7

Feasibility Assessment:
Overarching Themes and Conclusions

When the Department of Veterans Affairs (VA) charged the committee with this feasibility assessment, it asked the committee to focus on military veterans who had served on active duty at each of the 13 specified Manhattan Project locations between January 12, 1942, and August 15, 1947, excluding those who participated in atmospheric and nuclear weapons testing. Through examination of the historical record, targeted literature searches, information-gathering sessions, and responses to information requests, the committee determined that of the specified locations, only Oak Ridge, Hanford, Los Alamos, Uravan, Dayton, and Alamogordo (before the Trinity test) had a documented military presence during 1942–1947, and therefore, its assessment focused on these sites. Furthermore, although the committee identified several other sites in the continental United States associated with the Manhattan Project that had military presence and produced radiological or chemical wastes (see Chapter 3), these sites were not included in the statement of task and therefore were excluded. The three St. Louis, Missouri, sites listed in the statement of task that were the impetus for the congressional language calling for this feasibility assessment had no military presence during 1942–1947; furthermore, the uranium processing waste that led to the contamination at the West Lake Landfill, St. Louis Airport Project Site, and Coldwater Creek sites occurred after the time frame of the committee’s task.

This chapter considers the final portion of the statement of task: to provide conclusions regarding the committee’s overall assessment of the ability to conduct the congressionally requested study. Any assessment or research should consider what value it would add to the body of knowledge and at what resource cost (time, money, expertise, focus directed

away from other areas, etc.). The first section begins by summarizing and describing the many factors considered when assessing the feasibility of an epidemiologic study to understand the health outcomes of military veterans who participated in Manhattan Project activities, including time and budget estimates. In response to the statement of task, the next section describes other types of approaches and analyses that could be used to estimate the risk of health effects among those veterans who were exposed to radiological or chemical hazards. The chapter ends with final observations regarding the lasting legacy of the Manhattan Project and how concerns about exposures continue to have present-day relevance for communities around its sites.

ASSESSMENT OF FEASIBILITY

This section first describes how the availability and characterization of identified records and information relates to the five key elements required to conduct an observational epidemiologic study. A summary of such studies of workers at Manhattan Project sites are presented along with considerations of methodologic differences for studies of civilian versus military workers. Overarching bureaucratic and other factors that affect this feasibility assessment and the findings and conclusions presented in Chapters 4, 5, and 6 are also summarized. The committee then offers one study design approach that it considered the most feasible given the limited availability of health outcome information. Based on the observations and findings in those subsections, the committee presents its overall conclusion regarding the feasibility of using the information and records identified in Chapters 4–6 to conduct a study as requested in PL 117-168 section 506. After that conclusion, the committee provides an example of an epidemiologic study conducted among a cohort of veterans who served at a similar time as the Manhattan Project that offers insights into similar considerations that may affect a future study.

Key Characteristics of an Epidemiologic Study

An epidemiologic study is a statistical assessment using individual-level data of the relationship between an exposure and a health outcome. The five characteristics needed to conduct an observational epidemiologic study of exposures and health outcomes were outlined in Chapter 2:

• Identification of a population of interest from which a representative sample can be drawn;
• Individual-level exposure assessment of adequate quality;
• Individual-level health outcome assessment of adequate quality;

• Identification of an appropriate comparison group and ability to control for contributing factors that might be related to the exposures and the health outcomes of interest; and
• Sufficient sample size for precise estimation.

Consistent with other National Academies consensus committees that have used these characteristics for epidemiologic assessments of military and veteran populations, these characteristics are not intended to serve as a checklist or scorecard but rather considerations most applicable for epidemiologic research on the health outcomes of radiological and chemical exposures. They guided the committee’s approach to information gathering and provided the basis for identifying gaps in records or information related to the population, exposures, and health outcomes. The key characteristics inform critical thinking about limitations of any observational study to address the question of exposure to Manhattan Project wastes and health outcomes as posed in the statement of task.

Identification of a Representative Population of Interest

The committee found that identifying a veteran population is possible with substantial time, effort, and financial resources. The total veteran population for the specified Manhattan Project sites can be quantified but given the missing unit records and morning reports for several years, especially 1944–1946—the years of highest activity—identifying most individuals based only on military records from the National Archives at St. Louis of the National Archives and Records Administration (NARA) is likely not possible. Other supplemental sources, such as the Atomic Heritage Foundation, may be used to verify information about individuals identified through the National Archives at St. Louis or provide names from the missing records. As the three primary Manhattan Engineer District (MED) sites (Oak Ridge, Hanford, and Los Alamos) had the largest presence of active-duty service members (several thousand), most of the veteran population will have served at those sites. Sites with a smaller presence, such as the Dayton Project, which had a few dozen service members at most, would lack the characteristics of both a sufficient sample size (for analysis at the site level) and a representative sample for any analysis. The committee notes that most Manhattan Project veterans will be male (except the few hundred women in the Women’s Army Corps (WAC)), all will be White due to segregation of the armed forces, and nearly all will have been in an Army unit.

The lack of a roster directly impacts the committee’s ability to complete the fourth subtask in the statement of task—to determine other missions these veterans were involved in before or after the Manhattan Project that may have exposed them to toxic substances and may have contributed to

their overall health risks during their military service. Ideally, an official military personnel file could be used to determine a record of all duty stations and assignments; however, as described in Chapter 4, how many of these files were destroyed in the 1973 fire at the National Archives in St. Louis is unknown. It is probable, especially for technical personnel who served in the Special Engineer Detachment and the WACs, that most only served in the military during the Manhattan Project, so that there would not be any other military missions before or after the Manhattan Project. Without an individual’s official military personnel file, other supplementary sources as described in Chapter 4 may be used to track where an individual was stationed before or after the Manhattan Project. Once the individual left military service, determining their job history becomes even more difficult unless they remained within the energy worker complex. For these individuals, their job history may be requested through Department of Energy (DOE) records management or if they developed certain diseases the Energy Employee Occupational Illness Compensation Program Act (EEOICPA) Program. Without job type and history, it would be impossible to determine even general classes of potential exposures. However, without a complete roster, characterizing information on the primary outcomes of interest (exposures and health outcomes) was limited, resulting in a nearly impossible ability to address other missions and exposures these veterans may have had.

Individual-Level Exposure Assessment of Adequate Quality

The committee found that, although some radiation dosimetry records for 1942–1947 exist, they are specific to certain locations (Oak Ridge, Hanford, Los Alamos) and vary in format and type of information captured. Area monitoring records will be the primary source of exposure information for the early years of the Manhattan Project, as individual external radiation dosimetry and records were not standardized until 1944, and internal radiation monitoring and records were not standardized until after the Manhattan Project was concluded. Moreover, for those dosimetry records maintained by any site, the proportion associated with individual veterans is unknown. Records documenting individual chemical exposures, other than accident reports, which are not representative of typical exposures, are sparse or nonexistent. Similarly, area-level or job-type information to estimate radiological or chemical exposure would be nonspecific and limited for an epidemiologic investigation. Such aggregated exposures would assign everyone in an area or with a specific type of job a single level of exposure when, in reality, individuals would have experienced a range of exposure concentrations and frequencies, which could result in bias due to misclassification of individual exposures. Grouping all military veterans across the Manhattan Project sites as a single cohort will likewise result in misclassification. Therefore, based on

the limitations of the individual-level exposure information and records that are available, this criterion is not met. Without knowing the full job history of an individual, the magnitude and extent of any exposure (radiological, chemical, or combined) is uncertain.

Individual-Level Health Outcome Assessment of Adequate Quality

The only standardized source of health information is death records. Systematically collected and relatively high-quality information on cause of death can be obtained from the National Death Index for deaths after 1979 and individual death certificates from state agencies for deaths before 1979. Because the United States does not have a national health system, prospective longitudinal health records for individuals are not available, and so the incidence of most health outcomes cannot easily be studied. While state and regional cancer registries collect incidence data, most started decades after the Manhattan Project ended, making inclusion of the full population of veterans from the 1940s unlikely. Veterans have a variety of medical care options (VA, Department of Defense (DoD), private insurance), with no systematic way to collect their health data. VA stated that medical records for the Manhattan Project veterans are unavailable, but the committee suspects that some records may be available for those who used VA health or benefits. However, this subset is unlikely to be representative of all Manhattan Project veterans and using it alone would introduce bias into a study of this entire population.

Other sources of records explored by the committee are also not systematic or inclusive of all individuals of interest and would be of limited use in an epidemiologic study. These sources include records for DOE workers who were veterans and worked for DOE post–Manhattan Project and records for workers covered by compensation programs, such as EEOICPA, which collect medical records for specific cancer outcomes (but these records are not available for research).

Identification of an Appropriate Comparison Population or Group

An important consideration for any epidemiologic study is identifying an appropriate group (the fourth characteristic) with which to compare a hypothesized effect of an exposure. Potential comparison groups for the veteran population may be civilians who held similar jobs to those of the veterans, active-duty service members who held similar jobs but did not work at Manhattan Project sites, nonveteran members of surrounding communities, or even the general population during that era. The comparison group should be as similar to the target population as possible to account for contributing factors (confounders) in an exposure–health outcome pathway. Lack

of knowledge about or ability to control for confounding is an important limitation for any observational study. Examples of such contributing factors are smoking, work and job history, and—specific to the fourth point in the statement of task—other missions these veterans were involved in before or after the Manhattan Project that may have exposed them to toxic substances and contributed to their overall health risks during their military service.

Sufficient Sample Size for Precise Estimation

The final key consideration for an observational epidemiologic study is sufficient sample size to obtain adequately precise estimates of association relative to the expected magnitude of effect. Without conducting an actual review of unit records and morning reports to create a roster of individuals at Manhattan Project sites, the committee is not able to determine the exact number of active-duty military veterans at the sites, but it has estimated that at most 10,000 military personnel participated in the Manhattan Project overall. Likewise, without conducting a formal statistical power calculation for specific exposure–outcome associations, the committee is not able to estimate the power of a study in this population to detect an effect if there is one. However, prior epidemiologic studies of workers at the sites have examined a variety of cancer and noncancer mortality endpoints, and the imprecision of estimates from prior studies illustrates the likely low statistical power of studies of similar size to detect an effect if there is one, particularly for analyses of rare health outcomes, such as specific cancer types.

Limitations in exposure and outcome classification also can undermine an epidemiologic investigation’s ability to detect an adverse effect of exposure. The committee’s observations and findings in the preceding chapters highlight the difficulties in reconstructing individual-level estimates of specific exposures and concerns about the quality of mortality information. The statement of task lists multiple sites, each of which had markedly different activities (and types and frequencies of exposures) and numbers of personnel; however, to investigate the variations of exposure–outcome associations by site would result in smaller sample sizes for each site and less power and precision to detect an association. Similarly, individuals were recruited to and separated from military service throughout and after the period of interest of January 12, 1942–August 15, 1947, and an individual may have been stationed at both locations that are part of the 13 specified sites and others. Therefore, consideration of time-varying aspects of exposure and disease would involve smaller subsets of this population.

Epidemiologic Studies of Workers at Manhattan Project Sites

The committee did not identify any epidemiologic studies specifically of military veterans at any of the Manhattan Project sites. Epidemiologic studies of the larger workforce, which likely included some veterans, at the three primary sites (Oak Ridge, Hanford, and Los Alamos) have been published and may provide helpful guidance regarding methods and processes for cohort identification, dose reconstruction, and outcome ascertainment. While many of these studies examined exposure to radioactive materials or other Manhattan Project–related exposures, the inclusion criteria are usually much broader than the period of interest, and little to no information was available in the study methods regarding the inclusion or exclusion of military veterans, number of veterans followed, or specific health outcomes for veteran subgroups. Military or veteran status was not systematically collected on most exposure or employment records as far as the committee can determine from its information gathering. All sites had substantially more civilian than military personnel, so it is not surprising that this was not used as a stratification. Studies of nuclear industry workers, particularly for workers from the Manhattan Project era, note challenges to clearly identifying an increased risk for many diseases, including the healthy worker effect, inconsistent monitoring for internal and external ionizing radiation exposure, and study samples that include workers with substantially varied levels of exposure (Wakeford, 2009).

The epidemiologic studies of worker populations at the three primary Manhattan Project sites relied on dosimetry records that were more accurate and standardized and covered nearly every worker than those dosimetry records available for a subset of the population and the limited period in the early 1940s. Given the improved exposure assessments available after the 1940s, findings from studies of worker populations that likely included some veterans would be more accurate and generalizable to veterans who served as scientists and technical experts. As explained in Chapter 5, assessing chemical exposures was even more difficult than radiological exposures at these sites, even after the 1940s, and studies of chemical exposures and health outcomes in these worker cohorts used time-spent-in-job categories as a surrogate for these exposures (see, for example, Wing et al., 1993).

Table 7-1 summarizes the availability of information regarding the five subtasks in the statement of task by the 13 sites specified in it. As only six had a documented military presence, only they have relevant information related to each subtask. For elements other than military presence (the first column), the committee denotes whether the information is possible, limited, or unknown. Possible indicates that the committee identified evidence that records exist and are available, but the

TABLE 7-1 Availability of Each Statement of Task Element by Specified Site

NOTES: Possible indicates evidence that records exist, but the amount of time and effort required for individual assessment for the element is unclear. Limited indicates evidence of records, but these were incomplete and would not cover all military personnel or the full time period. Unknown indicates records with this type of information were not identified, and it remains unclear whether these exist or can be located.

^a Demographic information is likely available from a community cohort study conducted of all residents of Uravan, Colorado 1936–1984; some of them are likely veterans.

^b Internal dosimetry monitoring did not start until 1945 for most sites and was restricted to only individuals who were likely to have been exposed. More internal monitoring data is available 1947 onward.

^c Data for 1942–1943 are largely missing; measurements for 1944–1947 were more systematically captured.

^d Cause of death is feasible to obtain via linkage to the National Death Index for deaths after 1979 or by obtaining individual death certificates for deaths prior to 1979, assuming the availability of individual identifiers such as name, Social Security number, and date of birth.

^e Alamogordo, New Mexico, until July 15, 1945 only.

time and effort required to obtain and extract the information is unclear. Limited indicates that the committee found evidence of records, but these were incomplete and would not cover all military personnel or the full period. Unknown indicates that the committee did not find records, and it is unclear whether these exist or can be located. Much of the information is limited or unknown, as it was not systematically recorded. Some demographic characteristics, such as race and sex, may be assumed, as nearly all military personnel involved were White and male (with only a small number of women who were WACs), and birth date could be ascertained from official military personnel files or other supplementary sources; however, other demographic characteristics, such as education, would be unknown, as would other confounding factors, such as smoking and other military and job histories. Several military characteristics would also be known, especially if the approach in Chapter 4 were followed for requesting unit records and morning reports to generate a roster for the years that these records are available. Branch (all known to be Army) and units would be known as part of the request, and the records would provide rank and time-dependent factors for individuals, including dates of transfers, promotions, and separation from service.

In sum, the committee found that key information that would be needed to conduct an observational epidemiologic study on health outcomes in veterans resulting from radiological and chemical exposures during Manhattan Project activities is not available at the individual level or at a level of detail necessary for epidemiologic analyses.

Bureaucratic and Other Factors Affecting Feasibility

This section provides the committee’s overarching assessment of crosscutting issues and themes that informed and affected its information gathering. Central to many themes are issues of bureaucracy and complexity rooted in the historical record and postwar records stewardship. Committee efforts to learn about both factors yielded a variety of responses, ranging from bureaucratic silence to helpful and detailed overviews supported by organization leaders and shared by staff archivists, historians, and technical experts. However, one overarching observation arose: no single source exists at any location, geographic area, agency, or organization, or in any report or other documentation, that contains even minimal information about the individual participating service members, their potential exposures, or possible health outcomes. Moreover, even under the best of circumstances, as when leaders of federal agencies are receptive to information requests, access is challenging in large part due to incomplete records and the dispersal of exposure and health information on individual veterans across numerous agencies and even across offices within agencies. One

TABLE 7-2 Manhattan Project Sites and Department of Energy Offices That Oversee Records and Records Management

^a Alamogordo, New Mexico (through July 15, 1945) is one of the sites, but no DOE records office was identified. The Defense Threat Reduction Agency (Department of Defense) maintains the exposure records.

^b While these are the records management offices at the listed site, some facilities at these sites that closed, such as hospitals, transferred their records to DOE’s Office of Legacy Management.

example is DOE, the predominant agency that oversees former Manhattan Project sites. Table 7-2 shows a summary of its organizational oversight. One office does not always retain all records for a specific site. For example, records for Los Alamos are overseen by the National Nuclear Security Agency Records Office, but its 1940s hospital records have been transferred to DOE’s Office of Legacy Management. With no response from the Office of Legacy Management, the committee was unable to ascertain what information is in the records for several sites.

More broadly, the committee learned that the complex history of the Manhattan Project itself yields factors that impact the maintenance of records and the feasibility of access to them, including:

• Its top-secret character, which involved strict compartmentalization and varied recordkeeping across sites to maintain secrecy and security; the varied disposition of sites and their related records from after World War II to the present day and, in some cases, destruction of records.
• The multiplicity of qualitative and quantitative record formats, ranging from handwritten sources to microfiche and microfilm (some but not all of these surviving sources have been digitized to enable more efficient study).
• Military personnel were exposed to many toxic substances, with few of these recorded. (Therefore, epidemiologic research to determine what specific exposure(s) “caused” an outcome is often limited, and many exposures produce similar types of symptoms or outcomes, making it difficult to link one exposure to one health outcome.)
• Military unit records and many morning reports are missing for key years (1944–1946).
• Millions of individual official military personnel files were destroyed in the National Archives at St. Louis fire of 1973.
• The movement of military personnel among and between sites during 1942–1947 further complicates exposure assessment.
• Neither DoD nor VA hold records specific to military veterans that served on active duty at any of the Manhattan Project sites.

Of note, although the committee encountered several difficulties obtaining information from some agencies, that lack did not impede its ability to respond completely to its statement of task or form its conclusions.

One example of the effort required to search, identify, access, and synthesize hundreds of thousands of documents and records was the Los Alamos Historical Document Retrieval and Assessment (LAHDRA) Project. The primary purpose was to identify all available information concerning past releases of radionuclides and chemicals from Los Alamos National Laboratory (LANL) (Widner et al., 2010), beginning with the Manhattan Project. The LAHDRA report summarizes the historical processes at Los Alamos, materials used, materials likely released offsite, development of residential areas in Los Alamos, and the relative importance of identified releases in terms of public health. The project took a decade to complete (5 years each for initial research and report completion). It involved more than 30 full-time researchers, all of whom possessed security clearances, from a variety of fields, including history and different scientific fields. It demonstrates the massive and multimillion-dollar effort that may be needed to characterize relevant information for just one site in the statement of task, let alone all six that had a documented military presence.

During its information gathering at Los Alamos, the committee learned that of the 8,372 documents entered into the LAHDRA database, approximately 265 pertain to 1942–1947. Of those, it is unclear how many contain information about military veterans (Templeton and Gregonis, 2024). The need to identify military personnel at each Manhattan Project site even after creating a roster of individuals and identify and obtain their dosimetry and other exposure records and extract relevant information into a database, adds another layer of complexity and effort. Therefore, constructing a cohort spanning multiple sites would be a substantial undertaking that would most likely require a level of effort comparable to or more than that for LAHDRA.

A key observation from the committee’s information-gathering efforts and the LAHDRA experience is the complexity and importance of data and record management. The fragmentation of records and other sources of information among sites and over time and changes in site stewardship (and records management), including management by DOE headquarters, create additional barriers to conducting studies using information from these records. The committee stresses the importance of centralized repositories to maintain and archive records for research and other efforts, such as this feasibility assessment. As illustrated by the LAHDRA effort to search, identify, access, and synthesize hundreds of thousands of documents and records, the many complex bureaucratic, historical, and present-day factors mean that any study of Manhattan Project veterans would require an estimated budget of millions of dollars to support a years-long effort. It would necessarily involve extensive human resources (e.g., teams of professionals with required disciplinary knowledge, skills, and abilities in archival research and data collection), physical and digital record access (travel combined with limited and varying online access, and familiarity with and adherence to the many different policies and procedures of institutions holding these records collections), and extensive data analysis and synthesis. From a resource perspective, such a study would be equal to or perhaps much greater than LAHDRA.

Even within LANL, the committee’s information request and subsequent presentation took a concerted effort of multiple staff, performed in kind, totaling over 420 hours worked, across two groups: the Institutional Records Management, which assigns and tracks records disposition and links LANL to Federal Records Centers/NARA compliance, and National Security Research Center, which serves as a classified library and digitization program to LANL researchers, particularly in Weapons Programs. Institutional Records Management’s search process included keyword searching through records databases, indexes examined for date range, and calls to other organizations on site who generate records on relevant topics. The National Security Research Center’s search process included

LAHDRA report materials scrutinized, keyword search through databases cross-referenced to date range to quality check LAHDRA reports, investigated materials accessioned 2010–2024 not covered by LAHDRA, physical spot-checking throughout the vault, and a historical literature review of internal reports. Physical records of varying formats remain that the records managers believe could have information relevant to the committee’s tasking, but more time would be needed to go through the inventory should such a study be undertaken. This massive effort is possible because LANL has existing resources to perform such a task, including staffing and budget for responding to external and in-kind requests. However, not every site listed in the committee’s statement of task may have these available resources.

Given the many factors affecting the feasibility of obtaining identification, exposure, and health outcomes records for veterans who served at Manhattan Project sites 1942–1947, the committee made several findings and three conclusions, presented in Box 7-1 and discussed in greater depth in Chapters 4, 5, and 6.

Mortality Analysis

Military personnel who were scientists or had technical expertise and jobs were usually assigned to a technical detachment or unit within MED and worked side by side with civilians. Their exposures, therefore, cannot be differentiated, but nothing suggests that the associations with health outcomes would be different for civilian and military workers with similar jobs and at similar sites. As discussed in Chapter 4, identifying individual veterans would be difficult, time consuming, and incomplete, and their numbers are relatively few (less than 1%, or 10,000 at most) compared with civilians in the Manhattan Project. Moreover, a subset of individuals (number unavailable and difficult to identify) continued to work at Manhattan Project locations for DOE or other contracted companies after completing their military service (Seidel, 1993). Their experiences would be a composite of both military and civilian exposures over a different length of time (during the Manhattan Project and after) and location (physical proximity of post–Manhattan Project work). Although DOE employment records may be available for a proportion of veterans who also had careers in DOE, this would likely create a high risk of selection bias and exposure misclassification because those who remained would likely have had continued and higher cumulative exposures.

As concluded in Chapter 6, a mortality study is possible if a roster is developed with sufficient personal identifier information. In such an analysis, epidemiologists compare the observed number of deaths in a study population to the expected number of deaths from a reference population

to compute standardized mortality ratios for causes of death of interest. The resultant ratio can be standardized using basic demographic information (such as age, sex, and calendar period) and offers crude information as to whether mortality in the study population is higher, lower, or similar to the reference population. Although this type of analysis is possible, the committee determined that it would be of limited use in this context due to key limitations. First, reliance solely on standardized mortality ratios may mask any true adverse outcomes, as confounders are not accounted for in computing the standardized mortality ratio. Second, bias may arise if no comparable reference population is available to compute expected numbers of deaths; epidemiologists often use regional (i.e., county or state) or national reference populations, but for this widely dispersed group of veterans, such an approach will be less reliable than for a population that is more precisely defined by geography.

Conclusion Regarding Feasibility of an Epidemiologic Study of Manhattan Project Veterans

Overarching Conclusion 1: The committee considered several aspects of feasibility for conducting an epidemiologic study and concluded that such a study on the relationships between radiological and chemical exposures from Manhattan Project activities and health outcomes in veterans who worked on the project is not feasible. This conclusion is based on the following:

• the incomplete availability of unit records and official military personnel files to create a full roster of individual veterans who served at the Manhattan Project sites of interest;
• the lack of systematically collected exposure information that can be linked to individual veterans; what information is available is limited to only some radiological exposures and no chemical exposures (except for accident reports which are high dose);
• the only systematically available health outcome is cause of death; and
• finally, other key elements, such as individual-level demographics and confounders, that would be needed to conduct an observational epidemiologic study are not available or severely limited.

An Informative Comparison: The Five Series Study

Although military veterans who participated in nuclear weapons tests are out of scope, findings and conclusions from The Five Series Study: Mortality of Military Participants in U.S. Nuclear Weapons Tests (IOM, 2000) are relevant to the feasibility of performing both exposure assessments

and health outcomes studies of Manhattan Project veterans. The Five Series Study was primarily concerned with veterans exposed to radiation from atmospheric nuclear weapons testing during the 1950s and 1960s. Unlike Manhattan Project veterans, these veterans and their participation in these activities were compiled in a central location, the Nuclear Test Personnel Review, which includes the Trinity test. The Five Series Study was able to identify a cohort of 68,168 veterans and a referent cohort of similar size. The Nuclear Test Personnel Review database contains a dose assignment for each participant, usually derived through reconstruction based on duty assignments. Based on their examination of that dosimetry data, that committee did not believe that these data were appropriate for the individual-specific dose assignments necessary for epidemiologic analysis and found that future analysis and revisiting of the dose data with modern and standardized methodologies for reconstructing dose could be performed for some individuals but would be a major effort. That committee also found that even for veteran mortality from leukemia, the cancer that is most consistently linked with radiation, a study cohort four times the size of the one available would have been needed to detect a statistically significant effect.

The Five Series Study identified a cohort about seven times larger than the estimated total Manhattan Project veteran cohort and an existing database of individuals and their exposure data and still concluded that an epidemiologic study was not feasible. This further supports the committee’s conclusion that an epidemiologic study of Manhattan Project military veterans is not feasible.

RISK ASSESSMENT AS A POTENTIAL ALTERNATIVE STUDY DESIGN

As the committee concluded that an epidemiologic study as directed in PL 117-168 section 506 is not feasible given the lack of availability of necessary data, the final subtask of its statement of task requires that “the committee will explore possible alternative options for understanding the health effects on the veterans due to exposures from the Manhattan Project.” While alternative study design options to an epidemiologic study are presented here, the committee notes that these are not as rigorous as an epidemiologic study and are subject to many of the same limitations regarding availability of data and what types of analyses may be conducted. Based on the lack of complete exposure and health data, no type of study (epidemiologic or an alternative design) would permit evidence of a causal relationship between individual exposures and specific health outcomes. Furthermore, as discussed above related to LAHDRA and the Five Series Study, any research study would be expensive and resource intensive.

The committee considered alternative designs that could address similar issues of human health effects from exposures to Manhattan Project–related

activities as in the statement of task. While alternative study designs such as toxicologic-based approaches (cell- or animal-based studies) could be used to better understand mechanisms of potential exposures, such an approach is tangential to understanding human health effects. As described above in Epidemiologic Studies of Workers at Manhattan Project Sites, studies of workers may provide helpful guidance regarding methods and processes for cohort identification, dose reconstruction, and outcome ascertainment but they have generally used a much broader time period of exposure and without knowing the specific jobs and potential exposures of veterans or number of veterans with those jobs and how those jobs and safety measures changed over time, worker studies may only be representative of a small subset of veterans who had radiological exposures and who remained in the energy worker complex and not all veterans who participated in the Manhattan Project. Likewise, although the committee concluded that a mortality analysis is feasible if individuals were identified, it would still be quite limited and nonspecific to Manhattan Project–associated exposures. Therefore, to best address the driving association of interest in the statement of task and considering issues of feasibility and efficiency of resources, the committee finds that risk assessment is the most feasible alternative study design.

In a risk assessment, exposure data at the individual or group level are combined with known exposure-outcome relationships, obtained from other data sources and contexts, to predict the potential health impacts of the exposures. An important difference from an epidemiologic study is that no actual health outcome data specific to the exposed individuals are required or used in a risk assessment. A risk assessment may be used to estimate adverse health outcomes, given the state and accessibility of relevant records and information on Manhattan Project veterans and can often involve similar data gathering as for an epidemiologic study, including detailed information on historical site activities, understanding of nuclear and chemical processes that occurred, and associated exposures. Risks may also be discussed qualitatively, such as the relative contribution to risk of chronic diseases depending on magnitude as well as duration of exposure that would be required of an epidemiologic study and offer additional insight into the feasibility of conducting this type of study (e.g., NRC, 1994a). While limited in comparison, a risk assessment would still be able to provide some approximate bounds on potential health outcomes for veterans of the Manhattan Project, even if there are large uncertainties.

This approach to assess the health outcomes of exposure to radiological and chemical materials related to Manhattan Project activities on military veterans would draw on information regarding known exposure–health outcome associations and dose–response relationships from epidemiologic studies of nuclear energy workers and other workers covering sites and timeframes of interest. A research team could consult a variety of references

and guidance for (and examples of) human health risk assessments (e.g., EPA 2001, 2005; Fjeld et al., 2023; Killough et al., 1998; NRC, 1994b, 2009; Till et al., 2002; Till and Grogan, 2008). In brief, the overall process for conducting a risk assessment typically involves articulating the scope of the study, assembling information on the characteristics of the real or hypothetical population of interest, identifying potential sources and types of exposure, calculating the magnitude (and, potentially, distribution) of exposure (or dose) in the population, and leveraging existing exposure–health outcome associations to assess the anticipated health outcomes (i.e., risk) from the exposure (including evaluation of uncertainty) (NRC, 1994b, 2009). Risk assessments can range in rigor depending on the study objective and available data, models, and resources from conservative screening analysis to progressively more refined, realistic, and comprehensive site- and population-specific assessments; thus, a risk assessment is often possible, even with limited information and resources, although the level of confidence and, therefore, the scientific robustness of the result may vary.

Given what is known about the dose–response relationships between radiation and cancer, radiation and mortality, and exposure/dose and mortality/morbidity for many chemicals (see Chapter 6), a risk assessment approach may offer a way to calculate the estimated risk of adverse health outcomes associated with exposures to radiological materials and chemicals. Historical data (e.g., dosimetry records and chemical use) could inform dose reconstruction (NCRP, 2018), which can be combined with modern risk models for radiological and chemical exposure (e.g., Eckerman et al., 1999; EPA, 2011; ICRP, 2022) to estimate veterans’ risk. However, access to and availability of the historical exposure data will affect dose-reconstruction and impact the robustness of the analysis. Care must be taken in selecting a risk model and interpreting the results, given the various sources and magnitude of uncertainties encountered throughout the assessment process. As indicated in Table 7-1, exposure data for veterans are limited; however, valuable information can still be gained.

Overarching Conclusion 2: With careful consideration of the strengths and weaknesses of the exposure and health data, the committee concludes that a risk assessment could be conducted to estimate the potential health risks for Manhattan Project military veterans.

This conclusion of using risk assessment as an alternative study design to an epidemiologic study is not an indication or endorsement that it should be undertaken, especially without strong consideration of weighing both how a risk assessment will be used and the resources required, which was beyond the committee’s task.

FINAL OBSERVATIONS

While the Manhattan Project resulted in many scientific and engineering advancements, it left a profound legacy of human health and environmental impacts that are challenging to document and assess, particularly given that military veterans made up less than 1% of its personnel. The committee found that no single source or database contains all the information to address the statement of task, owing to the fragmented historical record of who participated, what exposures were known to be present, recognized health outcomes of concern, and physical disposition and stewardship of the individual personnel and site records over time. As historians and researchers discover more records from the Manhattan Project era and repositories make them available for research in digital form and using new technologies—dependent on federal schedules of disposal and local, regional, and national preservation interests—more could be understood about this landmark period of history beyond the current narrow task. The sources described in this report that identify the veteran population and individual exposures and health outcomes, and information about the custodial organizations of those records and the general processes that would be required to access them, may also be of utility to other scientific or historic endeavors.

The committee’s determination of the inability to carry out an epidemiologic study of health outcomes among veterans with Manhattan Project–related exposures should not be taken as an indicator that the Manhattan Project did not have long-term adverse impacts on the many workers and surrounding communities or that long-term studies of such exposures on communities are impossible. It is likely, however, that fully understanding and remedying the health and environmental impacts of the many Manhattan Project exposures may require approaches that differ greatly from those of classical epidemiology. The available exposure information for veterans may greatly limit the ability to disentangle the health outcomes arising from chemical, radiological, and mixed exposures. The committee’s experience with information gathering and substantial difficulties in contacting the appropriate offices or personnel for these information sources to understand the content and state of the many historical records—even with a congressional mandate—led it to find that obtaining records and extracting needed information would require substantial resources and interagency cooperation. For example, the paucity of knowledge about community health outcomes stands in sharp contrast to the more extensive information available on military veterans who participated in atmospheric and nuclear tests (“atomic veterans”) and workers engaged in Manhattan Project activities. Expanding our understanding of community impacts of Manhattan Project exposures is an important aspect to gaining a full understanding of its widespread and long-lasting health outcomes.

REFERENCES

This page intentionally left blank.