10126279632819J Expo Sci Environ EpidemiolJ Expo Sci Environ EpidemiolJournal of exposure science & environmental epidemiology1559-06311559-064X24022670406713510.1038/jes.2013.53NIHMS584512ArticleDevelopment of a job-exposure matrix for exposure to total and fine particulate matter in the aluminum industryNothElizabeth M.1Dixon-ErnstChristine2LiuSa1CantleyLinda3Tessier-ShermanBaylah3EisenEllen A.1CullenMark R.4HammondS. Katharine1 Division of Environmental Health Sciences, School of Public Health, University of California, Berkeley, CA, USA. Alcoa, Pittsburgh, PA, USA. Yale University School of Medicine, New Haven, CT, USA. Department of Internal Medicine, Stanford University, Stanford, CA, USA.Correspondence to: Dr. Elizabeth M. Noth, Division of Environmental Health Sciences, School of Public Health, University of California, Berkeley, 50 University Hall, #7360, Berkeley, CA 94720-7360, USA. Tel.: (510) 915-4907. Fax: (510) 642-0000. bnoth@berkeley.edu30520141192013Jan-Feb201401720142418999

Increasing evidence indicates that exposure to particulate matter (PM) at environmental concentrations increases the risk of cardiovascular disease, particularly PM with an aerodynamic diameter of less than 2.5μm (PM2.5). Despite this, the health impacts of higher occupational exposures to PM2.5 have rarely been evaluated. In part, this research gap derives from the absence of information on PM2.5 exposures in the workplace. To address this gap, we have developed a job-exposure matrix (JEM) to estimate exposure to two size fractions of PM in the aluminum industry. Measurements of total PM (TPM) and PM2.5 were used to develop exposure metrics for an epidemiologic study.

TPM exposures for distinct exposure groups (DEGs) in the JEM were calculated using 8,385 personal TPM samples collected at 11 facilities (1980-2011). For 8 of these facilities, simultaneous PM2.5 and TPM personal monitoring was conducted from 2010-2011 to determine the percent of TPM that is composed of PM2.5 (%PM2.5) in each DEG. The mean TPM from the JEM was then multiplied by %PM2.5 to calculate PM2.5 exposure concentrations in each DEG.

Exposures in the smelters were substantially higher than in fabrication units; mean TPM concentrations in smelters and fabrication facilities were 3.86 mg/m3 and 0.76 mg/m3, and the corresponding mean PM2.5 concentrations were 2.03 mg/m3 and 0.40 mg/m3. Observed occupational exposures in this study generally exceeded environmental PM2.5 concentrations by an order of magnitude.

Introduction

This paper describes the development of a job-exposure matrix (JEM) created to quantify personal exposures to two size fractions of particulate matter (PM) – total PM (TPM) and PM with an aerodynamic diameter of less than 2.5μm (PM2.5) – in the aluminum industry. Ultimately the JEM provides the basis of an exposure assessment linked to an epidemiologic study of possible work related health effects. To date, control of occupational exposure to particles has focused on the composition and specific toxicity of the constituents rather than the mass concentration or particle size. Occupational exposure limits for “particulates not otherwise regulated,” or PNORs, are orders of magnitude greater than daily environmental limits, which have evolved from total suspended particles (150 μg/m3, United States Environmental Protection Agency (USEPA), 1971) to PM10 (65 μg/m3, USEPA 1987) to PM2.5 (USEPA daily maximum 65 μg/m3 in 1997 lowered to 35 μg/m3 in 2006). By contrast, the Occupational Safety and Health Administration (OSHA) Permissible Exposure Limit (PEL) for PNORs is 15,000 μg/m3. Increasing evidence indicates that exposure to particles at environmental concentrations increases the risk of cardiovascular disease (1–11). The health impact of higher occupational exposures to particulate matter, however, has rarely been evaluated. To address this research gap, an epidemiological study was undertaken to assess the health effects of exposure to airborne PM for workers at an aluminum manufacturing company.

At each step of the modern aluminum manufacturing process there is occupational exposure to airborne PM (12). In mining the bauxite ore workers are exposed to particles from bauxite dust and to a lesser extent crystalline silica dust. During refining the PM exposures are primarily from inorganic dusts (bauxite, crystalline silica, alumina). Smelter workers are exposed to PM from many sources, including PM generated during the reduction of alumina to aluminum metal in the Hall-Heroult process. Although this reduction process takes place in carbon-lined steel pots that are hooded to decrease exposures to the mixture of dusts, metals, and fumes produced during smelting, these potroom exposures are among the highest PM exposures associated with aluminum manufacturing as employees work directly over the pots when replacing anodes. Following smelting, aluminum metal is fabricated into numerous diverse products, from aluminum used in can sheet to airplane parts, and workers may be exposed to metalworking fluids, lubricating oils and metal particles. These work processes are generally conducted at separate facilities, although smelting and fabrication sometimes take place in the same location. Research on PM exposures to workers throughout many stages of aluminum production is scant. Most field measurements reported in the literature have been taken in smelter potrooms and focused on exposures to PM constituents (e.g. fluorides, coal tar pitch volatiles) or metal exposures during welding or silica exposures (12–21).

Within the limited literature on PM exposures across the aluminum industry, there is very little information about the particle size distributions. What does exist is focused either on particle morphology or particle aging in smelters (22,23). Although respirable particulate, inhalable particulate, and total particulate concentrations have been reported, these have focused on a few potroom jobs (14,16). There have been no studies that present exposures of different particle sizes across multiple stages of the aluminum industry, nor with sufficient samples to construct a JEM for epidemiology studies.

The research we present in this paper fills both gaps. We measured concurrent personal PM2.5 and TPM exposure in aluminium workers and integrated these data with a large company database of IH measurements. Using an expert-based approach, we have developed a job-exposure matrix (JEM) with exposures for two sizes of particulate matter – total particulate matter (TPM) and PM2.5 – at facilities performing manufacturing operations as various as refining, smelting, and fabricating metal products.

Methods

The exposure assessment focused on PM exposure in 11 manufacturing facilities of a single aluminum manufacturing company in the United States (Table 1). Facilities were selected to encompass different manufacturing work processes throughout the company, from refining through fabrication. Of these 11 facilities, 1 is a refinery, 5 have smelters (all use the prebake technology), and 9 have fabrication units engaged in various processes, including rolling, extrusion, forging, and casting as well as lighter metalworking. Three facilities include both smelters and fabrication. Details of aluminum refining, smelting, and fabrication have been described elsewhere (12).

Within the company, industrial hygiene data have been collected for 60 years. Sampling conducted over the past 25 years has been compiled in an extensive industrial hygiene database, HYGenius (>300,000 samples). Samples were collected in each facility under the direction of certified industrial hygienists (CIH) and analyzed at an AIHA accredited IH laboratory (Clark Laboratories LLC, Jefferson Hills, PA). Samples were collected under one of the following three strategies: random, diagnostic, or worst case (as defined by the company). Random samples were meant to capture day-to-day regular work within the targeted job. Diagnostic and worst case samples were collected to answer specific questions about job exposures or to monitor exposures during specific tasks. The random samples form the basis for the JEM. However, sampling was generally targeted only for those jobs where 5% or more of the exposures were greater than 30% of the company's occupational exposure limit (OEL), of 10 mg/m3 throughout the company for the duration of sampling presented, as judged at each facility under the direction of the facility IH. In general, sampling was not performed for jobs where, after inspection by the facility IH, neither TPM nor any of the specific chemical exposures (e.g., fluoride, oil mist, metals) was over 30% of the OEL, as judged by the facility IH unless the toxicity of the agent of interest warranted sampling at lower exposure levels.

The HYGenius database contains detailed information including agent, purpose of sampling, duration of sampling, location of sampling (facility, department, job, task), whether personal or area sample, use and type of personal protective equipment, and sample result. The database contains over 100 agents of interest. Information in the database concerning particle mass concentration is limited to TPM and respirable particles (far fewer). Because each of the 11 facilities in the study was acquired by the company at different times, the dates of the earliest samples in HYGenius vary across facilities (Table 1).

The JEM was developed in the following five steps: standardization of job titles into distinct exposure groups (DEGs); categorization of DEGs into major manufacturing process categories; calculation of TPM from exposure from data in HYGenius; simultaneous PM2.5 and TPM measurement on a subset of workers at 8 facilities to determine the percent of TPM in each DEG that is composed of PM2.5 (%PM2.5); and, calculation of PM2.5 from the TPM and the percent PM2.5 in each DEG. Each of these steps is described in more detail below.

Creation of Distinct Exposure Groups (DEGs)

As is true in many workplaces, hundreds to thousands of job title/department combinations existed in HYGenius database for each facility, and these did not readily correspond to the job titles in the various human resources databases that track job changes for all employees. In order to reconstruct the job–exposure history of each individual in the epidemiologic study, jobs judged to have qualitatively and quantitatively similar exposures were aggregated into distinct exposure groups (DEGs) and mappings developed between the human resources databases and HYGenius.

A senior industrial hygiene manager (CDE) at the corporation led the aggregation effort. She created a team of site managers, industrial hygienists, and health and safety experts and worked closely with another researcher (LC). This team of experts began by defining the core work processes within the company. Within each core process and facility, the team created distinct exposure groups (DEGs) that aggregated jobs by department, job title, and job tasks based on similarity of the work performed. The DEGs were chosen by the team to be facility-specific, rather than pooled across facilities, because the organization of the tasks within similar jobs and departments was not always comparable at different facilities. The final step linked these DEGs to over 10,000 different human resources job titles contained in the human resources database across locations (24).

Categorization into major manufacturing process categories

In addition to the quantitative values for TPM and PM2.5 each DEG was assigned one of four qualitative major manufacturing process categories: smelting; fabricating; refining; or mixed smelting/fabrication (for DEGs in which an employee might work in either or both smelting and fabrication or be exposed to either operation type, e.g., electrician).

Generation of TPM Exposure for JEM

TPM sampling data from the HYGenius database was used to construct DEG-specific exposures for the JEM. Inclusion criteria for the TPM data were that samples had to be valid personal samples collected randomly (rather than as part of a specific diagnostic evaluation or as targeted worst case) for at least 70% of an employee's shift. We used only the random samples because they represent the day-to-day exposures of the workers, rather than specific events that diagnostic or worst case samples are designed to capture. TPM samples were collected using 37mm filters in traditional closed-face filter cassettes and analyzed gravimetrically (NIOSH Method 0500). Standard quality assurance methods were followed including calibrating pump flow before and after sampling, checking the integrity of the tubing and samplers, as well as following laboratory analysis protocols outlined in NIOSH 0500 (25). Samples analyzed prior to the first issue of NIOSH 0500 were analyzed in accordance with the NIOSH recommendations in the NIOSH Manual of Analytical Methods. The exposure metric in each cell was the arithmetic mean value of all the samples for each DEG. Because the exposure estimates are meant to capture annual average concentrations, arithmetic rather than geometric means are most appropriate for the JEM (26). However, there were 134 samples (approximately 1% of the total samples used, affecting a total of 42 DEGs) with extremely high values, >50 mg/m3. We considered three options for handling these extreme values in the JEM: include them without adjustment, omit them, or adjust them by some factor. In the table S1 (supplemental material) we show the comparison of these three methods for the 42 DEGs. Since these samples are valid measurements, we chose to adjust them by the respirator used during the sample collection as reported by the IH. Using the style and type of respirator, we applied the OSHA respirator protection factor (27) as an adjustment. The OSHA respirator protection factor varies from 10 to 10,000, depending on the type of the respirator. Another way to include all values without the high measurements overly influencing the average exposure is to use the geometric mean of all of the samples; the geometric mean of the unadjusted samples are included as part of the JEM as an alternative exposure metric.

If TPM samples were not available for a particular DEG, TPM samples from a similar DEG at the same or comparable facility were used. If information at the same or a comparable facility did not exist for a particular DEG, a default concentration of 0.10 mg/m3 was applied to the DEG cell of the JEM. We selected 0.10 mg/m3 as our default because it is higher than average environmental concentrations, in the lowest 5% of the TPM samples in our JEM, and is a simple number with one significant figure.

In order to preserve the information about the source of the data, each DEG-TPM cell in the JEM was assigned a ranking reflecting confidence in the source. The three categories of data sources for TPM were: measured data; surrogate measurements from a comparable DEG; default value when no other data were available. This information on data sources can be used in sensitivity analyses to examine the impact of potential exposure misclassification in an epidemiologic study and to guide future sampling.

Measurement of PM<sub>2.5</sub> and %PM<sub>2.5</sub>

Personal sampling for PM2.5 and TPM was conducted in 2010 and 2011 at 8 of the 11 facilities (Table 1); 3 facilities were not selected for sampling due to partial curtailment of operations at the facility or closure at the time of the monitoring campaign. This exposure monitoring campaign was designed both to measure personal exposures to PM2.5 and to derive the percent of TPM that is PM2.5 across jobs at these facilities. Two types of PM samplers were used to evaluate PM2.5 exposures: the traditional closed-face 37 mm cassettes (TPM) operated at 2 liters per minute and SKC Personal Modular Impactors (PMIs) operated at 3 liters per minute with 3 stages: > 10 μm; 2.5 −10 μm, and < 2.5 μm (PM2.5). These samplers were paired and worn simultaneously by each worker. Analysis of PMI filters was by NIOSH 0500, with the same procedures and quality controls listed above.

The percent of PM2.5 in the TPM samples was calculated for each sample by dividing the concentration of PM2.5 (from the PMI) by the concentration of paired TPM sample (cassette) in order to use the historical TPM cassette values to calculate the historical PM2.5 exposures in each DEG. The percentages within each DEG were averaged to generate the %PM2.5 for the JEM. For DEGs in which the percent of the total particles that are composed of fine particles (% PM2.5) was not measured, we used data from similar jobs at other facilities or jobs judged to have similar size distribution (measurements from comparable jobs). If no such comparable measurements were available, we estimated the % PM2.5 from an understanding of processes and associated particle size distribution; thus, some processes (e.g. welding, combustion) emit predominantly fine particles, while other processes (e.g. grinding) emit larger particles. Knowledge of the predominant source of particles in each job informed estimations of the % PM2.5 particles in TPM for the remaining DEGs: those in which the sources were predominantly fine particles were assigned 80% PM2.5, those in which the sources emit predominantly larger particles were assigned 20% PM2.5 and those with mixed sources, or unknown size distributions were assigned 50% PM2.5. In summary, the three methods used to assign %PM2.5: direct measurements in the DEG, measurements in comparable jobs, and estimations based on expert judgment. We did not estimate exposure at the three facilities that had no PM2.5 sampling.

Generation of PM<sub>2.5</sub> exposure concentrations for JEM

The average TPM concentration for each DEG was multiplied by the corresponding %PM2.5 to generate the PM2.5 concentration for the JEM. In order to preserve the information about the source of the data, ranked source codes were generated for PM2.5 values; these combined the rankings for the underlying TPM and %PM2.5 data. For the calculated PM2.5 concentration, we defined 5 ranks, with the highest (rank 1) defined as both TPM and %PM2.5 derived from sample measurements (TPM from HYGenius and %PM2.5 from 2010-2011 sampling campaign) in the given DEG, and the lowest (rank 5) where TPM was default value (regardless of %PM2.5 data source).

Influence of facility and DEG in PM exposure in smelters and fabrication facilities

To evaluate any increase in precision of the exposure estimates achieved by using facility-specific exposure groups, ie. DEGs, rather than exposure groups pooled across facilities, we systematically examined the sources of variability in the TPM measurement samples. To this end, we looked at the percent of total variance in TPM explained by facility and exposure group in a series of linear regression models. In models based on the measurement samples included in the development of the TPM exposure concentration estimate in the JEM, facility and exposure group were modeled as fixed effects. The coefficients of determination (r2) of models with each fixed effect alone were compared to that of a model with both fixed effects, and then to the full model with both fixed effects plus their interaction. The series of models were stratified by the two main manufacturing process categories – smelter and fabrication.

ResultsDEG creation and categorization into major manufacturing processes

In the 11 facilities in this study there were 2,780 unique job titles by department and facility in the industrial hygiene database and over 10,000 human resources job titles. These were reduced to 294 distinct exposure groups (DEGs). Of the 294 DEGs, 33% were assigned to smelting, 56% were assigned to fabricating, 7% to refining, and 4% to the mixed category of manufacturing processes (Table 1).

Generation of TPM Exposure for JEM

A total of 8,385 TPM personal samples were used to calculate TPM exposures for the DEGs. This represents 82% of personal TPM samples collected in DEGs of interest, excluding either specific diagnostic samples (15%) or worst case samples (3%). The TPM exposure estimates for most (210) the 294 DEGs in the JEM were calculated directly from TPM sample measurements, 55 were calculated from comparable DEGs, and 29 DEGs were given the default value (Table 2). Samples were collected from 1983 to 2011, with 50% collected from 2000-2011, 38% from 1990-1999, and 12% from 1983-1989. Approximately half of the TPM samples were collected in smelters (57%) and a third (36%) in fabrication units. Overall, TPM concentrations in smelters were higher than in fabrication units, with arithmetic means of 3.86 mg/m3 (SD 4.43 mg/m3) and 0.76 mg/m3 (SD 1.25 mg/m3), respectively (Figure 1); and geometric means of 1.63 mg/m3 (GSD 4.84) and 0.35 mg/m3 (GSD 3.22), respectively.

Measurement of PM<sub>2.5</sub> and %PM<sub>2.5</sub>

The PM2.5 personal sampling survey in 2010 and 2011 was conducted in 8 facilities; jobs in 2 facilities were resampled in a second season. There were 101 paired samples collected in smelter DEGs, 267 collected in fabrication DEGs, and 9 collected in refinery DEGs. The arithmetic mean PM2.5 concentration for all 377 paired PM2.5 and TPM personal samples was 0.50 mg/m3 (standard deviation: 0.92 mg/m3) and the geometric mean was 0.18 mg/m3 (geometric standard deviation: 4.27). The fabrication facilities had lower arithmetic mean PM2.5 concentrations than smelter or refinery facilities (0.21 mg/m3, 1.19 mg/m3, and 1.24 mg/m3 respectively) and geometric means (0.10 mg/m3, 0.73 mg/m3, and 0.54 mg/m3, respectively) (Figure 1). The percent of TPM that is PM2.5 (%PM2.5) was highly variable among the DEGs and ranged from 1% to 100% for all 377 paired samples; the interquartile range was 25% to 84%. Fabrication facilities had higher mean %PM2.5 compared to either smelter or refinery facilities (59%, 38%, 25%, respectively) (Figure 2). There was no significant seasonal difference in the observed PM2.5 concentrations or %PM2.5 when stratified by DEG.

The %PM2.5 values for a third of the 223 DEGs at these 8 facilities were directly measured during the 2010-2011 sampling campaign. An additional 48% of the DEGs were assigned values based on comparable measurements within the same facility or comparable facilities. Thus, the %PM2.5 values for 79% of the DEGs in the JEM were based on measurements, and 21% were based on more qualitative assessment.

Generation of PM<sub>2.5</sub> exposure concentrations for the JEM

PM2.5 exposure concentrations for each of 223 DEGs at the 8 facilities were derived by multiplying the TPM mean of the DEG by the corresponding %PM2.5 (Table 2). The TPM and PM2.5 exposures in the JEM are highly correlated, with a Spearman rank correlation coefficients of 0.93 in smelter DEGs and 0.82 in fabrication DEGs. Of the 223 DEGs at these 8 facilities, 30% were calculated directly using measured %PM2.5 and measured TPM; an additional 28% were calculated from TPM measured and %PM2.5 estimated. PM2.5 in DEGs with higher data source rankings had higher median PM2.5 concentrations (median PM2.5 in source ranks 1-5 is, in order, 0.29 mg/m3, 0.20 mg/m3, 0.19 mg/m3, 0.15 mg/m3, 0.04 mg/m3).

Influence of facility and DEG in PM exposure in smelters and fabrication facilities

The 7,531 samples used to calculate the TPM exposure for 187 measured facility-specific exposure groups (DEGs) in the JEM were used in linear models to evaluate the sources of variability in the sampling data. Exposure group explains more of the total variability than facility for both smelter and fabrication facility (Table 3). The full model, including the interaction (facility* exposure group), explained 27% of the variability in smelters and 36% in fabrication units. Because the r2 value increased 5% when the interaction term was added into the model, we conclude that there are facility-specific differences within TPM exposure groups. This finding corroborates the qualitative information that motivated the development of facility-specific exposure groups, i.e. DEGs, rather than pooling facilities within exposure groups.

Discussion

This paper presents a unique survey of personal exposures in aluminum manufacturing workers, specifically TPM and PM2.5. Few studies of particulate matter exposures in manufacturing have presented size distribution data or distinguished PM2.5 from TPM. Moreover, these measurement-based PM2.5 and TPM exposures were estimated across many parts of the industry, in contrast to previous studies, which focused on smelters. The exposure assessments for TPM and PM2.5 presented here have significant strengths. The TPM exposures in the JEM were based on 8,385 personal samples that were collected at 11 facilities and represent random full-shift exposures. The PM2.5 exposures were based additionally on 377 pairs of personal PM2.5 and TPM samples collected at 8 facilities. These are the first measured personal PM2.5 exposure data reported in this industry.

The occupational exposure limits set by both the company and OSHA are used as guidance for routine personal exposure sampling. Jobs that are likely to exceed 30% of the OEL at least 5% of the time are targeted for sampling. For TPM, with an OEL of 10 mg/m3, the 30% concentration is 3 mg/m3, however more than three quarters of the TPM samples in this study are under 3 mg/m3. This is because TPM was rarely the focus of sampling. It was collected when sampling for other contaminants (e.g. fluorides, metals), but this still did not generally capture low TPM concentrations. Thus in this study, there was less sampling of jobs with very low occupational exposures i.e. less than 0.150 mg/m3 (the highest environmental PM standard that USEPA ever issued) and therefore more uncertainty in the lower exposure estimates. This sampling strategy is reflected in the fact only 13% of TPM samples used in the TPM JEM were less than or equal to 0.150 mg/m3. Similarly, 21% of the PM2.5 samples used in the JEM were less than or equal to 0.035 mg/m3, the current USEPA daily PM2.5 standard. Although less important for industrial hygiene activities aimed at meeting OSHA regulations, this uncertainty may be important in epidemiologic studies that seek to distinguish risk among employees exposed to the lower end of the exposure range. This uncertainty was reflected in our data source rankings (based on the source of the exposure information, not the level of exposure), which indicated higher confidence in the higher exposure estimates.

The focus of previous TPM research in the aluminum industry has been on personal exposure in the potrooms. The personal exposures to TPM in smelters reported here are similar to those reported previously. Donohogue et al. (14) evaluated personal exposures to inhalable PM (similar to TPM) at 6 pre-bake smelters in Australia and New Zealand. The range of median values of the geometric mean exposure concentrations (mg/m3) from 1996-2006 was 2.17 – 4.50 mg/m3. This is comparable to the geometric mean TPM in our 5 pre-bake smelters of 1.63 mg/m3, with an interquartile range TPM of 0.60 – 4.48 mg/m3. Personal exposures to TPM as measured in 15 personal samples in a pre-bake potroom in Iran ranged from 0.1-5.90 mg/m3(16), which is also comparable to the range exposures we observed in our potrooms. Information on exposures in other departments in the smelters, fabrication units, refineries, and bauxite mines were unavailable in the literature. Information on the size distribution of particulate matter was limited to research on constituents in different size fractions in potrooms (17).

The three major limitations of this first assessment of PM2.5 exposure in aluminum manufacturing are lack of consideration of temporal trends, respirator usage, or constituents. Although the TPM measurements available for this study had been collected over a period of 30 years, there has been little change in the aluminum processes conducted in these facilities over the time period of interest, from early 1980s until present. There may, however, have been temporal changes in exposure across the company as a whole, as well as for particular processes. Changes in TPM and PM2.5 over the more recent past will be the subject of a subsequent, more formal, analysis. Second, we have not yet taken full advantage of information on respirator use. In this analysis we applied a respirator protection factor only to samples with extreme values (over 50mg/m3) with reported respirator use. A more thorough evaluation of reported respirator use will be forthcoming.

Third, this exposure assessment does not consider the composition of PM2.5, which is likely to be relevant to the toxicity of these exposures. Particles in the smelters are likely composed of inorganic materials, i.e. fluorides, alumina dust, metals and related fumes as well as coal tar pitch volatiles in some areas (13,18–21). The PM exposures in fabrication are predominantly water-based metalworking fluids and metals. The composition of both the TPM and PM2.5 fractions is clearly an important aspect of the personal exposures to these individuals. Analyses of the constituent exposures in each DEG are underway to develop a JEM for chemical-specific exposures.

Despite these limitations, the exposure assessment for PM2.5 presented in this report reflects a thorough examination of thousands of particle samples and contributes to our knowledge about the distribution of particle exposures in the US aluminum manufacturing industry. The ultimate objective of the exposure assessment described here was to provide the basis for an exposure-response analysis in an epidemiologic cohort study. Figure 3 presents the daily dose (mg) from a range of familiar sources of PM2.5, using a conversion method for transforming mg/m3 into units of daily dose (mg) recommended by Pope, et. al. to compare various epidemiologic studies of PM (8,28). Results from this study indicate that the range of PM2.5 exposures within the US aluminum manufacturing industry fill the important gap in PM2.5 intake identified by Pope between environmental air pollution and active smoking. The highest exposures in our study were equivalent to a daily PM2.5 dose slightly greater than actively smoking 1.5 cigarettes/day, although the dose rates and composition would obviously be quite different.

In conclusion, we have presented information on exposure to two fractions of particulate matter – TPM and PM2.5 – in 11 aluminum manufacturing facilities with different manufacturing operations. As anticipated, occupational exposures exceeded environmental PM and PM2.5 levels by an order of magnitude in most jobs. Additionally, both TPM and PM2.5 are highest in smelters and both vary significantly by distinct exposure group, even within the same facility. These differences underscore the importance of understanding the roles that different processes and sources may play in the PM exposure profile for aluminum workers.

Supplementary MaterialAcknowledgements

We wish to acknowledge the contribution of the following people: the Alcoa industrial hygienists and employees who participated in the sampling campaigns at each facility and Linda Maillet, Judi Caldwell Kuntz, Regi Jennings, Ralph Krobot, Bob Barr, and Jill Abston.

Funding: This work was supported by National Institutes of Health, Institute of Aging (5R01 AG026291-06 Disease, Disability and Death in an Aging Workforce: The Alcoa Study), Center for Disease Control and Prevention, National Institute of Occupational Safety and Health (5R01OH009939-02: Occupational Exposure to PM2.5 and Cardiovascular Disease), and by Alcoa, Inc.

This is the pre-publication version of a manuscript that has been accepted for publication. This version does not include post-acceptance editing and formatting. Readers who wish to access the definitive published version of this manuscript and any ancillary material related to it (eg, correspondence, corrections, editorials, etc) may do so at http://www.nature.com/jes/index.html or go to the print issue in which the article appears. Those who cite this manuscript should cite the published version, as it is the official version of record.

The citation for this paper is:

Noth EM, Dixon-ernst C, Liu S, et al. Development of a job-exposure matrix for exposure to total and fine particulate matter in the aluminum industry. J Expo Sci Environ Epidemiol. 2013;24(1):89-99.

Conflict of Interests: Dr. Noth, Dr. Liu, and Dr. Eisen declare no potential for conflict of interest. Ms. Dixon-Ernst is a senior industrial hygienist for Alcoa, Inc. Ms Cantley and Ms. Tessier-Sherman receive salary support from Alcoa, Inc. through contracts with Yale University. Dr. Cullen receives salary support from Alcoa, Inc. through contracts with Stanford University. Dr. Hammond has received compensation as a member of the scientific advisory board for Alcoa, Inc. She has also consulted for Alcoa, Inc. and received compensation.

NIA Data Sharing: As an alternative to providing a de-identified data set to the public domain, we allow access for the purpose of re-analyses or appropriate “follow-on” analyses by any qualified investigator willing to sign a contractual covenant with the host Institution limiting use of data to a specific agreed upon purpose and observing the same restrictions as are limited in our contract with Alcoa, such as 60-day manuscript review for compliance purposes.

ReferencesBrookRDRajagopalanSPopeCABrookJRBhatnagarADiez-Roux AVParticulate Matter Air Pollution and Cardiovascular Disease. An Update to the Scientific Statement From the American Heart Association.Circulation1052010DockeryDWPopeCAXuXSpenglerJDWareJHFayMEAn association between air pollution and mortality in six U.S. cities.The New England journal of medicine912199332924175398179653GanWQKoehoornMDaviesHWDemersP aTamburicLBrauerMLong-term exposure to traffic-related air pollution and the risk of coronary heart disease hospitalization and mortality.Environmental health perspectives420111194501721081301LawMRMorrisJKWaldNJEnvironmental tobacco smoke exposure and ischaemic heart disease: an evaluation of the evidence.BMJ (Clinical research ed.)18101997315711497380MillsNLDonaldsonKHadokePWBoonN aMacNeeWCasseeFRAdverse cardiovascular effects of air pollution. Nature clinical practice.Cardiovascular medicine1200961364419029991PetersADockeryDWMullerJEMittlemanMAIncreased particulate air pollution and the triggering of myocardial infarction.Circulation1262001103232810511401937PetersALiuEVerrierRLSchwartzJGoldDRMittlemanMAir pollution and incidence of cardiac arrhythmia.Epidemiology (Cambridge, Mass.)12000111117PopeCABurnettRTKrewskiDJerrettMShiYCalleEECardiovascular mortality and exposure to airborne fine particulate matter and cigarette smoke: shape of the exposure-response relationship.Circulation159200912011941819720932SmithKRPeelJLMind the gap. Environmental health perspectives.1220101181216435U.S. Department of Health and Human ServicesThe Health Consequences of Smoking: A Report of the Surgeon General2004U.S. Department of Health and Human ServicesThe Health Consequences of Smoking: Cardiovascular Disease.A Report of the Surgeon General1983BenkeGAbramsonMSimMExposures in the alumina and primary aluminium industry: an historical review.Annals of Occupational Hygiene1998423173899684558FritschiLSimMRForbesAAbramsonMJBenkeGMuskAWRespiratory symptoms and lung-function changes with exposure to five substances in aluminium smelters.International archives of occupational and environmental health2003761031012733082DonoghueAMFrischNIsonMWalpoleGCapilRCurlCOccupational Asthma in the Aluminum Smelters of Australia and New Zealand : 1991 – 2006.American Journal of Industrial Medicine2010SimMBenkeGWorld at work: Hazards and controls in aluminium potrooms.Annals of Occupational Hygiene2003601298993Akbar-khanzadehFExposure to Particulates and Fluorides and Respiratory Health of Workers in an Aluminum Production Potroom with Limited Control Measures.American Industrial Hygiene Association Journal20105610100815WeinbruchSBenkerNKochWEbertMDrabløsPASkaugsetNPHygroscopic properties of the workroom aerosol in aluminium smelter potrooms : a case for transport of HF and SO2 into the lower airways.Journal of Environmental Monitoring2010124485420145885KongerudJBoeJSùysethVNaalsundaMagnusPAluminium potroom asthma: the Norwegian experience.European Respiratory Journal11199471165728143817SundermanFWReview : Nasal Toxicity , Carcinogenicity, and Olfactory Uptake of Metals.Annals of Clinical & Laboratory Science2001311ArmstrongBGGibbsGExposure-response relationship between lung cancer and polycyclic aromatic hydrocarbons (PAHs).Occupational and environmental medicine1220096611740619546103ArmstrongBGTremblayCGCyrDThériaultGPThsriaultGPEstimating the relationship between exposure to tar volatiles and the incidence of bladder Estimating the relationship between exposure to tar volatiles and the incidence of bladder cancer in aluminum smelter workers.Scandinavian journal of work, environment & health198612548693ThomassenYKochWDunkhorstWEllingsenDGSkaugsetNJordbekkenLUltrafine particles at workplaces of a primary aluminium smelterw.Journal of Environmental Monitoring200681273316395469HöflichBLWWeinbruchSTheissmannRGorzawskiHEbertMOrtnerHMCharacterization of individual aerosol particles in workroom air of aluminium smelter potrooms.Journal of environmental monitoring52005754192415877161CullenMRVegsoSCantleyLGalushaDRabinowitzPTaiwoOUse of medical insurance claims data for occupational health research.Journal of occupational and environmental medicine / American College of Occupational and Environmental Medicine102006481010546117033505NIOSH Method of Analytical MethodsMethods1994Cincinnati, Ohio4th editionSeixasNSRobinsTGMoultonLHThe use of geometric and arithmetic mean exposures in occupational epidemiology. American journal of industrial medicine.1198814446577OSHAAssigned Protection Factors for the Revised Respiratory Protection Standard1200994AdamsWCMeasurement of breathing rate and volume in routinely performed daily activities1993185Sacramento, CA

Box-and-whiskers plot of the overall facility-wide distribution of the arithmetic mean (a) TPM (mg/m3) and (b) PM2.5 (mg/m3) by DEG in the 3 predominant aluminum manufacturing work processes (smelter, fabrication (note: abbreviated as Fab), and refinery), plotted with a lognormal scale. The bottom and top edges of the box indicate the intra-quartile range. The diamond inside the box indicates the mean concentration, the line inside the box indicates the median concentration.

Box-and-whiskers plot of the overall facility-wide distribution of mean %PM2.5 by DEGS in the 3 predominant types of DEGs for which %PM2.5 was determined (smelter, fabrication (note: abbreviated as Fab), and refinery), in the aluminum industry. The bottom and top edges of the box indicate the intra-quartile range. The diamond inside the box indicates the mean concentration, the line inside the box indicates the median concentration.

Distribution of exposure to estimated daily PM2.5 in different settings

Distribution of exposures to estimated daily PM2.5 (mg daily dose equivalent, calculated using 10 m3/day as the average daily breathing rate for a single-shift worker (Adams 1993)). White columns are environmental and SHS (second-hand smoke) exposures, black are aluminum manufacturing exposures, and black and white columns are active smoking exposures. Figure based on data from Pope, et. al. (2009).

The 11 facilities in the study that span the aluminum manufacturing industry in the United States from refining to fabrication.

FacilityTotal # of DEGs# Smelter DEGs# Fabrication DEGs# Smelter/Fabrication DEGs# Refinery DEGsYears of historical industrial hygiene exposure data used in TPM JEMSampled PM2.5 in 2010-2011
A33208501984-2011-
B16160001999-2008-
C382111601980-2011Yes
D22220001981-2008-
E30199201991-2011Yes
F17017001986-2011Yes
G22022001988-2010Yes
H40040002002-2011Yes
I21021001985-2011Yes
J36036002002-2011Yes
K19000191983-2010Yes

JEM of arithmetic and geometric mean TPM and PM2.5 exposure concentrations estimates (mg/m3) for all DEGs derived from random personal exposure samples TPM, in the aluminum industry. See footnote for explanation of combined data sources.

Distinct exposure groupsfacility id# TPM samplesTPM AM TPM STD TPM GM TPM GSD crude TPM GM* crude TPM GSD* %PM2.5PM2.5Combined data source
DEGs in smelters
Anode Assembly OperatorA1145.62 11.1 2.54 3.20 2.54 3.20
Anode Assembly OperatorB166.98 3.25 4.60 5.38 4.60 5.38
Anode Assembly OperatorC51.54 0.99 1.31 1.89 1.31 1.89 80%1.233
Anode Assembly OperatorD1232.01 2.51 1.50 2.08 1.50 2.08
Anode Assembly OperatorE191.61 2.22 0.95 2.62 1.07 3.82 80%1.293
Anode ChangerA969.32 12.0 5.35 3.09 5.35 3.09
Anode ChangerC4113.1 13.3 8.93 2.36 8.93 2.36 25%3.301
Anode ChangerD2907.58 9.11 4.62 3.18 4.73 3.28
Anode ChangerE1446.19 8.80 3.30 4.09 3.41 4.26 32%1.971
Baked Anode Furnace RepairerC815.18 10.4 2.62 2.92 2.62 2.92 72%3.762
Baked Anode Furnace RepairerE614.64 9.54 1.65 4.00 1.65 4.00 72%3.361
Baked Anode Furnace RepairersB121.35 0.55 1.24 1.57 1.24 1.57
Baked Anode OperatorA201.00 1.38 0.60 2.55 0.60 2.55
Baked Anode OperatorB582.34 7.41 0.71 3.10 0.71 3.10
Baked Anode OperatorC1982.36 2.88 1.44 2.89 1.44 2.89 31%0.721
Baked Anode OperatorD1371.59 2.17 0.80 3.53 0.82 3.85
Baked Anode OperatorE350.64 0.59 0.45 2.27 0.45 2.27 79%0.511
Caster Furnace OperatorA811.0 22.3 1.34 9.37 1.34 9.37
Caster Furnace OperatorD1230.91 2.24 0.42 3.24 0.42 3.24
Caster Furnace OperatorE353.36 7.80 0.89 4.53 0.89 4.53 31%1.042
Caster OperatorB130.39 0.38 0.29 2.17 0.29 2.17
Caster OperatorE211.83 1.87 1.41 1.92 1.41 1.92 56%1.032
Crane OperatorC1181.90 1.74 1.39 2.39 1.39 2.39 49%0.942
Crane OperatorD2032.65 8.81 0.76 4.69 0.76 4.69
Crane OperatorE998.08 17.9 1.78 7.26 2.04 9.28 49%3.981
Electrical MaintenanceA110.37 0.32 0.24 3.19 0.24 3.19
Electrical MaintenanceD210.59 0.74 0.40 2.56 0.40 2.56
Electrical MaintenanceE713.38 10.6 0.42 5.95 0.45 7.17 43%1.472
Facilities and Grounds OperatorsD140.63 0.91 0.26 4.03 0.26 4.03
Fume Control ServicerA104.87 3.92 3.16 3.08 5.00 6.31
Fume Control ServicerB347.96 9.65 3.94 3.56 3.94 3.56
Fume Control ServicerC196.02 8.80 2.35 4.23 2.35 4.23 41%2.481
Fume Control ServicerD15110.4 21.4 3.50 4.82 3.96 4.96
Fume Control ServicerE7812.0 21.3 4.07 5.13 7.15 9.69 41%4.952
Furnace OperatorA10.22.0.22.0.22
Furnace RepairerA392.73 3.03 1.78 2.57 1.89 3.04
Green Anode OperatorA851.55 2.49 0.62 3.94 0.65 4.50
Green Anode OperatorB161.08 1.92 0.44 3.32 0.44 3.32
Green Anode OperatorC593.27 4.08 1.69 3.93 1.69 3.93 26%0.841
Green Anode OperatorD1551.06 1.31 0.66 2.66 0.66 2.66
Green Anode OperatorE10812.4 35.9 1.86 7.30 2.87 11.1 37%4.581
Laboratory OperatorE60.49 0.34 0.41 1.88 0.41 1.88 72%0.352
MachinistB10.130.130.13
MachinistD80.22 0.08 0.21 1.47 0.21 1.47
Mechanical MaintenanceB471.12 1.00 0.85 2.07 0.85 2.07
Mechanical MaintenanceD1124.10 24.5 0.86 3.72 0.86 3.72
Mechanical MaintenanceE1665.97 23.6 0.73 6.44 0.80 8.09 43%2.592
Mobile Equipment OperatorA1614.70 17.6 1.00 4.83 1.01 5.02
Mobile Equipment OperatorB82.67 5.35 0.78 4.29 0.78 4.29
Mobile Equipment OperatorC193.02 3.52 1.63 3.12 1.63 3.12 64%1.922
Mobile Equipment OperatorD2589.83 39.5 3.69 4.16 3.75 4.24
Mobile Equipment OperatorE919.31 33.4 1.64 5.16 1.82 6.55 56%5.231
Potline RepairerA254.78 9.11 2.36 2.69 2.36 2.69
Potline RepairerB32.28 1.42 2.02 1.79 2.02 1.79
Potline RepairerC873.06 6.96 1.32 3.50 1.32 3.50 19%0.581
Potline RepairerD1034.23 17.7 1.33 3.19 13.6 3.51
Potline RepairerE7713.6 67.0 2.68 5.05 2.91 5.75 47%6.421
Potroom Bake In OperatorA180.80 0.66 0.49 4.06 0.49 4.06
Potroom Bake In OperatorC411.34 1.05 0.99 2.41 0.99 2.41 50%0.674
Potroom OperatorA383.08 2.86 1.63 3.90 1.63 3.90
Potroom OperatorB546.94 5.47 4.81 2.70 4.81 2.70
Potroom OperatorC172.14 1.21 1.85 1.75 1.85 1.75 28%0.601
Potroom OperatorD2644.92 6.82 2.72 3.02 2.72 3.02
Potroom OperatorE1014.73 6.14 2.40 3.33 2.51 3.66 41%1.961
Potroom Service OperatorA4915.0 52.6 2.37 4.78 2.37 4.78
Potroom Service OperatorC8712.3 24.7 3.99 4.26 3.99 4.26 42%5.201
Potroom Service OperatorD181.35 0.71 1.18 1.74 1.18 1.74
Power Plant MaintenanceD10.020.020.02
Power Plant OperatorC913.3 17.4 5.63 4.64 8.05 5.52 20%2.663
Power Plant OperatorD80.24 0.25 0.09 5.90 0.09 5.90
Production SupervisorA490.65 2.61 0.11 5.32 0.11 5.32
Production SupervisorD602.24 5.04 0.77 4.67 0.77 4.67
Raw Material OperatorE194.73 8.64 1.25 5.83 1.41 7.37 64%3.012
Reclamation Furnace OperatorA9511.9 58.0 1.29 5.17 1.40 6.44
TapperC294.63 9.52 2.44 2.59 2.44 2.59 38%1.751
TapperE5224.7 98.1 5.25 3.49 5.49 4.16 38%9.382
Utility ServicerA195.18 8.00 2.57 3.38 2.90 4.13
Utility ServicerB226.60 4.72 4.99 2.27 5.54 2.78
Utility ServicerC10.390.390.3950%0.204
Utility ServicerE70.22 0.12 0.20 1.66 0.20 1.66 50%0.114
DEGs in mixed smelter and fabrication
Electrical MaintenanceC210.79 0.81 0.55 2.29 0.55 2.29 43%0.342
Facilities and Grounds OperatorC11.071.071.0750%0.544
Facilities and Grounds OperatorsA40.31 0.13 0.28 1.84 0.28 1.84
MachinistA270.82 1.19 0.32 4.32 0.32 4.32
MachinistC60.33 0.19 0.28 1.91 0.28 1.91 43%0.142
MachinistE30.17 0.09 0.14 2.12 0.14 2.12 43%0.072
Mechanical MaintenanceA1192.96 7.51 0.86 4.99 0.88 5.25
Mechanical MaintenanceC554.13 6.44 1.64 4.32 1.64 4.32 43%1.792
Power House OperatorA49.35 17.5 1.64 7.83 1.64 7.83
Railroad EngineerA60.09 0.05 0.07 2.67 0.07 2.67
DEGs in fabrication
AdministrationJ10.200.200.2020%0.043
Air Melt OperatorH190.72 0.76 0.44 2.83 0.44 2.83 30%0.222
Alloy Control OperatorH200.43 0.40 0.35 1.79 0.35 1.79 30%0.131
Axis GrinderJ550.76 1.71 0.36 3.11 0.36 3.11 69%0.521
Caster Furnace OperatorC471.07 1.06 0.79 2.19 0.79 2.19 31%0.332
Caster Furnace OperatorG220.96 1.22 0.70 2.04 0.70 2.04 22%0.211
Caster Furnace OperatorI153.82 9.12 1.03 4.60 1.03 4.60 54%2.071
Caster OperatorH320.41 0.64 0.26 2.47 0.26 2.47 31%0.132
Caster OperatorI181.43 0.76 1.24 1.76 1.24 1.76 14%0.191
Caster OperatorJ190.22 0.18 0.16 2.78 0.16 2.78 31%0.072
Charge Prep OperatorH30.36 0.06 0.36 1.17 0.36 1.17 64%0.232
Chip and Trim OperatorF522.98 5.32 0.74 7.01 0.74 7.01 55%1.632
Coater OperatorE140.07 0.05 0.06 2.24 0.06 2.24 47%0.042
Cold Mill Oil AttendantG10.070.070.0743%0.032
Cold Mill OperatorA40.12 0.05 0.11 1.64 0.11 1.64
Cold Mill OperatorE430.15 0.11 0.12 1.88 0.12 1.88 47%0.072
Cold Mill OperatorG260.29 0.62 0.17 2.37 0.17 2.37 47%0.141
Crane OperatorA1003.84 15.4 1.18 3.47 1.18 3.47
Crane OperatorF50.64 0.19 0.62 1.38 0.62 1.38 100%0.641
Crane OperatorI90.19 0.23 0.11 3.33 0.11 3.33 94%0.181
Crucible Manufacturing OperatorJ30.28 0.07 0.28 1.26 0.28 1.26 50%0.143
Cut-OffH491.40 4.14 0.46 3.97 0.46 3.97 37%0.521
Cut-OffJ350.88 1.46 0.40 3.75 0.40 3.75 55%0.481
CVD Furnace OperatorJ70.39 0.51 0.17 4.12 0.17 4.12 75%0.292
Die OperatorC50.72 0.94 0.44 2.69 0.44 2.69 43%0.312
Dip OperatorJ90.14 0.12 0.09 3.44 0.09 3.44 50%0.074
Draw Bench OperatorI120.14 0.13 0.10 2.35 0.10 2.35 75%0.111
Electrical MaintenanceF230.39 0.30 0.30 2.23 0.30 2.23 43%0.172
Electrical MaintenanceG110.15 0.19 0.08 3.39 0.08 3.39 43%0.062
Extrusion Press OperatorI390.21 0.20 0.16 2.28 0.16 2.28 75%0.161
Facilities and Grounds OperatorG150.79 0.74 0.51 2.76 0.51 2.76 50%0.394
Final Finish OperatorH130.34 0.46 0.17 3.48 0.17 3.48 48%0.161
Foil Mill OperatorG270.27 0.70 0.13 2.77 0.13 2.77 47%0.132
Forge Press OperatorF2420.40 0.45 0.29 2.23 0.29 2.23 55%0.221
FPI OperatorH40.07 0.02 0.07 1.26 0.07 1.26 100%0.071
FPI OperatorJ30.11 0.01 0.48 1.02 0.48 1.02 30%0.031
Furnace OperatorF340.32 0.28 0.25 1.97 0.25 1.97 91%0.291
Furnace OperatorI340.28 0.23 0.19 2.60 0.19 2.60 67%0.191
Furnace OperatorJ480.22 0.36 0.13 2.67 0.13 2.67 91%0.202
Furnace RepairerG236.92 10.7 2.60 4.24 2.60 4.24 72%5.022
Hand GrindingJ200.87 0.84 0.54 3.20 0.54 3.20 69%0.601
Hot Mill Furnace OperatorE30.17 0.07 0.17 1.43 0.17 1.43 91%0.162
Hot Mill Oil AttendantA90.09 0.07 0.06 2.90 0.06 2.90
Hot Mill Oil AttendantE40.11 0.04 0.10 1.39 0.10 1.39 69%0.072
Hot Mill OperatorA50.74 0.46 0.51 3.43 0.51 3.43
Hot Mill OperatorE570.15 0.17 0.10 2.78 0.10 2.78 69%0.111
Hot Mill OperatorG130.36 0.27 0.28 2.12 0.28 2.12 69%0.252
InjectionJ90.16.0.15 1.47 0.15 1.47 100%0.161
Inspection OperatorF141.41 3.40 0.38 4.27 0.38 4.27 72%1.012
Inspection OperatorH100.15 0.06 0.13 1.84 0.13 1.84 55%0.081
Inspection OperatorI81.23 2.66 0.33 4.46 0.33 4.46 72%0.882
Inspection OperatorJ100.16 0.12 0.42 1.18 0.42 1.18 88%0.141
Inspection/Pack/ Ship OperatorF51.67 2.38 0.70 4.32 0.70 4.32 64%1.062
Laboratory OperatorG120.16 0.14 0.09 3.71 0.09 3.71 72%0.112
Laboratory OperatorH10.110.110.1172%0.082
Laboratory OperatorI10.280.310.3164%0.181
MachinistF140.56 0.62 0.41 2.05 0.41 2.05 43%0.242
MachinistF500.56 1.03 0.35 2.19 0.35 2.19 42%0.231
MachinistF381.27 3.16 0.37 4.55 0.37 4.55 52%0.661
MachinistG80.37 0.26 0.29 2.19 0.29 2.19 43%0.162
Mechanical MaintenanceH30.16 0.06 0.16 1.48 0.16 1.48 43%0.072
Mechanical MaintenanceH181.36 3.61 0.29 6.79 0.29 6.79 43%0.592
Mechanical MaintenanceI142.96 6.12 1.19 3.41 1.19 3.41 93%2.751
Mechanical MaintenanceJ60.29 0.28 0.20 2.57 0.20 2.57 43%0.122
Metal Cell OperatorH240.22 0.20 0.14 2.64 0.14 2.64 80%0.171
Metal Cell OperatorJ570.45 1.97 0.15 2.92 0.15 2.92 36%0.161
Metal Control OperatorJ110.56 0.71 0.26 3.75 0.26 3.75 80%0.452
Metal WeigherH50.43 0.15 0.41 1.39 0.41 1.39 50%0.221
Mobile Equipment OperatorG900.61 0.87 0.34 2.82 0.34 2.82 61%0.371
Mobile Equipment OperatorI90.26 0.29 0.18 2.34 0.18 2.34 100%0.261
Monoshell OperatorH222.86 4.32 1.16 4.53 1.16 4.53 20%0.581
Monoshell OperatorJ191.85 5.28 0.42 4.98 0.42 4.98 72%1.341
Pack/Ship OperatorA40.11 0.07 0.07 3.72 0.07 3.72
Plating TechnicianJ170.07 0.04 0.06 1.98 0.06 1.98 100%0.071
Powder Prep TechnicianJ143.34 10.6 0.57 4.18 0.57 4.18 13%0.431
Production SupervisorG30.13 0.04 0.12 1.38 0.12 1.38 47%0.062
Production SupervisorH30.12 0.14 0.07 3.22 0.07 3.22 80%0.102
Production SupervisorJ30.70 0.24 0.67 1.40 0.67 1.40 57%0.402
Raw Material OperatorC190.31 0.25 0.23 2.22 0.23 2.22 64%0.202
Repair OperatorF391.03 3.58 0.40 2.61 0.40 2.61 45%0.471
Repair OperatorI300.34 0.55 0.20 2.72 0.20 2.72 57%0.191
Repair OperatorI990.17 0.14 0.12 2.52 0.12 2.52 53%0.091
Roll Service OperatorA200.32 0.57 0.19 2.40 0.19 2.40
Roll Service OperatorE120.10 0.08 0.08 1.82 0.08 1.82 43%0.042
Roll Service OperatorG110.11 0.05 0.09 2.02 0.09 2.02 43%0.052
SandblastH70.44 0.47 0.33 1.99 0.33 1.99 36%0.161
SandblastJ50.18.0.12 8.70 0.12 8.70 62%0.111
Saw OperatorC100.22 0.15 0.18 1.88 0.18 1.88 57%0.122
Saw OperatorF170.43 0.33 0.34 1.92 0.34 1.92 56%0.241
Sheet/Plate Mill OperatorG590.35 0.70 0.18 2.62 0.18 2.62 35%0.121
Slitter OperatorA20.17 0.01 0.17 1.09 0.17 1.09
Slitter OperatorE130.16 0.23 0.09 2.92 0.09 2.92 47%0.072
StraightnerH30.14 0.03 0.14 1.31 0.14 1.31 80%0.112
Strander OperatorC10.220.220.2269%0.152
Thermatech OperatorJ40.11 0.01 0.11 1.13 0.11 1.13 100%0.112
Tube Press OperatorI240.08 0.09 0.06 2.51 0.06 2.51 70%0.061
Vacuum Furnace OperatorH197.24 8.44 2.03 8.71 2.92 10.6 11%0.831
Vacuum Furnace OperatorJ60.06 0.06 0.04 2.85 0.04 2.85 11%0.012
Wastewater Treatment OperatorH30.06 0.02 0.06 1.42 0.06 1.42 50%0.034
WaterblastJ50.67 0.80 0.71 1.46 0.71 1.46 21%0.141
Wax Cell OperatorJ100.15 0.08 0.38 1.54 0.38 1.54 96%0.141
WelderC105.14 7.16 1.70 6.24 1.70 6.24 20%1.031
WelderF72.97 3.27 1.27 6.08 1.27 6.08 43%1.292
WelderG1021.04 4.22 0.31 3.37 0.31 3.37 43%0.452
WelderH20.570.36 1.68 0.36 1.68 35%0.201
Wire Coil OperatorC60.40 0.62 0.21 2.96 0.21 2.96 69%0.272
Wire Draw OperatorC330.29 0.11 0.26 1.50 0.26 1.50 69%0.202
DEGS in refining
Calcination OperatorK292.97 5.76 1.03 3.80 1.11 4.46 17%0.501
Chemical OperatorK391.38 1.36 0.83 3.69 0.83 3.69 20%0.283
Clarification OperatorK150.60 0.55 0.43 2.30 0.43 2.30 20%0.123
Digestion Heater/Cleaner RepairerK110.46 0.15 0.43 1.51 0.43 1.51 21%0.102
Digestion OperatorK120.75 0.55 0.47 3.40 0.47 3.40 20%0.151
Electrical MaintenanceK160.43 0.39 0.27 2.93 0.27 2.93 43%0.182
MachinistK70.39 0.55 0.24 2.56 0.24 2.56 43%0.172
Mechanical MaintenanceK681.63 2.75 0.69 3.52 0.69 3.52 34%0.561
Mobile Equipment OperatorK20.12 0.04 0.11 1.37 0.11 1.37 64%0.072
Mud Tank CleanerK60.38 0.12 0.37 1.36 0.37 1.36 20%0.083
Precipitation OperatorK242.31 4.12 0.67 4.59 0.67 4.59 21%0.492
Raw Material OperatorK594.65 8.80 1.38 4.91 1.49 5.54 16%0.761
Utility ServicerK247.31 9.08 3.05 4.47 3.05 4.47 50%3.664

Note: Combined data source for PM2.5: 1=both TPM and %PM2.5 measured (30% of PM2.5 DEGs); 2= TPM measured and %PM2.5 estimated (28% of PM2.5 DEGs); 3=both TPM and %PM2.5 estimated from same or comparable facilities (22% of PM2.5 DEGs); 4=both TPM and %oPM2.5 estimated from expert judgment (8% of PM2.5 DEGs). Not shown are DEGs with rank 5 for which TPM was given a default value and %PM2.5 was estimated (12% of PM2.5 DEGs).

TPM concentrations without adjustment for respirator use.

Percent of total variance in TPM (modeled as log-transformed TPM) explained by facility and exposure group in a series of linear regression models, stratified by work process type in the aluminum industry.

Smelter (N=5,172)Fabrication (N= 2,359)
ModelModel DFModel p-value r2 adjusted r2AICmodel DFModel p-value r2 adjusted r2AIC
Facility only40.00010.0040.00497378<0.00010.100.103382
Exposure group only27<0.00010.200.19868160<0.00010.270.252993
Facility + exposure group31<0.00010.210.20862368<0.00010.310.292891
Facility + exposure group + Facility * exposure group79<0.0010.270.268303103<0.00010.360.332785