The Health Physics Society 62nd Annual Meeting
July 9-13, 2017 - Raleigh, N.C.

Single Session

[Schedule Grid]

MPM-C - Special Session: Low Dose Occupational Epidemiology: The Importance of Domismetry and Statistics in the Million Worker Study and the Mallinckrodt Chemical Works (MCWL) Cohort

306 AB   15:00 - 17:30

MPM-C.1   15:00  The Past Informs The Future: The Million Worker Study And The Mallinckrodt Chemical Works Cohort JD Boice*, National Council on Radiation Protection; Vanderbilt University

Abstract: The Million Worker Study (MWS) started nearly 25 years ago, and continues in a stepwise fashion (one study at a time). It contains workers at DOE Manhattan Project facilities, NRC regulated nuclear power plants, industrial radiographers, DOD nuclear weapons test participants, and physicians and technologists exposed to medical radiation. The purpose is to fill the major gap in radiation protection: what is the risk when exposure is received gradually over time, and not briefly as was the case for the atomic bomb survivors? Studies published or planned this year include leukemia (and dosimetry) among atomic veterans, leukemia among nuclear power plant workers, mortality among workers at the Mallinckrodt Chemical Works, and a comprehensive NCRP Report on dosimetry for the MWS. Mallinckrodt Chemical Works has a singular place in history: the 40 tons of uranium oxide produced at MCW was used by Enrico Fermi on December 2, 1942 to produce the first man-made sustained and controlled nuclear reaction, and the atomic age was born. 76 years later, we followed the over 2,500 workers for mortality, and reconstructed dose from 5 sources of exposure: external gamma rays from the radioactive elements in pitchblende, medical x-rays from occupationally required chest examinations, intakes of pitchblende (uranium, radium and silica) measured by urine samples, radon breathe analyses and dust surveys overseen by Robley Evans and Merrill Eisenbud. The results are relevant to NASA and missions to Mars because radium deposited in the brain releases high-LET alpha particles: the only human analogue for high energy, high Z particles (galactic cosmic rays) traveling through space. The unique exposure reconstructions allow for multiple evaluations, including dust.

MPM-C.2   15:20  Dosimetry Is Key To Good Epidemiology: Dose Reconstruction at Mallinckrodt Chemical Works considered 5 Different Source Exposures and the inhalation of pitchblende dust ED Ellis*, ORAU ; JD Boice, National Council on Radiation Protection; Vanderbilt University; RW Leggett, Oak Ridge National Laboratory

Abstract: The Mallinckrodt Chemical Works (MCW) was the earliest uranium processing facility in the Manhattan Project starting in 1942. Even then, there was concern about the possible health effects resulting from exposure to radiation and to pitchblende dust. This concern was well founded since the facility processed Belgian Congo pitchblende ore that was up to 60 percent pure uranium with high uranium-235 content. This ore also contained up to 100 mg of radium per ton which increased to 300 mg per ton during processing. The workers were exposed to external gamma radiation as well as internal radiation from the inhalation and ingestion of pitchblende dust (uranium, radium and silica). Alpha particles were emitted during the decay of radium and its progeny. Multiple sources of exposure were recognized and data collected to allow organ dose reconstruction to a degree which is unique for an epidemiologic study. Personal film badge measures were available from 1945 to capture external exposures. Additional external exposure included 16,790 occupational medical x-rays required for employment, and 210 radiation exposure records from other facilities both before and after employment at MWC. The organ dose calculations considered the organ specific coefficients that accounts for photon energy and also the job-specific orientation of the workers to the radiation source during processing. Internal intakes of uranium and radium were based on 39,451 urine bioassays and 2,341 breath radon measurements, utilizing ICRP publication 68 biokinetic models. This comprehensive dosimetric approach follows methods outlined by the NCRP Scientific Committee 6-9 for the Million Worker Study. Annual doses from all sources were calculated for 7 organs: lung, brain, heart, kidney, colon and red bone marrow. The largest organ dose was to the lung (mean 59.9 mGy; max 829.5 mGy) followed by the heart (mean 47.5 mGy; max 737.6 mGy). Evaluation and adjustment for individual cumulative measures of pitchblende dust inhalation were made for lung and kidney diseases.

MPM-C.3   15:50  Dust Up: Is Kidney Disease Related To Radiation Dose, Inhaled Pitchblende And Silica, Or Both? AP Golden*, ORAU ; JD Boice, National Council on Radiation Protection; Vanderbilt University; ED Ellis, ORAU; SS Cohen, EpidStat Institute; MT Mumma, International Epidemiology Institute; DJ Girardi, ORAU

Abstract: An extended follow-up with comprehensive dose reconstruction was conducted of 2,514 white males employed at the Mallinckrodt Chemical Works in St. Louis, the earliest uranium processing facility in the United States, between 1942 through 1966. Workers processed pitchblende, a naturally occurring radioactive material containing uranium and silica. Over 75% of the workers died during the up to 70 years of follow-up with cause of death known for 99%. The diminution of the healthy worker effect over time was apparent with the standardized mortality ratio (SMR) 0.94 for all causes, 0.97 for all cancer and 0.89 for all heart disease.Organ dose reconstruction was unique for an epidemiologic study with the inclusion of five different sources of external and internal exposures. Further, the contributions of uranium dust and silica exposures to risk could be investigated because measurements were available for 1,230 workers to calculate cumulative dust exposure. Average doses to organs of interest (lung, red bone marrow, kidney, brain, heart, and colon) were 30-60 mGy, with the maximum dose (830 mGy) to the lung. The only significant radiation dose-response relationship was for kidney cancer (RR 1.88 [95% CI 1.13-3.11] at 100 mGy) and was suggested for nonmalignant kidney diseases(RR 1.36 [95% CI 0.98-1.89] at 100 mGy). No notable observations were seen for any other cancer, leukemia or heart disease. Conceivably, dust could distort the association between kidney disease and radiation, related to non-radiogenic properties of uranium and silica (both kidney toxins), but the evidence was inconclusive. Analyses between lung cancer and dust exposure are ongoing.

MPM-C.4   16:20  Statistical Consequences: Choice of model and adjustment factors matter AP Golden*, ORAU ; SS Cohen, EpidStat Institute; CL Chen, Vanderbilt University

Abstract: Comparing radiation risk estimates across epidemiologic studies can be challenging as various analytic approaches are used. Different statistical models are employed for time dependent analyses, and different software packages can provide different results depending on the model chosen: Cox proportional hazards modeling or Poisson regression (including stratified and piecewise exponential models). It is unclear whether these methods provide comparable risk estimates per unit absorbed dose for the hazards ratio (HR) and excess relative risk (ERR), i.e. risk estimates frequently reported in epidemiologic studies. Further, the choice or availability of adjustment factors, such as duration of employment and socioeconomic status, can change the estimates of radiation risk. To address these issues, a comparative analysis was performed to evaluate three statistical models (Cox, Poisson, and piecewise exponential), three software packages (Epicure, SAS, and R), and inclusion/exclusion of covariates (duration of employment and socioeconomic status (SES)). Considering comparable methods for tracing, vital status determination, cause of death ascertainment and comprehensive dose reconstruction, five cohorts from the Million Workers Study were selected: Mallinckrodt Chemical Works (MWS), Nuclear Power Plant workers, Industrial Radiographers, workers at the Mound Ohio facility, and Atomic Veterans. This presentation will focus on the MCW cohort. Preliminary results indicate that differences in risk estimates occur when different statistical models are chosen. Inclusion or exclusion of duration of employment and/or SES also can significantly change risk estimates, with each of the cohorts affected differently. Other key analytic decisions, which are often not well described in the study methods, were also identified as potential hindrances to comparing results across studies, i.e., the approach for lagging dose, the accumulation and binning of person-time, and the underlying timescale for the model. These findings inform optimal approaches for performing statistical analyses for epidemiologic studies of radiation health effects and for making valid comparisons across studies.

MPM-C.5   16:50  Panel Discussion

[back to schedule]