Estimating Historical Occupational Exposure to Airborne Hexavalent ...

Journal of Occupational and Environmental Hygiene, 1: 752–767 ISSN: 1545-9624 print / 1545-9632 online c 2004 JOEH, LLC Copyright DOI: 10.1080/15459620490523294

Estimating Historical Occupational Exposure to Airborne Hexavalent Chromium in a Chromate Production Plant: 1940–1972 D.M. Proctor,1 J.P. Panko,2 E.W. Liebig,3 and D.J. Paustenbach4 1

Exponent, Irvine, California ChemRisk, Pittsburgh, Pennsylvania 3 Shaw Environmental, Pittsburgh, Pennsylvania 4 ChemRisk, San Francisco, California 2

This article presents a retrospective exposure assessment for 493 workers who were occupationally exposed to airborne hexavalent chromium, Cr(VI), at a Painesville, Ohio, chromate production plant from 1940–1972. Exposure estimates were reconstructed using a job-exposure matrix approach that related job titles with area monitoring data from 21 industrial hygiene surveys conducted from 1943 to 1971. No personal monitoring data were collected. Specifically, airborne Cr(VI) concentration profiles for 22 areas of the plant, termed job-exposure group (JEG) areas, were constructed for three distinct time periods (1940–1949, 1950–1964, and 1965–1972), with cut points based on known major plant and process changes. Average airborne Cr(VI) concentrations were the highest for the bridge crane operators (5.5 mg/m3 ) prior to 1965, although only four cohort members held this job title. Airborne concentrations for the rest of the production areas of the plant ranged from 1.9 mg/m3 for packers in the 1940s to 0.012 mg/m3 for ore mill operators after 1964. For nearly all JEG areas, exposures decreased over time, particularly after 1964. For example, average airborne concentrations in production areas of the plant decreased from 0.72 mg/m3 in the 1940s to 0.27 mg/m3 from 1950 to 1964, and the average was 0.039 mg/m3 after 1964. Former workers were interviewed to determine activity patterns in the plant by job title. This information was combined with Cr(VI) monitoring data to calculate cumulative occupational exposure for each worker. Cumulative exposures ranged from 0.003 to 23 (mg/m3 ) × years. The highest monthly 8-hour average exposure concentration for each worker ranged from 0.003 to 4.1 mg/m3 . These exposure estimates have been combined with mortality data for this cohort to assess the lung cancer risk associated with inhaled Cr(VI), and a positive dose-response relationship was observed for increases in lung cancer mortality with measures of cumulative exposure and highest monthly exposure. Keywords

exposure reconstruction, hexavalent chromium, jobexposure matrix, occupational exposure limit, risk assessment

Address correspondence to: Deborah M. Proctor, 320 Goddard, Suite 200, Irvine, CA 92618; e-mail: [email protected].

752

O

ccupational exposure in the historical chromate chemical production industry has been associated with an increased risk of lung cancer mortality for more than 50 years.(1–3) Understanding the relationship between exposure to airborne hexavalent chromium, Cr(VI), and lung cancer has been hindered primarily by limitations in the exposure assessment. Exposure reconstruction and dose-response studies have been conducted for several Cr(VI)-exposed occupational cohorts;(4–11) however, for most studies, historical exposure information was incomplete, qualitative, or could not be defined on a worker-specific basis.(4,6–8) In other cases, where the exposure data are more robust because the industrial hygiene surveys were more recent, the cohort lacked sufficient latency for observation of an increased cancer risk if it existed among these cohorts.(9,11) To date, the U.S. Environmental Protection Agency and the Occupational Safety and Health Administration (OSHA) have based quantitative cancer risk assessments for Cr(VI) (i.e., cancer potency factors or unit risks) on the dose-response and exposure analyses published by Mancuso(7) for workers at the same Painesville, Ohio, chromate production plant during the very earliest years of operation (workers who started during 1931–1937). However, the Mancuso study includes considerable uncertainties in both the exposure and mortality assessments.(2,12–14) For example, exposures were based on a single air monitoring survey conducted in 1949, 12 to 18 years after exposures began in 1931–1937, and the mortality assessment consisted of only crude lung cancer mortality rates, which were not referenced to a standardized population. In an effort to provide better exposure information for the workers at the Painesville plant, we recently obtained and evaluated historical records to identify additional exposure data for these workers.(15) While no data were identified that could be used to quantify the exposures for workers during the 1930s (when Mancuso’s cohort started), more than 800 measurements of airborne Cr(VI) from 23 newly identified

Journal of Occupational and Environmental Hygiene

November 2004

surveys, which were conducted from 1943 to 1971, were located. Most of the data were of sufficient quality and specificity that they could be used to characterize Cr(VI) exposures of a later Painesville cohort. This article describes the exposure assessment process for workers who started after 1939 in the Painesville plant and quantifies their occupational exposures to airborne Cr(VI) by two different exposure metrics: highest monthly average exposure and cumulative exposure. Previous quantitative exposure assessments for Cr(VI) have focused on lifetime cumulative exposure (units of mg/m3 years or [mg/m3 ] × years). However, as has been suggested for silicosis due to silica exposures,(16) it is plausible that another exposure measure, such as the highest daily or monthly exposure, or exposure over a threshold, could be a more appropriate exposure metric than the traditional measure of “cumulative exposure.” Because Cr(VI) is reduced to the essentially nontoxic and noncarcinogenic trivalent state, Cr(III), by the constituents and conditions of the lung,(17) it is plausible that respiratory cancer can be induced only at airborne concentrations of Cr(VI) that overwhelm the reductive capacity of the lung.(18) The risk of lung cancer, therefore, could be more strongly associated with some measure of the highest short-term exposure, or the area under the curve for concentrations above a certain threshold concentration (e.g., cumulative exposure above 0.2 mg/m3 ), rather than with lifetime cumulative exposure. Although the highest daily or monthly exposures are difficult to know definitively for the Painesville workers, this is the first study that has sought to quantify the highest average monthly exposure. The objective of this investigation was to construct a jobexposure matrix (JEM) that could be used to quantify the cumulative and highest monthly exposures for the workers from the Painesville plant who began employment after 1939. This cohort provides valuable data for cancer risk assessment because of the age of the cohort (most workers were born between 1910 and 1930), and because sufficient time has passed since initial exposures to accommodate the long latency of lung cancer. This study is particularly significant for understanding the risks posed by airborne Cr(VI) because of the reasonably large number of industrial hygiene samples collected during the time period of interest. The exposure estimates derived here were correlated with lung cancer mortality data to derive a cancer potency estimate for airborne Cr(VI).(3,19) BACKGROUND History and Description of Painesville, Ohio, Chromate Production Plant Cr(VI) is produced by the oxidation of Cr(III), which occurs naturally in chromite ore, bound tightly to magnesium, aluminum, and iron oxides.(1,20) The Painesville chromate production plant operated from 1931 until 1972, when operations were moved to a new facility in North Carolina. The primary operation at the Painesville plant was the production of sodium dichromate from chromite ore, lime, and ash. From sodium dichromate, other Cr(VI) chemicals were produced, including

potassium dichromate and chromic acid. The production process is summarized in Figure 1, and details of the operations have been described previously.(15) Changes in Chromate Manufacturing Processes (1940–1972) The sodium dichromate production process at the Painesville plant experienced two significant changes between 1940 and 1972, as well as minor ongoing improvements during this time period.(15,21,22) In 1949–1950, the plant was substantially renovated to install new dust and mist control equipment, convert the leaching operation from a batch operation to a continuous operation, and expand the plant—moving less dusty operations to a new building (the Evaporation Building). These changes were thought to have generally resulted in reduced exposures for most of the plant workers.(21) In addition, engineering improvements designed to reduce airborne exposures were implemented throughout the plant starting from approximately 1962 to 1964. Former workers noted that the installation of quench tanks, which received the roast directly from the kiln and fed into the classifiers for leaching of Cr(VI), substantially reduced airborne particles. Local exhaust ventilation systems were also installed during this time period to reduce emissions from the kilns.(22) These improvements resulted in a notable reduction in airborne Cr(VI) concentrations in the workplace starting in 1964.(15) METHODS

R

econstruction of exposures for the cohort was based on historical air monitoring data, historical company records, and information collected in interviews with former Painesville plant workers. Exposure was estimated based on industrial hygiene data for operational areas of the plant collected over time.(15) Historical written data sources were used to define each worker’s job history (i.e., job titles by month for the duration of employment). Job title assignments over time were used to relate individuals, by job title, to location-specific air monitoring data. Assumptions (described below) regarding time-activity patterns were used to quantify representative daily exposures (i.e., 8-hour time-weighted averages [TWAs]). The JEM consisted of job titles correlated with airborne concentrations of Cr(VI) that were averaged for specific time intervals, as discussed further below. Historical Exposure Data The reconstruction of historical exposures for the Painesville plant involved an exhaustive review of all historical hygiene records. The type of information and high level of detail documented in the hygienists’ historical records were impressive by current standards, and these were invaluable in helping to ensure the accuracy of the exposure estimates. In all, usable data were collected over a period of 28 years— in 1943, 1945, 1948, and every year between 1957 and 1971 (with the exception of 1958 and 1962). Our review of the available historical industrial hygiene data identified airborne


November 2004

753

FIGURE 1.

Sodium dichromate production process at the Painesville chromate production plant

Cr(VI)-speciated data from 26 surveys. All the information on airborne concentrations is based on area samples, and most of the operating areas of the plant were surveyed annually. Data from five of these sampling surveys were not used in the exposure reconstruction because they either lacked Cr(VI)speciated data, provided only ranges of airborne concentrations, or provided industry-wide exposure measures from which the Painesville data could not be discerned from those of other facilities. The resulting data set used for the exposure reconstruction included more than 800 samples collected throughout the plant and specifically analyzed for Cr(VI). Although exposure to Cr(III) occurred in the plant, no samples were identified to determine the airborne concentration of total chromium or Cr(III), other than those that have been reported previously.(23,24) All samples used in the reconstruction were collected using Greenburg-Smith impingers and were analyzed for Cr(VI) colorimetrically by reaction with diphenylcarbazide reagent. These methods are similar to validated methods that are currently used to measure Cr(VI) in air.(25) For the industrial hygiene surveys conducted in the 1940s, only one impinger was used to capture airborne Cr(VI), and thus the data collected in the 1940s was corrected in the evaluation for impinger “breakthrough” using the original data from later studies conducted by the plant industrial hygienists. The hygienists sought to determine the number of impingers required to capture airborne Cr(VI) in the plant by examining Cr(VI) in impingers that were connected in series. Using the 754

original data from these studies, the fraction of Cr(VI) that was transferred from the first to the second impinger was determined. On average, 14% of each sample was measured in the second impinger (standard deviation: 21%, range: 0–73%), and the amount of breakthrough was not concentration dependent. Thus, to account for uncaptured, and thus unmeasured, Cr(VI) in samples collected with the single impinger technique, the reported airborne concentrations from the 1940s surveys were increased by a factor of 16% (100/86 = 1.16). Quality of the Industrial Hygiene Data The air samples from the Painesville plant were collected generally for the purpose of assessing typical exposures during normal working conditions and thus can be used for estimating long-term average exposures.(15) Sample collection and analytical methods were assessed as part of our companion analysis(15) and were found to be reasonably accurate and reproducible. While only a few quality control procedures could be found in the historical record, the available information and discussions with one former company hygienist suggested that the air sampling results were accurate and sufficiently reliable for reconstructing worker exposures.(15) Data trends and patterns were consistent with expectations (concentrations were lower outdoors than indoors and decreased following process improvements) and measured concentrations of Cr(VI) were relatively consistent year after year at the same location. No co-located (duplicate) samples were reported in the historical records, but for samples collected at the same location


November 2004

FIGURE 2. Comparison of geometric mean hexavalent chromium airborne concentrations in the Painesville, Ohio, chromate production plant by process area and year

over the course of a single day, variability was relatively low (coefficients of variation averaged 33%). Finally, a variety of steps were taken by the company hygienists to ensure that the data were of high quality and not affected by contamination.(15) The airborne concentration data are summarized in Figure 2. Reconstruction of Worker Histories Cohort Identification Approximately 175 to 250 employees worked in the plant at any one time. The cohort included all workers who worked in the plant for at least 1 year from January 1940 until the plant closed in April 1972, and who had not worked in either of two other Diamond Alkali Company plants that manufactured chromium-containing chemicals in New Jersey. These selection criteria ensured that: (1) airborne Cr(VI) exposures

of the resulting cohort could be estimated with a reasonable degree of accuracy using the available industrial hygiene data; (2) a large cohort would be generated, which would increase power in the analysis; and (3) exposures of the cohort were of sufficient duration to have the potential to affect mortality. Around 1972, the operations of the Painesville plant were moved to a new production plant in North Carolina. Seventeen individuals who started in Painesville but moved to the North Carolina plant were included in the cohort because exposures had been estimated previously and could be incorporated into their exposure profiles.(10) Workers employed for less than 12 months (not necessarily consecutively) were excluded from the analysis because company records lacked sufficient information on these individuals


November 2004

755

to reconstruct their work histories and for consistency with previous studies of Cr(VI)-exposed workers that included only workers with more than 1 year of exposure.(7–9,11,26,27) Furthermore, short-term workers may differ in their risk profile for various health outcomes due to chemical exposures at other jobs as well as in socioeconomic status.(28) The risk of lung cancer resulting from exposures of less than 1 year was considered sufficiently small as not to compromise the results of this study. Data Sources/Reconstruction of Job Titles Over Time A concerted effort was made to locate and use any available information to reconstruct worker histories. Due to changes in company ownership and location since plant closure, documentation of worker histories was difficult, and the records that could be obtained varied in completeness. Complete, detailed individual work histories (e.g., personnel records) were available for approximately 30% of the workers included in the cohort. Therefore, union records, medical records, industrial hygiene survey reports, and other miscellaneous information (e.g., letters to and from Dr. Mancuso) were used as primary sources of data to reconstruct worker histories for the remaining 70%. From these data sources, more than 1800 individual pieces of information were identified and used to reconstruct worker exposures. Information extracted from the data was maintained in a Microsoft Access database and compiled for each cohort member (i.e., worker history database). The research team established criteria and followed a standardized procedure for reviewing the historical information to address gaps and ensure consistency. Historical data were reviewed with former long-term workers from the Painesville facility for verification of plant operations and job descriptions. To control for selection bias, cohort members were not identified from sources of information compiled on the basis of health status, such as medical files or disability claims. However, these sources were used to fill data gaps in job histories or demographics. To further control for possible investigator bias, the reconstruction of job titles and exposure levels was performed by different researchers (in different locations) from those who ascertained mortality data. For each cohort member, a detailed work history was reconstructed including start and end dates for occupational tenure in the plant, and job title for each month of employment. A total of 117 different job titles were identified over the course of operations at the plant. It was determined, however, that many of these job titles signified a promotion and thus no significant change in exposure to airborne Cr(VI). For instance, a Pipefitter II had seniority over a Pipefitter I, and a Kiln Operator 2 had seniority over a Kiln Helper 2, but any difference in exposure based on the promotion could not be discerned from the available data. Workers for whom no first job title was recorded in their first month(s) of employment were assumed to have the “departmental laborer” title, based on conversations with former 756

plant supervisors. Because they worked throughout the plant, the airborne concentrations to which departmental laborers were exposed were estimated using the time-specific average concentration for all operating areas of the plant, nearly all of which were indoors. Other missing job titles for periods of employment of typically no more than a few months, which was known to be in the middle of working tenure (for example, a specific worker was known to work in the plant from January 1963 to March 1965, but his job title was unknown for June and July of 1964) were assumed using professional judgment. Unless information suggested otherwise, when data gaps in job title were identified for production workers, exposure was assumed to be the average concentration in the production areas of the plant at that time. Job-Exposure Matrix (JEM) Job-Exposure Group (JEG) Areas Former plant workers and historical industrial hygiene surveys provided detailed descriptions of where in the plant workers spent their workday by job title. The areas of the plant in which workers of at least one job title spent at least a portion of their time during a typical 8-hour workday were termed “job-exposure group” (JEG) areas. JEG areas were used to relate air monitoring locations in the plant to job titles. Where exposures were reasonably consistent, JEG areas were collapsed to increase the robustness of the data. A few JEG areas were defined by almost identical sampling locations, but since plant operations moved within the plant with process changes or were relocated following renovations, they were kept as distinct areas in the JEM. Twenty-two JEG areas were identified and designated with the letters A through V for the JEM, and each job title was assigned to at least one JEG area (Table I). One JEG area was designated to represent all operating areas of the plant (JEG A) and was used in conjunction with other JEG areas to describe the exposures of workers whose job titles placed them throughout the plant (e.g., departmental laborer). All but one operating area included in JEG A was indoors, and approximately 96% of the airborne samples were from indoor sampling locations. Exposure Levels by JEG Area Exposure levels (mg/m3 ) for each JEG area were assumed to be equal to the arithmetic mean concentration of measured airborne Cr(VI) divided into three distinct time periods, 1940– 1949, 1950–1964, and 1965 through plant closure in 1972, chosen for reasons discussed earlier. In cases where the sample size was inadequate (arbitrarily defined as fewer than three samples per JEG area per time period), data from other time periods were used to estimate exposure. JEG areas with fewer than three samples per time period were generally those where exposures were perceived to be low (e.g., outdoor areas, locker rooms). The distributions of airborne concentrations in the two- and three-kiln areas of the plant (JEGs D and E), and in the all-plant operations area (JEG A) were lognormally distributed; thus,


November 2004

TABLE I.

Job Titles for Each Job Exposure Group (JEG)

Job Exposure Group

Description

JEG A

All plant production areas

JEG B JEG C

Locker rooms Indoor non-production areas

JEG D JEG E JEG F JEG G JEG H JEG I JEG J JEG K JEG L JEG M JEG N JEG O JEG P JEG Q JEG R JEG S JEG T JEG U JEG V

Two-kiln areas Three-kiln areas Ceiling level above classifiers Chromic acid plant Soda wringer Leaching operations Filtering operations Dry-mix operations Neutralization operations Liquor operations Mud transport operations Potassium dichromate operations Outdoor non-production area Sulfate basket operations Bagging and shipping dichromate Sulfate wringing operations Sulfuric acid-treating operations Boiler-tending operations Ore mill operations

A Engineers,

Job Titles Laborer, reliefman, foreman, project engineer, all maintenance titles A All production workers Locomotive crane fireman, ore hoist operator, chrome analyst assistant, project engineer, foreman, shipping clerk, all maintenance titles A 2 kiln helper/operator, roast cleaner 3 kiln helper/operator Bridge crane operator Chromic acid operator #1, #2, #3, #4, chromic acid reliefman Soda wringer Press helper/operator, leaching operator, shriver press helper Filter operator Raw mix operator, soda ash and lime unloader Hydrate helper/operator Liquor operator Mud crane operator Potash operator Raw liquor operator Sulphate basket operator Shipper (packer) Sulphate wringer Treating operator Boiler fireman Ore mill operator

foremen, and all maintenance titles split their time between JEGs A and C.

the arithmetic means provide an exposure measure that is positively biased. However, because the primary objective of this study was to assess lifetime cumulative exposures for cancer risk assessment, and the arithmetic mean is preferred for these purposes,(16,29,30) the arithmetic mean was selected as the exposure measure for the analysis. Of the more than 800 industrial hygiene samples available, only 39 had nondetectable concentrations of Cr(VI), with limits of detection ranging from 0.0035 to 0.09 mg/m3 . In instances when Cr(VI) could not be measured in a sample, a value of one-half the limit of detection for that survey was used when calculating the mean. Time Weighting of 8-Hour Average Exposures The calculation of TWA airborne exposures was based on two factors: (1) variability of airborne concentrations during the 8-hour shift in some JEG areas, and (2) expected movement of workers by job title throughout the plant during the course of a workday. It is noteworthy that concentrations of airborne Cr(VI) for two parts of the plant were recognized by the industrial hygienists who conducted the surveys to vary substantially depending on the work activity being performed. Airborne concentrations

in the bagging and shipping departments, where sodium dichromate was loaded into bags for shipping, were weighted for 3 hours per 8-hour workday when bagging operations actually occurred. In all surveys except the one conducted in 1943, samples were collected during both bagging and nonbagging operations in that area. To adjust the 1943 data, the nonbagging data from 1945 were combined with the bagging data from 1943 to calculate an 8-hour TWA concentration. Similarly, in the chromic acid plant (JEG G) that started operations in 1951, differences in exposure were recognized to exist when the “melt” was being discharged and when it was not. The industrial hygiene survey reports indicated the length of time involved in each discharge, the number of discharges made per shift, and the locations of workers when not making the discharge. The 8-hour TWA airborne concentrations in the chromic acid plant were weighted using this information in a similar manner to that for bagging operations discussed above. Workers of most job titles spent time in more than one JEG area; thus, when calculating 8-hour TWA exposures, multiple JEG areas where workers spent their time over the course of a typical shift were taken into consideration. No written records were located to assign time-weighting assumptions, so the recollections of former workers were relied on to obtain estimates


November 2004

757

of time-activity patterns. Because no personal monitoring data existed, the validity of these assumptions could not be established. However, because concentrations were not highly variable within individual operating areas, the use of simple assumptions regarding time-weighting were concluded to be adequate for estimating long-term exposures. The time-weighting generally consisted of assigning all production workers to 6 hours in the area of the plant where they primarily worked (e.g., kiln operators spent 6 hours in the kiln buildings [JEGs D or E]) and 2 hours in the locker rooms or break rooms (JEG B). Maintenance workers were assigned 4 hours in the maintenance building (indoor nonproduction areas including maintenance building, JEG C) and 4 hours in the indoor operating areas of the plant (JEG A) to reflect their typical daily work activities. Average exposures for maintenance workers were likely the most variable and less accurately quantified as compared with exposures of the operations workers. Salaried engineers and foremen were assigned 4 hours in the office (indoor nonproduction areas, JEG C) and 4 hours in the indoor operating areas of the plant (JEG A). Although foremen were assigned to specific areas of the plant, historical records did not provide this information. In some cases, the plant telephone directories or industrial hygiene reports could be used to place specific foremen in the area of the plant where they worked. Office, laboratory, and storeroom workers were assigned to 8 hours in the indoor nonproduction area (JEG C) because they did not use the locker rooms inside the plant. Only two job titles had time-weighting assignments that varied from those described above—the crane operator and the soda ash and lime unloader. The crane operator could have operated either the bridge crane or the mud crane, and thus was assigned 3 hours in the bridge crane (ceiling level above the classifiers, JEG F) and 3 hours in the mud crane (mud transport operations, JEG N) as well as 2 hours in the locker rooms (JEG B). There were also bridge crane operators who were assigned only to the bridge crane (JEG F). The other exception was the soda ash and lime unloader who was assigned to 1 hour in the dry mix operations area (JEG K), 5 hours in the indoor nonproduction areas (JEG C), and 2 hours in the locker room (JEG B). All of these assumptions are based on information that was initially obtained through interviews with former workers and finalized during a meeting between the authors and six former long-time plant workers. Calculation of Individual Cumulative Exposures The JEM linked individual worker job-title histories with estimated Cr(VI) concentrations by JEG to calculate cumulative exposure [units of (mg/m3 ) × year, or mg/m3 -years]. Time-weighted exposures were calculated for each month of each worker’s occupational tenure by combining airborne concentrations by JEG with time-weighting assumptions as described below. Cr(VI)t(n) = [(%Ta × Conca ) + (%Tb × Concb ) + · · ·] (1) 758

where Cr(VI)t(n) = time-weighted Cr(VI) concentration %Tx = percent of 8-hour day spent in JEG X Concx = Cr(VI) airborne concentration in JEG X Individual estimates of cumulative exposure were calculated by summing time-weighted concentrations over occupational tenure. Estimation of Highest Monthly Exposures In addition to evaluating cumulative exposure (e.g., [mg/m3 ] × years), we calculated highest monthly exposures to Cr(VI) for each worker included in the cohort. We hypothesized that highest monthly exposure could be the metric that best predicts the likelihood of developing lung cancer. This was based on the assumption that short-term exposures to high concentrations of airborne Cr(VI) might be sufficient to exceed the reductive capacity of the lung and result in DNA damage, increasing the risk of lung cancer. By comparison, the lung is better able to detoxify lower concentrations of Cr(VI) through its capacity to reduce Cr(VI) to Cr(III). As noted by Stewart et al.,(31) exploring different exposure measures may be useful when the mechanism of disease is not consistent with a monotonic increase in mortality with cumulative exposure. Uncertainties Associated with Exposure Reconstruction Inconsistencies in the historical records had to be reconciled to define job histories for each worker. For example, for 153 cohort members, the company’s “tenure lists,” which provided the number of years and the months that each worker was employed, but not the actual dates of employment, were found to be inconsistent with union seniority lists, medical records, and/or urine specimen collection lists. In this case, these sources, which included specific dates and periods of employment, were considered more reliable than the tenure lists and were used in their place. The historical records did not include work starting or ending dates for a fraction of the workers. For example, 3% of the cohort (18 workers) either began or ended their employment in the plant between May 1960 and June 1965, when almost no worker history records were found. If no other information could be identified to define their start or ending dates, it was assumed that they started or ended their career at the Painesville plant in January 1963—the midpoint of these dates. Furthermore, 16% of the cohort (77 workers) was still noted as working in the chromate plant at the end of 1971, and records do not support that they transferred to a different plant. For these workers, we assumed that their employment in the chromate plant ended in April 1972 when the plant closed. Time away from work because of medical disabilities was accounted for in each employee’s history based on the detailed medical records kept by the company physicians. However, information regarding overtime or time off due to strikes, vacation, or sick leave was not available. Former employees recalled that time off due to strikes was generally made up with overtime (e.g., additional shifts per week), and the strikes


November 2004

FIGURE 3.

Cumulative distribution of occupational tenure for the Painesville chromate production worker cohort (1940–1972)

were not long in duration (the longest recalled strike lasted 2 months). The workers recalled very specifically that the shifts were kept to 8 hours, even in the 1940s, to be compliant with the 8-hour occupational exposure limit for Cr(VI), which dates back to 1943 (1 mg CrO3 /10 m3 ).(32) Therefore, we concluded that strikes and overtime did not contribute significantly to the calculation of cumulative work hours, and we decided to work with the assumption that workers were in the plant 8 hours per day, 5 days per week while employed, with the single exception of known time off from work due to medical leaves for which there were written records. Once the worker job histories were entered into the database, a quality assurance/quality control procedure was implemented to verify that the information entered into the database was correct. Independent double input procedures were used, so we were assured of virtually 100% accuracy of the information in the database used to estimate exposures. RESULTS Cohort Identification Of the 1034 chromate production plant workers initially identified from the available data sources, 200 were excluded because they began work prior to January 1940. Another 284 workers were excluded because they were employed for less than 1 year, and an additional 57 were excluded because they had been employed at one of the New Jersey chromium chemical production facilities. This left a total of 493 workers who were included in the exposure assessment.

Occupational Tenure The occupational tenure of workers in the cohort ranged from 1 year to 32 years (Figure 3). Slightly more than 20% worked less than 2 years, and approximately half of the cohort worked for at least 6 years. The distribution of working tenure also included a significant proportion that worked for long durations (e.g., 15% worked for more than 20 years). Exposure Estimates by JEG The average Cr(VI) airborne concentrations in each JEG area of the plant for the time periods of 1940–1949, 1950– 1964, and 1965–1972, are presented in Table II. As an example of the exposure profile for a JEG area, the airborne Cr(VI) concentrations used to describe JEG D for the two-kiln areas of the plant are the arithmetic mean concentrations of samples collected in these areas of the plant, separated into three time periods (Figure 4). During most surveys, two samples were collected in the JEG D area; however, five samples were collected in this area in the 1943 survey (two samples had the same concentration), and five samples were collected in the 1961 survey. Average airborne concentrations of Cr(VI) for all indoor operating areas (JEG A) decreased nearly twentyfold from the 1940s to the period after 1964, and airborne concentrations decreased over time in most areas of the plant (Table II). Bridge crane operators, who worked at ceiling level above the classifiers, were exposed to the highest airborne concentrations, averaging 5.5 mg/m3 , which is more than a hundredfold higher than


November 2004

759

TABLE II.

Mean Cr(VI) Airborne Concentration (mg/m3 ) for Each Job Exposure Grouping (JEG) Area Cr(VI) Air Concentration (mg/m3 )

Job Exposure Group (JEG) JEG A JEG B JEG C JEG D JEG E JEG F A JEG G B JEG H JEG I JEG J JEG K JEG LC JEG M JEG N JEG O D JEG P JEG Q E JEG R F JEG S JEG T JEG U JEG V

1940–1949 Description All plant Locker rooms Indoor non-production area Two-kiln areas Three-kiln areas Ceiling level above classifiers Chromic acid plant Soda wringer Leaching operations Filtering operations Dry-mix operations Neutralization operations Liquor operations Mud transport operations Potassium dichromate operations (potash) Outdoor non-production area Sulfate basket operations Bagging and shipping dichromate Sulfate wringing operations Sulfuric acid-treating operations Boiler-tending operations Ore mill operations

1950–1964

1965–1972

N

Mean

N

Mean

N

Mean

61 0 2 9 16 1 0 6 8 11 13 2 6 3 2 0 4 3 4 2 13 4

0.72 — 0.0018 0.65 0.11 0.11 — 0.18 0.25 1.5 1.3 0.052 0.18 1.1 1.8 — 0.14 1.9 0.14 0.13 0.16 0.048

336 21 11 15 62 3 31 14 45 8 39 0 82 10 14 14 7 7 55 55 27 11

0.27 0.024 0.0099 0.40 0.030 5.5 0.091 1.4 0.86 0.086 0.74 — 0.11 0.35 0.17 0.040 0.041 0.14 0.046 0.046 0.052 0.024

362 24 16 16 72 0 27 16 28 8 15 32 85 8 23 16 0 15 51 51 32 8

0.039 0.0078 0.0026 0.060 0.033 — 0.019 0.032 0.040 0.035 0.063 0.055 0.033 0.031 0.14 0.025 — 0.048 0.029 0.029 0.024 0.012

Notes: The permissible exposure limit and threshold limit value are both 0.05 mgCr(VI)/m3 . N = number of air samples for each time period. If N < 3 for any time period, data for the closest time period were extrapolated forward or back in time, with the exception of JEG T, sulfuric acid-treating area, because these operations were moved in 1950 from Buildings 1 and 2 to the evaporation building; exposures were notably higher in Buildings 1 and 2. Dashes indicate that a mean was not calculated because there were no samples. A One sample from the 1940s was collected from an elevated level, but we do not know if it was collected from the ceiling level where the bridge crane operators worked; samples from 1950 to 1964 were collected to assess exposures to the bridge crane operators. The bridge crane was not used after 1966. B Chromic acid plant started operating in 1951. N is the number of averaged concentrations representative of 8-hour TWA exposure estimates. C All samples were averaged together because airborne concentrations were nearly equal in the 1940s and after 1965. D No cohort members worked in the potash plant until 1968; one worker, who started in the 1930s, ran the potash operations alone prior to 1968. E Sulfate basket operations, for drying sulfate, were conducted in Buildings 1 and 2 in the 1940s and in the sulfate conveyor building after 1950. Airborne exposures from 1950–1964 were used to estimate exposures for these workers because no data were collected in this building after 1964. F N is the number of averaged concentrations representative of 8-hour TWA exposures.

R

the current threshold limit value (TLV , American Conference of Governmental Industrial Hygienists’ effective date of 1994), and the OSHA permissible exposure limit (PEL).(33,34) This finding is consistent with the recollection of the former workers, who stated that exposures incurred by the bridge crane operator were thought to be the worst in the plant. Only four cohort members held this job title during the course of plant operations, and two of the four died from lung cancer. No cohort member held this job title after 1966 because the bridge crane was no longer used. Average measured airborne concentrations were generally above the occupational exposure limit (8-hour TWA TLV) of 0.05 mg/m3 in most areas of the plant until the mid-1960s, and 760

concentrations in the potash plant were never below the TLV. In the 1940s, the average airborne concentrations were below the TLV in only three areas of the plant: indoor nonoperation areas (e.g., office, maintenance building), neutralizing (which was always conducted outside the main plant in the Alumina Filtration Building), and ore mill operations, which primarily handled Cr(III) in chromite ore. Most cohort members worked in the kiln areas of the plant, and beginning in the early 1950s, airborne concentrations in the three-kiln area were, on average, less than the TLV. However, concentrations in the two-kiln area were substantially higher than the TLV until the mid-1960s. Airborne concentrations in the buildings/areas with two kilns were always higher than


November 2004

FIGURE 4. All hexavalent chromium air concentrations and exposure profile for the two-kiln areas (JEG D), Painesville chromate production plant (1940–1972)

those in the areas with three because of operations that took place in the buildings with two kilns. For example, mud drying was conducted in a two-kiln building, and this process was recognized as a significant source of particulate emissions. Worker-Specific Exposure Estimates Cumulative exposures for the cohort ranged from 0.003 to 23 (mg/m3 ) × yrs (Figure 5). The mean and median for the cohort were 1.5 (mg/m3 ) × yrs and 0.65 (mg/m3 ) × yrs, respectively. As described in detail in the mortality assessment for this cohort,(3) two-sided Poisson trend tests for the relationship between cumulative exposure and increases in standardized mortality ratios (SMRs) for lung cancer demonstrated a very high degree of correlation (p = 0.00002).(3) The highest monthly 8-hour TWA exposure for each cohort member ranged from 0.0026 to 4.1 mg/m3 (Table III). The mean and median for the cohort were 0.34 mg/m3 and 0.21 mg/m3 , respectively. The exposure groups for highest monthly exposure are presented in terms of the current PEL as a practical reference point in Table III. The current PEL is a ceiling value of 0.052 mg/m3 (100 µg CrO3 /m3 ); however, the occupational exposure limit when the Painesville plant was operating was 0.052 mg/m3 as an 8-hour TWA. Based on our estimates of the 8-hour TWA airborne concentration for these workers, 103 of the 493 workers were not exposed above the PEL. Slightly less than half of the cohort (242 workers) had highest monthly exposures that were greater than five times the current PEL (> 0.26 mg/m3 ).

An analysis of correlation was conducted on workers’ total cumulative exposure to Cr(VI) (in [mg/m3 ] × yrs) and their highest monthly exposure (in mg/m3 per month). Pearson’s correlation coefficient was used to measure the linear relationship between total cumulative exposure and highest monthly exposure. The results of the analysis demonstrate that total cumulative exposure and highest monthly exposure were highly correlated (R = 0.698) and that the correlation was statistically significant (p < 0.0001). This finding is consistent with the observation that workers who started in the 1940s tended to stay at the plant for longer durations than workers who started in the 1950s and 1960s, resulting in the former generally having both the greatest cumulative and the highest monthly exposures. Because the JEM uses average exposures over three specific time periods and workers held the same job title for many months, the highest monthly exposure for most workers is assessed to occur for more than one month (e.g., monthly monitoring data do not exist, which would allow for the differentiation of consecutive monthly exposures unless the individual changed jobs), for example, see Figure 6. Thus, in this reconstruction, highest monthly exposures occurred on average for 28.2 months. The worker with the highest cumulative exposure was also one of the workers with the highest TWA exposure. Specifically, this worker’s highest monthly exposure was 4.1 mg/m3 while he held the job of bridge crane operator (e.g., TWA exposure to the bridge crane operator was assumed to be 4.1 mg/m3 accounting for 6 hours per day in JEG F and


November 2004

761

FIGURE 5. Cumulative exposures for the Painesville chromate production cohort to 10 (mg/m3 ) × years (four workers had exposures greater than 10 (mg/m3 ) × years and are not presented in this figure)

2 hours per day in JEG C, and this individual held that job for 39 months) (Figure 6). Mortality Assessment The methods and results of the mortality assessment for this cohort have been presented previously.(3) SMRs using both U.S. and Ohio reference rates were calculated for selected cause-specific categories of death that included lung cancer. Lung cancer mortality was investigated further by calculation of SMRs stratified by year of hire, duration of employment, time since hire, and categories of cumulative exposure to Cr(VI). Cohort members were followed for mortality through 1997. Sixty-three percent of the cohort members were deceased; there were 51 deaths due to lung cancer. A dosedependent relationship between lung cancer death and measures of both cumulative and highest monthly exposure was observed (Table IV). Significantly elevated lung cancer SMRs were also found for year of hire before 1960, 20 or more years of exposed employment, and latency of 20 or more years. DISCUSSION

T

o conduct a reliable cancer risk assessment for workers exposed to airborne Cr(VI): (1) the cohort should include workers with sufficient occupational exposure such that an increased risk, if present, can be observed and distinguished from background; (2) there should be at least 25 years of

762

follow-up from initial exposure to account for the long latent period for Cr(VI)-induced lung cancer; (3) the cohort should be old enough that there are a significant number of decedents to observe a statistically significant change, if one exists; and (4) the mortality assessment should be based on age-, sex-, race-, and location-specific reference rates. Ideally, the exposure reconstruction should be based on measures of airborne exposure that are worker-specific, reliable, and preferably contain measures of both peak and cumulative exposure. For quantitative cancer risk assessment, the exposure assessment and mortality assessment should demonstrate a consistently positive dose-response relationship for increasing exposure levels. It is desirable to have consistency among data derived from various historical sources; a large number of long-term workers, such that the analysis represents risks from chronic exposures; complete smoking data; no confounding exposures; and no data gaps in worker histories. Unfortunately, the “ideal” situation rarely occurs in retrospective mortality and exposure assessments, and researchers must rely on less-than-perfect data, recollections of former workers, and professional judgment to fill data gaps.(35,36) Determining whether the results of a retrospective exposure reconstruction are meaningful can be a difficult task with uncertain results. However, one measure of the validity of an exposure reconstruction is that with increasing exposures there is an accompanying increase in the incidence of disease in the cohort.(37,38) When the results of this exposure


November 2004

763

Highest Monthly Airborne Cr(VI) Exposures, 1940–1972 (N = 493)

11*PEL 103 38 110 71 137 34

B PEL

are 8-hour time-weighted averages. = permissible exposure limit (0.052 mg/m3 ). C No workers had highest 8-hour TWA exposures between 3 and 4 times the PEL.

A Exposures

0.57 34 39 25 34 19 33

0.0026 0.074 0.21 0.27 0.49 0.65

0.048 0.142 0.21 0.42 0.56 4.1

0.027 0.12 0.21 0.33 0.55 1.3

0.095 0.53 0.69 1.7 2.4 6.0

4.6 8 6.7 11 12 15

Highest Range of Highest Average Number of Maximum Exposure (mg/m3 ) Cumulative Exposure Occupational Monthly Exposure Exposures in Terms of Number of Workers Months at Highest (mg/m3 × years) Tenure (years) (mg/m3 ) A the Current PEL B,C Within the Range Exposure Minimum Maximum Mean Mean Mean

TABLE III.

FIGURE 6. Cumulative exposures by month for the Painesville chromate production worker with the highest cumulative exposure and some of the job titles held over time

reconstruction were combined with mortality data, a strong and positive correlation was observed between increasing measures of exposure by both metrics (lifetime cumulative exposure and highest monthly exposure) and lung cancer mortality

TABLE IV. Observed and Expected Number of Lung Cancer Deaths Exposure Metric

Observed Lung Expected Lung Observed/ Cancers (O) Cancers A (E) Expected

Cumulative Exposure (mg/m3 × yr) 0–0.19 3 4.5 0.2–0.48 8 4.4 0.49–1.04 4 4.4 1.05–2.69 16 4.4 2.7–23 20 4.3 Highest Monthly Exposure (mg/m3 )