Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.
7 Cohort Descriptions T his report builds on the findings reported in Gulf War and Health, Volume 1: Depleted Uranium, Pyridostigmine Bromide, Sarin, Vaccines (IOM, 2000), hereafter referred to as Volume 1. The present chapter describes the published scientific literature on potential health effects of uranium in humans. Three major categories were constructed to organize the relevant studies: those of workers occupationally exposed to uranium in uranium-processing plants, those focusing on depleted-uranium exposure of deployed populations (some of which were exposed to depleted uranium through friendly-fire incidents), and those assessing exposure to uranium from environmental sources, including drinking water. The studies on the occupationally exposed workers generally had better study design and methods, especially for assessing exposure to uranium and disease outcomes. Studies of deployed populations, in contrast, had limited or no exposure data other than data on deployment itself and on the possibility of exposure to depleted uranium, but the committee chose to include these studies because of their relevance to the Gulf War and Operation Iraqi Freedom veterans. The chapter also includes studies that assessed health outcomes in people who lived near uranium-processing facilities or had high concentrations of uranium in their drinking water; these residents may have exposure conditions similar with veterans who received level III exposures (see Chapter 5). The chapter first presents an overview of the cohort studies of process- ing workers examined in Volume 1 and summaries of derivative studies pub- lished after that report. A summary of new cohorts introduced into the literature since 2000 follows. The chapter then describes studies that assessed mortality 117
118 updated literature review of depleted uranium p Â atterns and health outcomes in deployed service personnel. It ends with stud- ies of environmental exposures. In each instance, the study populations and methodsâincluding study design, measures of exposure, and assessment of outcomesâused by the investigators are described. Tables that summarize the studies are included at the end of the chapter. The traditional 5% level of statistical significance is used in reporting findings. Results that did not reach the 5% level of statistical significance are described as nonsignificant. Uranium-Processing Cohorts Studies of workers in the uranium-processing and uranium-machining indus- try are essential for understanding the long-term health effects of uranium expo- sure. Cohort studies assessing mortality patterns in processing workers have been conducted for some time. The studies of interest include uranium millers and other processors working in plants that process and refine uranium ore into met- als for commercial and nuclear use. During refinement and enrichment, workers are exposed to a number of hazardous substances, including chemical toxicants and potential carcinogens. Processing workers are exposed primarily to uranium oxides and derivative uranium compounds produced during the refinement pro- cess and other substances that contribute to adverse health outcomes. The studies described present a picture of diverse work histories and varied levels of exposure to enriched uranium, soluble and insoluble uranium compounds, other radioactive elements (such as radium and thorium), and other potentially hazardous industrial chemicals (such as sulfuric acid and fluorocarbons). In occupational settings, exposure is often prolonged, occurring over a period of several months to years in contrast with the shorter periods of expo- sure experienced by Gulf War veterans in friendly-fire incidents. Exposure also was greatest in the early years of the procurement and processing initiative in the United States, when safety measures were not as stringent. In occupational studies, exposure is generally assessed through work histories using cumulative measurement of exposure. Inhalation of dust that contains uranium compounds was the primary route of entry of uranium in processing plants, a route analo- gous to that of many Gulf War veterans exposed to depleted uranium during friendly-fire incidents. This section first details the cohorts reviewed in Volume 1, including updates on the cohorts published after the release of the report in 2000. That informa- tion is summarized in Table 7-1. In general, cohorts that did not have updates since that report are not included here, but they are included if there are data on health outcomes in them that were not considered in Volume 1. The section then describes new processing cohorts, including studies of uranium processors in the United Kingdom.
COHORT DESCRIPTIONS 119 Colorado Plateau Uranium-Mill Workers In assessing long-term health effects of uranium in the processing and machining industry, exposure to uranium and thorium-230 in mill workers is of particular interest (Waxweiler et al., 1983). Before World War II, uranium min- ing in the Colorado Plateau states was on a relatively small scale; efforts were directed primarily to recovery of vanadium contained in the ore. The establish- ment of a domestic uranium-procurement program sparked growth and expan- sion of uranium mining and milling in the United States after the war (Pinkerton et al., 2004). In that effort, uranium mills carried out extraction and purification of uranium ore for commercial use. The enrichment process exposed workers to a number of substances, including dust that contained vanadium, thorium, silica, and radium radionuclides in addition to uranium (IOM, 2000). Concerns about health risks associated with uranium milling were raised as early as 1949, when Colorado health officials submitted a formal request to the US Surgeon General to examine the health of uranium workers (Wagoner et al., 1964). That request, combined with reports from central Europe that documented an increased inci- dence of pulmonary malignancies in miners and millers, prompted the US Public Health Service to initiate a program to monitor health hazards in the uranium- mining and -milling industries (Wagoner et al., 1964; Archer et al., 1973). As radon exposure in the mines emerged as the primary health issue, the mortality experience of mill workers received little attention. As a result, there was little information on hazards in the uranium-milling industry. The studies described below sought to investigate the potential health effects associated with uranium milling in workers in the Colorado Plateau region. Wagoner et al., 1964 In the earliest study of the Colorado Plateau mill workers, Wagoner and col- leagues assessed mortality in 5,370 white male uranium miners and mill workers. The study population included three subcohorts, one consisting of 611 millers with no reported mining experience. The workers were prospectively identified and had volunteered for at least one physical examination. Followup of the cohort included triennial physical examinations, an annual uranium-mining industry census, and collection of personal data through mail and other methods. Census takers conducted annual interviews with mine and mill workers, and correspondence was sent to those who could be located. Vital status of 95% of the study group through December 31, 1962, was determined, and mortality was compared with that in the male population of the Colorado Plateau states. Death certificates for 317 workers known to be deceased were obtained â The Colorado Plateau region includes Arizona, Colorado, New Mexico, and Utah.
120 updated literature review of depleted uranium and classified according to the sixth edition of the International Classification of Diseases (ICD-6). The authors examined a number of health outcomes, includ- ing cancers of the digestive, respiratory, and lymphatic systems; nonmalignant respiratory diseases; and cardiovascular-renal disease. Workers were categorized by duration and type of employment (milling or mining) through July 1960 to ascertain exposure. They were initially assigned to three cohorts on the basis of type of work experience. Radiation exposure of a small group of miners was computed on the basis of months of underground experience, estimated dose rate, and cumulative dose. Milling experience of workers with mixed industry experience was discounted, and person-years starting with the date that mining experience began were added. The modified life-table technique was used in the analysis of mortality, and age-, race-, and cause-specific mortality was compared with that in the male population of the Colorado Plateau area by using standard- ized mortality ratios (SMRs). The authors found no significant differences in deaths between the mill subcohort and the general population (56 observed vs 55.8 expected; SMR, 100). Deaths from all forms of cancer were fewer than expected (6 observed vs 8 expected; SMR, 75). However, there was a slight excess of deaths from c Â ardiovascular-renal disease (28 observed vs 25.3 expected). The strengths of this study include a well-defined cohort, inclusion of smok- ing history, and a sound method of followup through annual interviews. In addi- tion, the mortality analysis was conducted by using a local reference population rather than the US population. Despite efforts to measure exposure on an indi- vidual level, biologic monitoring was not carried out; exposure was represented only by work classification. Furthermore, exposure assessment was limited by the lack of specificity of exposure because of the possibility of concomitant exposure, and the cohort was relatively small. Archer et al., 1973 In a second prospective cohort study of Colorado Plateau mill workers, Archer and colleagues extended the followup of mortality in mill workers with a group of 662 white male millers. The men had worked in one of six mills during 1950-1953 and were available for medical examinations (including blood and urine tests, a physical examination, and chest films) in 1950, 1951, and 1953. Occupational and social histories were also documented by the authors. Vital status through December 31, 1967, was ascertained through Social Security Administration (SSA) records and several other sources. Death certificates were collected, and underlying causes of death were coded according to ICD-6. Only 1% of the study population was lost to followup. Person-years by 5-year age group and calendar year were calculated with the modified life-table technique, and cause-specific mortality was compared with that in the white male population in the Colorado Plateau region (Colorado, Utah, New Mexico, and Arizona) by using SMRs.
COHORT DESCRIPTIONS 121 Overall mortality and mortality from major cardiovascular diseases were consistent with the numbers of deaths expected (104 observed vs 105.11 expected and 52 observed vs 47.72 expected, respectively). Mortality from all other causes was significantly less than expected (12 observed vs 22.42 expected; SMR, 54; pÂ < 0.05). However, there was a nonsignificant excess in the number of deaths due to all cancers (20 observed vs 18.11 expected; SMR, 110). A significant excess in deaths from lymphatic and hematopoietic malignancies other than leukemia was also noted (4 observed vs 1.02 expected; SMR, 392; p < 0.05). None of the four people who died had worked in or around the furnace area where exposure to uranium and vanadium was greatest. The study exhibited many of the same strengths as the previous study by Wagoner and colleagues (1964), including the use of a regional comparison group in the mortality analysis and a small sample (662). In contrast with Wagoner et al., Archer and colleagues did not conduct annual followup of the study group. Waxweiler et al., 1983 In this retrospective cohort study, the authors continued through 1977 fol- lowup of the Colorado Plateau mill-worker cohort first studied by Wagoner and colleagues (1964). Microfilmed personnel records of workers in seven uranium mills were used in the selection of 2,002 men who were employed at least 1 day after January 1, 1940; had worked for at least 1 year in uranium mills; and had no work experience in an underground uranium mine. Demographic data and work histories through 1971 were collected and coded for study use. Researchers found that about half the cohort was employed before 1950 and that only a small number had worked longer than 5 years. Vital status was determined through SSA records and other sources for all but 2% of the cohort. Death certificates were obtained for 515 (97%) of the 533 deceased workers. A total of 43,252 person-years were observed for the study group. Mortality was analyzed with the National Institute for Occupational Safety and Health (NIOSH) modified Life Table Analysis System (LTAS), and cause-, age-, race-, sex-, and calendar-period-specific SMRs were calculated. Mortality was compared with national rates. The authors sought to assess the association between mill employ- ment and lymphatic and pulmonary malignancies, toxic effects in the kidneys due to uranium exposure, and all-cause mortality. There were statistically significant deficits in total mortality (533 observed vs 605.2 expected; SMR, 88; 95% CI, 81-96) and mortality from all malignant neoplasms (82 observed vs 109.8 expected; SMR, 75; 95%, CI 59-93). There were also fewer deaths due to lung cancer than expected (26 observed vs 31.4 expected; SMR, 83; 95% CI, 54-121), but this difference did not reach the 5% level of statistical significance. The authors reported a statistically nonsignificant excess in mortality from Hodgkin lymphoma (SMR, 231; 95% CI, 48-675). Of the nonmalignant outcomes, there was a significant excess in deaths due to
122 updated literature review of depleted uranium nonmalignant respiratory disease (SMR, 163; 95% CI, 123-212) as a result of increases in the âother nonmalignant respiratory diseasesâ category. The authors also observed a nonsignificant increase in deaths due to chronic renal disease (SMR, 167; 95% CI, 60-353), but in all cases employment in the mill was rather brief and induction time ranged from 4 to 19 years. Moreover, death certificates in at least half of the six cases also indicated either prostatic obstruction or pros- tatic cancer. Pinkerton et al., 2004 Pinkerton and colleagues examined mortality in 1,484 men who worked in one of the seven uranium mills in the Colorado Plateau area. The cohort was drawn from personnel records of uranium-mill workers previously described by Waxweiler and colleagues (1983) that included 2,002 men employed at least 1 day after January 1, 1940, worked for at least 1 year, and had never worked in an underground uranium mine. The authors reviewed records from the Waxweiler et al. study to ensure the inclusion of workers who met the original criteria but were omitted from the study, and they recoded all work histories to remedy any inaccuracies. The resulting subcohort of 1,485 included men who satisfied the original cohort criteria, had never worked in an aboveground or underground mine, and had worked for at least 1 year in the seven uranium mills at the time of uranium or vanadium concentrate recovery. One person was excluded from analysis because his work history was incomplete. Of the 1,485 men, 97% (1,438) were members of the original Waxweiler et al. cohort. All workers included in the study were followed through the end of 1998. Vital status was determined on the basis of records of SSA, the Inter- nal Revenue Service (IRS), the US Postal Service, and state bureaus of motor vehicles. Death certificates and data from the National Death Index (NDI) were collected to ascertain cause of death of 794 (98%) of the workers known to be deceased and recoded in accord with ICD-9. Less than 2% of the study population had been lost to followup. The End Stage Renal Disease Program Management and Medical Information System (ESRD) was used to assess the risk of death from ESRD and renal disease in the cohort. The NIOSH modified LTAS was used in the analysis of mortality. The num- ber of person-years at risk was calculated and stratified into 5-year intervals by age and calendar time. SMRs were calculated by using US adjusted mortality and observed deaths. In addition, mortality was stratified by duration of employment, time since first employment (latency), and year of first employment. Mortality was also compared with that in the Colorado Plateau states. In general, mortality from all causes and mortality from all malignant neo- plasms were less than expected in comparison with the US population. Cancer mortality was consistent with findings reported in previous studies of this cohort
COHORT DESCRIPTIONS 123 (Archer et al., 1973; Waxweiler et al., 1983). Mortality from tracheal, bronchial, and lung cancer for the first time exhibited a statistically nonsignificant increase (78 observed vs 68.93 expected; SMR, 113; 95% CI, 89-141) that was not found in earlier studies of this cohort. Mortality from tracheal, bronchial, and lung can- cer was higher in those employed before 1955 (SMR, 134; 95% CI, 102-174) than in those hired in 1955 or later (SMR, 79; 95% CI, 49-121), but a reverse associa- tion was observed between tracheal-, bronchial-, and lung-cancer mortality and duration of employment (that is, longer employment was associated with lower mortality). The excess based on regional rates (75 observed vs 49.73 expected; SMR, 151; 95% CI, 119-189) was statistically significant and greater than the excess based on US rates since 1960. Pinkerton and colleagues also reported nonsignificant increases in deaths from lymphatic and hematopoietic cancers other than leukemia (16 observed vs 11.08 expected; SMR, 144; 95% CI, 83-235) and from chronic renal disease (8 observed vs 5.91 expected; SMR, 135; 95% CI, 58-267). The increase in lym- phatic cancers was less than the excess observed by Archer et al. (1973) (SMR, 392; 95% CI, 194-590), which reflected an excess of deaths from Hodgkin lym- phoma and lymphosarcoma and reticulosarcoma, but greater than that found by Waxweiler et al. (1983) (SMR, 119; 95% CI, 21-217). In addition, significantly fewer deaths from all digestive cancers than expected were observed (SMR, 62; 95% CI, 43-87). The authors also observed a significant increase in mortality from non- malignant respiratory disease (SMR, 143; 95% CI, 116-173) that was due to a significant excess in mortality from emphysema (SMR, 196; 95% CI, 121-299) and pneumoconiosis and other respiratory disease (SMR, 168; 95% CI, 126-221). Mortality from emphysema was higher in workers employed before 1955, when exposure to uranium, silica, and vanadium was thought to be highest (17 observed; SMR, 222; 95% CI, 129-356), than in those employed in 1955 or later (4 observed; SMR, 130; 95% CI, 36-333). However, there were no cor- responding differences in mortality from pneumoconiosis and other respiratory diseases. Fewer deaths from nonmalignant digestive disease than expected were observed (SMR, 62; 95% CI, 39-94). The strengths of this study include a long followup period (vital status through 1998 was determined) and the use of ESRD data that provide useful details on mortality from chronic renal disease associated with uranium milling. Like previous studies of the cohort, this one lacked assessment of individual exposure to uranium and other substances in the milling environment. Fernald Feed Materials Production Center Workers From 1951 to 1989, the Fernald Feed Materials Production Center (FFMPC) in Ohio chemically processed uranium-ore concentrate and uranium of low
124 updated literature review of depleted uranium enrichment grade into uranium-metal products. The process involved the use of hydrofluoric acid, ammonia, nitric acid, sulfuric acid, tributyl phosphate, trichloroÂ ethylene (TCE), and cutting fluids through dissolution, evaporation, and denitra- tion to produce pure uranium metal. During operation, the facility monitored internal and external radiation exposure. The Comprehensive Epidemiology Data Resource (CEDR) created by the Department of Energy (DOE) warehoused information regarding workers at multiple nuclear processing facilities, including FFMPC, for 30 years. Boiano et al., 1989 In 1985, NIOSH investigators conducted a cross-sectional medical assess- ment of workers at FFMPC. The study followed requests for investigations due to concerns about potential exposure to reported releases of uranium oxides from dust collectors in November and December 1984. About 850 hourly workers were employed at FFMPC at the time of the evaluation. The study focused on evaluating hazards related to lung and renal toxicity after consultation with plant management and workers. Of the 208 eligible long-term employees (147 hourly and 61 salaried), 146 (70%), identified through employee rosters, participated in the study. The study population consisted of 142 men and four women with mean age and median age of 56 and 58 years, respectively. The employees had worked at the facility for 10-34 years (median, 32 years). Study participants included hourly and salaried employ- ees who had worked at the facility continuously for 10 years, salaried employees who had previously been hourly and who had worked there for at least 10 years, and former hourly employees who had retired within the preceding 2 years after working there continuously for at least 10 years. A medical and occupational questionnaire was carried out with tests that included blood and urine analysis, chest radiography, and pulmonary-function tests (standard screening spirometry). The questionnaire was self-administered and collected details on an employeeâs medical history with emphasis on respi- ratory and renal conditions, occupational and job exposure history, cigarette and alcohol use, and basic demographic information. Questions on respiratory conditions were extracted from the American Thoracic Society questionnaire. On the basis of responses to relevant questionnaire items, the investigators cat- egorized breathlessness in five grade levels; 1-second forced expiratory volume (FEV1) and forced vital capacity (FVC) served as indicators of pulmonary func- tion. Blood and urine samples were collected for a number of glomerular and tubular biomarkers. Blood and urine tests were used as dichotomous variables in evaluating renal effects. Urinary uranium concentration was also measured. â The markers included beta-2-microglobulin, retinol-binding protein, albumin, total protein, creatinÂ ine, N-acetyl glucosaminidase, gamma glutamyl transpeptidase, and alanine aminopeptidase.
COHORT DESCRIPTIONS 125 Personnel records were used to establish work histories, and high-, medium-, low-, and no-exposure groups were created on the basis of job titles. Urinary- uranium data were used to construct exposure histories; the data were used to verify exposure groups and to weight the exposure categories. A cumulative uranium-exposure index was created for each participant by multiplying dura- tion of employment and the potential for uranium exposure in the job held. Radiation measurements (whole-body radiation counts) were used to determine uranium lung burden. Urinary uranium concentrations varied up to 13 Âµg/L, and 109 (92%) of the participants had concentrations under the detection limit of 5 Âµg/L. No associations between glomerular and tubular markers and measures of uranium exposure were observed. The ratio of FEV1 to FVC was associated with the job- historyâderived uranium-exposure index after adjustment for smoking. Short- ness of breath was significantly associated with self-reported uranium-exposure incidents. Ritz, 1999 A cohort of workers employed at FFMPC during its period of operation (1951-1989) was assembled through CEDR (Ritz, 1999). Some 4,014 white male workers were identified; most of them were employed before 1960. The cohort was followed through 1979 with SSA records and from 1979 to 1989 with the NDI to determine vital status; 1,064 had died, and death certificates were obtained. Internal exposure to uranium was measured through urinalysis and extrapola- tion from environmental sampling, and external exposure was measured with film badges. Internal exposure, measured as annual lung dose, was mostly exposure to various insoluble forms of uranium, from depleted through enriched in 235U. Because urinalysis reflects inhalation and internal transport of soluble uranium, air sampling was incorporated to provide a rough measure of the risk of inhala- tion of insoluble uranium as well. Internal exposure from radionuclides was responsible for the bulk of the radiation doses recorded; most monitored workers (68.9%) received cumulative external radiation doses of less than 10 mSv, only 2.6% had doses of over 100 mSv, and none exceeded 300 mSv. Because CEDR maintained records of exposure to TCE and cutting fluids, these were controlled, as was socioeconomic status (SES). SMRs were calculated by using Monsonâs life-table analysis. The numbers were too small to conduct separate dose-response analyses for specific cancers, so only the organ systems with the highest likelihood of exposureârespiratory, transport (blood and lymph), excretory, and upper gastrointestinalâwere examined individually. Cumulative dose was lagged by 0, 10, and 15 years to allow for cancer latency. The risk-set approach (similar to the nested case-control approach) of Breslow and Day was used for dose-response comparisons; each cancer death was matched to all cohort
126 updated literature review of depleted uranium members who were still alive at the calendar time of the index subjectâs death (on average 3,300 survivors per death). All-cause mortality was lower in the workers than in the US white male pop- ulation (SMR, 84; 95% CI, 79-89). Risk of death from all malignant neoplasms was nonsignificantly higher than in the general population (SMR, 109; 95% CI, 98-122) and lung-cancer mortality was similar (SMR, 101; 95% CI, 83-121). External radiation doses in excess of 100 mSv (which occurred in only 2.6% of the cohort) increased mortality from all cancers, all radiosensitive cancers, and lung cancer, but the numbers were too small for precise estimates (all were non- significant). Cumulative external radiation showed a dose-response relationship when lagged by 10 or 15 years and adjusted for chemical exposure and internal dose for all cancers (10-year lag: rate ratio [RR], 1.79; 95% CI, 1.12-2.86; 15- year lag: RR, 1.92; 95% CI, 1.11-3.32), all radiosensitive cancers (10-year lag: RR, 1.88; 95% CI, 1.06-3.32; 15-year lag: RR, 2.0; 95% CI, 1.02-3.94), and lung cancer (10-year lag: RR, 2.13; 95% CI, 1.08-4.18; 15-year lag: RR, 2.77; 95% CI, 1.29-5.95), but not for hematopoietic and lymphopoietic cancers. Cardiovascular mortality was lower in the workers than in the US white male population (SMR, 78; 95% CI, 71-86), as was emphysema mortality (SMR, 21; 95% CI, 4-60). Differences in mortality from all other causes were nonsignificant. The healthy-worker effect probably accounts for those lower SMRs. The SMR for all-cause mortality remained lower when Fernald workers were compared with NIOSH-Computerized Occupational Referent Population System (CORPS) workers (SMR, 81; 95% CI, 76-86); the comparison should have reduced bias caused by the healthy-worker effect. The strengths of this study are that it used one of the largest cohorts with monitored external and internal exposures at the individual level, had a long fol- lowup period, allowed for a lag period for development of radiation-related solid tumors, and adjusted for other exposures, such as to TCE and cutting fluid, and for socioeconomic status. Oak Ridge Nuclear Facilities Workers Oak Ridge, Tennessee, was the home of several nuclear facilities involved in nuclear-weapons production during World War II. Two of the facilities (Y-12 and K-25) were dedicated to uranium enrichment for use in atomic weapons, and a third (X-10, also called Clinton Laboratories) was an experimental laboratory designed to produce plutonium for further research. The Y-12 uranium-processing plant was run by the Tennessee Eastman Corporation TEC) in 1943-1947. With an electromagnetic separation process, uranium was enriched in 235U for use in atomic weapons. In 1947, ownership passed to Union Carbide Corporation, and the plant shifted toward weapons fabrication and research and development for the separation of isotopes. K-25 was a gaseous-diffusion plant for producing enriched uranium.
COHORT DESCRIPTIONS 127 Polednak and Frome, 1981 In the 4 years that TEC operated the Y-12 plant, about 45,000 people worked there; 38,000 left the plant when Union Carbide assumed management. Demographic and payroll data were submitted to SSA in 1974 for validation of decedents. In 1981, a subset of the 38,000 was examined, and mortality assessed (Polednak and Frome, 1981). The researchers excluded women (47%) from the sample because of incomplete recordkeeping at SSA. An additional 307 tempo- rary workers (those who worked less than 2 days) were excluded, as were 543 because of inaccurate or incomplete data at SSA or the plant. Minority-group members were also excluded (no reason given), leaving a total of 18,869 white men for analysis. Death certificates through 1977 were verified, and all underly- ing causes of death recorded on the certificates (according to ICD-8 classification) were recorded for analysis. Electromagnetic separation involved the use of mass spectrograph units in a two-step enrichment process. Both steps exposed workers to uranium dust but little external radiation; the exposure that posed a risk was inhalation of radioÂ active compounds. Air-sampling records obtained from TEC showed higher than normal readings at concentrations above acceptable limits. In 1945, the first step was eliminated, and this reduced levels of uranium dust, but the enrichment was higher. Company records indicate that particle size varied, some particles being smaller than 1 Î¼m. Film badges were rarely used, because exposure to external radiation was low owing to the nature of the operation; respirators were required but might not have been used with great frequency. Polednak and Frome examined whether mortality was associated with lon- ger periods of employment and exposure to higher levels of airborne uranium. Workers were categorized by stage of enrichment and other job classifications to determine an approximate level of exposure. SMRs were calculated by using both an internal comparison group (workers in buildings where uranium was not being processed) and external comparison group (the US white male population). SMRs were below 100 for all-cause mortality and all-cancer mortality; most of the results were nonsignificant. After correction for incomplete ascertainment of cause of death (access to death records was possible for only 95% of deaths), mor- tality from lung cancer had an SMR of 122 (95% CI, 11-136). Considering type and length of employment resulted in no discernable differences among strata. Checkoway et al., 1988 Researchers examined 6,781 white men who worked at the Y-12 plant for at least 30 days during the period May 31, 1947-December 31, 1974 (Checkoway etÂ al., 1988). They excluded those who had worked at the plant before May 4, 1947 (that is, before it was turned over to Union Carbide), those who had worked at other facilities, and those with unknown employment dates; they also excluded
128 updated literature review of depleted uranium nonwhite workers and women. The median age of cohort members at hire was 27.6 years, and the median duration of employment was 9.2 years. The cohort was retrospectively followed through 1979, allowing at least 5 years of followup (median followup time, 20.6 years). SSA and Tennessee Motor Vehicles Department records and state death indexes were used to assess vital status, and death certificates were coded according to ICD-8. Vital status of 96% of the cohort was determined, and person-years of missing cohort members were added up to the date of last contact. A total of 133,535 person-years were observed, and death certificates for 846 (98%) of the 862 deaths were obtained. Monitoring for external radiation began in 1948 with film badges and switched to thermoluminescent dosimeters in the late 1970s. Whole-body dose equivalents were determined with both methods and summed to provide cumula- tive exposure values. Researchers assumed that periods in which a worker was not monitored were associated with a low likelihood of exposure and assigned a value of zero to them. The mean cumulative external dose for monitored workers was 0.96 rem over a 31-year period. Internal radiation exposure was determined with urinalysis (which began in the early 1950s) and in vivo counting of depos- ited uranium (which began in 1961). Lung dose equivalents were determined with metabolic models and accounted for level of uranium enrichment. Internal- exposure monitoring data were available on 3,490 workers. The mean internal dose to monitored workers was 8.21 rem. All 6,781 cohort members were included in the overall analysis of mortality patterns. The dose-response analyses, however, were restricted to 5,278 workers on whom external-dose data were available and 3,490 (51.5%) workers on whom internal-dose data were available. Cause-specific mortality was compared to US and Tennessee mortality; SMRs were calculated with a modified life-table analysis with adjustment for age and calendar year in 5-year intervals. Poisson regression analysis was used to determine RRs with adjustment for dose category, age, calendar year, and duration of followup. An internal referent category was used that consisted of the lowest cumulative-dose category for both internal and external exposure. In addition, 10-year latency was assumed. No significantly increased SMRs were observed for deaths from any cause except lung cancer in comparison with the US population (89 observed; SMR, 136; 95% CI, 109-167). Dose-response trends were detected for lung-cancer mortality with respect to cumulative alpha and gamma radiation. Nonsignificant excesses were observed for cancers of the brain and central nervous system, kidney, and lymphatic system. Frome et al., 1990 Workers employed at the Oak Ridge facilities during World War II were selected for long-term health-status followup (Frome et al., 1990). The cohort
COHORT DESCRIPTIONS 129 was defined as white men who worked at the facilities for at least 30 days from the initial date of operation through December 31, 1947. Information was found on 28,008 of those who met the criteria. Monitoring for radiation exposure was infrequent during the period of employ- ment, so exposure was categorized on the basis of likelihood. Job codes and titles with a reasonable likelihood of radiation exposure were categorized as âYâ (yes), and those with no discernible reason for exposure as âNâ (no). Workers were also categorized according to the facility they worked at, SES, length of employment, period of followup, and age. Workers contributed person-years from 1950 to 1979 or until loss to followup. Analyses took two approaches: modified SMR (tradi- tional SMR and SMR-trend analyses over 30 years), which allowed the ratio of death rates in the index and reference groups to change over time; and multivariate analyses with Poisson regression modeling to account for the effects of multiple factors (birth year, duration of employment, SES, employment facility, period of followup, and radiation exposure level) on cancer mortality simultaneously. Unadjusted (crude) analyses showed a statistically significant excess in deaths from all causes (SMR, 111); no healthy-worker effect was observed in this cohort. The SMR for all malignant neoplasms was 105 (nonsignificant). Mortality from respiratory neoplasms was significantly increased (SMR, 125), as was mortality from lung cancer (SMR, 127 based on 850 observed deaths vs 667.99 expected). The increase in lung-cancer mortality was observed after 5 years of followup and increased at 1.44% per year. In the adjusted analysis, the strongest predictor of lung-cancer death was SES, and lung-cancer mortality was not associated with exposure to radiation. The trend statistics (the rate of change in SMR over a 30-year period) for all causes (0.74), all cancers (1.05) and respiratory cancers (1.36), tuberculosis (3.44), benign neoplasms (6.05), circulatory diseases (1.05), respiratory diseases (1.53), and suicide (2.45) are all significant in an upward direction. Loomis and Wolf, 1996 The cohort studied by Checkoway et al. (1988) was followed for an addi- tional 10 years until 1990 (Loomis and Wolf, 1996). The study population included workers employed for at least 30 days at Y-12 in the period January 1, 1947-December 31, 1974. Those who had worked at the plant before 1947 were also included in the cohort but analyzed separately. Of those who had worked at the plant in 1947 and later, 8,116 were eligible: 6,591 white men, 922 white women, 449 black men, 149 black women, and five men and women of other racial groups. Of those who had worked at the plant before 1947, 2,841 were eli- gible: 1,764 white men, 562 white women, 85 black men, 69 black women, and one man of another racial group. The total cohort consisted of 10,597 workers. The researchers noted that the cohort was not identical with that of Checkoway et al. (1988), because of corrections in records.
130 updated literature review of depleted uranium White men, nonwhite men, and women were analyzed separately, and those who worked at the plant before 1947 were also considered separately. SMRs were calculated and adjusted for age and calendar time. Employment was the only measure of exposure. The mortality experience of this cohort was compared with that of the US population. Lung-cancer mortality was significantly increased in all workers at the Y-12 plant (SMR, 117; 95% CI, 101-134) and in white men (SMR, 120; 95% CI, 104- 138). Lung-cancer SMRs were highest in workers hired before 1955 and those with 5-19 years of employment. No other significant excess or deficit of deaths from any other cancers was observed. Richardson and Wing, 2006 All Y-12 plant workers who were employed for at least 30 days during the period May 1947-December 1974 were assessed for lung-cancer mortality due to ionizing radiation (Richardson and Wing, 2006). For internal radiation exposure, annual lung-dose estimates were based on results of urinalysis through the 1950s and results of in vivo monitoring that was begun in 1961. The degree of enrichment was included in calculations that involved in vivo monitoring. For years in which monitoring was incomplete, dose was estimated on the basis of exposure potential associated with the department or job title or on the basis of comparison with a monitored equivalent period for the same person. In 1961, a plantwide policy for external monitoring with film badges was implemented; in the late 1970s, thermo- luminescent dosimeters began to be used. Exposure in years when there was no monitoring was estimated on the basis of a standard exposure potential associated with the department for the period before 1961 and dose estimates for the period after 1960. Estimates of lung dose from both internal and external radiation during the period 1947-1985 were calculated for each year of employment for each worker in the cohort, and vital status as of 1990 was obtained from SSA records and the NDI. The study focused on lung cancer as an underlying or contributory cause on the basis of ICDA-8 162 (code 162 of the International Classification of Diseases Adapted for Use in the United States). Because smoking history was not avail- able for analysis, two proxies were used: nonlung smoking-related cancers and noncancer smoking-related conditions. The analysis used a nested case-control method that constructed risk sets by matching noncases to cases on the basis of selection date (that is, a control is cho- sen at the point when he or she has reached the age of death for the matched case). To ensure comparability, controls were also matched to cases on year of birth, sex, race, SES, length of employment as of the selection date, and employment status on the selection date. Because exposure data were available only through 1985, the analysis assumes a 5-year lag. Conditional logistic regression was used to evaluate the association between lung-cancer mortality and radiation exposure.
COHORT DESCRIPTIONS 131 RRs for lung-cancer mortality were not significant for either internal or external dose. Joint internal-external exposure calculations showed an increase in mortality as dose increased, but they were not statistically significant. External radiation had a higher correlation with lung-cancer mortality than external and internal radiation combined or internal radiation alone. This study had several strengths, including exclusion of workers who had never been monitored for internal radiation exposure to reduce bias due to mis- classification of exposure and selection bias, direct measurement of both external and internal exposure at the individual level, and sound methodology and analytic approach. Limitations of the study include uncertainty in estimating workersâ âmiss- ingâ external and internal radiation doses and the lack of information on smoking (a strong risk factor for lung-cancer mortality). Inclusion of lung cancer as a contribu- tory cause of mortality created a potential for overestimating the association in that lung cancer is a common site of metastasis from other primary cancers. Frome et al., 1997 Researchers examined 106,020 workers at all four plants at the Oak Ridge site who were employed for at least 30 days in 1943-1985. Workers were cat- egorized by employment at one of the four facilities or were included in a fifth group if they worked at more than one plant. External radiation exposure was estimated for employees on the basis of limited monitoring; internal exposure was categorized in three groups: âeligible for monitoring and monitored,â âeligible for monitoring but not monitored,â and ânot eligible for monitoring.â For dose- response analysis, exposure was lagged. Workers were further stratified by length of employment, SES, birth year, and age. Vital status was determined from SSA files, and participants contributed person-years until death or loss to followup. SMRs were generated on the basis of mortality in the US general population. Groups were also compared for mortality for internal comparisons. Finally, dose- response analysis was included but only for white men. All-cause mortality and all-cancer mortality in white men were similar to those in the US population (SMR, 100 and 98, respectively). None of the cancers of interest resulted in an SMR that was significantly different from 100. Mortality from most nonmalignant diseases was lower than US population rates: diseases of the blood (SMR, 52), nervous system (SMR, 70), circulatory system (SMR, 95), digestive system (SMR, 80), and genitorurinary system (SMR, 83). The only nonmalignant condition with a reported excess of deaths in white men was respiratory disease (SMR, 112). Mallinckrodt Chemical Workers From 1943 to 1966, the Mallinckrodt Chemical Company in St. Louis, Mis- souri, processed large amounts of uranium ore for production of pure Âuranium
132 updated literature review of depleted uranium t Â etrafluoride and uranium metal. During that time, the plant also processed imported ore that contained 70 times the uranium content of North American ore. Processing workers in poorly ventilated areas were potentially exposed to internal and external radiation, in addition to a number of toxicants, including possible carcinogens, silica, and sulfuric acid. Using external radiation dose, Dupree-Ellis and colleagues (2000) sought to assess the link between occupational exposure and mortality patterns in Mallinckrodt processing workers. Lung-cancer deaths in this cohort were included in Dupree et al. (1995). The study population consisted of 2,514 white male processing-plant workers who were employed in Mallinckrodt during 1942-1966. Workers were selected on the basis of plant records, and vital status was ascertained through the SSA, Pension Benefit Information, and NDI databases. The cohort was retrospectively followed through 1993, with a mean followup time of 34.6 years (median, 36 years). Death certificates for 1,012 of the 1,013 who died during followup were obtained. The study excluded 745 workers whose prior exposure to external radiation was minimal (this group included 556 women [race not specified] and 43 nonwhite men). Person-years were calculated from 30 days after the date of first hire until the earliest of death, loss to followup, or the end of the study. A total of 87,757 person-years were observed. SMRs were calculated by using underlying cause of death, and the mortality experience of the cohort was compared with that of white men in the United States. Dose-response analyses were stratified on age and calendar period and included nonunderlying cancer causes. An internal com- parison group and time-dependent cumulative-dose groups (latency, 10 years for solid tumors and 2 years for leukemia) were used for this purpose. From the middle of 1945, workers were monitored for external radiation exposure with film badges. Individual annual doses were determined from deep dose-equivalent analysis of film badges, and an algorithm was assigned for years on which data were unavailable. The calculated mean cumulative dose was 47.8 mSv (median, 15.3 mSv). The total population dose was measured at 120,063 mSv. All-cause mortality was significantly lower than expected (SMR, 90; 95% CI, 85-96), probably because of the healthy-worker effect. There was a nonsig- nificant increase in all-cancer mortality (SMR, 105; 95% CI, 93-117). Several site-specific SMRs were increased, but none reached statistical significance. The authors reported a nonsignificant increase in excess relative risk of renal-cancer death of 10.5 per sievert (90% CI, 0.6-57.4), and observed:expected ratios of cases of renal cancer by dose were as follows: less than 5 mSv, 3:2.4; 5-9 mSv, 0:0.9; 10-19 mSv, 0:1.3; 20-39 mSv, 2:1.4; 40-79 mSv, 1:1.5; 80-159 mSv, 0:1.3; and 160 mSv and higher, 4:1.2. However, a possible excess risk associated with internal radiation, chemical exposure, or chance could not be ruled out. Chronic nephritis was the only nonmalignant outcome associated with an observed excess in mortality (SMR, 188; 95% CI, 75-381), but this effect was nonsignificant.
COHORT DESCRIPTIONS 133 Workers at Four Uranium-Processing Operations In this retrospective case-control study, authors combined data from stud- ies (Polednak and Frome, 1981; Cookfair et al., 1983; Dupree et al., 1987; Checkoway et al., 1988) of workers at four DOE uranium-processing or fabrica- tion operations to explore the relationship between uranium-dust exposure and lung-cancer mortality (Dupree et al., 1995). Study subjects included eligible workers at uranium-processing operations that were managed by TEC from 1943 to the middle of 1947 (TEC operations) and from 1947 onwards (Y-12 opera- tions), workers at the Mallinckrodt Chemical Workers Uranium Division (MCW) at two sites in Missouri from 1942 until operations ceased in 1966, and workers at FFMPC, where production activities ran from 1951 to 1989. The MCW and FFMPC sites also processed uranium-ore concentrate into metal. Study authors identified 787 lung-cancer cases by mortality followup of the employee cohorts through the end of 1982, which allowed followup of at least 30 years for each cohort. Cases included workers who were employed at any of the facilities for at least 183 days and who died before January 1983 and had lung cancer listed on the death certificate. One control was selected for each case and matched on race, sex, date of birth, and hire date within 3 years. Of the 787 workers, 567 were employed at TEC, 142 at Y-12, 27 at MCW, and 51 at FFMPC. Most of the employees were white men (92%); there were 44 white women, 13 black men, and four black women. Data on complete work history, smoking history (never or ever smoked), and SES (first pay code) were collected from employment and occupational radiation-monitoring records. Smoking data were collected on 48% of the cases and 39% of the controls, with 91% of the cases and 75% of the controls identified as smokers. Health physicists used uranium air-monitoring data and other environmental data to estimate individual annual radiation lung doses from deposited uranium. Cumulative internal and external doses were lagged for 10 years, and smoking status and pay code (monthly or nonmonthly) were accounted for. Cumulative lung doses ranged from 0 to 137 cGy in cases and from 0 to 80 cGy in controls. In general, there was little evidence of a relationship between internal radiation dose and lung-cancer mortality. The authors reported increased risk in those exposed to 25 cGy or more, with an odds ratio (OR) of 2.0; however, the CI was wide (95% CI, 0.20-20.70) and the result was not statistically signifi- cant. Dose-response analyses limited to cases hired at the age of 45 years or more showed higher ORs for exposed workers, but no trend was evident. Portsmouth Uranium Enrichment Facility Workers Brown and Bloom (1987) conducted a retrospective cohort study to exam- ine causes of death in 5,773 employees of the Portsmouth Uranium Enrichment facility. The subjects were primarily white men who were employed for at least
134 updated literature review of depleted uranium 1 week during the period September 1954-February 1982. The facility, in Pike County, Ohio, used gaseous diffusion processes to enrich uranium up to 98% 235U. The primary chemical of concern at the plant was uranyl fluoride, which is highly soluble and is a known renal toxicant. The plant had conducted routine (monthly) urine bioassays since 1954 of employees who had the potential to be exposed to toxicants. However, uranium is excreted quickly, and results of the bioassays were not useful in identifying cumulative exposure to the chemical. Instead, the results were used to classify jobs and rank departments on the basis of relative potential for exposure to uranium. The authors identified two cohorts: one with the greatest potential uranium exposure and the other with any potential exposure. Person-years at risk (PYARs) based on each employeeâs time at the plant until death were stratified by 5-year calendar periods, age groups, length of employment, and time since first employment. PYARs were multiplied by US white male cause-specific mortality to determine expected numbers of deaths. It was also calculated on the basis of Ohio mortality. Statistically significant (p < 0.05) deficits in mortality from all causes (SMR, 68; 95% CI, 62-75) and from diseases of the respiratory system (SMR, 42; 95% CI, 23-70), the nervous system (SMR, 40; 95% CI, 21-68), the circu- latory system (SMR, 72; 95% CI, 62-82), and the digestive system (SMR, 54; 95% CI, 32-86) were identified. The SMR for all malignant neoplasms was 85 (95% CI, 71-102). There were nonsignificant increases in mortality from stom- ach cancer (SMR, 169; 95% CI, 81-310) and lymphatic and hematopoietic can- cers (SMR, 146; 95% CI, 92-218). The authors found a nonsignificant increase in stomach-cancer mortality in those with more than 15 years of employment and 15 years of latency. The study is limited by its relatively short observation period and its lack of exposure information. Phosphate-Fertilizer Production Workers Stayner and colleagues (1985) conducted a retrospective cohort mortal- ity study of 3,199 phosphate-fertilizer production employees in Polk County, Florida. The plant, which produced primarily diammonium and dicalcium phos- phates, began operation in 1953. From 1953 to 1958, it also recovered uranium from phosphate ore. NIOSH received reports of a number of lung cancers in nonsmoking workers in 1976 and conducted a survey of the facility to determine exposure to uranium. At that time, all samples from the analysis were below occu- pational standards. NIOSH also began a study to evaluate mortality in workers at the site. Personnel records were obtained and reviewed; however, in most cases, no detailed job-specific information was available except dates of employment and job titles. Vital status was ascertained by using data from SSA, the Florida Department of Motor Vehicles, and IRS.
COHORT DESCRIPTIONS 135 PYARs for death were calculated by using the date of hire to December 31, 1997, date of death, or date lost to followup. They were calculated for each race, sex, 5-year age group, and calendar period. Expected deaths were calculated for each 5-year age groups, 5-year calendar period, race, sex, and cause of death. SMRs were calculated for each race and sex. Cause-specific SMRs for all study subjects and race- and sex-specific groups were not significant at p < 0.05. However, the authors note that when lung-cancer deaths were stratified by duration of employment and length of followup, they found an excess in black male workers who had over 10 years of employment and followup (3 observed vs 0.73 expected; SMR, 411; p < 0.05). The study is limited by the lack of job-specific information, information on exposure, and relatively short followup (93% of the cohort had less than 20 years), particularly if one considers that the outcome of interest, lung cancer, has a long latency period. Nuclear-Fuels Fabrication Workers Hadjimichael and colleagues (1983) conducted a retrospective cohort study to examine mortality and cancer incidence in 4,106 nuclear-fuels fabrication plant workers in Connecticut. The plantâs fuel-fabrication process included receiving enriched uranium, fabricating uranium fuel, encapsulating the fuel in a corrosion- resistant metal covering, and assembling it into larger components for reactors. The subjects had been employed at the plant for at least 6 months during 1956- 1978. Personnel records were used to determine job classifications (40 job titles were combined into 16 groups on the basis of similarity of industrial exposure in the manufacturing process). Exposure groups were also developed on the basis of discussions with industrial-hygiene and safety personnel and interviews with supervisors and employees. External-exposure information was obtained from film badges worn by all employees who worked in designated radiation- c Â ontrolled areas; internal exposure was measured with urine bioassays. SMRs and standardized incidence ratios (SIRs) were calculated for each exposure group, and cause-specific SMRs were calculated for all industrial employees. Con- necticut rates were used for indirect standardization. Mortality data were obtained from SSA and from the Connecticut Department of Health Services. Connecticut Tumor Registry data were used to assess the incidence of cancer. The overall cancer incidence in all male employees (industrial and office employees) was significantly lower than expected (SIR, 0.81; 95% CI, 0.65- 0.99). However, brain cancer was marginally significantly higher than expected in industrial male employees (SIR, 2.70; 95% CI, 0.99-5.88). SMRs for male industrial employees were significantly lower than expected for all causes (SMR, 0.83; 95% CI, 0.71-0.97) and lower but nonsignificant for all cancer deaths (SMR, 0.88; 95% CI, 0.62-1.20). Significantly more deaths were observed than expected from chronic obstructive pulmonary disease (SMR, 3.03; 95% CI,
136 updated literature review of depleted uranium 1.11-6.59) and from central and peripheral nervous system diseases (SMR, 3.46; 95% CI, 1.26-7.53). The study had a number of limitations, including an inability to account for multiple exposures of the study population, the lack of detailed smoking information, and the lack of information on exposures that occurred at previous places of employment. The authors noted that the cause-specific ratios may be underestimated because of the inability to account for 7% of death certificates. Similarly, the SIRs may be underestimated inasmuch as some employees were lost to followup or moved outside the catchment area of Connecticut. Finally, followup was short and may not have accounted adequately for cancers that typi- cally have a long latency. United Kingdom Processors Studies of UK processors evaluated cancer incidence and disease-related mortality. Investigating cancer incidence has the advantage of capturing data on people who had malignancies that did not result in death. McGeoghegan and Binks, 2001 In 1959, the Chapelcross nuclear plant in Scotland, a gas-cooled reactor plant, began operation. In 1980, British Nuclear Fuels PLC (BNFL) began the production of tritium. Researchers conducted a study of the employees of this plant from its inception through 1995 (McGeoghegan and Binks, 2001). The cohort consisted of 2,628 people ever employed at the Chapelcross site before January 1, 1996; there were 63,967 person-years of followup and a mean fol- lowup time of 24 years. Subjects were classified as industrial (hourly) or nonin- dustrial (salaried), and this served as a proxy for SES. Workers were also divided into radiation and nonradiation workers; radiation workers routinely carried film badges. The collective radiation dose received by the radiation workers was 185.1 person-sieverts, and the mean cumulative external dose was 83.6 mSv. The mean annual dose was 8.7 mSv. Cancer diagnosis (registration) and vital status of each member of the cohort were determined through the National Health Services Central Register (NHSCR), and death certificates were verified; participants contributed person- years until the date of death, the beginning of missing status, or emigration. Age-, sex-, and calendar-year-specific SMRs and SRRs were calculated on the basis of national population statistics for England and Wales and for Scotland obtained from the Office for National Statistics (ONS) and the Information and Statistics Division, respectively. Because Scotlandâs population is smaller, the Scottish comparisons are not as precise or robust as the English and Welsh com- parisons. Adjustments were also made for industrial status (a proxy for SES), worker status, year of joining, length of exposure, length of service, and length
COHORT DESCRIPTIONS 137 of followup. RRs were calculated to determine differences between radiation and nonradiation workers. Mortality from benign and unspecified neoplasms were significantly higher when compared to English and Scottish populations for radiation workers (SMR, 348) and all workers (SMR, 348), and p values were below 0.05. Endocrine and nutritional diseases, respiratory system diseases, and bronchitis all had higher incidences in radiation workers and all workers than compared to English and Scottish populations, but the SMRs were below 100. No RRs comparing radiation with nonradiation workers were significantly different from 1.0. For the trend analyses of cumulative external exposure and lag, there was a significant increase in RRs only for bronchitis, with p values below 0.05. McGeoghegan and Binks, 2000b The Springfields site at BNFL in Lancashire, UK, was originally a poison- gas factory; in 1948, it was converted to the production of uranium metals. The plant used a chemical separation process to convert yellowcake either to uranium metal or to uranium hexafluoride for further enrichment to uranium oxide fuel. All 19,589 employees (72% of whom were radiation workers) of the plant were studied; they contributed 479,146 person-years of followup through 1995 (2000b). The vital status of each member of the cohort was determined through the NHSCR, and death certificates were verified. Age-, sex-, and calendar-year-specific SMRs and SRRs were calculated on the basis of national population statistics and those for the Lancashire area obtained from ONS. Adjustments were made for industrial status (a proxy for SES), worker status, year of joining, length of exposure, length of service, and length of followup. RRs were calculated to determine differences between radia- tion and nonradiation workers. External radiation at the site was measured with film badges. The maximum cumulative dose was 769.3 mSv, and the median was 9.3 mSv; 95% of all indi- vidual cumulative doses were found to be less than 89.4 mSv. Deaths from all causes differed significantly between radiation workers and all workers, but the SMRs were below 100. The RR comparing radiation and nonradiation workers was 0.88 (p < 0.05). Deaths from all cancers (SMR, 88) and lung cancer (SMR, 86) were significantly decreased and deaths from uterine cancer (SMR, 201) were significantly increased in all workers. Deaths from smoking-related and respiratory system cancers were significantly lower in radiation workers and all workers with SMRs below 100. Significant deficits in deaths were observed for a number of cancers, includ- ing cancers of the stomach (SMR, 71; p < 0.01), colon (SMR, 68; p < 0.001), liver and gall bladder (SMR, 53; p < 0.01), pancreas (SMR, 67; p < 0.05), lung (SMR, 72; p < 0.001), prostate (SMR, 79; p < 0.05), bladder (SMR, 77; p < 0.05),
138 updated literature review of depleted uranium and kidney and ureter (SMR, 59; p < 0.05). However, there was an increased incidence of uterine cancer in all workers (SMR, 168; p < 0.05) and of testicular cancer in nonradiation workers (SMR, 263; p < 0.05). Regarding noncancer mortality, trend analysis with a lag of 15 or 20 years showed significant (pÂ = 0.037 and 0.043) positive results for diseases of the ner- vous and sense organs and with a time lag of 10, 15, or 20 years showed positive results for cerebrovascular disease (pÂ = 0.023, 0.011, and 0.037, respectively). The authors also found significant trend results with a lag of 20 years for dis- eases of the digestive system (pÂ = 0.034), prostatic hyperplasia (pÂ = 0.042), and violence and accidents (pÂ = 0.008). McGeoghegan and Binks, 2000a In 1953, the UK Ministry of Supply opened a gaseous-diffusion plant at Capenhurst for the production of enriched uranium; in 1962, the high-enrichment portion of the plant was closed, but nuclear-fuels production continued; in 1973, a gas-centrifuge process was implemented at a second plant; and in 1982, the diffusion plant was officially closed. Researchers assembled a cohort of all Capenhurst workers employed from the initiation of the plant through 1995 (McGeoghegan and Binks, 2000a). The cohort consisted of 12,543 employees (26% of whom were radiation workers) who contributed a total of 334,473 person-years and had a mean followup of 26.7 years. The NHSCR was used to determine the vital status of the study participants, and participants contributed person-years until the date of death, missing status, or emigration. Age-, sex-, and calendar-year-specific SMRs and SRRs were cal- culated on the basis of national population statistics and those for the Capenhurst area, obtained from ONS. Adjustments were made for industrial status (a proxy for SES), worker status, year of joining, length of exposure, length of service, and length of followup. RRs were calculated to determine differences between radiation and nonradiation workers. External radiation at the site was measured with film badges. The personnel recorded as being at risk for radiation exposure (classified by job status) had a mean cumulative external whole-body dose of 9.85 mSv. Deaths from all causes in radiation and nonradiation workers showed sig- nificant reductions in risk, with SMRs all below 100. The RR between radiation and nonradiation workers was 0.9 (p < 0.05). Mortality from all cancers was significantly lower in radiation workers and in all workers than that of the local population of Capenhurst but not lower than the national (England and Wales) population. Lung-cancer mortality in radiation workers and all workers was also significantly lower than that in the Capenhurst population (SMR, 69 and 85, respectively). Mortality from cancer of the pleura was significantly increased in radiation workers (SMR, 496) and in all workers (SMR, 236).
COHORT DESCRIPTIONS 139 SMRs for nonmalignant diseases of the respiratory and circulatory systems were significantly lower than 100 for radiation workers, nonradiation workers, and all workers. Rocketdyne/Atomics International Workers Ritz et al., 2000 Researchers examined 2,297 nuclear-fuel assembly and disassembly workers who were employed at Rocketdyne/Atomics International (RAI), in California, after 1950. The participants were chosen because they had been part of a moni- toring program at the site from 1950 to 1993, and extensive internal radiation- exposure data were available. Followup was conducted until death or December 31, 1994. Vital status was ascertained from company records or through SSA, the NDI, or California vital-statistics files, and underlying and contributory causes of death were examined. Internal radiation was measured with bioassays, in vivo counting, and whole- body counting for the period 1963-1983. Before that, limited internal monitoring was conducted; and after 1983, all radionuclide operations had ceased. External radiation was estimated from RAI records. Demographic and lifestyle factors for each employee were obtained from personnel records, and workers were categorized. Pay type (hourly, salaried, or managerial) was used as a surrogate for SES, and job location, employment period, or job title was used as a proxy for chemical exposure. Smoking was adjusted for only in a subgroup of the cohort (658 subjects) on whom detailed information was available and was used to assess potential confounding in the larger cohort. The cohort was followed for an average of 25.4 years. Only 0.7% of the workers received an estimated internal radiation dose to the lung greater than 30 mSv, and slightly more than half the workers had recorded doses of 0 mSv. Two types of analyses were conducted: a comparison with the general popu- lation to determine SMRs, which were calculated on the basis of rates in the US white male population; and dose-response analysis of selected combinations of cancer sites (because of low incidence), which used the risk-set approach. Cumulative doses were lagged by 0, 2, and 10 years and adjusted for external radiation exposure. Mortality for all causes of death was significantly lower than expected (SMR, 72; 95% CI, 66-80). No SMR for cancer achieved significance. The test for trend was significantly positive for lymphopoietic and hematopoietic can- cers (pÂ = 0.0001) and upper aerodigestive tract cancers (pÂ = 0.0001). Only the upper aerodigestive tract cancers achieved significance when lagged at 0 years (pÂ = 0.01), 2 years (pÂ = 0.01), and 10 years (pÂ = 0.04), but no clear dose-response relationship was observed.
140 updated literature review of depleted uranium The authors observed significantly fewer deaths from diseases of the circula- tory system (SMR, 68; 95% CI, 58-78) and digestive system (SMR, 41; 95% CI, 21-72), and from all external causes (SMR, 62; 95% CI, 43-86) in the RAI cohort. SMRs for the remaining nonmalignant causes were not significant. A later paper (Boice et al., 2006a) re-examined the methodology of Ritz and colleagues and noted several limitations: the population was small, the low occupational doses limited the analysis, the exposure information was restricted to radiation doses received at Rocketdyne, and the cumulative lung dose from inhaled radionuclides was used as a surrogate of dose to other organs. The authors tried to remedy some of those methodologic limitations in their followup study (see the next section). Boice et al., 2006a Boice et al. (2006a) conducted a retrospective cohort mortality study of 5,801 radiation workers employed for at least 6 months during 1948-1999 at the facility. The approach to identifying the study cohort and the dose-reconstruction methodology were described in detail in another study by Boice and colleagues (Boice et al., 2006b). The cohort was identified primarily by using records from the Radiation Health and Safety Department. All available records were reviewed to determine whether a worker was monitored for radiation exposure externally or internally. External radiation exposure was determined annually, and bioassay data on radionuclide intake were estimated for 16 organs or tissues by using the International Commission on Radiological Protection models. Those models also accounted for delayed dissolution of inhaled material in the respiratory tract. Annual radiation doses received before and after employment were obtained from a variety of databases, including those of the Nuclear Regulatory Commission and DOE. The data were combined to estimate organ doses for each worker. After conducting a dose reconstruction to estimate exposure, Boice and col- leagues (Boice et al., 2006a) compared the observed and expected numbers of deaths on the basis of mortality in the general population of California. To deter- mine person-years of followup, they used the date that was 6 months after the first date of radiation monitoring or July 1, 1948, depending on which was later, and the date of death, December 31, 1999, the date of reaching the age of 95 years, or the date when lost to followup, depending on which was first. The average obser- vation period was 27.9 years. The mean dose from external radiation was 13.5 mSv, and the mean lung dose from combined external and internal radiation was 19.0 mSv. The authors conducted internal comparisons in an attempt to minimize bias generated by comparing rates with those in the general population. Relative risks were generated by using Cox proportional-hazard models. The authors found that mortality from all cancers (SMR, 0.93; 95% CI, 0.84- 1.02) and from all leukemia, excluding chronic lymphocytic leukemia (SMR, 1.21; 95% CI, 0.69-1.97), was not significantly increased. Nor was mortality from
COHORT DESCRIPTIONS 141 any other cancer or other causes of death. No dose-response trends (as determined by Cox regression analyses) were found for any cancers. RRs were calculated for exposure at 100 mSv. For all cancers except leukemia, the RR was 1.00 (95% CI, 0.81-1.24); for all leukemia, excluding chronic lymphocytic leukemia, the RR was 1.34 (95% CI, 0.73-2.45). The strengths of the analysis include the extensive approach to the dose- reconstruction methodology, including the ability to estimate specific organ doses; the nearly complete followup of the workers; and the large comparison group of unexposed workers. Limitations include the small sample, the low doses recorded, and the incomplete availability of workersâ smoking history. Savannah River Plant Workers The Savannah River Plant in Aiken, South Carolina, has been in production since 1952 and is engaged in a variety of processes: uranium processing; nuclear- fuel fabrication; nuclear-reactor operation; nuclear-reactor overhaul, modifica- tion, and maintenance; nuclear-reactor refueling; and nuclear-fuel reprocessing. The occupational radiation dose to workers at this site was 85% external and 15% internal. In 1988, researchers examined records of employees of the Savannah River Plant who had worked there from 1952 to 1975 (Cragle et al., 1988). They identi- fied 17,922 people with complete data records. When the cohort was limited to white men who had worked for more than 90 consecutive days and were either salaried or hourly, a total of 9,860 subjects remained. Person-years were calcu- lated from the date of first hire plus 90 days. Those with unknown status at the time of the study contributed person-years until the date of last contact. Study participants were categorized in three ways: type of employment, date of first employment, and duration of employment. Type of employment served as a proxy for SES and contained three groups: salaried, hourly, and combined. Date of first employment stratified the cohort into those who were hired before 1955 and those were hired in or after 1955 to identify a group with the longÂ est followup. Duration of employment served as a surrogate for exposure and included those who were employed for less than 5 years, for 5-14 years, and for more than 14 years. Mortality was compared with that in the US white male population, adjusted for age and calendar year, and stratified by the three categories described above. No SMRs for combined hourly and salaried workers were significant. How- ever, mortality from all causes and from all cancers was significantly lower than expected (p < 0.05) in hourly workers (all-causes SMR, 8; all-cancers SMR, 72) and in salaried workers (all-causes SMR, 64; all-cancers SMR, 68). There were also significant deficits in respiratory cancer deaths for both hourly and salaried workers and brain and central nervous system cancer deaths in hourly workers:
142 updated literature review of depleted uranium SMRs were all below 100. All deaths from infective and parasitic diseases and diabetes were significantly lower in hourly but not salaried workers. Deaths from respiratory, gastrointestinal, circulatory, and genitourinary diseases were all sig- nificantly increased in both hourly and salaried workers, but SMRs were below 100. External causes of death were significant in salaried workers only, but again SMRs were below 100. Atomic Weapons Establishment Workers Mortality in 22,552 employees of the Atomic Weapons Establishment in the UK was studied by Beral et al. (1988). All employees who worked at the Alder- maston, Fort Halstead, Orfordness, Foulness, and Woolwich Common facilities at any time from January 1, 1951, to December 31, 1982, were included in the study. The average followup time was 18.6 years. Exposure of 9,389 workers to uranium and other radiation sources was mea- sured with dosimeters. The average cumulative whole-body exposure to external radiation was 7.8 mSv. Internal radiation dose was not estimated, because the dose probably would have varied from organ to organ and absorption and deposi- tion of radionuclides are âoften difficult to assess from external measurements.â PYARs were calculated from the date of first hire or January 1, 1951, if the worker was recruited before then (records were incomplete before 1951). The data were stratified by age, sex, calendar period, and social class. Overall mortality in the employees was lower than that in the general popula- tion. Mortality in the employees with radiation records was similar to that in other employees. However, after a 10-year lag, mortality from prostatic cancer (RR, 2.23; 95% CI, 1.13-4.56) and mortality from cancers of ill-defined and secondary sites (RR, 2.37; 95% CI, 1.23-4.56) were significantly increased. A significant dose-response trend was also noted for prostatic cancer when uranium exposure and cumulative whole-body exposure to external radiation were monitored. The study used a relatively large cohort but was limited in that fewer than half the workers had individual monitored exposure and smoking information was lacking. Egyptian Processors In this study of uranium workers in Egypt, Shawky and colleagues (2002) monitored external radiation exposure at two uranium-processing sites. The study population consisted of 86 processors at milling, monazite-production, and Âyellow-cake production locations who handled ores and materials that had high concentrations of naturally occurring radioactive materials. Work histories and descriptions were recorded. Dust monitoring and bioassays were conducted to determine radiation exposure of workers. Hematologic and renal-function measures were assessed in a clinical evaluation in which all study subjects
COHORT DESCRIPTIONS 143 underwent a complete blood count and measurement of serum creatinine and urea, and urinary uranium. Thirteen study subjects provided spot urine samples for urinary-uranium analysis because timed and 24-hour urine samples were difficult to ascertain. Uranium concentration, expressed in micrograms per liter, was measured with laser fluorimetry. Air samples were collected to measure air uranium concentration. Linear regression was used in the analysis. Mean urinary uranium concentration was 17.8 Âµg/L in the 13 participants who provided spot urine specimens; urinary uranium ranged from 8-29 Âµg/L. There was a correlation between urinary uranium and serum creatinine in the 13 specimens, and mean uranium excretion was more than 20 times the occupational-exposure decision level of 0.8 Âµg/L. Depleted-Uranium Studies This section describes studies that examined the health outcomes related to exposure to depleted uranium as a result of military deployment; the studies are also summarized in Table 7-2. The literature focuses on veterans deployed to conflicts in the Balkans and the Persian Gulf region. This section begins with a case series of US Gulf War veterans involved in friendly-fire incidents who received fragments of depleted-uranium shrapnel. Next, it summarizes the cohort studies that examined the mortality experience and cancer outcome of UK Gulf War veterans, followed by studies that assessed cancer incidence primarily in European service personnel deployed to the Balkans. Finally, it summarizes a study on workers exposed to depleted uranium at the FFMPC in Ohio. Baltimore Veterans Affairs Medical Center Studies Since 1993, the Depleted Uranium Follow-up Program at the Baltimore Vet- erans Affairs Medical Center (Baltimore VAMC) has sought to provide clinical surveillance of Gulf War veterans exposed to depleted uranium through friendly- fire incidents. Depleted uranium was first used by US and other military during the first Gulf War as material for tank armor and in weaponry (McDiarmid et al., 2004). During the course of that conflict, soldiers in or on vehicles and tanks âwere mistakenly fired on and struck by munitions containing DU [depleted uranium]â (McDiarmid et al., 2000) and are thought to have inhaled or ingested airborne depleted-uranium particles or to have experienced wound contamination by depleted uranium. In addition, some soldiers had multiple tiny fragments of depleted uranium scattered throughout muscle and soft tissue. As a result, the Department of Veterans Affairs established a medical surveillance system to determine health effects in depleted-uraniumâexposed veterans, evaluate tech- niques to measure uranium, and assess possible surgical management of shrapnel (McDiarmid, 2007). The results of the surveillance program are detailed in a number of studies by researchers at the Baltimore VAMC.
144 updated literature review of depleted uranium Since the start of the Depleted Uranium Follow-up Program, investigators have prospectively evaluated 74 of the estimated 100 survivors of the friendly- fire incidents during the Gulf War (McDiarmid et al., 2006). The program has started biologic monitoring of soldiers and veterans of Operation Iraqi Freedom (McDiarmid, 2007). In 1993-1994, the first group of depleted-uraniumâexposed veterans was evaluated by the Baltimore VAMC team. Of the 33 examined, nearly half were confirmed through skeletal examination as having uranium fragments embedded in a number of locations throughout the soft tissue. They also had much higher mean urinary uranium concentrations than those without retained fragments (4.47 vs 0.03 Âµg/g of creatinine), but no other effect was detectable (McDiarmid et al., 2000). Those veterans were examined every 2 years to assess functioning of the major target organ systems likely to be affected by uranium (primarily the kid- neys, the central nervous system, and the reproductive system). The surveillance protocol, based on uraniumâs known and presumed toxic properties, consists of a detailed questionnaire to document medical history, socioeconomic background, and occupational exposure and extensive laboratory testing that includes hema- tologic and clinical-chemistry measures, urinalysis, seminal and blood uranium, renal markers, seminal analysis, and reproductive endocrine measures; neurocog- nitive testing; and chromosomal analysis to test for chromosomal aberrations and hypoxanthine-guanine phosphoribosyl transferase (HPRT) (McDiarmid et al., 2001). A urinary uranium concentration of 0.10 Âµg/g of creatinine was used as a cutpoint to compare mean values between âhighâ and âlowâ uranium-exposure groups. The studies discussed below examine the health effects of depleted uranium in a group of Gulf War veterans examined at the Baltimore VAMC in 1997, 1999, 2001, 2003, and 2005. The studies concerned primarily people who had retained fragments of depleted-uranium shrapnel and those who suffered inhalation expo- sure. Results of the 1997 evaluation were discussed in Volume 1; a brief summary of that study is provided. McDiarmid et al., 2000 Of the 33 Gulf War veterans with retained fragments of depleted-uranium shrapnel examined in 1993-1994, 29 were re-examined in 1997 for clinical health effects associated with friendly-fire exposure, and results were compared with those of examinations of 38 veterans not exposed to depleted uranium. Age and military rank were used to recruit unexposed veterans. The authors used several sources, including Army and National Guard units, advertisements, and a Depart- ment of Defense hospitalization database. Exposure status was determined from â Human genotoxic outcomes have been explored in greater detail in Chapter 4; therefore, little attention is given here to measures used in the Baltimore VAMC surveillance.
COHORT DESCRIPTIONS 145 medical records, telephone screening, and a series of questions about military experience. Clinical evaluation included a complete history and physical examination and a series of laboratory tests to assess hematologic and renal-function mea- sures. Urinary and seminal uranium concentrations and whole-body radiation counting were used to determine exposure. For total-uranium analysis, 24-hour and spot urine samples were collected. The resulting values were expressed in micrograms per gram of creatinine. Kinetic phosphorescence analysis was used to measure seminal uranium concentrations. The authors measured a number of clinical elements in relation to urinary and seminal uranium concentrations. Tra- ditional (paper and pencil) and automated neurocognitive testing batteries were used, and two impairment indexes (one based on traditional measures and one on automated measures) were created for analysis. The ratio of the total number of below-expectation scores to the scores obtained for each battery was used to determine impairment indexes. When t scores were not available, decision cutpoints were used. A number of reproductive-health measures were analyzed by using concentrations of follicle-stimulating hormone (FSH), luteinizing hor- mone (LH), thyroid-stimulating hormone (TSH), free thyroxine, prolactin, and testosterone. In addition, semen characteristics (volume, sperm concentration and total count, and functional measures of motility) were evaluated by using World Health Organization (WHO) criteria for semen normality. Peripheral blood lym- phocytes were cultured to examine frequencies of chromosomal aberrations and sister-chromatid exchange. On the basis of 24-hour and spot urinary-uranium values, the participants were divided into high- and low-exposure groups. The high-exposure group con- sisted of 14 veterans with urinary uranium greater than 0.10 Âµg/g of creatinine. The low-exposure group consisted of all subjects with spot urinary uranium of less than 0.10 Âµg/g of creatinine; 15 depleted-uraniumâexposed and 38 unex- posed veterans were in this category. Researchers used correlation and regres- sion analyses to evaluate exposure measures, using 24-hour urinary uranium as the primary measure of exposure. Results were stratified at the median to create low- and high-result groups and compared with the results in the low- and high- exposure groups to determine an association between higher and lower median tendencies. In a separate analysis, neurocognitive indexes were modeled as a function of urinary uranium concentration with adjustment for intelligence (the Wide Range Achievement Test 3 Reading, WRAT-3 Reading) and psychiatric status (the Beck Depression Inventory, BDI). â Traditionalneurocognitive measures included the Wide Range Achievement Test 3 Reading, the National Adults Reading Test, the California Verbal Learning Test, the Trail Making Test Parts A and B, the Shipley Institute of Living Scale, and the Digit Span, Arithmetic, and Digit Symbol subsets of the Wechsler Adult Intelligence Test-Revised. Automated measures included Automated NeuroÂ psychological Assessment Metrics, the Nonverbal Selective Reminding Test, and the Kay Continuous Performance Test.
146 updated literature review of depleted uranium Seven years after first exposure, veterans who had retained depleted- u Â ranium shrapnel fragments had higher urinary uranium concentrations than those who did not have retained shrapnel. Urinary uranium ranged from 0.01 to 30.74 Âµg/g of creatinine in veterans with retained fragments vs 0.01 to 0.05 Âµg/g of creatinine in unexposed veterans. Renal-function measures (serum cre- atinine, beta-microglobulin, retinol-binding protein, serum uric acid, urinary creatinine, and urinary protein) were quite similar between exposure groups. No significant differences were observed between high- and low-exposure groups in hematologic, reproductive, and genotoxicity measures. In general, outcome measures in the depleted-uraniumâexposed were within normal clinical lim- its. Results suggested a statistically significant relationship between increased urinary uranium and poor performance on automated neuropsychologic tests regardless of the regression model used (24-hour urinary uranium for depleted- uraniumâexposed veterans, pÂ = 0.01; and spot urinary uranium for all veterans, pÂ = 0.01). McDiarmid et al., 2001 In March-July 1999, researchers at the Baltimore VAMC continued their assessment of clinical health effects in Gulf War veterans exposed to depleted uranium by friendly fire. The study population consisted of 50 male depleted- u Â raniumâexposed Gulf War veterans who had retained fragments and were excreting uranium at increased rates 8 years after first exposure. Of the 50, 21 had participated in previous Baltimore VAMC studies, and 29 were newly iden- tified by the study team. Using published estimates of mean urinary uranium concentrations in unexposed groups (11-22 ng/L) and upper dietary limits as a result of naturally occurring uranium in groundwater (up to 0.35 Âµg/L), the authors established lowâ and highâurinary-uranium groups. Of the 50 veterans, 13 veterans had urinary uranium greater than 0.10 Âµg/g of creatinine and were therefore in the high-exposure group; the remaining 37 were below the cutpoint of 0.10 Âµg/g of creatinine. Three of the 29 new participants had urinary uranium greater than 0.10 Âµg/g of creatinine. The 1999 clinical assessment replicated the protocol from the previous study and consisted of a laboratory examination to evaluate hematologic and renal- function measures, reproductive function (semen quality and neuroendocrine function), and genotoxic measures and a detailed questionnaire history and com- plete physical examination. Test batteries used for neurocognitive-performance measures were similar to those used in the 1997 surveillance (McDiarmid etÂ al., 2000). The traditional neuropsychologic-test measures were used to create an index score. Three impairment-index scores were obtained from automated mea- sures based on response accuracy, median response time for correct response, and number of correct responses per minute. Reproductive endocrinologic values included measurements of TSH, free thyroxine, and the hormones previously
COHORT DESCRIPTIONS 147 assessed (prolactin, FSH, LH, and testosterone). Only 44 of the 50 samples were considered in the semen analysis because six veterans were azoospermic. Enzyme treatment was used for 17 samples (12 with low and 5 with high urinary ura- nium). In the assessment of genotoxicity, cultured peripheral blood lymphocytes were tested for chromosomal abnormalities and sister-chromatid exchange for baseline measurements. The cells were subjected to two concentrations of bleo- mycin (4 and 8 Âµg/mL). A number of potential confounders were adjusted for in the regression analysis, including current smoking status and use of psychotropic and antidepressant drugs. Robust regression analysis was used to account for highly influential observations of neurocognitive function in adjusting for intel- ligence (WRAT-3) and depression (BDI). Urinary uranium ranged from 0.018 to 39.1 Âµg/g of creatinine in the depleted- uraniumâexposed veterans with retained fragments and from 0.002 to 0.231 Âµg/g of creatinine in depleted-uraniumâexposed veterans without fragments. In gen- eral, clinical tests revealed hematologic, renal, and neuroendocrine measures that were within normal limits with slight differences between highâ and lowâurinary- uranium groups. When veterans were assessed for active medical problems, those in the high-uranium group were found to suffer a higher proportion of injuries than those in the low-uranium group (76.9% vs 45.9%; pÂ = 0.05). Hematologic measures had statistically significant differences between exposure groups that were not observed in the previous surveillance. The highâurinary-uranium group had a lower mean lymphocyte count (32% vs 37%; pÂ = 0.04), a higher mean neutrophil percentage (55% vs 49%; pÂ = 0.03), and a lower mean monocyte percentage (7.6% vs 9.1%; pÂ = 0.01). The authors did not detect any clinically important changes in renal function due to depleted-uranium exposure. Urinary creatinine concentration was slightly lower in the highâurinary-uranium group, but the difference only marginally significant. Results of neurocognitive tests were not consistent with those in past evalu- ations. The relationship between urinary uranium and performance on automated measures observed in the 1994 and 1997 surveillance appeared to weaken and was only marginally significant when WRAT-3 and the BDI were adjusted for high and low urinary uranium. There were no statistically significant differences in concentrations of FSH, LH, prolactin, testosterone, TSH, or thyroid hormones between low and high groups. Of the 44 sperm samples included in the analysis, three were designated subnormalâthat is, as having values of at least three of the five clinical measures below normal, as defined by the WHO standards. The highâurinary-uranium groups showed increases in mean total sperm count (583.5 Â± 106.1 vs 286.6 Â± 44.8), total progressive sperm (220.9 Â± 44.0 vs 108.2 Â± 19.2), and total rapid pro- gressive sperm (155.5 Â± 31.1 vs 81.3 Â± 15.4), and the differences were significant (p < 0.02, 0.03, and 0.04, respectively).
148 updated literature review of depleted uranium McDiarmid et al., 2002 In this study that revisited their previous results, McDiarmid and colleagues identified 30 new members of the originally exposed group. Urinary uranium concentrations were measured, and correlation analyses were conducted to deter- mine the relationship between excretion measures in the 1994, 1997, and 1999 surveillance groups. An increase in urinary uranium (24-hour urinary uranium concentrations higher than 0.05 Î¼g/g of creatinine) was observed in four of the 30 newly identified veterans. Urinary uranium showed a high correlation (R-squared [rsq] = 0.8623) between the 1994 assessment and the 1997 assessment. A strong correlation (rsq = 0.8764) was also observed between the 1994 and the 1999 assessments. McDiarmid et al., 2004 In the third surveillance, 39 Gulf War veterans were examined at the Bal- timore VAMC during April-July 2001. Of the 39, eight were identified as new participants, and the remaining 31 had participated in the followup program at least once before. As in earlier studies, the authors investigated a number of clini- cal outcomes as related to urinary uranium concentrations 10 years after the initial exposure to uranium. In addition to the clinical measures assessed previously (McDiarmid et al., 2000, 2001), the evaluation considered immunologic measures and mutagenic effects related to depleted-uranium exposure by assessing HPRT mutation frequency. The 29 participants with no history in the Depleted Uranium Follow-up Program were also evaluated for posttraumatic stress disorder and substance abuse. The exposure groups were defined on the basis of participantsâ 2001 urinary-uranium results. Thirteen participants were in the high-exposure category (with concentrations greater than 0.10 Î¼g/g of creatinine), and 26 in the low-exposure category. Neurocognitive test batteries were similar to those used previously (McÂDiarmid et al., 2001). Automated measures were selected from the Automated Neuropsychological Assessment Metrics Test Library. The authors constructed four indexes of impairment based on the traditional tests and the automated measures (response accuracy, median response time for correct response, and number of correct responses per minute or throughout). The indexes represent the proportion of scores that fell 1 standard deviation below the mean. The BDI was used to evaluate emotional status. Genotoxic tests were adjusted for potential confounders (age, smoking, exposure to genetic toxicants, and cloning efficiency). Urinary uranium ranged from 0.001 to 78.125 Î¼g/g of creatinine. The pres- ence of retained depleted-uranium shrapnel appeared to be associated with higher urinary uranium concentration. In addition, most urinary-uranium results were fairly consistent throughout a given personâs history in the group.
COHORT DESCRIPTIONS 149 The percentage of veterans who reported suffering injuries during friendly-fire incidents showed some significant differences between high- and low-Âexposure groups. Mean values of all hematologic and renal-function markers were within normal clinical limits with few statistically significant differences between highâ and lowâurinary-uranium groups. Differences in hematocrit (42.59% in the high group and 44.60% in the low group) and hemoglobin (14.79 vs 15.40 g/dL) lev- els were not observed in the 1997 and 1999 surveillance groups. Renal-function measures showed movement toward decreased protein reabsorption and increased glomerular filtration of protein: serum creatinine concentrations were higher in the low-uranium group (0.85 vs 0.95 mg/dL; pÂ = 0.03); values for urinary retinol- binding protein and total urinary protein concentrations were higher in the high- uranium group (retinol-binding protein, 65.68 vs 46.13 Âµg/g of creatinine; pÂ = 0.06; and total protein, 78.69 vs 54.63 mg/g of creatinine; pÂ = 0.01, respectively). As in past years, neurocognitive measures did not differ between exposure groups. Overall neuroendocrine function was normal, but mean free thyroxine was higher in the low-uranium group (1.66 vs 1.08 ng/dL)âa result not observed in the 1997 and 1999 surveillances. Semen measures were higher in the high-uranium group, but the differences did not achieve statistical significance. Immunologic measures revealed a higher proportion of CD4+ T cells in the high-uranium group (65.98% vs 60.83%) and a lower proportion of CD8+ T cells (26.55% vs 31.28%). The authors reported a statistically significant difference between groups with respect to changes in mean baseline measurements of chromosomal aberrations (0.01 Â± 0.004 in the high-uranium group and 0.003 Â± 0.001 in the low-uranium group; pÂ = 0.027). There were no statistically significant differences in HPRT mutation frequencies between exposure groups. McDiarmid et al., 2006 This evaluation of the friendly-fire group took place 12 years after first expo- sure to depleted uranium. The authors reported on the same health outcomes examined in the 1999 and 2001 evaluations. They examined 32 Gulf War veterans in April-July 2003 for hematologic and blood-chemistry measures, renal function, neurocognitive function, genotoxic measures, and reproductive neuroendocrine function and semen characteristics. Urinary-uranium results obtained in the 2003 examination were used for group composition, and a concentration of 0.10 Âµg/g of creatinine served as the cutpoint for high- and low-exposure groups. The high- exposure group consisted of 13 with urinary uranium of at least 0.10 Î¼g/g of creati- nine, and the low-exposure group 19 with less than 0.10 Î¼g/g of creatinine. Robust regression and polynomial transformation analysis were applied to account for possible outliers in the linear-regression model for the neurocognitive evaluation. Results for all health measures were within normal clinical limits. The dif- ference in serum phosphate concentration was the only measurable difference (pÂ = 0.03) between the high-exposure group (3.75 mg/dL) and the low-exposure
150 updated literature review of depleted uranium group (4.11 mg/dL). Higher mean values of semen characteristics were observed in the highâurinary-uranium group. Despite the persistently increased urinary uranium concentrations, no clinical abnormality or dysfunction was observed. McDiarmid et al., 2007 In the 2005 surveillance, 34 members of the depleted-uraniumâexposed Gulf War veteran group were examined 14 years after first exposure. The authors used clinical assessments that had been used in previous evaluations to determine uri- nary uranium concentrations, renal function, hematologic and blood-Âchemistry characteristics, neuroendocrine measures, semen quality, genotoxicity, and neu- rologic function. Fluorescent in situ hybridization assay analysis was carried out to detect low-level chromosomal abnormalities. As in past evaluations, data on urinary uranium were divided into low and high groups on the basis of a cut- point of 0.10 Î¼g/g of creatinine. In addition, investigators measured cumulative uranium exposure for each participant to account for the duration and intensity of exposure, bearing in mind urinary uranium concentrations for each surveil- lance visit and the time between measurements. The latter metric had a cutpoint of 10 Î¼g/g of creatinineÂ·years (based on the distribution of data) and resulted in a group composition consistent with the current uranium cutpoint of 0.10 Î¼g/g of creatinine. As in previous surveillance years, robust regression was used to account for outliers. Age and cloning efficiency were adjusted for in the analysis of mean HPRT mutation frequencies. Urinary uranium concentrations ranged from 0.002 to 44.1 Î¼g/g of creatinine for total 24-hour urinary uranium concentration in participants known to have embedded depleted-uranium shrapnel fragments and specific indicators of depleted uranium at or above 0.10 Î¼g/g of creatinine. The results showed a high correlation between current and cumulative uranium exposure, with an R2 value of 0.827. Results regarding health outcomes were fairly consistent with past evalua- tions. There were no statistically significant differences between high and low uri- nary uranium in hematologic, blood-chemistry, and neuroendocrine measures, and they were generally within normal clinical limits. Mean serum uric acid reached significance (pÂ = 0.03) when high and low groups were compared for cumula- tive uranium exposure. Despite that finding, the values were within the normal clinical range, and the difference was rather small: the high group registered 5.22 mg/dL and the low 6.19 mg/dL. Other renal measures revealed no significant dif- ferences regardless of the exposure metric used. Results of neurocognitive testing were similar to those in past years. The authors found no statistically significant differences between exposure groups in all neurocognitive indexes when either exposure metric was used. Mean values of semen characteristics also showed no significant difference; however, values of percent progressive sperm and percent rapid progressive sperm were lower in the high-uranium group (pÂ = 0.15 and 0.12, respectively) when the current exposure metric was used.
COHORT DESCRIPTIONS 151 Unlike many studies, the Baltimore VAMC study examined targeted health outcomes in the study group that might have resulted from continuous exposure from embedded depleted-uranium shrapnel. The study had a long-term and con- sistent followup, particularly in relation to uranium excretion, which allowed one to see the chronic health effects in this population. In the most recent study (McDiarmid et al., 2007), the use of measures of cumulative exposure and depleted uranium enhanced the specificity of exposure. However, the small and select sample and the absence of a control group limited its ability to detect effects in the various outcomes. In addition, the selection of the urinary-uranium cutpoint of 0.10 Î¼g/g of creatinine was not based on a generally accepted standard for urinary uranium. UK Gulf War Studies All UK military personnel who were deployed to the gulf region from September 1990 to June 1991 were evaluated in a retrospective cohort study (Macfarlane et al., 2000, 2003). The cohort consisted of 53,462 service members and an age-, sex-, service-, and rank-matched comparison group of 53,462 service members who were not deployed to the gulf region in the same period. Personnel were linked to the NHSCR to determine cancer diagnosis and vital status, and deaths were coded according to ICD-9. After exclusion of those who had died before the end of the Gulf War, those who had emigrated from the UK during the study period, and those whose vital status could not be determined, the Gulf Warâdeployed group consisted of 51,721 participants, and the nondeployed group consisted of 50,755 participants. Initial cancer diagnoses in the registry through July 2002 were included in the analysis, and person-years at risk were calculated from April 1, 1991, to the earliest of either the date of emigration from the UK, the date of death, the date of first diagnosis of cancer, or July 31, 2002. Of the 51,721 deployed to the gulf, 2,092 reported an exposure to depleted uranium. No excess risk of cancer overall was observed in the Gulf War veterans: there were 270 incident cancers in the Gulf Warâdeployed and 269 in the nondeployed (incidence rate ratio, 0.99; 95% CI, 0.83-1.17). No excess risk of any site-Âspecific cancers was found, and adjustment for lifestyle factors and other potential con- founders did not change the results. Balkans Studies Gustavsson et al., 2004 Swedish military and civilian rescue personnel deployed to the Balkans in 1989-1999 were studied to determine whether they had a higher incidence of cancer (Gustavsson et al., 2004). Swedish Armed Forces and Swedish Rescue
152 updated literature review of depleted uranium Services Agency registries were used to assemble the cohort; most subjects served 6-month missions. Each person was matched to the Swedish Cancer Reg- istry, and 99.9% of the subjects could be followed up; this resulted in a cohort of 8,347 military men, 433 military women, 403 civilian men, and 5 civilian women. Person-time was calculated through the end of followup (1999) or until death, emigration, or cancer diagnosis. SIRs were calculated on the basis of cancer incidence in the general population, and adjustments were made for sex, age (5- year age groups), and period. No measurement or modeling for depleted-uranium exposure was included. There were 34 incident cases of cancer diagnosed during the followup period compared with 28.1 expected in the cohort (SIR, 120; 95% CI, 90-170). Eight cases of testicular cancer were identified in military men compared with 4.3 expected. The authors reported no statistically significantly increased incidence of cancer but recognized that the followup period was too short to assess the long-term risk of cancer. Nuccetelli et al., 2005 On the basis of reports of possible depleted-uraniumârelated cancer risk, the Italian Ministry of Defense examined a large portion of the Italian military deployed to the Balkans during December 1995-January 2001. The cohort con- sisted of about 40,000 soldiers 20-59 years old who had been deployed at least once in that period and contributed about 80,000 person-years. Cancer incidence was calculated for 5-year age groups for all cancers and specific cancers: Hodg- kin lymphoma, non-Hodgkin lymphoma, acute lymphocytic leukemia, and solid tumors. SIRs were calculated on the basis of cancer registries for the general Italian male population. No measurement or modeling for depleted-uranium exposure was included. The overall incidence of cancer was significantly lower than expected. The only cancer that was significantly increased was Hodgkin lymphoma (SIR, 236; 95% CI, 122-436). A limitation of the study is that the followup period was too short to assess cancer outcomes. Storm et al., 2006 After reports of increased cancer incidence in military personnel deployed to the Balkans, Danish Defence Health Services and the Danish Cancer Society undertook a study of Danish military (Storm et al., 2006). From January 1992 to December 2001, 15,091 persons were deployed to the Balkans. After exclusion of those deployed to other conflicts, those with errors in their files, and those with a previous diagnosis of a cancer, the cohort contained 14,012 people. The entire cohort was followed through December 2002 or until death, emigration, or loss to followup. SIRs were calculated for the personnel by using corresponding incidences in the Danish population.
COHORT DESCRIPTIONS 153 No significantly increased SIRs were observed for all cancers or site-Âspecific cancers except bone cancer; of these, there were four cases, three of which occurred in the first year. The SIR for all bone cancers was 600 (95% CI, 160-1,530); if the first year was excluded, it was 170 (95% CI, 0-1,010). Sumanovic-Glamuzina et al., 2003 During the Bosnian War, civilians were potentially exposed to environmen- tal contaminants that might have resulted in increases in malignant diseases and other adverse health effects. However, the magnitude of exposure and the result- ing health outcomes remained unclear. In 2000, researchers sought to assess the prevalence of major congenital malformations in two 1-year cohorts of neonates born immediately after the war (in 1995) and in 2000, 5 years after military activities. The study population included all live-born neonates and stillborn fetuses in the maternity ward of the Mostar University Hospital in western Her- zegovina. Malformations were documented by using the EUROCAT Protocol during physical examination of live-born and stillborn neonates. Autopsies were not performed on the stillborn. For the 1995 cohort, data on prenatal and perina- tal complications were collected from medical records and interviews with the mothers. Interviews were not conducted for the 2000 cohort. Prevalence was analyzed with consideration of the relevant organ systems, sex, and gestational age, and chi-square tests were conducted. Aborted fetuses were not included in the analysis. In 1995, 40 of 1,853 neonates had major malformation, a prevalence of 2.16% (95% CI, 1.49-2.82%). In 2000, 33 of 1,463 (2.26%) had major malforma- tions (95% CI, 1.50-3.01%). Anomalies of the cardiovascular and central nervous systems were significantly higher in the 2000 cohort than in the 1995 cohort. Environmental-Exposure Studies The studies reviewed below examine health outcomes in persons who lived near uranium-processing facilities or in households in Finland where well water with high uranium content was the primary source of drinking water. The studies are summarized in Table 7-3. Residential Studies Bithell and Draper, 1999 Greenham Common US Air Force base in Berkshire in the UK was the site of a B-47 jet fire in 1958. Residents living around the base expressed concern that depleted-uranium contamination from the fire might have resulted in an increased incidence of cancer, particularly leukemia, in the area. In a 1961 declassified document, excess concentrations of plutonium and uranium were modeled on a
154 updated literature review of depleted uranium contour map. The concentrations peaked in a line along the runway and at two points about 2 km from each end of the runway, resulting in a dumbbell shape. The concentrations were determined on the basis of the ratio of 235U to 238U in 26 evergreen leaf samples taken from the area. Bithell and Draper (1999) analyzed the information and determined that some errors had occurred in the environmen- tal modeling. They estimated that there was only a 1% increase in uranium in the area and a 0.1% increase in uranium activity on the basis of estimated level of enrichment. The researchers then used an existing set of data on the incidence of child- hood leukemia and non-Hodgkin lymphoma that was constructed specifically to show the spatial distribution of cancers around nuclear facilities. In 1966-1987 throughout Britain, 11,283 cases were diagnosed; expected numbers of cases were calculated, allowing for various socioeconomic and demographic factors. The researchers looked at cancer incidence within 6 km of the runway, counted the children who resided in that area, and compared the cases with estimated values. Within 6 km of the study area, 15 cases of leukemia were observed compared with 13.4 cases expected. There was no evidence that the observed cases were closer to the airfield than expected; that is, there was no clustering of cancers around the airfield. Boice et al., 2003b Reports of cancer clusters around nuclear facilities have prompted a number of population-based descriptive studies to determine whether cancer mortality has increased. In 2003, researchers examined cancer rates in people who lived around two nuclear-materials processing facilities in Pennsylvania. The two facilities were 3 miles apart on the Kiskiminetas River, which forms the boundary between Armstrong County and Westmoreland County. The Apollo facility, which pro- cessed uranium fuel, opened in 1957; the Parks facility, which processed uranium and plutonium, opened in 1960. Uranium processing ceased in 1983, plutonium processing ceased in 1980, and both sites have been decommissioned and decon- taminated. Both facilities were in Armstrong County, and Westmoreland County is just across the river; researchers examined cancer-mortality data on the two counties. Most of the population lived within 20 km of both sites, although wind modeling suggested that Armstrong County had a greater risk of exposure. Six counties were chosen as controls on the basis of similarities in race, geo- graphic density, employment status, poverty level, age, education, mean family income, and population size. Other cancer risk factors, such as diet and smoking, could not be controlled for with this method. Death rates in 1950-1995 based on National Center for Health Statistics data were collected for each county and categorized by cause, sex, race, and 5- year age group. SMRs were calculated for each county on the basis of national
COHORT DESCRIPTIONS 155 statistics. Rates were also divided into three periods to capture deaths before, during, and after the operation of the plants. RRs were calculated from the SMRs between the study counties and the control counties. No significant excess in mortality was found in the study counties in com- parison with the control counties or the US population for all cancers and cancers of a priori interest (lung, bone, hepatic, and renal cancers). Mortality rate ratios for all malignancies were similar in the three periodsâbefore, during, and after plant operation (RR, 0.96, 0.95, and 0.98, respectively). Boice et al., 2003a The researchers who studied the Apollo and Parks sites also surveyed cancer incidence in a smaller region around the facilities. They looked at eight boroughs and municipalities, identifying 16,772 people who lived in the area in 1990 in whom cancer incidence could be determined. They then examined cancer inci- dence in 1993-1997. The population was categorized by age, sex, and race, and person-years in each category were calculated. Expected numbers of incident cancer cases were determined by multiplying person-years by incidence of spe- cific cancers in Pennsylvania. The researchers then determined cancer incidence in each municipality and constructed SIRs. No significant SIR was found for all cancers or a number of cancer sites, such as lung (SIR, 88), kidney (SIR, 105), non-Hodgkin lymphoma (SIR, 110), liver (3 observed vs 4.91 expected), and bone (2 observed vs 1.19 expected). However, a statistically significant excess of cervical cancer was observed (SIR, 235; 95% CI, 113-433). Boice et al., 2003c A similar study was conducted in Karnes County, Texas, where uranium milling and mining began in 1954 (Boice et al., 2003c). The three mills and 40 in situ mines (also known as solution mining) and surface mines were in opera- tion until the early 1990s and created concern about environmental exposure in the resident population. In 1961, the Texas Department of Health began environ- mental monitoring; in 1988, it began sampling residential areas (including water supplies, homes, and food items). Researchers examined cancer mortality in the general population of Karnes County. They chose four control counties in the same region in Texas as com- parison areas on the basis of similarities in race, geographic density, percentage employed in manufacturing, poverty level, age, education, mean family income, and population size. Numbers of deaths in each county in 1950-2001 were obtained and categorized by cause, sex, race, and 5-year age group. Expected numbers were calculated for each county and compared with the observed values to determine SMRs. In addition, numbers of deaths were combined in the periods
156 updated literature review of depleted uranium 1965-1979, 1980-1989, and 1990-2001 to allow comparisons before, during, and after milling and mining operation. RRs for cancer mortality were computed between Karnes County and the control counties and for three periods (1965- 1979, 1980-1989, and 1990-2001). Overall, 1,223 cancer deaths occurred compared with 1,392 expected (SMR, 88). There was no statistically significant increase in SMRs for cancers of a priori interest (lung, bone, renal, and hepatic cancer). The only exception was colon and rectal cancer in the early periodâbefore and in the early years of operation. Pinney et al., 2003 During 1951-1988, the FFMPC, in Fernald, Ohio, processed uranium ore and other uranium feed materials for nuclear-weapons production. Uranium was released into the atmosphere and groundwater of the surrounding area, and large amounts of radon and other decay products are thought to have been dispersed into the surrounding air. Household drinking-water sources included reservoirs, such as cisterns, where rainwater was collected from roof gutters. People who lived near the site were potentially exposed to radiologic and nonra- diologic toxicants from groundwater, soil contamination, and plant emissions. A m Â edical-surveillance program, the Fernald Resident Medical Monitoring Program (FMMP), was later created to monitor the health of Fernald residents. Pinney and colleagues conducted a study to determine the prevalence of chronic disease in residents who lived near the Fernald facility. The study cohort was selected from the FMMP, which included adults who lived within 5 miles of the site for 2 years or more during 1952-1984. Former plant workers were excluded, as were 21 FMMP participants who were under- going chemotherapy at the time of the first FMMP examination. The final study population consisted of 8,464 persons. Medical history (which included current and past medical problems, hospitalizations, surgical and medical procedures, and family history) was ascertained from records of the first FMMP medical examination. Medical conditions were coded according to ICD-9. Additional medical records were consulted to verify some health outcomes. A question- naire was distributed to gather information on lifestyle risk factors, location of residences and drinking water sources, and other measures of exposure. The questionnaire also included four questions on current medical conditions, such as heart problems, diabetes, cancer, chronic bronchitis, and emphysema. Residential history by location was used to develop a crude exposure metric, which divided participants according to whether they had lived within 2 miles of the facility, lived in the direction of groundwater runoff from the facility, or obtained their drinking water from a well or cistern. The authors conducted interviews with area medical practitioners, examining physicians of the FMMP, local and state health officials, and community residents to obtain information about perceived disease excess.
COHORT DESCRIPTIONS 157 Selected health outcomes of interest included goiter, other thyroid disease, chronic bronchitis, asthma, emphysema, nephritis, other renal disease, and dia- betes mellitus. Data from the US National Health Interview Study (NHIS) and the Third National Health and Nutrition Examination Survey (NHANES III) were used for comparison. Age- and sex-specific rates were calculated for ICD-9âcoded health outcomes by using the FMMP database. Those with more than one code for a specific category were counted once for that category. Separate rates were cal- culated by using data on the four-question set on the FMMP questionnaire. Both rates were compared with similar rates for white non-Hispanics in the NHANES III and NHIS samples. Sampling weights were applied to account for survey design and nonresponse. The authors found statistically significant excesses in the FMMP population in renal disease (standardized prevalence ratio [SPR], 215; 99% CI, 186-248), bladder disease (SPR, 132; 99% CI, 111-156), and thyroid diseases (SPR, 155; 99% CI, 133-179). Those outcomes included increases in a few subcategories, such as kidney stones (SPR, 398; 99% CI, 336-468) and chronic nephritis (SPR, 203; 99% CI, 76-435). Residents who obtained their drinking water from a well or cistern had higher urinary microalbumin, hematocrit, and red-cell counts. Resi- dents had significantly fewer cases of asthma (SPR, 85; 99% CI, 73-98), chronic bronchitis (SPR, 19; 99% CI, 14-24), and emphysema (SPR, 61; 99% CI, 41-68) compared with NHIS rates. Finnish Well-Water Studies Uranium occurs naturally throughout Earthâs crust. Highly concentrated ura- nium granitoids and granites can become soluble in soft, slightly alkaline bicar- bonate groundwater. That is the case in various parts of Finland, where uranium concentrations in drilled wells can reach 12,000 Î¼g/L. The proposed global limit for uranium in drinking water is 2 Î¼g/L. Several animal and occupational studies have documented uraniumâs toxic effect on the kidneys, specifically on tubular and glomerular function, but few studies have assessed renal toxicity due to ura- nium exposure through drinking water (Kurttio et al., 2002). Furthermore, animal models show that ingestion of uranium through drinking water increases urinary calcium and phosphate excretion and thus affects bone metabolism (Kurttio et al., 2005). In six successive studies, Kurttio and colleagues assessed the health effects of naturally occurring uranium in drinking water in 28 municipalities in southern Finland. The municipalities selected had the highest uranium concentrations. The study population was derived from a drinking-water database that contained radioactive measurements for over 5,000 drilled wells (Kurttio et al., 2002). Study subjects were selected through surveys sent to households throughout the area. The first questionnaire, mailed to 798 households, collected details on well-
158 updated literature review of depleted uranium water use, purification methods and equipment, and medical history. A second questionnaire, sent to 436 individuals, limited the cohort to people who resided in households that had no more than two persons and had used a well as the main source of drinking water for at least 1 year. The authors collected details on residential history, well-water consumption and use, education, disease his- tory, smoking history, occupation, and use of medication and herbal products. Of the 436 individuals, 78% consented to participation in the study, and urine and blood samples were collected from them. A third questionnaire was used to obtain additional medical details on bone-fracture history, estrogen use, and previously mentioned items. Uranium exposure was defined with four measures: uranium in drinking water, daily intake of uranium from drinking water, uranium in urine, and estimated cumulative intake from drinking water. Uranium exposure and outcome variables were modeled by using linear regression to establish associa- tions. A summary of the investigations follows. Kurttio et al., 2002 Researchers investigated possible renal effects by evaluating relevant bio- markers in connection with uranium exposure through drinking water in Finland. The study population (n = 325) consisted of people living in 28 municipalities where uranium concentrations were highest and measurement was frequent. As described above, on the basis of responses from the 798 households, 436 indi- viduals were selected; investigators restricted the population to people older than 15 years who lived in households of no more than two persons that had used a well as the main source of drinking water for at least 1 year. Of the 436 people, 340 (78%) agreed to participate in the study and returned a second questionnaire that detailed SES, smoking history, medical history, drug use, and heavy-metal exposure. The study was limited to 194 wells in 24 municipalities that served as the main source of drinking water for an average of 13 years (range, 1-34 years). The authors excluded 14 respondents who had diabetes mellitus or used methotrexate or sodium aurothiomalate chronically, pregnant women, and those in households that had efficient purification systems. The study population included 163 women (56% of the group), and the mean age of the group was 52 years (range, 15-82 years). Some 15% of the study population were current smokers; 56% had never smoked cigarettes, cigars, or pipes; and 29% were ex-smokers. Study participants provided overnight and spot urine and blood samples at least a week after consumption of drilled-well water and water samples for analysis. Investigators also measured blood pressure, height, and body weight. To ensure quality, blinded samples were submitted to different facilities for uranium- isotope measurement, and results were comparable. Four metrics were used in the measurement of uranium exposure: daily intake (of uranium from drinking water), uranium in urine, uranium in drinking water, and estimated cumulative
COHORT DESCRIPTIONS 159 uranium intake from drinking water. A number of metrics were chosen to assess renal function on the basis of previous findings on uranium toxicity. Proximal tubular markers included urinary and serum concentrations of calcium, phos- phate, glucose, and beta-2-microglobulin; glomerular function was measured by using creatinine and urinary albumin concentrations. Fractional excretion of calcium and phosphate was used as the primary outcome metric. The exposure measures were analyzed as both continuous and categorical variables adjusted for age, sex, and body-mass index (BMI) by using general linear-regression models. In determining uranium dose-response relationships, separate analyses were conducted for calcium and phosphate fractional excretion end points for sample points above and below the median. The authors used exist- ing or suggested standards for drinking water to establish cutpoints for uranium in drinking water and approximate values (quintiles) for urinary uranium and daily intake. Uranium concentration in water ranged from 0.001 to 1,920 Âµg/L. The median daily intake of uranium from water was 39 Âµg. Uranium exposure through drinking water was associated only with calcium excretion (pÂ = 0.03). Urinary uranium was associated with fractional excretion of calcium for all exposure metrics. The authors also observed an association between fractional phosphate and uranium concentration in urine (pÂ = 0.03). There was no association between uranium exposure and measures of glomerular function. Kurttio et al., 2005 In this study, researchers evaluated biochemical markers of bone turnover in 146 men and 142 women who obtained drinking water from drilled wells in high-uranium areas for an average of 13 years. The study population was a subset of the cohort described in Kurttio et al. (2002). Samples were collected and ana- lyzed as previously described. A third questionnaire was distributed to ascertain additional details on bone fracture, menopause, and physical activity. Details were also collected on estrogen use by 26 women who reported use of oral con- traceptives or hormone-replacement therapy. The authors excluded 43 subjects who were less then 25 years old; had diabetes mellitus; reported long-term use of glucocorticoids, thiazide diuretics, methotrexate, or sodium aurothiomalate; were currently pregnant; or had effective water purification equipment. The remaining group consisted of 288 people in 179 households. Uranium exposure metrics were similar to those used in the 2002 study (daily intake, uranium in urine, uranium in drinking water, and estimated cumulative intake from drinking water); uranium concentration in drinking water served as the primary measure. Indicators of bone formation included serum osteocalcin and amino-terminal propeptide of type 1 procollagen (P1NP) based on an imu- noradiometric assay. Bone resorption was assessed according to serum type 1 collagen carboxy-terminal telopeptide (CTx). Urinary calcium and phosphate
160 updated literature review of depleted uranium were also measured. Linear and weighted robust regression methods were used to account for highly influential observations. Separate analyses were conducted for men and women and accounted for age, smoking status, and estrogen use. Through robust and linear-regression analyses, uranium exposure was shown to be associated with increased CTx in men (uranium in water, pÂ = 0.05 and 0.01; daily intake, pÂ = 0.16 and 0.02; and cumulative intake, pÂ = 0.16 and 0.03, respec- tively). Uranium concentrations in drinking water appeared to be associated with increased osteocalcin (pÂ = 0.19; pÂ = 0.04 in linear-regression analysis). Uranium exposure was not related to any biomarkers of bone metabolism in women. P1NP was not associated with uranium exposure. Kurttio et al., 2006a In 2003, Kurttio and colleagues continued their assessment of nephrotoxic effects of naturally occurring uranium in well water, focusing on measures associ- ated with renal-cell toxicity and renal tissue damage. The study population was based on a previous study group in which 325 participants were selected in the evaluation of basic renal function (Kurttio et al., 2002). A third questionnaire was sent to the previous study participants; 222 responded, of whom 202 provided samples for a number of tubular and glomerular markers. Details on residential history, daily well-water consumption and use, medical history, use of medica- tion, and smoking history were collected, and only current well-water users were selected. The authors excluded one person who had recently taken methotrexate. The final study population consisted of 95 men and 98 women in 124 households in which drilled wells were the primary source of drinking water for an average of 16 years (range, 5-40 years). The mean age was 56 Â± SD 12 years. Some 6% were smokers, and 56% reported never having smoked. The average BMI was 26 Â± 4 kg/m2. Thirty-nine study participants (20%) reported regular use of analgesics 1 year before study enrollment. Urine, blood, nail, hair, and household water (kitchen-tap) samples were collected from study participants in households that had consumed well water for at least 1 week continuously. Blood pressure, body weight, and height were also measured. As in the previous study, urinary uranium was measured, as were daily intake, daily intake per unit body weight, intake from drinking water, and cumulative intake. A number of biomarkers were measured in urine and serum samples selected as indicators of cell toxicity and renal dysfunction. Cytotoxic measures included N-acetyl-Î³-D-glucosaminidase, lactate dehydrogenase, alka- line phosphatase, and Î³-glutamyltransferase. Concentrations of Î±-glutathione- S-Âtransferase, calcium, phosphate, and glucose served as indicators of proximal tubular and glomerular dysfunction. As in the previous study, the authors calcu- lated values for calcium and phosphate fractional excretion, in addition to glucose excretion and creatinine clearance. Linear regression was used to model outcome
COHORT DESCRIPTIONS 161 variables and uranium exposure with adjustment for sex, age (specified as a linear and/or quadratic effect), BMI, smoking, and use of analgesics. Urinary uranium concentrations were an average of 44% greater than in prior sampling. In general, markers of renal function were within clinical limits. Bio- markers of cytotoxicity, renal proximal tubular function, glomerular function, and other exposure indicators were not statistically significantly associated with urinary uranium concentrations. There were statistically significant associations between cumulative uranium intake and glucose excretion (pÂ = 0.02) and between uranium exposure and increased blood pressure (diastolic, pÂ = 0.01; systolic, pÂ = 0.07). The studies described above provide useful information on the health effects of uranium exposure from drinking water. The well-water studies have a number of strengths, including the relatively large population, the specificity of the mul- tiple exposure metrics, and the use of individualized clinical evaluation. In addi- tion, the long-term, internal nature of the exposure is analogous to that of veterans exposed to depleted uranium. Despite those strengths, the cross-sectional nature of the exposure assessment limits the ability to ascertain details of chronic health effects. Furthermore, there were large variations in the duration of exposure, and dietary intake was not taken into account in the data-collection process. Because of some characteristics of granite bedrock, groundwater in Finland has higher amounts of naturally occurring radionuclides (up to 100 or even 1,000 times as high as in other populations). Researchers identified small municipal units in which 90% of residents obtained their water from wells outside the municipal supply. The cohort consisted of those residents of the units who had maintained residence from January 1967 to December 1980 and were born in 1900-1930. The researchers then conducted three nested case-control studies to analyze cancer incidence in correlation with this well-water use. Auvinen et al., 2002 To assess the effect of natural uranium and other radionuclides in drinking water (internal exposure) on leukemia, researchers identified all leukemia cases in the cohort (well-water drinkers in Finland) who were diagnosed in 1981-1995 (Auvinen et al., 2002). A portion of the noncases in the cohort were chosen for comparison and matched by age and sex to the cases. Well-water use was deter- mined by surveying the subjects and from information from local health officials. Totals of 371 controls and 41 cases were known to have relied extensively on well water before 1981. Samples of drinking water from wells drilled in July- November 1996 were obtained for 274 controls and 35 cases. Using a modified proportional-hazard model, the authors calculated hazard ratios (HRs) by matching cases at date of diagnosis with controls at risk and assigning controls a weight to represent the entire cohort. The median uranium concentrations in drinking water were comparable in cases (0.08 Bq/L) and controls (0.06 Bq/L). The median radon concentrations
162 updated literature review of depleted uranium were 80 and 130 Bq/L, respectively. The corresponding median concentrations of radium were 0.01 and 0.03 Bq/L. The HRs were not statistically significant for water uranium concentration (HR, 0.91; 95% CI, 0.73-1.13 per Bq/L), radon (HR, 0.79; 95% CI; 0.27-2.29 per Bq/L), or radium (HR, 0.80; 95% CI, 0.46-1.39 per Bq/L). Calculations for the groups with the highest exposures also did not show an increased risk of leukemia. Auvinen et al., 2005 All stomach-cancers in the cohort diagnosed in 1981-1995 were identified by the researchers (Auvinen et al., 2005). Controls were randomly selected from the cohort and matched by age and sex. Some 371 (8.1%) cohort members and 107 (7.2%) stomach cancer cases had used drinking water from drilled wells before 1981. Well-water samples for 274 controls and 88 cases were obtained in July-November 1996. HRs were calculated by matching cases at date of diagnosis with controls at risk and assigning controls different weights (to reflect more accurately the composition of the entire cohort). Median total alpha activity in the drinking water was comparable in cases (0.08 Bq/L) and controls (0.05 Bq/L), and uranium concentrations were also comparable (0.07 Bq/L in cases and controls). The HR was not statistically sig- nificant for uranium (HR, 0.76; 95% CI, 0.48-1.21 per Bq/L), radium (HR, 0.69; 95% CI, 0.33-1.47 per Bq/L), or radon (HR, 0.68; 95% CI, 0.29-1.59 per 100 Bq/L). The groups with the highest concentrations of radionuclides did not show a statistically significant correlation with stomach cancer. Kurttio et al., 2006b From 144,627 persons born between 1900 and 1930 who had lived out- side the municipal water supply from 1967 to 1980, researchers identified 884 b Â ladder-cancer and 644 renal-cancer cases, and they selected 4,590 controls ran- domly from the remaining cohort. Through surveys and local health authorities, they identified 371 (8%) controls, 79 bladder-cancer cases, and 65 renal-cancer cases as having used drilled well water during the specified period. Well-water samples were obtained for 274 (74%) of the controls, 61 (77%) of the bladder- cancer cases, and 51 (78%) of the renal-cancer cases, who made up the study population of this nested case-control investigation. Mean followup time was 19 years. The effective dose to the kidneys and bladder were calculated on the basis of a consumption rate of 2 L/day; ingestion dose and effective dose were derived from other sources, such as the National Research Council, International Com- mission on Radiological Protection, and other previously published studies.
COHORT DESCRIPTIONS 163 Smoking and BMI are known risk factors for renal and bladder cancers, so both were adjusted for in subanalyses (that is, for subjects on whom these data were available). HRs were calculated by matching cases at date of diagnosis with controls at risk and assigning the controls weights to reflect the composition of the entire cohort. The median effective radiation doses, close to 100 ÂµSv, of all three radio- nuclides (uranium, radon, and radium), were comparable among bladder-cancer cases, renal-cancer cases, and controls. No statistically significant associations were observed between any of the radionuclides and renal or bladder cancer. When subjects without information on smoking or BMI were excluded from the analysis, the results did not change significantly. The strengths of this series of nested case-control studies include long-term internal exposure, blinded analysis of the water samples, applicability to the entire cohort (because of weighting of the controls), individual measure of exposure, and data on confounders (for example, smoking history and BMI of some study subjects), allowed assessment of potential bias. Limitations of the study include the lack of data on the amount of water consumed individually, of information on other sources of drinking water (for example, in the workplace), of data on other risk factors (for example, radiation exposures from medical treatment and other sources), and of sufficient statistical power to detect a small risk. Summary In assessing the potential health effects of uranium and depleted uranium in humans, the committee examined a number of studies that focused on people who were occupationally exposed to uranium through processing activities, on depleted-uranium exposure in deployed populations, and on uranium exposure through residence and drinking-water use. Each study was carefully reviewed, and the evidence presented was used to elaborate on the key findings and draw conclusions about the relevant malignant and nonmalignant health outcomes in the next chapter.
164 updated literature review of depleted uranium TABLE 7-1â Studies of Uranium Processors Person-Years Study Design Population Observed Cohorts Evaluated in Gulf War and Health, Volume 1 Colorado Plateau (mill workers) Wagoner et al., 1964 Cohort 5,370 uranium miners and millers; 6,390 followup 1950-1962 Archer et al., 1973 Cohort 662 male uranium-mill workers; followup 1950-1967 Waxweiler et al., 1983 Cohort 2,002 male uranium millers 43,252 employed at least 1 day after January 1, 1940, worked at least 1 year in uranium mills, never worked in uranium mining; followup through 1977 Pinkerton et al., 2004 Cohort 1,484 men who worked in uranium mill at least 1 day after January 1, 1940, worked for at least 1 year in uranium mills, had never worked in a uranium mine; followup through 1998 Cincinnati, OH (Fernald Feed) Boiano et al., 1989 Cross- 146 (70%) of 208 eligible long- sectional term employees at Feed Materials Production Center after releases of uranium oxide from dust collectors in November-December 1984
COHORT DESCRIPTIONS 165 Exposure Outcomes Adjustments Comments Exposure determined Mortality Age, race by duration and type of employment (miner or miller); for subset of miners, radiation exposure calculated from months of underground experience, estimated dose rate, cumulative dose Occupational exposure Mortality Age, race, calendar to vanadium and long- years lived members of uranium radioactive decay series: uranium- 238, uranium-234, thorium-230, radium- 228, lead-210 Exposure defined by Mortality Age, race, sex, work history, duration calendar period of employment in mills Exposure defined by Mortality from NIOSH Stratified analyses Limitations include work history, duration modified life-table by duration of lack of smoking of employment analysis system employment, time data, small cohort, through 1998 since first employment limited power to (latency), year of first detect moderately employment increased risk of some outcomes of interest, inability to estimate individual exposures to uranium, silica, vanadium Self-reported exposure Lung, renal disease Smoking Limitations in incidents; job history; exposure based partly assessed urinary- on recall; crude and uranium data imprecise exposure categories (low, medium, high) Continued
166 updated literature review of depleted uranium TABLE 7-1â Continued Person-Years Study Design Population Observed Ritz, 1999 Cohort 4,014 white male uranium-processing 124,177 workers employed in 1951-1989; followup through January 1, 1990 Oak Ridge, TN Polednak and Frome, Cohort 18,869 white male workers at 1981 uranium conversion and enrichment plant, employed in 1943-1947; followup until 1977 Checkoway et al., Cohort 6,781 white male employees at 133,535 1988 nuclear-weapons materials fabrication plant in May 4, 1947-December 31, 1974; followup through 1979
COHORT DESCRIPTIONS 167 Exposure Outcomes Adjustments Comments External radiation Cancer mortality Controlled for Also ran further exposure derived from Social Security exposure to analyses by radiation from film-badge Administration, trichloroethylene, dose measurements; National Death Index cutting fluids internal radiation exposure based on combination of individual urine bioassays, environmental area sampling; assessed exposure to uranium, thorium, radium compounds Uranium-dust Mortality Age, calendar year exposure defined by departments and averages where employee worked Range of average concentrations of uranium in air: 25-300 Âµg/m3 Radiation exposure Mortality Age, calendar year assessed with film badges, estimates of dose equivalents delivered to lungs obtained from urinalysis measurements, in vivo counting of internally deposited uranium Mean cumulative alpha-radiation dose to lung 8.21 rem, mean cumulative external whole-body penetrating dose from gamma radiation 0.96 rem Continued
168 updated literature review of depleted uranium TABLE 7-1â Continued Person-Years Study Design Population Observed Frome et al., 1990 Cohort 28,008 World War II nuclear-plant workers, white males employed at least 30 days in 1943-December 31, 1947; followup January 1, 1950- January 1, 1979 Loomis and Wolf, Cohort 6,591 white men, 922 white women, 1996 (continuation 449 black men, 149 black women, of Checkoway 5 men and women of other racial et al., 1988) groups who had worked at plant after 1947; 1,764 white men, 562 white women, 85 black men, 69 black women, 1 man of other racial group who had worked at plant before 1947 Richardson and Wing, Cohort; nested 3,864 nuclear-materials fabrication- 2006 case-control plant workers employed at least 30 days in May 4, 1947-December 31, 1974; followup through 1990 Frome et al., 1997 Cohort 106,020 workers at four plants in Oak Ridge, TN, employed at least 30 days in 1943-1985; followup through 1984
COHORT DESCRIPTIONS 169 Exposure Outcomes Adjustments Comments Indexes for radiation Mortality SES, duration of by specific job employment, facility, title-department birth year, period of combination followup Radiation exposure Mortality Age, calendar year assessed with film badges, estimates of dose equivalents delivered to lungs obtained from urinalysis measurements, in vivo counting of internally deposited uranium Alpha radiation based Lung-cancer mortality Matched for age, on in vivo monitoring, from National Death year of birth, sex, urinalysis results, Index, Social Security race, SES, length estimates based on Administration, of employment, exposure potential, Health Care Financing employment status given department in Agency, Tennessee which employed Department of Motor Vehicles Internal radiation dose: 10-100+ mSv (exposed), <10 (unexposed) External radiation Mortality Stratified by length exposure based on of employment, SES, limited monitoring; birth year, age internal radiation exposure categorized into three levels: eligible for monitoring but not monitored, eligible for monitoring and monitored, not eligible for monitoring Continued
170 updated literature review of depleted uranium TABLE 7-1â Continued Person-Years Study Design Population Observed Four Uranium-Processing Operations Dupree et al., 1995 Case-control 787 lung-cancer cases employed at least 183 days at one of four uranium-processing facilities with one control match; followup through 1982 St. Louis, MO (Mallinckrodt) Dupree-Ellis et al., Cohort 2,514 white male uranium-processing 87,757 2000 plant workers employed in 1942- 1966; followup through 1993 Portsmouth Uranium Enrichment Facility, Pike County, OH Brown and Bloom, Retrospective 5,773 white male employees who 1987 cohort had worked for at least 1 week in September 1954-February 1982
COHORT DESCRIPTIONS 171 Exposure Outcomes Adjustments Comments Alpha radiation Lung-cancer mortality Matched for race, sex, Two operations from airborne from death certificates birth, hire date within at Y-12 facility in dust containing (ICD8A, codes 3 years, smoking Oak Ridge, TN: uranium compounds 162.0-163.9) status, SES (pay code) one Mallinckrodt from employment Chemical Works and occupational Uranium Division, monitoring records; one Feed Materials gamma radiation from Production Center in available personal Fernald, OH monitoring data Film badges to Mortality from Age, calendar period Cohort overlaps with monitor beta, gamma National Death Index, Dupree et al., 1995 radiation exposure; Social Security 20.8% of workyears Administration, had no monitoring Pension Benefit results, and exposure Information through was estimated with 1993 algorithm Median cumulative whole-body exposure 15.3 mSv Urinary-uranium Mortality Age, calendar time Short followup data (collected since (maximum 17 years) 1954) used to identify limited ability departments with to assess cancer potential exposure; outcomes departments ranked by relative degree of Study included potential exposure in Gulf War and Health, Volume 1, Other data included but additional cancer continuous air outcomes included in sampling, personal current report sampling of external radiation with film badges, dosimeters, in vivo counting Continued
172 updated literature review of depleted uranium TABLE 7-1â Continued Person-Years Study Design Population Observed Polk County, FL (phosphate-fertilizer production workers) Stayner et al., 1985 Retrospective 3,199 ever employed at phosphate- cohort fertilizer production facility in Polk County, FL; subjects followed from first date of hire to December 31, 1977, or date of death Connecticut (nuclear-fuels fabrication plant) Hadjimichael et al., Retrospective 4,106 employees of plant who had 1983 cohort ever worked for more than 6 months in 1956-1978 New Cohorts (Not Evaluated in Gulf War and Health, Volume 1) UK (Springfields) McGeoghegan and Cohort 19,454 current and former British 479,146 Binks, 2000b Nuclear Fuels workers ever employed before 1996; followup 1946-December 31, 1995
COHORT DESCRIPTIONS 173 Exposure Outcomes Adjustments Comments Mortality Age, sex, race, Lack of exposure calendar year information Study included in Gulf War and Health, Volume 1, but additional cancer outcomes included in current report Exposure groups Mortality from Age, sex, calendar Potential exposure based on film badges Social Security year, cause of death or overlap between worn by employees, Administration cancer site groups due to job classification, records, Connecticut multiple exposures; consultation with mortality records; did not quantify industrial hygiene, cancer incidence from degree of exposure safety personnel, Connecticut Tumor for individuals or supervisors, employees Registry groups who had been at plant since operation began Study included in Gulf War and Job experience Health, Volume 1, obtained from but additional cancer personnel records outcomes included in current report Radiation dosimetry Mortality, cancer Age, sex, calendar Additional analyses collected for morbidity from 1971- year, industrial by dose, lag time compliance with December 31, 1995 status (industrial statutory radiation- or nonindustrial), protection guidelines; worker status, year of used film badges joining, time from first and recorded mSv; exposure, length of included doses service acquired from other sites for workers transferred into plant Mean individual cumulative external whole-body dose: 22.8 mSv (analysis A), 20.5 mSv (analysis B) Continued
174 updated literature review of depleted uranium TABLE 7-1â Continued Person-Years Study Design Population Observed UK (Chapelcross) McGeoghegan and Cohort 2,628 nuclear-fuels workers ever Binks, 2001 employed in 1955-1995; followup 1955-December 31, 1995 UK (Capenhurst) McGeoghegan and Cohort 12,543 workers at British Nuclear 334,473 Binks, 2000a Fuels plant ever employed before 1996; followup 1946-December 31, 1995 Rocketdyne (Atomics International) Ritz et al., 2000 Retrospective 2,297 male (2,218) and female (79) 58,837 cohort employees enrolled in companyâs health-physics radiation-monitoring program in January 1, 1950- December 31, 1993; followup through 1994
COHORT DESCRIPTIONS 175 Exposure Outcomes Adjustments Comments Radiation dosimetry Mortality, cancer Age, sex, calendar Additional analyses collected for morbidity through year, industrial by dose, lag time compliance with December 31, 1995 status (industrial statutory radiation- or nonindustrial), protection guidelines; workers status, year used film badges of joining, time from and recorded mSv; first exposure, length included doses of service acquired from other sites for workers transferred into plant Mean cumulative external whole-body dose: 83.6 mSv Radiation dosimetry Mortality, cancer Age, sex, calendar Additional analyses collected for morbidity 1971- year, industrial by dose, lag time compliance with December 31, 1995 status (industrial statutory radiation- or nonindustrial), protection guidelines; workers status, year used film badges of joining, time from and recorded mSv; first exposure, length included doses of service acquired from other sites for workers transferred into plant Mean cumulative external whole-body dose: 9.85 mSv Doses estimated Mortality Pay categories used by urine or feces as proxy for SES, bioassay, in vivo smoking status whole-body counts, lung counts Company did not collect information Estimated internal on race before 1972, cumulative dose to although 96% of lung of each employee deceased workers were was calculated white Mean lung dose: 2.1 mSv Continued
176 updated literature review of depleted uranium TABLE 7-1â Continued Person-Years Study Design Population Observed Boice et al., 2006a Retrospective 5,801 radiation workers monitored 161,605 cohort for radiation and employed on or after January 1, 1948, for at least 6 months; person-years of followup began 6 months after date of first radiation monitoring or July 1, 1948, and stopped at date of death, December 31, 1999, age of 95 years, or date lost to followup Savannah River Plant Cragle et al., 1988 Retrospective 9,860 white male employees who 232,061 cohort worked at plant in 1952-1981 and were hired before 1975; followup through 1980 Atomic Weapons Establishment, UK Beral et al., 1988 Retrospective 22,552 employees who worked cohort at establishment in January 1, 1951-December 31, 1982; followup through 1982 Egyptian Processors Shawky et al., 2002 Cross-sectional 86 processors at two sites in Egypt working in three locations, 13 of whom also participated in urinary- uranium analysis
COHORT DESCRIPTIONS 177 Exposure Outcomes Adjustments Comments Annual doses of Mortality Race, age, calendar Low doses, small internally deposited year, sex study, incomplete radionuclides smoking histories, estimated by positive imperfect bioassay data, in vivo categorization of lung counts, incident pay type, dosimetry reports sources not evaluated Internal monitoring of 2,232 of 5,801 workers; measurements included radionuclides in urine, supplemented with fecal measurements, lung counts Exposure defined by Mortality Stratified by duration No exposure work history, duration of employment, date assessment, lack of of employment of first employment, generalizability (average exposure, hourly vs employment 13 years) type (hourly vs salaried) Exposure measured Mortality Age, sex, calendar with dosimeters worn period, social class externally Average exposures per radiation worker, 7.8 mSv (whole-body exposure), 14.4 mSv (surface exposure) Concentration of Clinical measurements Tested air uranium uranium in air: 22.6 of hematologic, renal concentration Ã 10â7 â 11.1 Ã 10â5 function Bq/cm3 Exposure: 1-80 Î¼Sv/h
178 updated literature review of depleted uranium TABLE 7-2 Studies of Depleted-UraniumâExposed Persons Study Design Population Exposure Baltimore VA Medical Center Study McDiarmid Case series 29 exposed GW veterans Exposure to DU by means of et al., 2000 exposed to DU during friendly fire; may have inhaled friendly-fire incidents in or ingested airborne DU February 1991, particles, and/or experienced 38 DU-nonexposed wound contamination by DU; veterans; examined in assessed urine and semen March-June 1997, 7 years uranium concentration after first exposure Veterans with DU fragments: 0.01-30.7 Âµg/g creatinine vs nonexposed: 0.01-0.05 Âµg/g creatinine McDiarmid Case series 50 exposed GW veterans Exposure to DU by means of et al., 2001 divided into low-uranium, friendly fire; may have inhaled high-uranium groups; or ingested airborne DU examined in March-July particles, and/or experienced 1999, 8 years after first wound contamination by exposure DU; assessed urine uranium concentration Veterans with DU fragments: 0.018-39.1 Âµg/g creatinine vs DU-exposed veterans without fragments: 0.002-0.231 Âµg/g creatinine McDiarmid Case series 29 exposed GW veterans, 38 Exposure to DU by friendly et al., 2002 nonexposed GW veterans, fire; may have inhaled 30 newly identified exposed; or ingested airborne DU examined in spring 1999 particles, experienced wound contamination by DU; assessed urinary uranium McDiarmid Case series 39 GW veterans exposed Exposure to DU by friendly et al., 2004 to DU during friendly-fire fire; may have inhaled, incidents in February 1991; ingested airborne DU examined in April-July; particles, experienced wound followup in 1994-2001 contamination by DU; assessed urinary uranium Low-uranium group, <0.1 Âµg/g of creatinine; high- uranium group, â¥0.1 Âµg/g of creatinine
COHORT DESCRIPTIONS 179 Outcomes Adjustments Comments Hematologic, renal function; Stratification at median into Small sample neurocognitive, psychiatric low-, high-result groups measures; genotoxicity measures; concentrations of follicle stimulating hormone, prolactin, testosterone; semen characteristics (volume, concentration, morphology, motility) Hematologic, renal, Race, education, age, Small sample, no true neurocognitive, genotoxic marital status, military rank, comparison group outcome measures; injuries; intelligence (WRAT-3), concentrations of thyroid- depression (BDI), smoking stimulating hormone, free status, use of prescription thyroxine; reproductive psychotropic, antidepressant neuroendocrine indicators, drugs, recent X-ray exposure semen characteristics reported in McDiarmid et al., 2000 Urinary uranium determinations; Small sample clinical laboratory values; psychiatric, neurocognitive assessment Hematologic, renal function; Age, smoking, exposure to No comparison group, small immunologic measures; genetic toxicants, cloning sample genetoxicity; neurocognitive, efficiency psychiatric assessment; reproductive characteristics reported in McDiarmid et al., 2001 Continued
180 updated literature review of depleted uranium TABLE 7-2 Continued Study Design Population Exposure McDiarmid Case series 32 GW veterans exposed Exposure to DU by friendly et al., 2006 to DU during friendly fire; fire; may have inhaled, examined in April-July 2003 ingested airborne DU particles, experienced wound contamination by DU; assessed urinary uranium Low-uranium group, <0.1 Âµg/g of creatinine; high-uranium group, â¥0.1 Âµg/g of creatinine McDiarmid Case series 34 GW veterans exposed Exposure to DU by friendly et al., 2007 to DU during friendly-fire fire; may have inhaled, incidents in 1991; examined ingested airborne DU in April-June 2005 particles, experienced wound contamination by DU; assessed urinary uranium; both current and cumulative exposure measures reported Low-uranium group, <0.1 Âµg/g of creatinine; high-uranium group, â¥0.1 Âµg/g of creatinine UK Gulf War Veterans Macfarlane Cohort 51,721 UK GW veterans, Deployment to GW; self- et al., 2003 50,755 nondeployed UK reported exposure to DU service personnel; followup April 1, 1991-July 31, 2002 Balkans Cohorts Gustavsson Cohort 9,188 Swedish military Deployment to UN missions in et al., 2004 personnel deployed to UN Balkans missions in Balkans in 1989-1999; followed up through December 31, 1999; 39,816 person-years Nuccetelli Summary of 40,000 Italian soldiers Deployment to Balkans et al., 2005 data presented deployed to Balkans at in Italian least once in 1995-2001 Defence (followup time not reported) Ministry study published in Italian in 2002
COHORT DESCRIPTIONS 181 Outcomes Adjustments Comments Hematologic characteristics; Age, intelligence, emotional Small sample, no true renal function; neurocognitive status, smoking, exposure comparison group measures; genotoxicity; to genetic toxicants, cloning reproductive characteristics efficiency reported in McDiarmid et al., 2001 Hematologic, renal function; Age, IQ, depression, cloning Small sample, no true neurocognitive, psychiatric efficiency comparison group measures; genotoxicity measures; reproductive characteristics reported in McDiarmid et al., 2001 Cancer DU analysis adjusted Latency of many cancers is for smoking, alcohol beyond time of study consumption; matched for age, sex, rank, service, level of fitness Cancer Sex, age, period Short followup period for cancer Cancer Continued
182 updated literature review of depleted uranium TABLE 7-2 Continued Study Design Population Exposure Storm et al., Population- Danish military deployed to Deployment to Balkans 2006 based Balkans (13,552 men, 460 retrospective women); followup from cohort January 2002 to December 2002 Sumanovic- Pre-post All liveborn and stillborn Living in western Herzegovina Glamuzina comparison neonates in Maternity after military activities et al., 2003 Ward of Mostar University Hospital in western Herzegovina, part of Bosnia and Herzegovina, immediately (1995) and 5 years after (2000) 1991- 1995 military activities
COHORT DESCRIPTIONS 183 Outcomes Adjustments Comments Cancer Age-, sex-, period-specific Few cases, wide CIs, young SIRs cohort Major congenital malformations Not known whether DU was used in region
184 updated literature review of depleted uranium TABLE 7-3â Studies of Environmental Exposure to Uranium Reference Design Population Exposure Residential Studies Bithell and Modeling Those living within 6 km of Environmental uranium (for Draper, 1999 Air Force base example, in leaves) Boice et al., Cross-sectional 16,722 people living in Residential proximity to 2003a 8 municipalities in PA; Apollo, Parks nuclear cancer-mortality records facilities in PA from Pennsylvania Cancer Registry for 1993-1997; reference, PA or national SEER registry Boice et al., Ecologic Mortality data from NCI, Residential proximity to 2003c mortality Texas Department of Health uranium-processing site survey for 1950-2001 Cases, Karnes County (site of uranium mining); controls: 4 counties matched on various characteristics; reference, US general population Boice et al., Mortality Cases: 3 comparison Residential proximity to 2003b survey counties matched on Apollo, Parks nuclear demographics; facilities controls, 2 counties in PA from NCHS for 1950-1995; reference, US general population Pinney et al., Cohort 8,464 people from FMMP; Residential proximity (less 2003 comparison rates, NHIS, than 2 miles) to Fernald NHANES uranium-processing plant in direction of groundwater runoff or possible well or cistern contamination in January 1952- December 1984
COHORT DESCRIPTIONS 185 Outcomes Adjustments Comments Childhood leukemia Authors modeled uranium exposure around US Air Force base and compared it with distribution of childhood leukemia in same area; no correlation found between two plots Cancer incidence Age, sex, calendar year No adjustment for diet, smoking, other cancer risk factors; no determination of length of residence (and hence exposure) Cancer mortality Control counties chosen No adjustment for diet, on basis of matched smoking, other cancer risk demographics factors; no determination of length of residence (and hence exposure) Cancer mortality Matched on demographics No adjustment for diet, smoking, other cancer risk factors; no determination of length of residence (and hence exposure) Goiter, other thyroid disease, Age, sex Study questionnaires not chronic bronchitis, asthma, directly comparable; FMMP emphysema, nephritis, other is self-selected volunteer renal disease, diabetes mellitus group Continued
186 updated literature review of depleted uranium TABLE 7-3â Continued Reference Design Population Exposure Finland Well-Water Studies Kurttio et al., Cross-sectional 325 people in Finland Median drinking-water 2002 who obtain drinking water uranium concentration, 28 from drilled wells used an Î¼g/L (interquartile average of 13 years range, 6-135 Î¼g/L; maximum, 1,920 Î¼g/L) Median urinary uranium concentration, 13 ng/mmol of creatinine (range, 2-75 ng/mmol) Median daily uranium intake, 39 Î¼g (range, 7-224 Î¼g) Kurttio et al., Cross-sectional 146 men, 142 women in Median drinking-water 2005 southern Finland who obtain uranium concentration, 27 (same drinking water from drilled Î¼g/L (interquartile range, population as wells used an average of 6-116 Î¼g/L) Kurttio et al., 13 years 2002) Median daily uranium intake, 36 Î¼g (range, 7-207 Î¼g) Median cumulative intake, 0.12 g (range, 0.02-0.66 g) Kurttio et al., Cross-sectional 95 men, 98 women in Median drinking-water 2006a Finland who obtain drinking uranium concentration, 25 (same water from drilled wells Î¼g/L (interquartile population as used an average of 16 years range, 5-148 Î¼g/L; Kurttio et al., maximum, 1,500 Î¼g/L) 2002) Auvinen et al., Nested 35 leukemia cases, 274 Well-water samples 2002 case-control stratified randomly sampled collected blind people from subcohort who in July-November 1996 obtained well water before 1981 Median activity uranium concentration for leukemia cases, 0.08 Bq/L; for reference group, 0.06 Bq/L
COHORT DESCRIPTIONS 187 Outcomes Adjustments Comments Renal function (based on urinary, Uranium exposure adjusted Uranium exposure measured serum concentrations of calcium, for age, sex, BMI by daily intake, uranium in phosphate, glucose, albumin, urine, uranium in drinking creatinine, beta-2-microglobulin water, cumulative intake as biomarkers) from drinking water Indicators of bone formation: Age, smoking, estrogen use Used two types of regression serum osteocalcin, amino- (women) analysis (linear regression, terminal of type 1 procollagen weighted robust regression) (P1NP); indicators of bone to account for highly resorption: serum type 1 collagen influential observations carboxy-terminal telopeptide (CTx), urinary calcium, urinary phosphate Renal-cell toxicity, renal Sex, age (linear-quadratic), Reference group is general dysfunction (based on BMI, smoking, use of population concentrations of various analgesics enzymes, creatinine, calcium, phosphate, glucose as indicators) Exposure assessment: uranium concentration in drinking water, hair, nails, urine Association with uranium, radon, Age, sex No dose-response radium assessment; no adjustments for other risk factors Continued
188 updated literature review of depleted uranium TABLE 7-3â Continued Reference Design Population Exposure Auvinen et al., Nested 88 stomach-cancer cases, Well-water samples 2005 case-control 274 stratified randomly collected blind sampled people from in July-November 1996 subcohort who obtained water from drilled wells Median activity before 1981 concentration for both cases and reference group, 130 Bq/L Kurttio et al., Nested Cases, 61 bladder-cancer Drilled well water outside 2006b case-control cases, 51 renal-cancer cases municipal water supply diagnosed in 1981-1995; obtained in 1967-1980 controls, 274 randomly sampled people stratified by Uranium concentrations: sex, age bladder cancer, 0.08 Bq/L; renal cancer,0.07 Bq/L; reference, 0.06 Bq/L
COHORT DESCRIPTIONS 189 Outcomes Adjustments Comments Association with uranium, radon, Age, sex No dose-response radium assessment; no adjustments for other risk factors Association with radon, radium, Bladder cancer: age at Exposures measured only up uranium exposure followup, sex, smoking status to 10 years before diagnosis to account for cancer latency Renal cancer: age, sex, smoking, BMI
190 updated literature review of depleted uranium References Archer, V. E., J. K. Wagoner, and F. E. Lundin, Jr. 1973. Cancer mortality among uranium mill work- ers. Journal of Occupational Medicine 15(1):11-14. Auvinen, A., P. Kurttio, J. Pekkanen, E. Pukkala, T. Ilus, and L. Salonen. 2002. Uranium and other natural radionuclides in drinking water and risk of leukemia: A case-cohort study in Finland. Cancer Causes and Control 13(9):825-829. Auvinen, A., L. Salonen, J. Pekkanen, E. Pukkala, T. Ilus, and P. Kurttio. 2005. Radon and other natu- ral radionuclides in drinking water and risk of stomach cancer: A case-cohort study in Finland. International Journal of Cancer 114(1):109-113. Beral, V., P. Fraser, L. Carpenter, M. Booth, A. Brown, and G. Rose. 1988. Mortality of employees of the Atomic Weapons Establishment, 1951-82. British Medical Journal 297(6651):757-770. Bithell, J. F., and G. J. Draper. 1999. Uranium-235 and childhood leukaemia around Greenham Com- mon airfield. Journal of Radiological Protection 19(3):253-259. Boiano, J. M., C. E. Moss, and G. A. Burr. 1989. Health hazard evaluation report HETA 83-144-2001, feed materials production center. Fernald, OH: Westinghouse Materials Company of Ohio. Boice, J. D., Jr., W. L. Bigbee, M. T. Mumma, and W. J. Blot. 2003a. Cancer incidence in munici- palities near two former nuclear materials processing facilities in Pennsylvania. Health Physics 85(6):678-690. âââ. 2003b. Cancer mortality in counties near two former nuclear materials processing facilities in Pennsylvania, 1950-1995. Health Physics 85(6):691-700. Boice, J. D., Jr., M. Mumma, S. Schweitzer, and W. J. Blot. 2003c. Cancer mortality in a Texas county with prior uranium mining and milling activities, 1950-2001. Journal of Radiological Protection 23(3):247-262. Boice, J. D., S. S. Cohen, M. T. Mumma, E. Dupree Ellis, K. F. Eckerman, R. W. Leggett, B. B. Boecker, A. B. Brill, and B. E. Henderson. 2006a. Mortality among radiation workers at RocketÂ dyne (Atomics International), 1948-1999. Radiation Research 166(1 Pt 1):98-115. Boice, J. D., Jr., R. W. Leggett, E. D. Ellis, P. W. Wallace, M. Mumma, S. S. Cohen, A. B. Brill, B. Chadda, B. B. Boecker, R. C. Yoder, and K. F. Eckerman. 2006b. A comprehensive dose reconstruction methodology for former Rocketdyne/Atomics International radiation workers. Health Physics 90(5):409-430. Brown, D. P., and T. Bloom. 1987. Mortality among uranium enrichment workers. Cincinnati, OH: National Institute of Occupational Safety and Health. Checkoway, H., N. Pearce, D. J. Crawford-Brown, and D. L. Cragle. 1988. Radiation doses and cause- specific mortality among workers at a nuclear materials fabrication plant. American Journal of Epidemiology 127(2):255-266. Cookfair, D. L., W. L. Beck, C. Shy, C. C. Lushbough, and C. L. Sowder. 1983. Lung cancer among workers at a uranium processing plant. Presented at Epidemiology Applied to Health Physics. In Proceedings of the Health Physics Society. Pp. 398-406. Cragle, D. L., R. W. McLain, J. R. Qualters, J. L. S. Hickey, G. S. Wilinson, W. G. Tankersley, and C. C. Lushbaugh. 1988. Mortality among workers at a nuclear fuels production facility. Ameri- can Journal of Industrial Medicine 14:379-401. Dupree, E. A., D. L. Cragle, R. W. McLain, D. J. Crawford-Brown, and M. J. Teta. 1987. Mortality among workers at a uranium processing facility, the Linde Air Products company ceramics plant, 1943-1949. Scandinavian Journal of Work, Environment and Health 13(2):100-107. Dupree, E. A., J. P. Watkins, J. N. Ingle, P. W. Wallace, C. M. West, and W. G. Tankersley. 1995. Uranium dust exposure and lung cancer risk in four uranium processing operations. Epidemiol- ogy 6(4):370-375. Dupree-Ellis, E., J. Watkins, J. N. Ingle, and J. Phillips. 2000. External radiation exposure and mortality in a cohort of uranium processing workers. American Journal of Epidemiology 152(1):91-95.
COHORT DESCRIPTIONS 191 Frome, E. L., D. L. Cragle, and R. W. McLain. 1990. Poisson regression analysis of the mortality among a cohort of World War II nuclear industry workers. Radiation Research 123(2):138-152. Frome, E. L., D. L. Cragle, J. P. Watkins, S. Wing, C. M. Shy, W. G. Tankersley, and C. M. West. 1997. A mortality study of employees of the nuclear industry in Oak Ridge, Tennessee. Radia- tion Research 148(1):64-80. Gustavsson, P., M. Talback, A. Lundin, B. Lagercrantz, P. E. Gyllestad, and L. Fornell. 2004. Inci- dence of cancer among Swedish military and civil personnel involved in UN missions in the Balkans 1989-99. Occupational and Environmental Medicine 61(2):171-173. Hadjimichael, O. C., A. M. Ostfeld, D. A. DâAtri, and R. E. Brubaker. 1983. Mortality and cancer incidence experience of employees in a nuclear fuels fabrication plant. Journal of Occupational Medicine 25(1):48-61. IOM (Institute of Medicine). 2000. Gulf War and health, volume 1: Depleted uranium, sarin, pyr- idostigmine bromide, vaccines. Washington, DC: National Academy Press. Kurttio, P., A. Auvinen, L. Salonen, H. Saha, J. Pekkanen, I. Makelainen, S. B. Vaisanen, I. M. Penttila, and H. Komulainen. 2002. Renal effects of uranium in drinking water. Environmental Health Perspectives 110(4):337-342. Kurttio, P., H. Komulainen, A. Leino, L. Salonen, A. Auvinen, and H. Saha. 2005. Bone as a pos- sible target of chemical toxicity of natural uranium in drinking water. Environmental Health Perspectives 113(1):68-72. Kurttio, P., A. Harmoinen, H. Saha, L. Salonen, Z. Karpas, H. Komulainen, and A. Auvinen. 2006a. Kidney toxicity of ingested uranium from drinking water. American Journal of Kidney Disease 47(6):972-982. Kurttio, P., L. Salonen, T. Ilus, J. Pekkanen, E. Pukkala, and A. Auvinen. 2006b. Well water radioac- tivity and risk of cancers of the urinary organs. Environmental Research 102(3):333-338. Loomis, D. P., and S. H. Wolf. 1996. Mortality of workers at a nuclear materials production plant at Oak Ridge, Tennessee, 1947-1990. American Journal of Industrial Medicine 29(2):131-141. Macfarlane, G. J., E. Thomas, and N. Cherry. 2000. Mortality among UK Gulf War veterans. Lancet 356(9223):17-21. Macfarlane, G. J., A. M. Biggs, N. Maconochie, M. Hotopf, P. Doyle, and M. Lunt. 2003. Incidence of cancer among UK Gulf War veterans: Cohort study. British Medical Journal 327(7428):1373. McDiarmid, M. 2007 (unpublished). Depleted uranium (DU) follow-up program update. Presenta- tion to the IOM Committee on Gulf War and Health: Depleted Uranium. College Park, MD: University of Maryland. McDiarmid, M. A., J. P. Keogh, F. J. Hooper, K. McPhaul, K. Squibb, R. Kane, R. DiPino, M. Kabat, B. Kaup, L. Anderson, D. Hoover, L. Brown, M. Hamilton, D. Jacobson-Kram, B. Burrows, and M. Walsh. 2000. Health effects of depleted uranium on exposed Gulf War veterans. Envi- ronmental Research 82(2):168-180. McDiarmid, M. A., K. Squibb, S. Engelhardt, M. Oliver, P. Gucer, P. D. Wilson, R. Kane, M. Kabat, B. Kaup, L. Anderson, D. Hoover, L. Brown, and D. Jacobson-Kram. 2001. Surveillance of depleted uranium exposed Gulf War veterans: Health effects observed in an enlarged âfriendly fireâ Cohort. Journal of Occupational and Environmental Medicine 43(12):991-1000. McDiarmid, M. A., F. J. Hooper, K. Squibb, K. McPhaul, S. M. Engelhardt, R. Kane, R. DiPino, and M. Kabat. 2002. Health effects and biological monitoring results of Gulf War veterans exposed to depleted uranium. Military Medicine 167(2 Suppl):123-124. McDiarmid, M. A., S. Engelhardt, M. Oliver, P. Gucer, P. D. Wilson, R. Kane, M. Kabat, B. Kaup, L. Anderson, D. Hoover, L. Brown, B. Handwerger, R. J. Albertini, D. Jacobson-Kram, C. D. Thorne, and K. S. Squibb. 2004. Health effects of depleted uranium on exposed Gulf War veter- ans: A 10-year follow-up. Journal of Toxicology and Environmental Health A 67(4):277-296.
192 updated literature review of depleted uranium McDiarmid, M. A., S. M. Engelhardt, M. Oliver, P. Gucer, P. D. Wilson, R. Kane, M. Kabat, B. Kaup, L. Anderson, D. Hoover, L. Brown, R. J. Albertini, R. Gudi, D. Jacobson-Kram, C. D. Thorne, and K. S. Squibb. 2006. Biological monitoring and surveillance results of Gulf War I veterans exposed to depleted uranium. International Archives of Occupational and Environmental Health 79(1):11-21. McDiarmid, M. A., S. M. Engelhardt, M. Oliver, P. Gucer, P. D. Wilson, R. Kane, A. Cernich, B. Kaup, L. Anderson, D. Hoover, L. Brown, R. Albertini, R. Gudi, D. Jacobson-Kram, and K. S. Squibb. 2007. Health surveillance of Gulf War I veterans exposed to depleted uranium: Updat- ing the cohort. Health Physics 93(1):60-73. McGeoghegan, D., and K. Binks. 2000a. The mortality and cancer morbidity experience of workers at the Capenhurst uranium enrichment facility 1946-95. Journal of Radiological Protection 20(4):381-401. âââ. 2000b. The mortality and cancer morbidity experience of workers at the Springfields uranium production facility, 1946-95. Journal of Radiological Protection 20(2):111-137. âââ. 2001. The mortality and cancer morbidity experience of employees at the Chapelcross plant of British Nuclear Fuels plc, 1955-95. Journal of Radiological Protection 21(3):221-250. Nuccetelli, C., M. Grandolfo, and S. Risica. 2005. Depleted uranium: Possible health effects and experimental issues. Microchemical Journal 79(1-2):331-335. Pinkerton, L. E., T. F. Bloom, M. J. Hein, and E. M. Ward. 2004. Mortality among a cohort of uranium mill workers: An update. Occupational and Environmental Medicine 61(1):57-64. Pinney, S. M., R. W. Freyberg, G. E. Levine, D. E. Brannen, L. S. Mark, J. M. Nasuta, C. D. Tebbe, J. M. Buckholz, and R. Wones. 2003. Health effects in community residents near a uranium plant at Fernald, Ohio, USA. International Journal of Occupational Medicine and Environmental Health 16(2):139-153. Polednak, A. P., and E. L. Frome. 1981. Mortality among men employed between 1943 and 1947 at a uranium-processing plant. Journal of Occupational Medicine 23(3):169-178. Richardson, D. B., and S. Wing. 2006. Lung cancer mortality among workers at a nuclear materials fabrication plant. American Journal of Industrial Medicine 49(2):102-111. Ritz, B. 1999. Radiation exposure and cancer mortality in uranium processing workers. Epidemiol- ogy 10(5):531-538. Ritz, B., H. Morgenstern, D. Crawford-Brown, and B. Young. 2000. The effects of internal radiation exposure on cancer mortality in nuclear workers at Rocketdyne/Atomics International. Environ- mental Health Perspectives 108(8):743-751. Shawky, S., H. A. Amer, M. I. Hussein, Z. el-Mahdy, and M. Mustafa. 2002. Uranium bioassay and radioactive dust measurements at some uranium processing sites in Egyptâhealth effects. Journal of Environmental Monitoring 4(4):588-591. Stayner, L. T., T. Meinhardt, R. Lemen, D. Bayliss, R. Herrick, G. R. Reeve, A. B. Smith, and W. Halperin. 1985. A retrospective cohort mortality study of a phosphate fertilizer production facil- ity. Archives of Environmental Health 40(13):133-138. Storm, H. H., H. O. Jorgensen, A. M. Kejs, and G. Engholm. 2006. Depleted uranium and cancer in Danish Balkan veterans deployed 1992-2001. European Journal of Cancer 42(14):2355-2358. Sumanovic-Glamuzina, D., V. Saraga-Karacic, Z. Roncevic, A. Milanov, T. Bozic, and M. Boranic. 2003. Incidence of major congenital malformations in a region of Bosnia and Herzegovina al- legedly polluted with depleted uranium. Croatian Medical Journal 44(5):579-584. Wagoner, J. K., V. E. Archer, B. E. Carroll, D. A. Holaday, and P. A. Lawrence. 1964. Cancer mortality patterns among U.S. uranium miners and millers, 1950 through 1962. Journal of the National Cancer Institute 32(4):787-801. Waxweiler, R. J., V. E. Archer, R. J. Roscoe, A. Watanabe, and M. J. Thun. 1983. Mortality patterns among a retrospective cohort of uranium mill workers. In Epidemiology Applied to Health Physics, Proceedings of the Sixteenth Midyear Topical Meeting of the Health Physics Society, Albuquerque, New Mexico, January 9-13:428-435.