model because other causes of cancer also are known to vary regionally. Nevertheless this comparison will make the point that the model is used by the authors to extrapolate to magnitudes of risk that are far outweighted by risks due to other factors. Putting aside thyroid dose due to 131I, the current estimates for the other radionuclides yield population external dose estimates that are only a small fraction of the external doses due to background radiation. Assuming a conservative LNT model, translating 8   This differs from the 20% cited above because that refers to risk of death, whereas the 37% refers to total risk of cancer (not necessarily death from cancer.

OCR for page 15
Exposure of the American Population to Radioactive Fallout from Nuclear Weapons Tests these doses into increases in the number of cancer cases in the US population yields a total number of expected excess cases that may appear large in absolute numbers but is very small when compared either with the number of total cancers in the population or with cancers attributable to background exposure according to the same LNT model. A more detailed dose reconstruction is very unlikely to change that result even if internal deposition is included. On the basis of those considerations, fine tuning of the calculation of population dose estimates is an academic exercise with little obvious public-health significance. The committee is sensitive to the fact that other reasons than public health significance may motivate further refinement of estimates of total population dose—e.g. to show, by dint of good effort, that the US government is not attempting to keep veiled in secrecy the true consequences of nuclear testing. We do not think, however, that detailed dose reconstruction, at anything like the level of detail given for 131I in the 1997 NCI report, would be scientifically meaningful, when applied to radionuclides other than 131I. In the single case of thyroid cancer, however, the situation is markedly different. The attributable fraction of thyroid cases in the most exposed cohort may be a substantial portion of total thyroid-cancer risk and is considerably larger than the perhaps 8% of cases that may be attributed to background exposure. Land’s calculations in 1997 (see Appendix B in IOM-NRC, 1999) estimated a potential increase of 40% in thyroid-cancer incidence in the most exposed people. Unlike increases in the other cancers, such increases in thyroid cancer should be detectable in two types of studies: those which compare rates of thyroid cancer by geographic region and those which compare rates in the most exposed birth cohort with rates in the least exposed birth cohort. However, fallout is not the sole cause of increases in thyroid-cancer rates that have been seen in recent decades. Between the 1930s and 1950s, the medical use of radiation directed to the head and neck was common. The procedure delivered a high dose at a high dose rate to the thyroid, particularly of children. The practice was largely discontinued during the 1950s, once the dangers were recognized. Such medical exposures, rather than fallout, appear to be principally responsible for the increases in thyroid cancer seen beginning in the 1930s (Pottern et al., 1980). If the thyroid doses from fallout can be estimated with sufficient accuracy, the committee believes that it would be useful to compare these doses at the individual or collective dose level or both with the thyroid doses from some of the earlier therapeutic procedures. Although the draft report mentions the thyroid-cancer, thyroid-adenoma, and hypothyroidism risks in Chernobyl-exposed children several times, it does not mention the Hanford results. The Hanford Thyroid Disease Study had doses similar in magnitude and duration to those projected in the draft report, and it is the best-conducted study of 131I (with exceptional subject location and participation rates and uniform, high-quality screening), so its results merit a summary. The feasibility study apparently used a linear model to estimate leukemia risk. It is unclear, however, whether the basis of the linear model is a linear extrapolation from high dose down to zero dose or the linear component of a linear-quadratic model (in which the linear component dominates at low doses); the latter is much preferable for the feasibility study. A risk discussion is needed that describes the low-dose extrapolation issue and that notes that it is uncertain whether there is any increased risk of leukemia. At least one analysis of the atomic-bomb survivor data finds evidence of a zero-effect threshold below 0.2 Sv (Hoel and Li, 1998)

OCR for page 15
Exposure of the American Population to Radioactive Fallout from Nuclear Weapons Tests and thus disagrees with the analyses produced by the Radiation Effects Research Foundation (Pierce et al., 1996; Preston et al., 1994), which estimate that 14% of the leukemia cases in survivors exposed below 0.2 Sv were due to exposure on the basis of a linear-quadratic model fit. The committee notes also that the term leukemia encompasses a variety of hematologic disorders that involve the white cells and that have seemingly different dose-response relationships (Preston et al., 1994). The discussion of risk estimates for other cancers in the draft report (Section 4.2.1.3) needs to be elaborated. There is no discussion of the degree of concordance among the primary studies of the principal tumor sites, nor are any risk estimates or any references given. The discussion is therefore not very informative or convincing to readers who are unfamiliar with radiation epidemiology. The discussion of noncancer-risk estimation indicates in passing that the estimate of a dose threshold of 0.1–0.2 Gy for hypothyroidism is likely to yield an overestimate of risk in that most data show hypothyroidism risk only at appreciably higher doses (UNSCEAR, 1993; IOM-NRC, 1999); that it is an overestimate ought to receive more emphasis. Given that these were both protracted and fractionated exposures, a dose threshold as low as 0.1–0.2 Gy for hypothyroidism seems implausible. In addition, it could be stated more strongly on the basis of available data on the radiation-related risk of major nonmalignant chronic diseases (such as cardiovascular, respiratory, digestive, and genitourinary), that no excess risk of these diseases is expected to be posed by NTS and global fallout. THE VALUE OF FURTHER REFINEMENTS OF THE 131I NEVADA TEST SITE CALCULATIONS AND UNCERTAINTY ANALYSIS Dosimetric Refinements Improvements in some aspects of the 1997 NCI report on 131I exposures from the NTS are possible based partly on what has been learned from the Chernobyl experience. In particular, parts of the pasture-milk pathway of exposure are probably better known now than they were in the 1980s and early 1990s (when the dose reconstruction for the NTS 131I exposure was performed), because of analyses of the database of thyroid-dose measurements in many Chernobyl-exposed people. In addition, it may be technically possible, partly because of the Chernobyl experience, to conduct a more thorough analysis of additional aspects of the uncertainty in thyroid dose than was included in the 1997 report. The uncertainty analysis also might be improved by providing a better differentiation between uncertainties that are shared and those which are unshared from county-to-county and from person-to-person. The review of the 1997 NCI report by a 1999 Institute of Medicine (IOM)-National Research Council committee included many specific comments regarding the uncertainty analysis, some of which could be reexamined in more detailed studies along with other sources of uncertainty. The following are some examples:

OCR for page 15
Exposure of the American Population to Radioactive Fallout from Nuclear Weapons Tests During the estimation of uncertainties in dose due to the interpolation of deposition, the kriging methods used to interpolate the gummed-film measurements could have been modified to include a reasonable distribution of errors in the gummed-film measurements (see page 31 of IOM-NRC, 1999); the incorporation of such methods would reduce to some degree the between-county variability in average-dose estimates and lead to a “flatter”, probably more realistic, map of exposure as a function of geographic location. This is a standard consequence of statistical approaches for measurement error: estimates of true exposure are pulled together toward a common mean when measurement errors are factored into the calculations. A better distinction could be made between the variability in herd averages (over large numbers of the various breeds of milk cows) of milk-transfer coefficients and the variability in milk-transfer coefficients estimated from a very small number of animals of a particular breed (see pages 34–35 of IOM-NRC, 1999). Uncertainty in the estimates of milk distribution between producer and consumer counties could potentially be included in the analysis (see the discussion of the Vol(i,j) terms on page 37 of IOM-NRC, 1999). Allowing considerations of this source of measurement error would probably tend to smooth the estimates (pulling them toward a common mean), the amount of smoothing being dependent on the degree of uncertainty considered in the Vol(i,j) terms. Information about interindividual variability in dose-conversion factors could be included (see page 39 of IOM-NRC, 1999) perhaps again on the basis of the Chernobyl experience. That said, it is probably the case that improvements are apt to be only at the margins in the sense that estimates of the doses and uncertainties are unlikely to be markedly more precise. The net effects of additional uncertainty analysis are likely to be a decrease in the variability in the county-dose estimates between countries (this is expected from points 1–3 above) and a small increase in the already wide uncertainties in individual dose estimates. What is very unlikely to be achieved as a result of further work on dose reconstruction is a substantial increase in the correlation between true dose (the average for either a county or an individual) and estimated doses, because ultimately the exposure estimates must be based on data that have fundamental limitations—such as the gummed-film network for estimating deposition—and because there are gaps in our knowledge of the milk-distribution patterns that existed in the 1950s and early 1960s. Only if the correlation between true and estimated dose were substantially improved would the re-examination of the 131I exposures lead to better risk estimates of the effect of the NTS exposures on public health. Epidemiologic Refinements The only radionuclide whose exposure is examined in the draft report and for which the combination of dose estimate and assumed dose-response relationship may produce a number of excess cancer cases that is potentially detectable in epidemiologic work in the United States

OCR for page 15
Exposure of the American Population to Radioactive Fallout from Nuclear Weapons Tests appears to be 131I—specifically as released from the NTS and as related to the risk of thyroid cancer. Several unique aspects of the exposure and of thyroid cancer contribute to that specificity. The existence of a highly specific pasture-milk pathway, concentration of iodine in the thyroid gland, a remarkable dependence of risk of thyroid cancer on age at exposure, and apparent continuing risk decades after exposure add up to a pattern of risk in the US population that may still be detectable. Exposure to 131I from global fallout was estimated in the feasibility study to be much less important as a cause of thyroid cancer because, owing to its short half-life, much of the 131I had decayed before reaching the United States. Other cancers are much harder to study because the potential excess is spread out over a much wider group of organs and because doses to other organs were generally much lower than doses to the thyroid. Even for thyroid cancer, many uncertainties remain. Thyroid cancer clearly was induced by 131I exposure in the Chernobyl studies, but the exposures there were higher than those produced by Nevada Test Site fallout. Estimation of an excess of thyroid cases relies on interpolation, assuming that linear models are valid down to levels of excess risk undetectable by epidemiology. In addition, the overall magnitude of exposure (total dose to the population) is uncertain, as are the details of the geographic distribution of 131I throughout the country. The Utah Thyroid Disease Study found evidence of excess thyroid disease (cancer and benign tumors) in areas relatively close to the NTS (Kerber et al., 1993). At least two other sources of information regarding NTS exposure and its influence on thyroid-cancer risk have to be considered. The first is the broader geographic distribution of 131I from the NTS and its relationship to the incidence of and mortality from thyroid cancer, examined using passive registry systems in place today. The second is comparison of age-specific rates of thyroid cancer in groups of US residents whose age at the time of the NTS releases placed them into a “high-risk” or “low-risk” exposure-age category. Gilbert et al. (1998) exploited the wider geographic distribution of dose, relating county-average exposures to the geographic distribution of thyroid-cancer cases and deaths; the IOM-NRC review of the 1997 NCI report of exposure to 131I did a simple overall comparison of age-specific rates of thyroid cancer. Although the analyses all have substantial limitations, they are useful and are highly complementary in that the information contained in each analysis is almost independent of that in the others. All those analyses found some evidence of increases in thyroid-cancer rates that suggest effects of exposure. The Utah study indicated statistically significant increases only when thyroid cancers were combined with benign thyroid tumors (Kerber et al., 1993). The Gilbert et al. (1998) study, using the 1997 NCI estimates of NTS 131I dose, found evidence that thyroid-cancer mortality, but not incidence, was related to county-specific thyroid-dose estimates. And the IOM-NRC review of the 1997 NCI report found roughly a 10% increase in age-specific thyroid-cancer incidence when rates at ages 25–29 years or at ages 35–39 years in an “exposed” age cohort (born in 1948–1952) were compared with rates in two “relatively unexposed” age cohorts (born in 1959–1963 or born in 1938–1942). However, temporal variations in other sources of radiation exposure (e.g., dental x-rays; therapeutic x-rays for common benign conditions) were not evaluated in the IOM-NRC review, although such temporal variations could be the cause of the observed increase. Further “inexpensive” epidemiologic work with existing data seems to be warranted by

OCR for page 15
Exposure of the American Population to Radioactive Fallout from Nuclear Weapons Tests the results seen so far. Specifically, the approach of the Gilbert et al. (1998) study cited above could be expanded to include comparisons by birth-cohort of age-specific incidence and mortality at the county level. It is particularly important to see whether the apparent excess of incident cases in the “exposed” birth cohort (born in 1948–1952) holds up when data from other sources (including non-SEER registries) are included in the analysis and whether the apparent increase in mortality by geographic region holds up in age-specific comparisons. The statistical power of such analyses is an important issue, especially if null results are found since thyroid cancers are frequently misdiagnosed or undiagnosed partly because detection of disease is to some extent dependent upon screening practices which may differ both geographically and temporally. However, a simplistic calculation provided by the IOM-NRC review of the 1997 NCI report indicates that reasonable power is expected to detect the large potential increases in age-specific incidence estimated by NCI (see Appendix B in IOM-NRC, 1999). Moreover, the power of these analyses continues to increase with additional follow-up by the tumor-registry system in the United States. Those attempting to interpret such data, however, should keep in mind the limitations of ecological data, the inaccuracies and surveillance biases in diagnosing thyroid cancer, and the fact that mortality is a poor surrogate for thyroid cancer incidence because of the low case-fatality rate. In addition, care would need to be taken to ensure that radiodiagnostic or radiotherapeutic exposures do not confound the analysis. Although not directly incorporated as one of the feasibility study’s proposals, it is worth commenting here on the feasibility of further screening-based epidemiologic investigation of NTS fallout. The committee is not in favor of such investigation directed at the effects of 131I fallout, except perhaps for a re-screening of participants in the Utah study. An extension of screening to other areas of the country would face a dilemma. If people of the same general geographic region and age at exposure are used in the screening, there will be little ability to distinguish between those who received low exposure and those who received high exposure, because estimation of individual dose is known to be highly uncertain (with uncertainty of at least a factor of 3 within an age and a geographic region). However, if people from various regions are compared, dose estimates would depend primarily on geographic region alone. That implies not only that regional differences in underlying rates of disease would confound the results, but also that a study large enough to have good statistical power would probably only recapitulate results that are much more easily obtained by comparing regional differences in age-specific rates with existing tumor-registry data. Moreover, no screening of disease prevalence could be designed that would exploit the predicted differences in age-specific rates of disease between different birth cohorts. Such birth-cohort comparisons may, in fact, have more power than even screening-based geographic comparisons. COMMUNICATION WITH THE PUBLIC ABOUT EXPOSURE AND CANCER RISK The draft report provides important information about the possible health consequences to the American population arising from exposure to radiation resulting from the NTS and global fallout. At the public session of the committee’s meeting on September 12, 2002, in Des Moines, Iowa, members of the public spoke about their continuing concerns about the health effects of fallout. They asked for a full and timely disclosure of information related to their and

OCR for page 15
Exposure of the American Population to Radioactive Fallout from Nuclear Weapons Tests their family members’ exposure. In so doing, they invoked a legacy of denial, secrecy, and injustice on the part of government agencies and officials. In his statement to the committee, Senator Tom Harkin (Democrat, Iowa) expressed the concerns of his constituents: “Many people are wondering and asking me, could their cancer, or that of their relatives, have been caused by fallout? . . Is there anything they can do to protect themselves now?”9 Senator Harkin emphasized the need for public education and attention to these issues. Overview of the Proposed Communication Plan Plans to communicate with the public occupy a major chapter and a long appendix in the draft report. Various communication issues were evaluated, and alternative plans were presented. The most thoroughly discussed plan would accompany Option 5 for future work and includes a major educational effort that would build on a communication plan already under development, the 131I/Nevada Test Site Communication Plan. This effort, about exposures to radioactive iodine at the NTS, is sponsored by NCI and has been in development for more than 2 years. More modest communication goals are outlined for Options 3 and 4. Those options depend on adapting the plans, methods, and materials already being developed by the 131I/Nevada Test Site Communication Plan and adding information from the feasibility study. This would be done in phases that could include the addition of iodine doses from global fallout for Option 3 and additional radionuclides from both the NTS and global fallout for Option 4. The goal of these more limited communication plans seems achievable, although in Option 4 supporting information about different radionuclides would have to be added to the communication plan, particularly for affected populations in the eastern part of the United States. The committee is concerned about the lack of communication planning for Option 1. It appears that for this option only a highly technical final form of the report would be used to communicate the feasibility study’s findings to the public. The committee strongly believes that a public summary should be prepared for this option, and this is discussed in more detail below. The 131I/Nevada Test Site Communication Plan Much of the communication planning in the feasibility study is based on experience with the 131I/Nevada Test Site Communication Plan, which seeks to communicate to the public the results of NCI’s report Estimated Exposures and Thyroid Doses Received by the American Public from Iodine-131 in Fallout Following Nevada Atmospheric Nuclear Bomb Tests and which has been evolving and under test for more than 2 years. The plan involves providing materials for various public audiences and educating health professionals. In its thorough and well-developed outline, the plan has followed the basic rules for health and risk communication with attention to audience analysis, risk perception, social and political considerations, and the need for health follow-up advice and procedures. 9   Statement to the committee of Senator Tom Harkin on nuclear testing fallout report September 12, 2002.

OCR for page 15
Exposure of the American Population to Radioactive Fallout from Nuclear Weapons Tests To help to develop the plan, NCI sponsored a conference in January 2000 that involved many people from affected communities, activist groups, state and federal government agencies, and health and risk communication researchers. In particular, the conference included many stakeholders in the 131I fallout areas. It produced excellent recommendations for proceeding with the plan’s development, including major community involvement (see Appendix H of the draft report). The 2000 NCI conference was followed by research efforts with a series of lay focus groups (including representatives of various stakeholder groups) and interviews with experts to refine approaches and methods for the plan. According to NCI representatives at the committee’s open meeting in September 2002, the public communication plan will include a general brochure about 131I exposure, a brochure intended to serve as “a decision aid for people exposed to 131I”, and a PowerPoint presentation specially designed for American Indian audiences. Those will be placed on the NCI Web site and distributed in print. The committee saw them in draft form at the September meeting. NCI plans to distribute the materials through community groups, state government health agencies, and health professionals. In addition, an educational plan for health care professionals related to health effects of fallout exposure is described in the draft report of the feasibility study. Would Adapting the 131I/Nevada Test Site Communication Plan Work for the Feasibility Study? The 131I/Nevada Test Site Communication Plan could provide a good base on which to build a communication program for the feasibility study. The committee strongly suggests timely disclosure in the development of communication materials for the feasibility study. The draft report says that materials from the 131I/Nevada Test Site Communication Plan would be distributed, evaluated, and perhaps modified before materials from the feasibility study are added. The committee believes that this would result in too great a delay and that information from the feasibility study should be added as quickly as possible to the 131I/Nevada Test Site Communication Plan brochures and Web materials now under development in order to give citizens a more complete picture of their fallout exposure. The materials could be revised later if problems arose. The committee encourages CDC to take note of the frustration expressed by Senator Harkin and several members of the public about the delay in disseminating the results on health effects of fallout exposure and to make every necessary effort at timely disclosure. Communication Issues for Option 1 In the feasibility study, Option 1 is limited to the release of a final version of the report of the feasibility study. The draft report of the feasibility study is a generally well-written technical report. Efforts were made throughout to define technical terms as they were used and to explain complicated subjects. The draft report also includes helpful, brief summaries at the beginning of each chapter and a glossary. Those facts notwithstanding, a public summary of the study would

OCR for page 15
Exposure of the American Population to Radioactive Fallout from Nuclear Weapons Tests be essential for disseminating its findings to interested members of the public. The public summary should include the following elements: It should place the projected health effects of NTS and global fallout in the context of other risks. It should explain that, given the available data, the feasibility study cannot provide individuals with their estimated radiation doses and should explain why it cannot do so. It should describe the feasibility study’s limitations in a way that the nontechnical public can understand. It should discuss the uncertainties in the feasibility study, why they arise, and their implications for the data discussed in the report of the study. Accuracy, understandability, readability, and consistency of the message are important for the public summary. Tables, charts, and other illustrations could play a major role in helping members of the public to understand the radiation information. The committee recommends that a modest communication plan for the distribution of the public summary be developed to plan for its timing, form, distribution network, and possible public involvement, including that of stakeholders and third parties, such as state health agencies. COMMENTS ON THE OPTIONS FOR FUTURE WORK The draft report sets forth five options for future work. The authors make no recommendation as to which option they believe to be the most appropriate. Rather, they advance a rationale for the selection of each option, discuss the technical issues associated with it, and describe its probable cost and staffing requirements. The draft report does not identify the benefits associated with each option, nor does it indicate what policy or public-health decisions might be aided by selecting it. The five options identified in the draft report are summarized as follows: “Option 1. Do no additional fallout-related work. Option 2. Retrieve and archive the historical documentation related to radioactive fallout from nuclear-weapons tests conducted by the United States and other nations. Option 3. Conduct a more detailed dose reconstruction of radioactive fallout from global nuclear-weapons tests for Iodine-131, the most significant radionuclide identified in the feasibility study. Option 4. Conduct a more detailed dose reconstruction for multiple radionuclides in radioactive fallout from both Nevada Test Site and global nuclear-weapons testing.

OCR for page 15
Exposure of the American Population to Radioactive Fallout from Nuclear Weapons Tests Option 5. Conduct a detailed study of the health effects of fallout of nuclear-weapons testing fallout including, in a single project, dose estimation, risk analysis, and communication of the results to interested parties.” The committee notes that those options are neither mutually exclusive nor exhaustive. They encompass three decisions that need not proceed in the lock-step fashion that the options indicate. Each decision affects choices that can be made independently of the others. The committee suggests that the decisions and choices be framed in some such fashion as the following: Decision 1: Retrieve and archive documents Elective 1: Do nothing more. Elective 2: Archive documents from a certain number of sites believed to be the most likely sources of additional data. Elective 3: Conduct a comprehensive retrieval and archiving of information from many data sources. Decision 2: Expanded scientific study Elective 1: Do no further analysis of doses and risks (inasmuch as the risks from all the radionuclides except 131I are very low). Elective 2: Redo the 131I analyses of the 1997 NCI report to correct the data and programs where errors have been discovered and to include improvements in dosimetry methods or parameters and in uncertainty-analysis methods that have become available since the 1997 report. Elective 3: Await the results of the retrieval and archiving of documents. If new data are found that would appreciably change doses and risks, proceed to full-fledged dose and risk assessment. Elective 4: Proceed with more detailed modeling of doses and risks associated with all radionuclides. If important new data are discovered through the retrieval and archiving of documents, incorporate them. Model the dose and risk uncertainties, incorporating state-of-the-art statistical methods to provide a thorough assessment of uncertainties. Decision 3: Communication plan Elective 1: Provide no additional communication effort. The committee believes that this elective is unacceptable. Elective 2: If the draft report becomes the final product of the feasibility study, develop a public summary as rapidly as possible to add to the feasibility study, to put on the NCI and CDC Web sites, and to distribute as a stand-alone item. Elective 3: Update materials from the 131I NTS Communication Plan to include results of the feasibility study related to other nuclides at NTS and global fallout. The updating should be as rapid as possible.

OCR for page 15
Exposure of the American Population to Radioactive Fallout from Nuclear Weapons Tests Elective 4: Develop an extensive communication plan to make both the feasibility study itself and its results transparent and readily accessible to the American public. The NTS 131I communication project can be used as a blueprint, but both speed and accuracy are important.