=> DATA FILE SET-LEVEL: PNSMFLA1
|
=> indicates the hierarchical level of the current display Further description of the hierarchical organization of CEDR Structured Documentation
|
Description:
This analytic data file set consists of one analytic file generated for the leukemia mortality case-control studies by Mary Schubauer-Berigan et al., National Institute for Occupational Safety and Health (NIOSH) and published in Radiation Research (chronic myeloid and acute leukemia) and British Journal of Haematology (chronic lymphocytic leukemia) in 2007, of workers at the Department of Energy (DOE) facilities Hanford, Los Alamos National Laboratory (including Zia), Oak Ridge National Laboratory (ORNL), and Savannah River Site (SRS), as well as the Portsmouth Naval Shipyard (PNS) in Kittery, Maine. A record in this data set represents one study subject, either a case or a control, in the NIOSH multisite leukemia case-control study.
I. Background
The multi-site leukemia case-control study (LCCS) consists of 1284 study subjects from four DOE sites (Hanford, Savannah River Site, Oak Ridge National Laboratory and Los Alamos National Laboratory, including Zia Company employees), and one nuclear naval shipyard (the Portsmouth Naval Shipyard). The purpose of the study is to determine whether workplace exposure to ionizing radiation is associated with risk of leukemia. Although most of the 257 leukemia cases identified in this study have been included in previous cohort studies, a case-control study allowed incorporation of more-accurate estimates of exposure than is typically possible at the cohort level. For example, plutonium bone marrow doses were calculated for most workers who were monitored for Pu exposure, and doses from work-required medical x-rays were incorporated into the external dose assessment. In this multi-site study, where different dosimetry practices have been followed over time and across facilities, a very important advantage is the ability to calculate dose as consistently as possible. Adjustment for more confounding variables is also possible in a case-control study. For this study, exposure to leukemogenic solvents (benzene and carbon tetrachloride) and smoking status were estimated for each worker. In addition, follow-up for several cohorts was expanded (to 1994 for Hanford and SRS, and to 1996 for PNS). Information about the data collection and exposure assessment methods used in this study are available in the references that follow.
II. Main study outcomes
All leukemia cases in the study were identified as underlying cause of death (UCOD) in the original cohort studies or extended follow-up. Although statistical power might have been increased (particularly for chronic lymphocytic leukemia (CLL)) by including non-underlying leukemias, non-underlying causes of death were not available for LANL and Zia workers. The main study outcomes for this study consist of two mutually exclusive groupings: Leukemia excluding chronic lymphocytic, and CLL. To reduce the numbers of cases with indeterminate CLL status, information was searched in the patient medical records to more specifically identify the leukemia subtype. In addition, claim files from the NIOSH Office of Compensation Analysis and Support were searched to obtain more specific information for those whose leukemia subtype remained uncertain. CLL was defined as all cases with a UCOD of 204.1. Non-CLL was defined as cases that are acute (or other non-chronic) and myeloid, monocytic (or other non-chronic) leukemias. There were 43 cases defined as CLL (n=43), and 184 defined as known to be non-CLL (n=184). Indeterminate (n=30).
Because most leukemias of known cell type and severity are not CLL, the 20 leukemias of unknown cell type and severity were classified as non-CLL in the main analysis and were removed from this class in the sensitivity analysis. Because over half of all chronic leukemias of known cell type are chronic myeloid leukemia, the two chronic leukemias of unknown cell type were considered non-CLL in the main analysis (and were removed from this class in the sensitivity analysis). Thus, the main analysis contained 43 CLL cases and 206 non-CLL cases. As a result, two new variables were created to indicate case status for the matched case-control set. These identification of case type variables are casetypm and casetyps
III. Exposures.
A) Primary exposures of interest. The exposures of primary interest in the study included external exposures to occupational ionizing radiation (Daniels et al 2005) including work-related medical x-ray examinations (Anderson et al. 2006). At the time of study initiation, it was believed that bone marrow doses from plutonium exposures would be found to be trivial. However, for a subset of study subjects, Pu was found to comprise a substantial component of total bone marrow dose (Daniels et al. 2006). Tritium doses have been assessed for each worker, and were highest, on average, at the Savannah River Site. The exposure variables of primary interest in the study included the following: extall02, phtall02, neuw1002, puw20l02, extw0002, extw0205, extw0510, extw1020, extwge20, pumoncat, phtall00, phtall05, phtall07, phtall10, neuw0102, newuw0502, neuw2002, neuw100, neuw1005, neuw1007, neuw1010, puw01l02, puw05l02, puw10l02, puw20l00, puw20l05, puw20l07, puw20l10. In these variables, "dose" means red bone marrow (RBM) dose unless otherwise specified as HP(10) dose. wR=Kocher indicates that the weights were used as recommended in Kocher et al. 2005. Health Phys. 89:3-32.
B) Confounding variables and effect modifiers. Age was controlled in design phase by using age as the time scale in the incidence density sampling program. Controls were required to have been under observation at the age the case died (i.e., must have been hired younger than the case's age-at-death, and must have survived longer than the case). Cases were permitted to serve as controls for other cases (provided they met the above criteria), and resampling of cases was permitted. As a result of this selection, twelve controls were selected twice. However, cases were not permitted to serve as their own control. Age is controlled in the analysis phase in two ways: exposure of each control is truncated at the case's age of death (minus any lag), and the logistic regression analysis was conditional on the case-control set (i.e., the age stratum).
C) The following are variable definitions for factors to be analyzed as potential confounders or effect modifiers: sex, ethrace, smoking1, smkimses, smkmedrc, facility, bcohortt, hireyrt. The following potential confounders and/or effect modifiers were evaluated:
1. Time since exposure: To evaluate differences in the efficacy of radiation in producing leukemia risk over time, dose was divided into exposure windows based on the period before the cutoff date. External dose occurring 0-2 years before the cutoff date, and as 2-5 years, 5-10 years, 10-20 years, >20 years before the cutoff date were created. These variables are: extw0002, extw0205, extw0510, extw1020 and extwge20.
2. Sex, Race/ethnicity: Whites and males are expected to have higher leukemia rates compared to women and those of other races. These variables are: sex, ethrace. Because few persons of Hispanic ethnicity or of race other than white, Hispanic persons and non-whites were combined into a single category.
3. Smoking status as of the cutoff date: Smoking status estimates were derived for each of the study subjects. Although it would have been optimal to estimate smoking intensity or duration, this was not possible given the nature of the data available in the medical records, and only summary estimates of smoking status were obtained through the cutoff date. See variables: smoking1, skimses, and smkmedrc.
4. Plutonium monitoring status: Several studies have observed that plutonium-monitored workers may have been healthier on average than non-monitored workers. Therefore, differences in external radiation risk estimates were evaluated for workers ever and never monitored for plutonium, using the variable pumoncat.
IV. Factors or exposures to be evaluated as potential confounders
Several factors were evaluated additionally as potential confounders of the association between radiation and leukemia. These are benzene and carbon tetrachloride. Benzene and carbon tetrachloride variables for analysis in pnsmfla1LCCS are: bnz3l2, ccl3l2 and bncc3l2.
Number of data files: 1
Cohort Size: 1284
Exposure Type: External exposures to occupational ionizing radiation, including work-related medical x-ray, plutonium and tritium. As well as benzene and carbon tetrachloride.
Exposure Agent: Chemical agents and ionizing radiation
Method: ORNL: Between 1 May 1944 and September 1953, this site used a two-element film dosemeter as the primary beta-gamma dosemeter. A four-element dosemeter was introduced in late September 1953. In 1959, the pene
filter was modified to provide for accident level neutron exposures. During this time, the film packet was changed to the DuPont Type 554, then replaced by Eastman Kodak Type 2 between 1968 and 1969. It was used until replaced b
1975. SRS: The personnel radiation monitoring program began in November 1951 using the ORNL two-element film dosemeter until SRS introduced a dosemeter of its own design by November 1959. This design remained essentially uncha
became the principal means of beta-gamma measurement on 1 April 1970. Hanford: Between October 1944 and March 1957, Hanford used a two-element film dosemeter. A multi-element film badge replaced the two-element badge in Apri
in service until replaced by TLDs in January 1972. PNS: Film badge monitoring began at PNS on 1 July 1950 with Kodak Eastman. In 1957, this site began using a film badge incorporating the DuPont SX-233 film packet. This dos
remained essentially unchanged until July 1969 when the Kodak Type 3 Film Radiac Pack replaced it. TLD became the standard for personnel gamma exposure monitoring on 1 October 1974. LANL: Film badges were first distributed to
1944. By late 1947, the configuration changed and it remained in service until 1949, when a new badge housing was developed consisting of a brass clip which provided for an open window area of the film surface for beta radiatio
Between April and August 1950, a new two-filter dosemeter using lead and brass that was designed to provide for energy discrimination was issued. This badge was replaced in April 1951 by a similar dosemeter using two 0.5 mm filt
brass and the other made of cadmium. Appearing in October 1962, the Cycolac film badge was first used by approximately 100 persons who had histories of appreciable or extraordinary radiation exposure. It was replaced by TLDs in
Note: This information has been taken from the article "Bias and Uncertainty of Penetrating Photon Dose Measured by Film Dosemeters in an Epidemiological Study of US Nuclear Workers" by R. D. Daniels and M. K. Schubauer-Berigan.
citation 3689 above.
Sites: Hanford, Los Alamos National Laboratory (including Zia), Oak Ridge National Laboratory, and Savannah River Site (SRS), as well as the Portsmouth Naval Shipyard (PNS) in Kittery, Maine.
Contact Information:
Schubauer-Berigan, Mary
Industrywide Studies Branch. Division of Surveillance, Hazard Evaluations, and Field Studies National Institute for Occupational Safety and Health
Ph.D., Research Epidemiologist
NIOSH, MS R-15, 4676 Columbia Parkway
Cincinnati, OH 45226
(513) 841-4251
Fax: (513) 841-4470
Email: zcg3@cdc.gov
Notes: This is a pre-copy-editing, author-produced PDF of an article accepted for pub lication in the journal Radiation Protection Dosimetry following peer review. The definitive publisher-authenticated version [Radiation Dosimetry (2006), Vol. 118, No. 1, pp. 43-55] is available online at: http://rpd.oxfordjournals.org/