The U.S. Million Person Study of Low-Dose-Rate Health Effects

The Million Person Study of Low-Dose-Rate Health Effects (MPS) is one of the largest radiation occupational studies designed specifically to provide understanding of late health effects from chronic radiation exposures. It has been ongoing for 20 years and includes only U.S. workers and veterans, within 30 epidemiologic subcohorts. The MPS has published more than 100 articles to date in the peer-reviewed scientific literature and is implemented by a team of radiation scientists using consistent epidemiologic methods while following precise dosimetry guidelines for determining organ-specific radiation doses from low-LET (linear energy transfer) external radiation, neutrons, and intakes of radionuclides.

Innovations include the use of Medicare and Medicaid claims data to obtain individual information on smoking histories and disease occurrence, the use of the Virtual Pooled Registry Cancer Linkage System for cancer incidence, the development of open-source R-based software to handle big data with billions of data points, and the development of statistical approaches to optimize epidemiologic data harmonization.

Healthy American radiation workers and veterans are being evaluated for cancer and noncancer (e.g., heart disease) mortality and morbidity (incidence) following exposures to low-LET (e.g., gamma) and high-LET (e.g., alpha, neutron) radiations and intakes of radionuclides (e.g., uranium, plutonium, etc.). Importantly, rare cancers and the differences between women and men are being examined. The study is directly relevant to several U.S. worker populations: nuclear utility plant workers, medical radiation workers, industrial radiographers, air crews flying at high altitudes, astronauts, cleanup workers of contaminated environments, radioactive waste management workers, emergency responders to nuclear incidents, nuclear submariners, and military personnel engaged in radiation activities.

The MPS will provide evidence on appropriate risk models for radiation protection programs impacting worker safety, cleanup criteria, medical practice, and energy production. Different linear and nonlinear dose-response models are being evaluated. Evaluation of dose-response relationships are ongoing. Findings of the MPS are important input for evaluating, developing, and implementing science-based guidance and decision-making. The study provides information to workers and their families about the potential level of radiation-­related health effects incurred during their occupations or service. It also engenders worker trust through a commitment to long-term health assessments and transparency in results.

The study is being coordinated by the National Council on Radiation Protection and Measurements (NCRP) and is performed by a team of experts in epidemiology, biostatistics, and dosimetry led by Vanderbilt University Medical Center and the International Epidemiology Institute.

MPS population

The MPS seeks to combine the radiation dose and mortality/morbidity experiences from occupational groups. Workers and veterans have been identified through previously conducted epidemiologic studies, employment records, military records, and regulatory and commercial dosimetry databases.

It is nearly 12 times larger than the Japanese atomic bomb survivor study and covers a broad distribution of cumulative organ doses up to 1 Gy, but predominately up to 300 mGy. Cohorts evaluated to date include workers from Los Alamos National Laboratory, Tennessee Eastman Corporation at Oak Ridge, above-ground military nuclear weapons test sites, nuclear utilities, industrial radiography, and the medical radiation field.

Overall, 160,000 women have been included in the study. The MPS has published on 40 causes of death among more than 600,000 workers. Efforts continue to evaluate 200,000 workers with limited follow-up at large Department of Energy facilities such as the Hanford and the Savannah River Sites, as well as 200,000 U.S. Navy personnel, including nuclear submariners and naval shipyard workers. The MPS has reactivated the study of U.S. radium dial workers, one of the most important and influential radiation effects studies ever conducted.

Vital status

A hallmark of high-quality epidemiologic studies is a low percentage of subjects who are lost to follow-up and a low percentage of deaths with unknown cause. For the MPS cohorts, vital status has been confirmed for 90–98 percent of subjects, and the ascertainment of cause of death is 96 percent. Mortality records from state departments of vital statistics and the Social Security Administration Death Master File are matched to MPS workers by personal identifiers using deterministic and probabilistic algorithms. The National Death Index is used to obtain causes of death (available after 1978), and the SSA Service for Epidemiological Researchers is used to verify alive status. Other tools include online public records, ancestry services, death certificates, and specialized mortality sources.

Because death certificate information does not include a complete assessment of health outcomes, the MPS supplements mortality data with incidence (occurrence) data on medical conditions, not only cancer and heart disease but also on neurocognitive conditions such as Parkinson’s disease. Linkage with Medicare claims data has provided diagnosis codes for diseases related to hospitalizations, skilled nursing care, emergency department visits, office visits, hospice, and home health care.

Nursing facility and home health care assessment data include additional measures that capture cognitive impairment and functional status. Importantly, such linkages provide information on covariate, cofactor, and confounding variables such as smoking, alcohol ­consumption, blood pressure, diabetes, and other health and lifestyle characteristics. Medicare claims data provide information about drug prescriptions, which helps confirm disease diagnoses. Recent approval has been obtained from the Virtual Pooled Registry Cancer Linkage System, which will provide high-quality cancer incidence diagnoses. Linkages with large veteran databases are in the works.

A unique application: Mission Mars

The MPS study continues to provide guidance on radiation risks to astronauts on long-duration spaceflight missions. Nuclear submariners are being studied because they share unique similarities with astronaut crews: sleep deprivation, fatigue, disruption of circadian rhythms, confined spaces, isolation, workload, complex technologies, and protracted radiation exposures.

Noncancer outcomes among this population include heart and cerebrovascular disease, Parkinson’s disease, Alzheimer’s disease and related dementias, motor neuron disease, depression, anxiety disorders, and cognitive neurobehavioral disorders. The possible interaction between multiple stressors and low-dose radiation in causing cognitive outcomes is a focus of study. High-LET exposures to the brain from the intake of radionuclides (an assumed analog to cosmic ray exposures) at DOE facilities like Los Alamos National Laboratory are being quantified and linked to cancer and noncancer outcomes.

Nuclear industry engagement

In 1986, the National Cancer Institute contacted the U.S. Nuclear Regulatory Commission about an opportunity during the revision of 10 CFR Part 20 to lay the groundwork for a registry of radiation workers containing annual doses received. Ultimately, the final rule contained reporting requirements that allowed the collection of information necessary to establish such a registry: the Radiation Exposure Information Reporting System (REIRS). The NRC subsequently issued Generic Letter 94-01, requesting voluntary reporting of occupational radiation data from the industry to help populate early annual doses for individuals. The nuclear industry’s voluntary response in backfilling those key early exposure data was remarkable, and these data are now available for such a study.

Dosimetry, analysis, and findings

Accurate individualized doses are the foundation of high-quality epidemiological studies. The dosimetry for the MPS considers the unique aspects of each cohort regarding exposure scenarios, geometry, energy, radiation type, dosimetry monitoring methods (ion chambers, film badges, thermoluminescent dosimeters, and digital devices), shielding, and bioassays, including associated uncertainty ranges over decades. Comprehensive guidance reports and commentaries have been developed providing step-by-step methods for transparently reconstructing annual high- and low-LET radiation doses to specific organs or tissues for each individual subject across his or her entire employment history based on employment records, bioassays, incident notes, and occupational radiation monitoring. Statistical approaches to addressing dosimetry uncertainty will follow recent NCRP guidance.

Standardized mortality ratio analyses are employed to compare the observed number of deaths from specific causes with the number expected based on the general population rates in the U.S. for persons of the same age and sex over the same calendar years of follow-up. Internal (within-cohort) analyses are conducted to account for the healthy worker effect and to evaluate risk-related dose responses. Statistical approaches include Cox proportional hazards modeling that produces hazard ratios and Poisson regression modeling to estimate excess relative risk and excess absolute risk. Evaluated dose response models include linear, nonlinear, and multimodel. Because of the enormous quantity of data—literally billions of data points—a specialized open-source software called Colossus was developed to enable these and future analyses. When results appear in peer-reviewed publications, all analytic data are deposited in the DOE Comprehensive Epidemiologic Resource.

Early findings based on approximately 600,000 workers and veterans include an apparent association between occupational radiation and Parkinson’s disease (consistent with Russian data), a very low risk of lung cancer and no difference in the risk between males and females (consistent with tuberculosis-fluoroscopy patient data), a risk of leukemia excluding chronic lymphocytic leukemia (consistent with atomic bomb ­survivor data but with wide confidence limits), little evidence for an association with ischemic heart disease (IHD), inconsistent evidence of increases in all solid cancers combined due in part to the heterogeneity of the cancer sites and the influence of uncontrolled confounding due to asbestos exposure and cigarette smoking, little evidence that contemporary radiation workers had higher risks than earlier workers, and evidence that chromosomal damage could be detected 65 years after exposure to fallout from nuclear weapons testing.

The future of the MPS

In these uncertain times, the MPS has struggled with budget cuts that have hindered the timely completion of individual cohorts, data harmonization, and the ultimate pooling of data into one large cohort with enormous statistical power. The American Nuclear Society has joined other professional organizations in support of the MPS. With appropriate funding, the MPS could be completed within the next three to five years.

At current funding levels, it may take 10–15 years since researchers have been hamstrung into essentially conducting one cohort study at a time rather than conducting parallel efforts to get to the full pooled study. Once complete, there will be an incredibly rich dataset and the potential for future vital status follow-up and reevaluations to continue providing science-­based evidence for structured policy- and decision-­making for the nation, industry, military, workers, and public. The availability of the complete dataset will enable the answering of questions yet to be asked. We envision a National Center for Radiation Epidemiology and Biology.

See millionpersonstudy.org for more information and a comprehensive list of publications to date.

Lawrence Daur is a clinical physicist and corporate radiation safety officer at Memorial Sloan Kettering Cancer Center in New York City.