Industry-wide review of potential worker exposure to 1,3-butadiene during chemical manufacturing and processing as a reactant

Abstract

Among the first 20 high-priority chemical substances selected by USEPA to undergo risk evaluation as part of the Toxic Substances Control Act, as amended by the Frank R. Lautenberg Chemical Safety for the 21^st Century Act of 2016 is 1,3-butadiene (1,3-BD). Because much of the literature related to occupational exposure to 1,3-BD is associated with the use of the substance in synthetic rubber production and few data have been published for exposures to 1,3-BD manufacturing workers, existing industrial hygiene data collected at facilities where the substance is manufactured or processed as a reactant were compiled and analyzed. The dataset was comprised of personal air samples collected between 2010 and 2019 at facilities located throughout the United States and was compiled into a single database using a uniform data collection template. Data designated by the companies as full-shift were stratified by job group and one of three operational conditions of the workplace: routine, turnaround, and non-routine. Data designated by the companies as short-term and task-level were stratified by task description, sample duration, and operational condition. The final aggregated database contained a total of 5,676 full-shift personal samples. Mean concentrations of 1,3-BD for the job groups ranged from 0.012 ppm to 0.16 ppm. High-end estimates of 1,3-BD air concentrations for the job groups under routine operations ranged from 0.014 ppm to 0.23 ppm. The aggregated database also included 1,063 short-term and task-level personal samples. For short-term samples (< =15 min), mean concentrations ranged from 0.49 ppm to 3.9 ppm, with the highest concentrations observed for the cleaning and maintaining equipment tasks. For task samples with durations greater than 15 min, mean concentrations ranged from 0.49 to 3.6 ppm, with the highest concentrations observed for the unloading and loading task. In addition to the personal air sampling records, information on the use of PPE during various tasks was compiled and analyzed. This data set provides robust quantitative air concentration data and exposure control information for which occupational exposures to 1,3-BD in the Manufacturing and Processing as a Reactant condition of use can be assessed.

Keywords:

• 1,3-butadiene
• manufacturing
• risk
• risk evaluation
• worker

Introduction

The Toxic Substances Control Act (TSCA), as amended by the Frank R. Lautenberg Chemical Safety for the 21^st Century Act of 2016 (amended TSCA), authorizes the U.S. Environmental Protection Agency (EPA) to conduct risk evaluations on existing chemicals in commerce to determine whether a chemical presents an unreasonable risk of injury to health or the environment under the conditions of use (COUs). EPA defines the COUs as the circumstances under which a chemical substance is intended, known, or reasonably foreseen to be manufactured, processed, distributed in commerce, used, or disposed of (15 U.S.C. 2602(4)). Further, the amended TSCA specifically requires the Agency to evaluate the risk to “potentially exposed or susceptible subpopulations,” the definition of which includes workers (15 U.S.C. 2605(b)(4)(A)).

As part of the risk assessment, EPA conducts a quantitative assessment of worker and occupational non-user (ONU) exposure through the following six steps:

1. Review reasonably available exposure monitoring data for a specific COU, including what is available through the Occupational Safety and Health Administration (OSHA) and the National Institute for Occupational Safety and Health (NIOSH), as well as in the published literature.

2. Review reasonably available data for surrogate chemicals that have uses, volatility, and physical and chemical properties like the chemical substance under evaluation.

3. Review existing exposure models that may apply to estimating exposure levels if measured data are limited or not reasonably available.

4. Review reasonably available data to use in developing, adapting, or applying exposure models.

5. Map or group each COU to an occupational exposure assessment scenario.

6. Evaluate the weight of scientific evidence of occupational exposure data (U.S. EPA 2020).

EPA’s systematic review protocol describes how the agency will collect and evaluate data for risk evaluation, as well as their preference for using personal air monitoring data to characterize worker exposures under the relevant COUs (U.S. EPA 2021). EPA will assess the quality of the monitoring data across seven metrics: sampling and analytical methods, geographic scope, applicability to population or receptor, year(s) that samples were collected, sample size, metadata to inform job groups, tasks, exposure frequency, and duration as well as statistical analysis to characterize the variability and uncertainty associated with the data. Therefore, access to high-quality industrial hygiene (IH) data such as those collected in compliance with OSHA standards and using the American Industrial Hygiene Association’s (AIHA®) Exposure Assessment Strategies will be essential to EPA in conducting their risk assessments. Nevertheless, EPA’s chemical risk evaluation process differs from current occupational risk assessments, and the differences necessitate careful interpretation of the IH data, placing the data into the context of the worker and ONU exposure scenarios (i.e., exposure time and frequency), as well as the use of engineering controls and personal protective equipment (PPE).

Some key factors that can result in differences in occupational risk assessments performed by different organizations include the methods to determine exposure levels and the incorporation of different acceptable excess cancer risk probabilities (Deveau et al. 2015). In conducting the risk evaluations, EPA does not rely on existing occupational exposure limits (OELs) as a point of comparison to determine the potential for risk to workers and ONUs, rather the agency evaluates the toxicity and epidemiology data independently and derives their own reference values. Additionally, the definition of acceptable cancer risk for workers differs between EPA and OSHA; whereby EPA’s TSCA Assessments for workers and NIOSH’s policies of acceptable excess cancer is 1 in 10,000, and OSHA permissible exposure limits (PEL) for carcinogens have historically been based on an acceptable excess risk of 1 in 1,000. For these reasons, worker risk assessments conducted using the OSHA, NIOSH, or AIHA frameworks by some industrial hygienists may reach conclusions different from those reached by EPA via the TSCA Risk Evaluation.

Among the first 20 high-priority (HP) chemical substances selected by EPA to undergo risk evaluation is 1,3-butadiene (1,3-BD) (U.S. EPA 2020). 1,3-BD was prioritized for risk evaluation in part because of its acute and chronic health hazard potential for humans. Acute health effects include irritation to the eyes, throat, nose, and lungs and at high concentrations may cause central nervous system effects; skin exposure to liquified 1,3-BD can result in frostbite. 1,3-BD is designated as a Category 1 carcinogen by the International Agency for Research on Cancer (IARC) and some organizations also designate it as a reproductive toxicant (U.S. EPA 2019).

1,3-BD is a gas at ambient temperature and pressure; it is produced commercially by three processes: (1) steam cracking of paraffinic hydrocarbons (ethylene co-product process), (2) catalytic dehydrogenation of n-butane and n-butene (Houdry process), and (3) oxidative dehydrogenation of n-butene (Oxo-D or O-X-D process). These manufacturing processes are performed in closed systems (many of which operate at high pressure and low temperature). Each of these processes produces a stream, commonly referred to as crude butadiene, that is rich in 1,3-BD. Separation and purification of the butadiene stream typically are carried out by extractive distillation because the boiling points of the various C4 components are so close to each other. The final concentration in the purified 1,3-BD product is typically >99 wt% pure and is stored in liquid form in pressurized spheres. Liquefied 1,3-BD is shipped by pipelines, ships, barges, rail tank cars, tank trucks, and bulk liquid containers to industrial customers who use it as a reactant or ingredient. It is important to note that the storage and transport conditions are such that engineering controls must be implemented to minimize losses and ensure safe operations.

Worker exposure to 1,3-BD is regulated by OSHA under 29 CFR 1910.1051, which obligates employers, including manufacturers and processors, to determine the airborne concentration of 1,3-BD to which employees may be exposed, evaluate those exposures in comparison to the OSHA PEL of 1 ppm as an 8-hr time-weighted average (TWA) or short-term exposure limit (STEL) of 5 ppm as a 15-minute TWA, and reduce exposures, if the PEL is exceeded. A non-regulatory occupational exposure limit of 2 ppm as an 8-hr TWA has been established by the American Conference of Governmental Industrial Hygienists (ACGIH®), and similar to the OSHA PEL, it is based on cancer as the critical health effect (ACGIH 2022). However, the ACGIH does not develop TLVs® associated with specific levels of cancer risk and hence has resulted in a different OEL than OSHA (ACGIH 2020).

The OSHA 1,3-BD standard also obligates manufacturers and processors to implement a variety of exposure controls to eliminate or reduce the potential for worker exposure. These controls include regulated areas (allowing only authorized and trained personnel to have access to operations where air concentrations of 1,3-BD may exceed the PEL), engineering controls to reduce the potential for exposure, respirators when engineering controls are insufficient to reduce 1,3-BD exposures to below the PELs, and dermal protection to prevent eye contact and dermal exposure to 1,3-BD.

Much of the literature related to occupational exposure to 1,3-BD is associated with use of the substance in synthetic rubber production (Checkoway and Williams 1982; Fajen 1988; Fajen et al. 1990; Macaluso et al. 1996; Sorsa et al. 1996; Sathiakumar et al. 2007) and few data have been published for exposures to 1,3-BD manufacturing workers (CONCAWE 2000; Strandberg et al. 2014; Akerstrom et al. 2016; Almerud et al. 2017; Scarselli et al. 2017). The objective of the data collection and analysis provided herein was to evaluate the magnitude of air concentrations of 1,3-BD to which workers and ONUs may be exposed at the industry level during the manufacturing and processing (as a reactant) of 1,3-BD in the U.S. Additionally, information regarding the workplace exposure scenarios, such as engineering controls, PPE use, and typical exposure frequency and duration, are summarized to provide context for the measured air concentrations. The data were provided to authors J.P., L.M., and K.F. by the American Chemistry Council’s 1,3-Butadiene TSCA Risk Evaluation Consortium (Consortium).

Methods

Data collection

Ten Consortium member companies provided 1,3-BD occupational exposure data from 47 sites located in the U.S. The Consortium members represent 100% of the 1,3-BD manufacturing industry in the U.S., as well as 28% of others who process 1,3-BD as a reactant (ACC 2022). The data set comes from sites in the U.S. that have NAICS codes including 324110—Petroleum refineries, 325110—Petrochemical manufacturing, 325199—Other organic chemical manufacturing, 325210—Resin and synthetic rubber manufacturing, 325211—Plastic and resin manufacturing, 325991—Custom compounding of purchased resins, and 325998—All other chemical product and preparation manufacturing.

The data sets were generated based on the AIHA Exposure Assessment Strategy (EAS) guidance (AIHA 2015). The personal air samples collected between 2010 and 2019 at facilities located throughout the United States were compiled into a single database using a uniform data collection template developed in accordance with OECD guidelines and after discussions with various stakeholders (OECD 2021). The components of this template are shown in Figure 1. Two analytical methods were used to quantify 1,3-BD air concentrations: NIOSH 1024 (NMAM 1994) and OSHA 56 (OSHA 1985). Both methods involve the collection of an air sample onto a solid sorbent medium, followed by desorption using a solvent and analysis using gas chromatography (GC) coupled with flame ionization detection (FID). Passive sampling badges (e.g., 3 M 3520, 3530) were also used to collect personal air samples and have been demonstrated to meet OSHA’s accuracy requirements (33M 1996). All samples included in the Consortium data set were analyzed by AIHA-accredited laboratories.

Figure 1. Components of exposure data collection template.

Data quality review, aggregation, and stratification

On receipt of member-company data, the authors (J.P., K.F., L.M) reviewed each file for completeness and accuracy. Specific attention was focused on the verification of analytical methods and detection limits, sample durations, job categories, task descriptions, operational status, work-area descriptions, sample dates, and PPE. Duplicates, samples without values for 1,3-BD, and samples rejected by the respective member company industrial hygienists as being invalid or not representative of occupational exposure were excluded from the aggregated data set. Three types of samples may have been rejected by the company's industrial hygienists upon conferring with them on their datasets. The first type was samples where the company notes indicated that the sample was associated with emergency response actions. In this case, the company industrial hygienist was asked to verify whether it was associated with emergency response, and if so, it was excluded from the analysis. The second type was short-term/task samples that lacked a description of the task or work. In this case, the company industrial hygienist was asked to research the sample and determine whether there was a task associated with it, and if not, it was excluded from the analysis. The third type was air samples that were described as area source samples; potentially indicative of air samples used to characterize a process leak/emission rather than a worker exposure air concentration. For these samples, the industrial hygienist was asked to research the sample description and determine whether the sample was representative of potential worker exposures, and if not, it was excluded from the analysis. The excluded samples represented 1.8% of the total number of data points.

Following review and clarification, company-specific identifiers were eliminated to maintain confidentiality, and then the data were combined into one database. Additional data quality factors were assessed using the EPA’s approach for determining the quality of measured occupational exposure data for use in chemical risk evaluations (U.S. EPA 2021) as described previously.

Data designated by the companies as full-shift (with sampling durations greater than four hours) were stratified by job group and one of three operational conditions of the workplace: routine, turnaround, and non-routine, as shown in Figure 2. Each job group and the activities carried out by workers that may expose them to 1,3-BD are provided in the supplementary materials (Table S1). All full-shift personal air concentrations were used as-is in the analysis and were not transformed into 8-hr TWA concentrations, because the member companies indicated that the data were representative of full-shift exposures.

Figure 2. Full-shift job groups and workplace operational conditions.

Data designated by the companies as short-term and task-level were stratified by task description, sample duration, and one of the three operational conditions of the workplace, as shown in Figure 3. Data were available for each of the tasks identified by EPA in the Final Scope Document (U.S. EPA 2020), except for repackaging chemicals, formulations, or products containing 1,3-BD, because that task is not conducted in the manufacturing and processing (as a reactant) of 1,3-BD. It is important to note that, even though some tasks are conducted as part of routine operations, the frequency of the task varies and may be daily, weekly, monthly, quarterly, or even annually. Therefore, these kinds of data are used to make decisions on the appropriate PPE for any of the given tasks but should not be used to characterize long-term exposure without knowledge of the task frequency.

Figure 3. Short-term task designations, durations, and operational conditions.

Descriptions of each task, along with job-group mapping to task and exposure controls utilized while performing each task, are provided in the Supplementary Materials (Table S2). Air samples collected to characterize potential exposures during tasks include those for OSHA compliance monitoring with the STEL (1 to 15 min in duration) or task-level exposure characterization (16 to 240 min in duration). Generally, the compliance STEL samples are collected during periods of expected maximum or peak concentrations and compared to the OSHA STEL (5 ppm), which is the legal maximum average exposure for a 15-min period. The data associated with the durations of 16 to 240 min are representative of the average exposure over the total duration of the task. The task-level data are used to identify specific tasks that may contribute to a worker’s full-shift exposure throughout the workday and to target specific exposure controls and/or PPE selection to minimize exposure.

Data analysis

Data were analyzed using the statistical computing platform, R, and the analysis packages: NADA (Non-detects and Data Analysis for Environmental Data) and tidyverse (Wickham et al. 2019; Lee 2020; R Core Team 2021). For each stratified data set, summary statistics were calculated using the Kaplan-Meier (KM) method for left-censored data (Helsel 2012). KM is a nonparametric technique (i.e., data are not ascribed to a particular underlying distribution) that can handle data with multiple levels of left censoring (i.e., detection limits). The technique defines an empirical cumulative distribution function (CDF) in the shape of a step function that increases at each detected observation and is flat at each left-censored value. The area under the CDF defines the KM mean. Additionally, general statistics were computed (i.e., minimum; maximum; and 50^th, 90^th, and 95^th percentiles; and the proportion of non-detects reported between specific concentrations) using a standard substitution method of one-half the detection limit for non-detected samples.

The complete set of statistics (including the alternative analyses) for full-shift (stratified by job group and operational condition) and short-term (stratified by task, operational condition, and sampling time) personal air samples can be found in Supplementary Materials (Table S3). The results presented herein focus on KM methods to accommodate the left-censored data with various levels of detection frequencies and detection limits.

Results

The aggregated data set included 6,739 personal air-sample data points from which to characterize potential worker exposure to 1,3-BD under the Manufacturing and Processing as a Reactant COU. The data analysis revealed highly censored data sets for all job groups and tasks. For example, in the routine operations stratification, the full-shift data sets showed 69%–94% non-detect values, and the short-term task data sets showed 45%–100% non-detect values. An analysis of the detection limits associated with each full-shift sample indicated that they were less than 10% of the current OSHA PEL of 1 ppm and, therefore, were sufficiently low to have confidence in the levels of exposure that can be inferred.

The designation of “routine” for operational status reflects exposures that are likely to occur regularly over the long-term working lifetime for a particular job. Short-term samples designated as routine characterize peak exposures that may occur regularly, for short durations. Therefore, results for routine full-shift personal exposures for workers and ONUs, as well as routine short-term task-specific personal exposures, are presented below. Personal air concentration statistics for the other operational conditions (i.e., turnaround and non-routine conditions) and the alternative statistical approach are provided in Supplementary Materials (Table S3).

Full-shift personal air concentration data

The final aggregated database contained a total of 5,676 full-shift personal samples (workers and ONUs). Of the full-shift samples, 3,949 were collected under routine operations, 49 during non-routine conditions, and 1,678 during turnaround activities. A summary of the full-shift personal air sample results for routine operations, stratified by job group and operation type, is provided in Table 1. These data are most representative of potential chronic exposures incurred by workers in the Manufacturing and Processing as a Reactant COU. Central estimates of 1,3-BD air concentrations for the job groups, characterized by the KM-mean, ranged from 0.012 ppm to 0.16 ppm, with the lowest personal air concentration associated with ONU and the highest mean concentration observed for Safety Health and Engineering (SHE) workers. Use of the alternative statistical approach yields arithmetic means that are similar to or higher than the KM means, and geometric mean values that are lower than the KM means. High-end estimates of 1,3-BD air concentrations for the job groups under routine operations ranged from 0.014 ppm to 0.23 ppm, based on the 95^th upper confidence level on the mean (95 UCL Mean). It should be noted that the air concentrations reported in Table 1 do not necessarily represent the amount of 1,3-BD to which workers are actually exposed, because they do not account for the reduction in exposure from the use of respirators and dermal PPE (i.e., chemical-protective gloves and suit/coveralls) that are required for certain tasks.

Table 1. Summary of full-shift personal air concentrations—Routine operations.

Job Group	N Samples	% Non- Detects	% DL < 0.1 ppm	Full-Shift Personal Air Concentrations (ppm)—Kaplan Meier Statistics
				Min	50th	90th	95th	KM-Mean	SE	95LCL Mean	95UCL Mean	Max
Infrastructure/Distribution Operations	455	78%	72%	0.006	NA	0.21	0.45	0.12	0.038	0.045	0.19	16.4
Instrument and Electrical	313	91%	63%	0.008	NA	0.021	0.16	0.068	0.033	0.003	0.13	10.0
Laboratory Technician	215	73%	86%	0.006	NA	0.12	0.25	0.063	0.016	0.031	0.094	2.93
Machinery and Specialists Group	222	80%	97%	0.008	NA	0.060	0.28	0.087	0.023	0.042	0.13	3.31
Maintenance	354	69%	46%	0.001	NA	0.23	0.24	0.11	0.010	0.089	0.13	2.10
Occupational Non-User	39	77%	100%	0.008	NA	0.013	0.033	0.012	0.001	0.010	0.014	0.038
Operations Onsite	1952	88%	85%	0.0001	0.001	0.037	0.19	0.074	0.016	0.043	0.11	16.0
Safety Health and Engineering	21	71%	100%	0.038	NA	0.19	0.36	0.16	0.036	0.087	0.23	0.78
Missing Job Group Designation	378	94%	91%	0.002	NA	NA	0.037	0.024	0.004	0.016	0.032	1.3

Short-term and task personal air concentration data

The data set included 1,063 short-term and task-level personal samples. Of these samples, 1,001 were collected under routine conditions. The percentage of non-detects by task code and sample duration ranged from 45% to 100%. The distribution of detection limits by sample duration for the various task categories is provided in Table 2. For short-term and task-level samples the detection limits were generally less than 0.5 ppm (i.e., 10% of STEL) ranging from 54% to 95% of non-detected samples for the various tasks.

Table 2. Distributions of detection limits for short-term and task samples by sample duration.

Task	N Non-Detects	DL [<0.1 ppm]	DL [0.1–<0.5 ppm]	DL [0.5–<1 ppm]	DL [1–<5 ppm]	DL [> = 5 ppm]
Sample Duration < =15 minutes
Unloading and Loading	40	10%	55%	13%	23%	0%
Handling of Waste	7	14%	71%	0%	14%	0%
Cleaning and Maintaining Equipment	81	1%	65%	25%	7%	1%
Sampling and Analysis	167	17%	37%	37%	10%	0%
Performing Other Work	22	0%	82%	14%	5%	0%
Sample Duration >15 minutes
Unloading and Loading	71	11%	54%	11%	24%	0%
Handling of Waste	9	22%	33%	22%	22%	0%
Cleaning and Maintaining Equipment	109	19%	50%	15%	16%	1%
Sampling and Analysis	210	15%	62%	22%	0%	0%
Performing Other Work	17	24%	71%	0%	6%	0%

A summary of personal air concentrations measured during various routine short-term and task-level sampling events is provided in Table 3. For short-term samples (< =15 min), KM-mean concentrations ranged from 0.49 ppm to 3.9 ppm, with the highest concentrations observed for the cleaning and maintaining equipment tasks. High-end estimates for these short-term samples, characterized by the 95 (UCL) Mean, include 3.7, 6.7, 0.59, and 0.92 ppm for the unloading and loading, cleaning, and maintaining equipment, sample collection and analysis, and other tasks (e.g., routine rounds, tank gauging, opening process equipment), respectively. Central and high-end estimates for the handling waste task are not available (NA), because all samples for this task were below detection limits.

Table 3. Summary of short-term and task personal air concentrations—Routine operations.

Task Code	Sample Duration (minutes)	N Samples	% Non-Detects	Short-Term Personal Air Concentrations (ppm)—Kaplan Meier Statistics
				Min	50th	90th	95th	KM-Mean	SE	95LCL Mean	95UCL Mean	Max
Unloading & Loading	< =15	89	45	0.04	1.0	8.1	17	2.7	0.52	1.7	3.7	26
	>15	158	45	0.02	0.73	8.0	18	3.6	0.77	2.1	5.1	89
Handling Waste	< =15	7	100	0.06	NA	NA	NA	NA	NA	NA	NA	1.6
	>15	10	90	0.08	NA	0.69	0.69	0.69	NA	NA	NA	3.7
Cleaning & Maintaining Equipment	< =15	102	80	0.06	NA	2.7	15	3.9	1.4	1.1	6.7	120
	>15	159	73	0.02	0.06	2.6	8.3	1.8	0.77	0.28	3.3	110
Sample Collection & Analysis	< =15	187	89	0.03	NA	0.51	1.3	0.52	0.04	0.44	0.59	4.8
	>15	237	89	0.02	NA	0.36	2.10	0.49	0.12	0.25	0.74	18.0
Performing Other Tasks	< =15	31	71	0.2	0.20	0.54	2.1	0.49	0.22	0.06	0.92	6.5
	>15	21	81	0.02	NA	0.17	3.7	0.49	0.31	−0.11	1.1	4.7

For task samples with durations greater than 15 min, KM-mean concentrations ranged from 0.49 to 3.6 ppm, with the highest central estimate observed for the unloading and loading task. High-end estimates for task samples, characterized by the 95 UCL Mean, include 5.1, 3.3, 0.74, and 1.1 ppm for the unloading and loading, cleaning, and maintaining equipment, sample collection and analysis, and other tasks, respectively.

A summary of records in the database that simultaneously reported short-term, task-level personal air concentration data, along with PPE usage, is shown in Table 4. For each task and respirator use scenario (i.e., supplied air, full-face air-purifying respirator [APR], half-face APR, and no respirator), a wide range of personal air concentrations were observed. For example, workers used supplied-air respirator systems when performing cleaning and maintaining equipment tasks, and personal air exposure levels to 1,3-BD ranged from as low as 0.15 ppm to as high as 120 ppm. This demonstrates that respiratory protection is specified in anticipation of the potential for exposures over the OSHA STEL and not just when it is known that the task involves exposures greater than the STEL.

Table 4. Summary respiratory protection used during task activities.

Task	1,3-BD Workplace air concentration ranges (ppm) reported with respirator use
	Supplied Air	Full-Face APR	Half-Face APR	No Respirator
Unloading & Loading	<0.118–89	<0.06–36	<0.05–2.2	–
Handling Waste	–	<0.25–<3.7	<0.08–<0.1	–
Cleaning & Maintaining Equipment	<0.15–120	<0.02–110	<0.04–<0.7	<0.4–<0.7
Sampling Collection & Analysis	<0.52	<0.06–12	<0.09–7.3	<0.02–4.8
Performing Other Tasks	0.27–4.7	<0.24–<0.42	<0.2–<0.3	<0.39–<0.67

Note: APR = air-purifying respirator.

Discussion

The Consortium datasets were generated based on the AIHA EAS guidance which includes the use of qualitative assessments to prioritize exposure monitoring, development of similar exposure groups to allow for the collection of representative samples, and sample program design to ensure sufficient sample size for decision analysis-based on the collected data (AIHA 2015). Based on the exposure assessment strategies used by Consortium members, this data set represents air concentrations of 1,3-BD for workers who are more likely than not to experience exposure, rather than being representative of potential exposure to the whole worker population at a given facility. In addition to characterizing exposures during routine operations, industrial hygienists target worst-case exposure scenarios such as during turnaround operations (a planned maintenance shutdown of a processing unit) or process upsets (unplanned maintenance). These conditions occur infrequently and may not have engineering controls in place to minimize exposures. Thus, monitoring these exposure conditions allows for the selection of proper PPE or other control measures.

The Consortium data set provides robust quantitative air concentration data for which occupational exposures to 1,3-BD in the Manufacturing and Processing as a Reactant COUs can be assessed. The central-tendency measures of the data demonstrate that typical full-shift air concentrations to which workers in these COUs are potentially exposed are at levels one order of magnitude lower than the OSHA PEL 8-hr TWA of 1 ppm. Further, even the high-end of the air concentrations (i.e., 95^th percentile) is lower than the OSHA PEL and less than the average air concentration of 1.66 ppm that was determined by Valdez-Flores et al. (2022) to not exceed a 1-in-10,000 theoretical excess cancer risk over a 45-year work duration and 85-year lifetime. Additionally, the air concentrations in this data set are similar to those reported from European 1,3-BD manufacturing operations (CONCAWE 2000; Akerstrom et al. 2016; Almerud et al. 2017; Scarselli et al. 2017). It is recognized that since the Consortium dataset is representative of only 28% of workers in the Processing as a Reactant COU, additional data would be necessary to fully characterize 1,3-BD air concentrations in that COU, and in particular, those concentrations associated with synthetic rubber production.

OSHA’s Chemical Exposure Health Database contains results from the agency’s air sampling at various U.S. facilities, including industrial operations at plastics facilities, rubber and tire facilities, ornamental and architectural metal work manufacturing, and shelving manufacturing, where the 1,3-BD air concentrations have been largely non-detected (OSHA 2021). Similar to the 1,3-BD data in OSHA’s Chemical Exposure Health Database all of the full-shift data sets for the job groups in the Consortium data set are highly censored, meaning that 1,3-BD was not detected in most of the samples. This demonstrates the low exposure potential for workers in the Manufacturing and Processing as a Reactant COUs and the effectiveness of engineering and operational controls to minimize worker exposures without considering the use of respirators. Furthermore, the upper-bound values of the detection limits provide additional confidence that the non-detected values reflect very low exposure potential. In general, the NIOSH and OSHA methods for chemical substances are developed to provide a result that is accurate over a concentration range of 0.1 to 2 times the occupational exposure limit (OSHA 2010; NIOSH 2015) and, as such, are fit for purpose in terms of worker risk assessment. A non-detected full-shift sample value for 1,3-BD may represent an actual concentration within the range of 0–0.099 ppm. Challenges in interpreting occupational exposure data, particularly those that include a large number of non-detected values, arise when comparing these data to other exposure limits or health benchmarks that are lower than the OELs. Given the differences in the approaches used by various organizations and agencies to derive health benchmarks (Deveau et al. 2015; U.S. EPA 2020), it is possible that unacceptable health risk determinations could be made for workers currently assessed to be unexposed or in the low exposure category, based on available industrial hygiene data; and if so, analytical method detection limits will need to be lowered to recharacterize those workers who are currently considered unexposed.

The designation of “routine” for operational status reflects exposures that are likely to occur regularly over the long term. Therefore full-shift and short-term samples designated as routine are most useful for characterizing occupational exposures that may occur over a working lifetime, or to characterize peak exposures that may occur regularly, for short durations during a work week, respectively. It is important to note, however, that some routine task activities are not carried out on a daily/weekly basis, such as filter change-out, clearing/venting/opening process equipment, etc. These activities, though scheduled routinely, are conducted infrequently (monthly, quarterly, or annually). These infrequent tasks are characterized as part of the exposure assessment program, and administrative controls or PPE are implemented to reduce exposures. Given the variability in frequency and the challenges associated with achieving sufficiently low detection limits for short-term samples, respiratory protection is generally required, as demonstrated by the Consortium data and in compliance with the 1,3-BD OSHA Standard.

EPA’s evaluation of ONUs is meant to ensure that all workers, regardless of exposure potential, are evaluated. The agency has defined ONUs as workers who do not directly handle the chemical but perform work in an area where the chemical is present (U.S. EPA 2020). For the 1,3-BD manufacturing and processing industry, the ONUs are supervisory personnel associated with all of the worker job groups. The OSHA standard requires employers to establish regulated areas to restrict access to areas or operations where air concentrations of 1,3-BD may exceed the PEL (i.e., 8-hr TWA or STEL) to only authorized and trained personnel. Therefore, manufacturers and processors are required to anticipate, designate, and restrict entry to these areas. This required control significantly reduces the opportunity for 1,3-BD exposures by ONUs in the manufacturing and processing as a reactant COU; supervisory personnel are the only potentially exposed ONUs because administrative-type employees (e.g., accountants, salespersons, etc.) do not enter the operational parts of a facility. It is important to note that the use of AIHA’s exposure assessment strategy guidance typically results in a low-risk rating for ONUs. Therefore, most industrial hygiene programs do not prioritize exposure measurements for ONUs, and when measurements are made, it is generally to confirm the low potential for exposure. The Consortium database reflects the implementation of both the OSHA requirement for regulated areas and AIHA prioritization guidance, in that it contains very few personal air samples (N = 39) collected from workers who may be considered ONUs (i.e., supervisors). As expected, this job group had the lowest full-shift exposures.

The Consortium data set is substantial concerning both the coverage of potentially exposed workers by job group and the various tasks that they perform. The data are relatively recent (i.e., 2010 to 2019) and are specific to U.S. 1,3-BD manufacturing and processing as a reactant COU. The data set includes documentation of the sampling and analytical methods used to collect and analyze the samples, as well as operating conditions at the time that the samples were collected. Additionally, information regarding the use of respiratory protection to minimize worker exposures is also available for the task activities.

In using this data set, it is important to keep in mind that the data are inherently biased toward those activities or worker groups who are considered to have potential exposures, to determine the need for PPE and/or engineering controls, and also to assess the effectiveness of those controls. As such, the data are representative of potential exposure for those workers who are more likely than not to incur exposures. Job groups are aggregated across companies that submitted the data, but because the different companies handle process streams with low or high concentrations of 1,3-BD, the variability in the data is much more than would exist at any given facility. Additionally, job groups are aggregated by similar names across facilities, but their levels of exposure may be different, depending on the site-specific activities. At an individual facility, the job groups better represent similar exposure groups (SEGs) and have smaller geometric standard deviations. As such, the data presented herein may be biased high because the tail of the distribution is longer than what would be reflected for an individual site. Nevertheless, the standard errors (0.001–0.038) for the full-shift samples within each job group demonstrate that, quantitatively, there is good precision around the mean exposure concentrations. Similarly, tasks are aggregated across industries and job groups, and while production operators at some facilities might carry out certain frequent maintenance activities (i.e., filter change-outs), at other facilities, maintenance operators might be involved in opening process equipment.

It should be noted that sample duration is not necessarily representative of the duration of the task. In some instances, the part of the task that leads to exposure is short, but it might be impractical to start/stop the sampler at those times. On the other hand, the industrial hygienist might focus on the specific aspect of the full task that leads to the exposure (e.g., just the actual connection/disconnection of hoses during truck loading), which might bias the estimated exposure for the whole task when there is no exposure. Finally, the database does not provide enough information to determine, for a particular task, how often the task occurs at facilities, and of those occurrences, how often they are characterized by the high-end exposure value.

The focus of this analysis has been on the inhalation pathway; however, EPA has also included dermal exposure as a relevant, albeit minor, pathway for the risk evaluation. The potential dermal exposure of certain workers who may contact liquid streams with trace amounts of 1,3-BD has not been assessed quantitatively; however, streams with trace amounts of BD are likely to be hydrocarbon mixtures. Safe practices in the workplace require the use of dermal protection to prevent contact with hydrocarbon mixtures. The use of gloves that are resistant to hydrocarbons would provide sufficient protection for low concentrations of BD.

A variety of statistical values may be used to characterize occupational exposure potential; however, for chronic exposures, a central tendency value such as the mean is typically the best descriptor over a working lifetime. EPA’s risk evaluations are typically done assuming chronic exposures that are consistent from day to day, with infrequent spikes. Therefore, the statistical values from the upper percentiles are unlikely to represent chronic occupational exposure. The reasonable maximal exposure (RME) (i.e., the highest exposure that could reasonably be expected to occur) is often evaluated to account for both uncertainties in the contaminant concentration and variability in exposure parameters such as exposure frequency (U.S. EPA 1992). The concentration term used in a theoretical dose calculation for the RME is the 95^th UCL of the mean (U.S. EPA 1992). In this data set, for the full-shift routine samples, the KM 95^th UCL of the mean is up to two-fold higher than the KM mean and two-fold lower than the 95^th percentile. The European Chemical Agency (ECHA) also cautions risk assessors about using upper-percentile values to characterize occupational exposures in guidance provided to chemical manufacturers who register chemicals under the REACH program. In this guidance, ECHA highlights the need to apply occupational hygiene expertise, rather than applying traditional environmental risk assessment conventions or the use of rigid statistical methods (ECHA 2009).

Conclusion

This dataset fills the gap of publicly available data regarding the air concentrations of 1,3-BD to which workers may be exposed during the manufacturing and processing (as reactant) of 1,3-BD in the United States. The mean and 95^th percentile air concentrations associated with full-shift exposures are lower than OSHA’s 8-hr TWA PEL of 1 ppm, OSHA’s action level of 0.5 ppm as an 8-hr TWA, and other existing or proposed OELs. The mean air concentrations associated with the short-term and task samples are lower than the OSHA STEL of 5 ppm, however at the 95^th percentile, air concentrations associated with unloading/loading and the cleaning and maintaining equipment tasks may exceed the OSHA STEL, such that respiratory protection is often required by the companies, even if precautionary for some sites.

Occupational health professionals have been collecting exposure data in the workplace for many decades, to assess and characterize potential exposures, determine risk management measures, and verify the effectiveness of such controls. The data presented here are of high quality because they were collected using best practices for quality assurance and quality control, as required by OSHA and recommended by NIOSH and AIHA, including using AIHA-certified labs. At the industry level, the air concentrations and associated jobs and work tasks reported in this analysis provide a robust dataset for informing health risks to workers. It is recommended that any future, industry-wide exposure assessment should investigate and characterize the exposure frequency and duration for the various job tasks.

Acknowledgments

The authors sincerely thank the Consortium members for providing the data used in this analysis and for their assistance during the QA/QC review of the data set.

Disclosure statement

The American Chemistry Council’s 1,3-Butadiene TSCA Risk Evaluation Consortium is comprised of companies that manufacture or process 1,3-butadiene (1,3-BD) and was formed to develop relevant data to support USEPA’s risk evaluation of 1,3-butadiene. The Consortium members are prevented from seeing each other’s data and as such, authors J.P., L.M., and K.F. were retained by the consortium to serve in an independent third-party role to compile the individual company industrial hygiene datasets and provide an analysis of the data across the Manufacturing and Processing as a Reactant conditions of use.

Additional information

Funding

The data analysis and preparation of this manuscript were funded by the Consortium.