Skip to main content
NCBI home page
Annals of Work Exposures and Health logoLink to Annals of Work Exposures and Health
. 2018 Jun 13;62(7):840–851. doi: 10.1093/annweh/wxy042

An Evaluation of the Relationship among Urine, Air, and Hand Measures of Exposure to Bisphenol A (BPA) in US Manufacturing Workers

Cynthia J Hines 1,✉, Annette L Christianson 1, Matthew V Jackson 2, Xiaoyun Ye 3, Jack R Pretty 1, James E Arnold 1, Antonia M Calafat 3
PMCID: PMC6093326  PMID: 29905773

Abstract

Background

Exposure to bisphenol A (BPA) can be assessed using external and internal exposure measures. We examined the relationship between two measures of external BPA exposure (air and hand-wipe samples) and one of internal exposure (total BPA in urine) for a group of US manufacturing workers.

Methods

During 2013–2014, we recruited 78 workers from six US companies that made BPA or made products with BPA. We quantified BPA in seven urine samples, two full-shift air samples and in pre- and end-shift hand-wipe samples collected from workers over 2 consecutive days. We examined correlations between creatinine-corrected urinary concentrations of total BPA (total BPACR) and BPA levels in air and hand wipes using Pearson’s correlation coefficient. We also applied mixed-effects regression models to examine the relationship between total BPACR with BPA in air (urine~air model) and with BPA in end-shift hand wipes (urine~hand model), separately and together (urine~air+hand model), after adjusting for covariates.

Results

End-shift total BPACR strongly correlated with BPA in air (rp = 0.79, P < 0.0001) and nearly as strongly with BPA in end-shift hand wipes (rp = 0.75, P < 0.0001). In mixed-effect models, BPA air concentration and end-shift hand-wipe BPA level were significantly and positively associated with end-shift total BPACR (P < 0.0001 each). We found a significant effect of the Day 1 BPA air concentration on Day 2 total BPACR (P = 0.0104). When BPA air concentration and end-shift hand-wipe BPA level were in the same model, the air concentration (P < 0.0001) was more significant than the hand-wipe level (P = 0.0106).

Conclusion

BPA levels in air and end-shift hand wipes strongly correlated with total BPACR, suggesting that both inhalation and dermal contract were likely exposure routes; however, inhalation, on average, appeared to be a more dominant exposure route than dermal contact for these manufacturing workers.

Keywords: air sampling, biological monitoring, bisphenol A, dermal exposure assessment, exposure assessment, exposure assessment – mixed models, manufacturing, urine

Introduction

Bisphenol A (BPA) (CAS 80-05-7, 4,4ʹ-isopro pylidenediphenol) is used extensively as a monomer in the production of polycarbonate, epoxy, and phenolic resins (Kopf, 2003; Pham and Marks, 2004; Brunelle, 2014). BPA is also used as a filler in certain investment casting waxes (Carney, 2014) and as a developer in thermal paper although the latter use has declined as substitutes for BPA have been introduced (USEPA, 2014).

After ingestion, BPA is rapidly conjugated in the human liver and gut to mainly BPA glucuronide (BPA-G), a water-soluble compound that is eliminated in the urine with a half-life of 5.4–6.4 h (Völkel et al., 2002; Thayer et al., 2015). Unconjugated BPA (i.e. ‘free BPA’) can also be detected in urine, but at a much lower percentage than BPA-G, typically <1% of total BPA (Thayer et al., 2015). BPA that is inhaled or absorbed through the skin largely enters the circulatory system without undergoing first-pass metabolism although metabolism of BPA by the skin has been reported in an ex vivo model (Zalko et al., 2011; Toner et al., 2018). In vitro, the dermal absorbed dose for BPA is generally in the range of 2–13% (Kaddar et al., 2008; Mørck et al., 2010; Demierre et al., 2012; Toner et al., 2018). A few observational studies have suggested a longer elimination half-life for BPA in certain human populations than reported in oral-dosing studies (Stahlhut et al., 2009; Christensen et al., 2012; Hines et al., 2017a). It is unclear, however, if the suggestion of a longer half-life is related to exposure routes, dosing patterns, prolonged elimination of BPA after dermal uptake (Liu and Martin 2017), fat storage, or other factors.

The toxicity of BPA has been extensively investigated in both experimental and observational studies. BPA is weakly estrogenic and has low acute toxicity (Dodds, 1936; European Union, 2008; NTP, 2008). BPA-G, unlike free BPA, does not exhibit estrogenic activity (Snyder et al., 2000; Matthews et al., 2001). In laboratory animals, changes in numerous biological endpoints have been reported after exposure to BPA and a range of health effects have been associated with BPA exposure in epidemiological studies, most of which were cross-sectional in design (as reviewed in WHO, 2011; Cantonwine et al., 2013; Peretz et al., 2014; Mínguez-Alarcon et al., 2016). Endocrine system disruption is hypothesized to underlie many of the effects reported in animal and human studies.

In the USA, BPA (conjugated plus free) has been detected in the urine of more than 92% of individuals ≥6 years of age, an exposure thought to be largely diet-related (Calafat et al., 2008; NTP, 2008). Published reports of occupational BPA exposure are limited. Initial investigations of occupational exposure to BPA were conducted largely among manufacturing workers in Asia (Hanaoka et al., 2002; Xiao et al., 2005, Cha et al., 2008; He et al., 2009; Ren et al., 2012; Wang et al., 2012; Zhuang et al., 2015). More recently, BPA exposures have been reported for cashiers in the USA and France (Ndaw et al., 2016; Thayer et al., 2016), factory workers in Finland (Heinälä et al. 2017), and US manufacturing workers (Hines et al., 2017a). Changes in male reproductive health in several functional domains (sexual function, hormone levels, and semen quality) were reported for workers in Chinese factories making BPA or BPA-based epoxy resins in the largest health study of BPA-exposed workers to date (Li et al., 2010a,b, 2011; Zhou et al., 2013; Liu et al., 2015). This study was cross-sectional and its findings have not been confirmed in a similarly exposed population.

While the concentration of BPA in urine provides an estimate of body burden, it does not by itself indicate route(s) of exposure. Understanding exposure routes (i.e. inhalation, dermal, oral) and their relative contributions is important for controlling BPA exposure in the workplace, especially when designing engineering controls, selecting personal protective equipment, and determining the need for workplace cleanliness regimens.

In 2013–2014, the National Institute for Occupational Safety and Health (NIOSH) conducted a study to assess the BPA exposure of US manufacturing workers who made BPA, made products using BPA, or used a BPA-containing product. We previously reported BPA levels in urine, personal air, and hand-wipe samples collected from these workers (Hines et al., 2017a,b). On average, workers had urinary BPA concentrations ~70 times higher than in US adults in the National Health and Nutrition Examination Survey (NHANES) 2013–2014 (Hines et al., 2017a). BPA was detected in 95% of the air samples, and BPA levels in end-shift hand wipes averaged 10 times higher than at pre-shift (Hines et al., 2017b). Our aim in this article is to explore the relationship between BPA levels in personal air and hand-wipe samples, and BPA concentrations in urine. We also evaluate these data for evidence of the relative importance of inhalation and dermal contact as exposure pathways for BPA among these workers.

Methods

Company and participant recruitment

Methods for recruiting companies and participating workers have been described previously (Hines et al., 2017a,b). Relying mainly on the 2010 and 2011 US EPA Toxic Release Inventory (USEPA, 2018), we identified 73 companies potentially making or using BPA. Of these 73 companies, 15 did not respond to inquiries, 15 no longer produced or used BPA, 37 had few workers handling BPA, infrequent BPA use or could not be scheduled within the study period, and six participated in the study. We selected companies that represented a range of BPA manufacturing processes. We visited each company to identify BPA-related jobs and invited workers performing these jobs to participate in the study. The NIOSH Institutional Review Board approved this study. Participants gave written informed consent and were reimbursed $70 for the time and inconvenience of providing samples.

Sample collection

We collected urine, air, and hand-wipe samples from participants over 2 consecutive days (Day 1 and Day 2) (Fig. 1). We scheduled sampling to begin after the participant had been off work for at least 24 h to allow time for BPA to approach baseline concentrations in the urine. Details on sampling methods are provided in Hines et al. (2017a,b). Briefly, we asked participants to provide a total of seven urine samples (time points 1–7): pre-shift (baseline), mid-shift (±30 min of the participant’s shift mid-point), end-shift, and post-shift (4–6 h after leaving work) urine samples on Day 1 and pre-shift, mid-shift, and end-shift urine samples on Day 2. Participants collected urine samples in sterile polypropylene specimen cups after washing hands with water only (to avoid potential interferences). Participants kept post-shift samples on refrigerant packs between shifts. We aliquoted the urine into polypropylene cryovials followed by immediate freezing on dry ice. We also collected quality control (QC) field blanks and blind duplicates of participants’ samples; these samples were aliquoted and handled in the same manner as other participants’ samples. Urine samples were shipped and stored frozen (−80°C) until analysis.

Figure 1.

Figure 1.

Open in a new tab

Urine, air, and hand-wipe sample collection over 2 consecutive work days.

We collected a full-shift breathing zone air sample for each participant on each of the 2 days. We used an IOM sampler (SKC, Inc.) with a 25-mm quartz fiber filter in a stainless steel cassette at a nominal flow rate of 2 l min−1 to collect inhalable particles. Sampling pumps were pre- and post-calibrated. Upon completion of sampling, filter cassettes were placed in polypropylene containers, and kept cold (4°C) until analysis.

On Day 2, we collected a pre-shift and an end-shift hand-wipe sample from each participant. We collected hand wipes on Day 2 to minimize interference with concurrent biological monitoring. Because of a schedule change, pre- and post-shift hand wipes were collected on Day 1 for one worker. We used four Large Alpha® Swabs (TX715, ITW Texwipe) moistened with 100% high-performance liquid chromatography (HPLC) grade isopropanol (Thermo Fisher Scientific) to systematically wipe the palms, backs and fingers of a participant’s hands (two swabs per hand). We instructed participants to delay their final hand washing of the shift until the end-shift hand-wipe sample had been taken; otherwise they could wash their hands as needed. We inserted the four swabs into a polypropylene vial to form a single sample. Field blanks were prepared for both air and hand-wipe samples (10% each) and handled in a manner similar to the participant samples.

Sample analysis

Sample analysis details and QC results are given in Hines et al. (2017a,b). Briefly, we quantified urinary concentrations of total (free plus conjugated) BPA by online solid phase extraction-HPLC-isotope dilution tandem mass spectrometry (Zhou et al., 2014). The limit of detection (LOD) was 0.1 μg l−1. We measured urinary creatinine using a Vitros® 250 Chemistry Analyzer (Ortho-Clinical Diagnostics). To adjust for urine dilutions, BPA concentrations were divided by creatinine (units = μg g−1 creatinine). The NIOSH contract laboratory Bureau Veritas North America quantified BPA in air and hand-wipe samples. Briefly, air and wipe samples were extracted with acetonitrile and analyzed for BPA by HPLC with ultraviolet detection. The LOD across seven sample batches ranged from 0.03 to 0.1 μg m−3 (air) and from 0.05 to 0.3 μg per sample (hand wipe).

Statistical analysis

The distributions of the levels of creatinine-corrected total BPA in urine (total BPACR), BPA in air, and BPA on hand wipes were each skewed to the right, and a natural log transformation was applied to each variable prior to all statistical analyses. BPA detection was 100% (urine), 95.2% (air), 93.2% (pre-shift hand wipe), and 100% (end-shift hand wipe) (Hines et al., 2017a,b). We assigned LOD/2 to air and hand-wipe samples where BPA was not detected (Hornung and Reed, 1990). All statistical analyses were performed in SAS v. 9.3 (SAS Institute, Inc.).

We initially used Pearson’s correlation coefficient to evaluate the relationship between the end-shift total BPACR and (1) the BPA air concentration and (2) the end-shift BPA hand level. Day 1 and Day 2 BPA air concentrations were paired with the end-shift total BPACR for the corresponding day. We also averaged the mid-shift and end-shift total BPACR on each day and re-ran the correlations.

We next applied mixed-effects linear regression models using the PROC MIXED procedure in SAS to evaluate the relationship between total BPACR and (1) BPA air concentration (urine~air model), (2) end-shift hand-wipe BPA level (urine~hand model), and (3) BPA air concentration and end-shift hand-wipe BPA level in the same model (urine~air+hand model). The base models are summarized in Table 1. In these models, the dependent variable was ln(total BPACR) at mid- and end-shift on Day 1 (time points 2 and 3) and on Day 2 (time points 6 and 7).

Table 1.

Parameters of base models.

Dependent variable Base modela
Urine ~air Urine~hand Urine~air+hand
ln(total BPACR) at time points 2, 3, 6, 7 ln(total BPACR) at time points 2, 3 (for hand wipes on Day 1), 6, 7 (for hand wipes on Day 2) ln(total BPACR) at time points 2, 3, 6, 7
Random effect Worker Worker Worker
Covariance structure AR(1) CS CS
Fixed effects
 Day X
 Shift time X X X
 Day × shift time X
 ln(total BPACR) at baseline X X X
 BMI X X X
 ln(BPA air) X X
 ln(BPA hand at pre-shift) X X
 ln(BPA hand at end shift) X X
Open in a new tab

AR(1), first-order autoregressive.

aModel variations explained in Methods.

In the urine~air model, worker was treated as a random effect; fixed effects included day (Day 1 or 2), shift time (mid-shift or end shift), an interaction term for day and shift time (day × shift time), ln(total BPACR at baseline), body mass index (BMI) computed from self-reported height and weight, and ln(BPA air concentration). We used a first-order autoregressive AR(1) covariance structure in this model. The air concentration measured on Day 1 was associated with time points 2 and 3; the air concentration on Day 2 with time points 6 and 7. BMI and ln(total BPACR at baseline) were included in the model based on earlier modeling results with total BPACR (Hines et al., 2017a).

We also evaluated alternate urine~air models. Specifically, the model was re-run with the dependent variable ln(total BPACR) including only end-shift urine samples (time points 3 and 7), with all post-baseline urine samples (time points 2–7), and with post-baseline Day 1 through pre-shift Day 2 urine samples (time points 2–5) to capture mostly Day 1 exposure. We also re-ran the air model with the dependent variable restricted to mid- and end-shift urine samples on Day 2 only (time points 6 and 7), but with the Day 1 and Day 2 air concentrations included as separate covariates, then re-ran this model after removing the Day 1 air concentration (all other terms remained) in order to assess if BPA air concentrations on Day 1 had an effect on total BPACR on Day 2. Optimum covariance structures for these alternate models varied by model. We used Akaike’s Information Criterion (AIC) to compare models. A lower AIC indicated better model fit.

In the urine~hand model, worker was treated as a random effect. Fixed effects included shift time (mid-shift or end shift), ln(total BPACR at baseline), BMI, ln(BPA hand at pre-shift), and ln(BPA hand at end shift). We used a compound symmetric (CS) covariance structure as each worker in this model had only two urine samples. We re-ran the urine~hand model excluding two participants with particularly high end-shift hand-wipe BPA levels (19000 and 12000 µg per sample, 76- and 48-times higher, respectively, than the 95th percentile of 250 µg per sample for 148 hand-wipe samples). We also re-ran the urine~hand model restricting the dependent variable to end-shift urine samples [time points 3 (participant with Day 1 hand wipes) and 7 (all other participants)].

In the urine~air+hand model, we treated worker as a random effect, included shift time (mid-shift or end shift), ln(urinary total BPACR at baseline), BMI, ln(BPA air concentration), ln(BPA hand at pre-shift) and ln(BPA hand at end shift) as fixed effects, and used a CS covariance structure. We then ran the model without the BPA air concentration, then re-ran the model after restoring the BPA air concentration but excluding the pre- and end-shift hand-wipe BPA levels. This latter analysis allowed us to evaluate the relative effect of the air and hand-wipe data on model fit using AIC to compare models.

Results

A total of 78 workers participated in the study. One worker was excluded from all analyses because of possible urine sample contamination with BPA; 77 workers remained after exclusion (Fig. 2). Demographic characteristics of these 77 workers are described in Hines et al. (2017a). Briefly, the workers were predominately male (98.7%) and white (89.6%), with a median age of 44 years. Median BMI was 29.8 kg m−2 (range 21.0–44.3 kg m−2). Most workers (84.4%) had been off work at least 24 h before collecting the first urine sample. Seven of the 12 workers with less than 24 h off work were maintenance workers.

Figure 2.

Participants included in regression models with urine, air, and/or hand BPA exposure data.

aExclude 1 worker (2 worker-days) with possible BPA urine contamination.

bExclude 7 worker-days on 7 different workers with a missing air sample; each worker had an air sample on the non-missing day.

cExclude 4 workers (6 worker-days) missing one or both hand samples; exclude 72 worker-days because hand samples collected on only 1 day.

dExclude 4 workers (6 worker-days) missing one or both hand samples; exclude 4 additional workers (4 worker-days) missing an air sample on the day hand samples were collected; exclude 72 worker-days because hand samples collected on only 1 day.

Figure 2.

Open in a new tab

We found a strong correlation between the BPA concentration in air and end-shift total BPACR (rp = 0.79, P < 0.0001, Fig. 3). The correlation was nearly as strong when the average of mid- and end-shift total BPACR was used (rp = 0.77, P < 0.0001). A strong correlation was also found between the end-shift hand-wipe BPA level and end-shift total BPACR (rp = 0.75, P < 0.0001, Fig. 4), as well as with the average of mid- and end-shift total BPACR (rp = 0.76, P < 0.0001).

Figure 3.

Figure 3.

Open in a new tab

Scatter plot of urinary total BPACR (μg g−1) at end shift versus BPA air concentration (μg m−3) stratified by Day. Pearson’s r = 0.79 (on natural log-transformed data), P < 0.0001.

Figure 4.

Figure 4.

Open in a new tab

Scatter plot of urinary total BPACR (μg g−1) at end shift on day of hand-wipe collection versus end-shift hand-wipe BPA level (μg per sample). n = 73 (n = 1 on Day 1; n = 72 on Day 2). Pearson’s r = 0.75 (on natural log-transformed data), P < 0.0001.

In the urine~air model, after adjusting for day (P < 0.0001), shift time (P < 0.0001), the interaction of day and shift time (P = 0.0121), total BPACR at baseline (P < 0.0001), and BMI (0.0268), the BPA air concentration was significantly and positively associated with total BPACR at end shift (P < 0.0001, Table 2). The association remained significant when the dependent variable consisted only of end-shift urine samples (time points 3 and 7) (Table S1, Supplementary data, available at Annals of Occupational Hygiene online), all post-baseline urine samples (time points 2–7) (Table S2, Supplementary data, available at Annals of Occupational Hygiene online), or urine samples thought to be most associated with exposure to BPA on Day 1 (time points 2–5) (Table S3, Supplementary data, available at Annals of Occupational Hygiene online). When we included Day 1 and Day 2 BPA air concentrations in the model as separate independent variables and restricted the total BPACR to Day 2 mid- and end-shift samples, we saw a significant effect of Day 1 air concentrations on Day 2 total BPACR (P = 0.0104); Day 2 BPA air concentration was, as expected, highly significant (P < 0.0001) in this model (Table S4, Supplementary data, available at Annals of Occupational Hygiene online). The negative interaction between day and shift time indicated that the change in total BPACR from mid-shift to end shift on Day 1 was greater than on Day 2.

Table 2.

Urine~air model. Multiple regression results for creatinine-adjusted total BPA (µg g−1) at time points 2, 3, 6, and 7 regressed on BPA air concentration (µg m−3) and including covariates (144 worker-days, 77 workers).

Dependent variable: ln(total BPACR, µg g−1)a,bn = 286 β (SE) P-value Factorc
Intercept 0.7702 (0.5376) 0.1559
Day
 1
 2		 Ref. 0.6276 (0.1163) <0.0001 1.87
Shift time
 Mid-shift
 End shift		 Ref. 0.6486 (0.08804) <0.0001 1.91
Day × shift time −0.3379 (0.1333) 0.0121 0.71
ln(total BPACR at baseline, μg g−1) 0.5890 (0.05498) <0.0001 1.80
BMI, kg m−2 0.03514 (0.01556) 0.0268 1.04
ln(BPA air concentration, µg m−3) 0.2614 (0.02821) <0.0001 1.30
Open in a new tab

Ref, referent group.

aFirst-order autoregressive covariance structure, AR(1), AIC = 665.7.

bRegression equation: total BPACR, µg g−1 = (e0.7702 + 0.6276 (if Day=2) + 0.6486 (if Shift time=End shift) − 0.3379 (if End shift on Day 2) + 0.3514×BMI) × (total BPACR at baseline, µg g−1)0.5890 × (BPA air concentration, µg m−3)0.2614.

ceβ.

In the urine~hand model, after adjusting for shift time (P < 0.0001), total BPACR at baseline (P < 0.0027), BMI (P = 0.1932), and the pre-shift BPA hand-wipe level (P = 0.5703), the end-shift hand-wipe BPA level was significantly and positively associated with total BPACR at end shift on the day the hand-wipe sample was collected (Table 3). When we excluded the two participants with the particularly high end-shift hand-wipe BPA levels from the urine~hand model, the end-shift hand-wipe BPA level remained significant (Table S5, Supplementary data, available at Annals of Occupational Hygiene online). When the urine~hand model was restricted to urinary total BPACR at end shift only, the end-shift hand-wipe BPA level remained significant with a slightly higher effect estimate (β = 0.4340, Table S6, Supplementary data, available at Annals of Occupational Hygiene online) than in the urine~hand model with both mid- and end-shift total BPACR concentrations (β = 0.4036, Table 3).

Table 3.

Urine~hand model. Multiple regression results for creatinine-adjusted total BPA (µg g−1) at time points 2 and 3 (for hand wipes collected on Day 1) and 6 and 7 (for hand wipes collected on Day 2) regressed on BPA hand levels (µg per sample) and including covariates (73 worker-days, 73 workers).

Dependent variable: ln(total BPACR, µg g−1)a,bn = 144 β (SE) P-value Factorc
Intercept 1.3698 (0.8294) 0.0968
Shift time
 Mid-shift
 End shift		 Ref.
0.3481 (0.07898)		 <0.0001 1.42
ln(total BPACR at baseline, μg g−1) 0.3438 (0.1106) 0.0027 1.41
BMI, kg m−2 0.03019 (0.02298) 0.1932 1.03
ln (BPA hand, µg per sample at pre-shift) 0.05962 (0.1045) 0.5703 1.06
ln (BPA hand, µg per sample at end shift) 0.4036 (0.09127) <0.0001 1.50
Open in a new tab

Ref, referent group.

aCompound symmetric covariance structure. AIC = 376.1.

bRegression equation: total BPACR, µg g−1 = (e1.3698 + 0.3481 (if Shift time=End shift) + 0.03019×BMI) × (total BPACR at baseline, µg g−1)0.3438 × (BPA pre-shift hand level, µg per sample)0.05962 × (BPA end-shift hand level, µg per sample)0.4036.

ceβ.

When both the BPA air concentration and the end-shift hand-wipe BPA level were in the same model (urine~air+hand) and after adjusting for covariates, the BPA air concentration was significant at P < 0.0001, and the end-shift hand-wipe BPA level at P = 0.0106 (Table 4). When we re-ran this model with either the BPA air concentration or the end-shift hand-wipe BPA level in the model but not both, the AIC was lower for the model with the air concentration (AIC = 337.4) than the model with the end-shift hand wipe (AIC = 357.5) (Table 4).

Table 4.

Urine~air+hand model. Multiple regression results for creatinine-adjusted total BPA (µg g−1) at time points 2, 3, 6, and 7 regressed on BPA air concentration (µg m−3) and BPA hand level (µg per sample) and including covariates. Model 1: BPA air concentration and BPA levels on hand wipes both in model; Model 2: BPA air concentration excluded from model; Model 3: BPA levels on hand wipes excluded model (69 worker-days, 69 workers).

Dependent variable: ln(total BPACR, µg g−1)an = 136 β(SE) P-value Factorc
Model 1 (AIC = 336.7)b
 Intercept 2.0137 (0.7013) 0.0055
 Shift time
  Mid-shift
  End shift		 Ref.
0.3417 (0.08305)		 0.0001 1.41
 ln(total BPACR at baseline, μg g−1) 0.3022 (0.09162) 0.0016 1.35
 BMI, kg m−2 0.02279 (0.01970) 0.2518 1.02
 ln (BPA air concentration, µg m−3) 0.2725 (0.04977) <0.0001 1.31
 ln (BPA hand, µg per sample at pre-shift) −0.04015 (0.08851) 0.6516 0.96
 ln (BPA hand, µg per sample at end shift) 0.2205 (0.08374) 0.0106 1.25
Model 2 (AIC = 357.5)
 Intercept 1.5486 (0.8384) 0.0693
 Shift time
  Mid-shift
  End shift		 Ref.
0.3424 (0.08317)		 0.0001 1.41
 ln(total BPACR at baseline, μg g−1) 0.3231 (0.1103) 0.0047 1.38
 BMI, kg m−2 0.02705 (0.02372) 0.2584 1.03
 ln (BPA hand, µg per sample at pre-shift) 0.04822 (0.1049) 0.6472 1.05
 ln (BPA hand, µg per sample at end shift) 0.4034 (0.09250) <0.0001 1.50
Model 3 (AIC = 337.4)
 Intercept 2.0897 (0.6994) 0.0040
 Shift time
  Mid-shift
  End shift		 Ref.
0.3438 (0.08309)		 0.0001 1.41
 ln(total BPACR at baseline, μg g−1) 0.4015 (0.07412) <0.0001 1.49
 BMI, kg m−2 0.02890 (0.02021) 0.1575 1.03
 ln (BPA air concentration, µg m−3) 0.3320 (0.04418) <0.0001 1.39
Open in a new tab

Ref, referent group.

aCompound symmetric covariance structure.

bRegression equation model 1: Total BPACR in µg g−1 = (e2.0137 + 0.3417 (if Shit-time=End shift) +0.02279×BMI) × (total BPACR at baseline, µg g−1)0.3022 × (BPA air concentration, µg m−3)0.2725 × (BPA pre-shift hand level, µg per sample)−0.04015 × (BPA end-shift hand level, µg per sample)0.2205.

ceβ.

Discussion

In the general population, diet is thought to be the main source of BPA exposure (NTP, 2008). In this study, however, manufacturing workers were likely exposed to BPA mainly by inhalation and/or dermal contact with diet contributing minimally to their overall BPA exposure (Hines et al., 2017a,b). Measuring BPA in urine captures exposure by all routes, but does not provide direct information on exposure routes. In this analysis, we evaluated the relationship between total BPACR in urine and measures of exposure via inhalation (air samples) and dermal contact (hand wipes). We then constructed models to estimate total BPACR from air and hand-wipe data in order to explore the relative importance of inhalation and dermal exposure routes among workers in our study.

We found that both air and hand-wipe measures of BPA exposure strongly correlated with total BPACR. Further exploration using mixed-effects regression modeling indicated that model fit improved when the model included the BPA air concentration compared with when the model had only the BPA hand-wipe level (holding all other terms constant). This result suggests that inhalation was a more important exposure route, on average, than dermal contact for this group of workers although we note that both measures were statistically significant in all models. This observation is also consistent with higher BPA intake estimates based on inhalation than on dermal exposure in these workers (Hines et al., 2017b). At the BPA air concentrations measured in this study, we also found that worker exposure to BPA in air is reflected in total BPACR on both the day the air sample was taken and the following day. This finding indicates that information on the prior day’s BPA air concentration should be collected in order to better model total BPACR in exposed workers.

We did not have direct measures of workers’ oral exposure to BPA; however we had information on behaviors at work that could have contributed to oral exposure such as smoking, eating, chewing gum, and chewing tobacco and hand washing frequency. In our previous analyses (Hines et al., 2017a), these factors were not associated with increased total BPACR although the study may not have had sufficient power to detect such effects.

Using the BPA air concentration to predict total BPACR at various time points improved by having information on a worker’s BMI and total BPACR at baseline (pre-shift Day 1). With these two pieces of information, the urine~air model (Table 2) can be used to estimate total BPACR at select time points. For example, at a concentration of 6 µg m−3 (median air concentration for study workers), a baseline total BPACR of 25 μg g−1 (~ median for study workers), and a BMI of 30 kg m−2 (~median for study workers), the predicted end-shift total BPACR was 126 μg g−1 (Day 1) and 168 μg g−1 (Day 2). At the maximum air concentration measured, 920 µg m−3, the predicted end-shift total BPACR was 470 μg g−1 (Day 1) and 628 μg g−1 (Day 2).

Likewise, BMI and total BPACR at baseline were important factors in estimating end-shift total BPACR from end-shift hand-wipe BPA levels. Pre-shift hand-wipe BPA levels had essentially no effect on end-shift total BPACR (Table 3). For example, at a BMI of 30, a baseline total BPACR of 25 μg g−1, a pre-shift hand-wipe BPA level of 0.3 µg per sample (highest reported LOD), and an end-shift hand-wipe level of 2 µg per sample (median for study workers), the predicted end-shift total BPACR was 140 µg g−1. At the highest end-shift hand-wipe level, 19000 µg per sample, the predicted end-shift total BPACR was 2100 µg g−1.

Workers in this study had, on average, BPA concentrations in their Day 1 pre-shift urine 20 times higher than in NHANES 2013–2014, despite 84% of the workers having been away from work for at least 24 h and 71% for at least 48 h (CDC 2017; Hines et al., 2017a). Consequently, regression models estimating post-baseline total BPACR from air and hand-wipe data were adjusted for these relatively high baseline concentrations. Moreover, because the range of baseline total BPACR for workers in our study was quite wide, 0.78–1580 µg g−1, and variable (geometric standard deviation of 5.74), baseline urine samples would be critical to include in future studies of similarly exposed workers.

We collected the inhalable aerosol fraction anticipating mostly large particles when handling raw BPA. Workers in jobs involving molten BPA-filled wax may have inhaled respirable-sized particles if BPA vaporized with heat and then condensed to small particles with cooling. While the inhalable fraction includes small particles, we do not know the partitioning of the sample between large and small particles. Particle size affects deposition in the respiratory tract, and therefore, may affect uptake and possibly the correlation between BPA concentrations in air and in urine.

Results from this analysis may have implications for workplace controls. As a first step, controls should reduce inhalation exposure, initially through use of engineering or process controls, and then respirators. These efforts should be followed by reducing opportunities for dermal contact with BPA. Wipe sampling of work surfaces in this study that workers touched with ungloved hands indicated, not surprisingly, that objects and surfaces in production areas had higher BPA levels, on average, than eating areas and offices/control rooms (Hines et al., 2017b).

To our knowledge, this is the first analysis in a single study of the relationship among three measures of BPA exposure, urine, air, and hand wipes in either an occupational or a non-occupational population. He et al. (2009) examined the correlation between BPA concentrations in personal air samples and in pre- and post-shift urine samples from 131 epoxy resin- and BPA-manufacturing workers in China. They found Spearman coefficients of 0.525 (air with pre-shift urine) and 0.726 (air with post-shift urine) although air and urine samples were not always collected on the same day. We found a slightly higher correlation between BPA air concentration and end-shift total BPACR (rp = 0.79). Concentrations of BPA in urine and in air among workers in our study were similar to those of manufacturing workers in China for whom male reproductive health effects were reported (He et al. 2009; Li et al., 2010b). Additional health studies of similarly exposed workers are needed.

Strengths of this analysis include its well-defined temporality around the collection of urine, air, and hand-wipe samples, a detailed time course of BPA concentrations in urine over 2 consecutive days, and a sufficient sample size to both identify exposure determinants and to examine the correlation among BPA in urine, air, and on workers’ hands. Possible study limitations should also be noted. Participants in our study may not be representative off all manufacturing workers exposed to BPA. In addition, we may not have captured all important exposure determinants, which could affect estimates of urinary BPA concentrations derived from air and hand-wipe data, and we did not have direct measures of oral BPA exposure. The apparent weaker correlation of hand-wipe BPA levels with total BPACR might be due to a delayed appearance of BPA in urine after dermal uptake. Hand wipes were collected at end shift on the last day of urine sampling and if the elimination of BPA into the urine is slower after dermal contact than after inhalation or ingestion, then we may have not have captured all dermal-related exposure in the urine.

Conclusion

External measures of BPA exposure by inhalation (air samples) and dermal contact (end-shift hand wipes) were highly correlated with mid- and end-shift total BPACR. Although both inhalation and dermal contact likely contributed to exposure, our analyses suggest that inhalation, on average, appeared to be a more dominant exposure route than dermal contact for this group of manufacturing workers. This finding has implications for setting priorities for exposure control measures in manufacturing workplaces where BPA is handled.

Supplementary Data

Supplementary data are available at Annals of Work Exposures and Health online.

wxy042_suppl_Supplementary_Material
Click here for additional data file. (566.8KB, pdf)

Funding

This study was supported in part by an interagency agreement between the National Institute for Occupational Safety and Health and the National Institute of Environmental Health Sciences (AES12009) as a collaborative National Toxicology Program research activity. The authors declare they have no conflict of interest.

Acknowledgments

We gratefully acknowledge the contributions of participating workers, companies, and unions to this study.

References

  1. Brunelle DJ. (2014)Polycarbonates. Kirk-Othmer encyclopedia of chemical technology. John Wiley & Sons; pp. 1–30. Available at http://onlinelibrary.wiley.com/doi/10.1002/0471238961.1615122502182114.a01.pub3/abstract. Accessed 31 May 2018. [Google Scholar]
  2. Calafat AM, Ye X, Wong LY et al. (2008)Exposure of the U.S. population to bisphenol A and 4-tertiary-octylphenol: 2003-2004. Environ Health Perspect; 116: 39–44. [DOI] [PMC free article] [PubMed] [Google Scholar]
  3. Cantonwine DE, Hauser R, Meeker JD (2013)Bisphenol A and human reproductive health. Expert Rev Obstet Gynecol; 8: 329–35. [DOI] [PMC free article] [PubMed] [Google Scholar]
  4. Carney C. (2014)Advanced casting methodologies: investment casting, centrifugal casting, squeeze casting, metal spinning, and batch casting. In Hashmi S, editor. Comprehensive materials processing. Vol. 5, Chapter 5.03. Amsterdam, The Netherlands: Elsevier; pp. 39–47. [Google Scholar]
  5. CDC, Centers for Disease Control and Prevention (2017)Fourth National Report on Human Exposure to Environmental Chemicals, Updated Tables, January 2017, Vol. 1 Available at https://www.cdc.gov/exposurereport/pdf/FourthReport_UpdatedTables_Volume1_Jan2017.pdf. Accessed 31 May 2018. [Google Scholar]
  6. Cha BS, Koh SB, Park JH et al. (2008)Influence of occupational exposure to bisphenol A on the sex hormones of male epoxy resin painters. Mol Cell Toxicol; 4: 203–34. [Google Scholar]
  7. Christensen KL, Lorber M, Koslitz S et al. (2012)The contribution of diet to total bisphenol A body burden in humans: results of a 48 hour fasting study. Environ Int; 50: 7–14. [DOI] [PubMed] [Google Scholar]
  8. Demierre AL, Peter R, Oberli A et al. (2012)Dermal penetration of bisphenol A in human skin contributes marginally to total exposure. Toxicol Lett; 213: 305–8. [DOI] [PubMed] [Google Scholar]
  9. Dodds EC, Lawson W (1936)Synthetic oestrogenic agents without the phenanthrene nucleus. Nature; 137: 996. [Google Scholar]
  10. European Union (2008)Risk assessment report: 4,4’-isopropylidenediphenol (bisphenol-A), part 2, human health. Ispara, Italy: European Commission, Joint Research Centre, Institute for Health and Consumer Protection; p. 69. [Google Scholar]
  11. Hanaoka T, Kawamura N, Hara K et al. (2002)Urinary bisphenol A and plasma hormone concentrations in male workers exposed to bisphenol A diglycidyl ether and mixed organic solvents. Occup Environ Med; 59: 625–8. [DOI] [PMC free article] [PubMed] [Google Scholar]
  12. He Y, Miao M, Wu C et al. (2009)Occupational exposure levels of bisphenol A among Chinese workers. J Occup Health; 51: 432–6. [DOI] [PubMed] [Google Scholar]
  13. Heinälä M, Ylinen K, Tuomi T et al. (2017)Assessment of occupational exposure to bisphenol A in five different production companies in Finland. Ann Work Expo Health; 61: 44–55. [DOI] [PubMed] [Google Scholar]
  14. Hines CJ, Jackson MV, Christianson AL et al. (2017b) Air, hand wipe, and surface wipe sampling for Bisphenol A (BPA) among workers in industries that manufacture and use BPA in the United States. J Occup Environ Hyg; 14: 882–97. [DOI] [PubMed] [Google Scholar]
  15. Hines CJ, Jackson MV, Deddens JA et al. (2017a) Urinary Bisphenol A (BPA) concentrations among workers in industries that manufacture and use BPA in the USA. Ann Work Expo Health; 61: 164–82. [DOI] [PMC free article] [PubMed] [Google Scholar]
  16. Hornung RW, Reed LD (1990)Estimation of average concentration in the presence of non-detectable values. App Occup Environ Hyg; 5: 46–51. [Google Scholar]
  17. Kaddar N, Harthé C, Déchaud H et al. (2008)Cutaneous penetration of bisphenol A in pig skin. J Toxicol Environ Health A; 71: 471–3. [DOI] [PubMed] [Google Scholar]
  18. Kopf PW. (2003)Phenolic resins. Kirk-Othmer encyclopedia of chemical technology. John Wiley & Sons; pp. 1–54. Available at http://onlinelibrary.wiley.com/doi/10.1002/0471238961.1608051411151606.a01.pub2/abstract. Accessed 31 May 2018. [Google Scholar]
  19. Li DK, Zhou Z, Miao M et al. (2010a) Relationship between urine bisphenol-A level and declining male sexual function. J Androl; 31: 500–6. [DOI] [PubMed] [Google Scholar]
  20. Li DK, Zhou Z, Miao M et al. (2011)Urine bisphenol-A (BPA) level in relation to semen quality. Fertil Steril; 95: 625–30.e1–4. [DOI] [PubMed] [Google Scholar]
  21. Li D, Zhou Z, Qing D et al. (2010b) Occupational exposure to bisphenol-A (BPA) and the risk of self-reported male sexual dysfunction. Hum Reprod; 25: 519–27. [DOI] [PubMed] [Google Scholar]
  22. Liu J, Martin JW (2017)Prolonged exposure to bisphenol A from single dermal contact events. Environ Sci Technol; 51: 9940–9. [DOI] [PubMed] [Google Scholar]
  23. Liu X, Miao M, Zhou Z et al. (2015)Exposure to bisphenol-A and reproductive hormones among male adults. Environ Toxicol Pharmacol; 39: 934–41. [DOI] [PubMed] [Google Scholar]
  24. Matthews JB, Twomey K, Zacharewski TR (2001)In vitro and in vivo interactions of bisphenol A and its metabolite, bisphenol A glucuronide, with estrogen receptors alpha and beta. Chem Res Toxicol; 14: 149–57. [DOI] [PubMed] [Google Scholar]
  25. Mínguez-Alarcón L, Hauser R, Gaskins AJ (2016)Effects of bisphenol A on male and couple reproductive health: a review. Fertil Steril; 106: 864–70. [DOI] [PMC free article] [PubMed] [Google Scholar]
  26. Mørck TJ, Sorda G, Bechi N et al. (2010)Placental transport and in vitro effects of Bisphenol A. Reprod Toxicol; 30: 131–7. [DOI] [PubMed] [Google Scholar]
  27. Ndaw S, Remy A, Jargot D et al. (2016)Occupational exposure of cashiers to Bisphenol A via thermal paper: urinary biomonitoring study. Int Arch Occup Environ Health; 89: 935–46. [DOI] [PMC free article] [PubMed] [Google Scholar]
  28. NTP, National Toxicology Program (2008)NTP-CERHR monograph on the potential human reproductive and developmental effects of bisphenol A. NIH Pub. No. 08-5994 Available at https://ntp.niehs.nih.gov/ntp/ohat/bisphenol/bisphenol.pdf. Accessed 31 May 2018.
  29. Peretz J, Vrooman L, Ricke WA et al. (2014)Bisphenol a and reproductive health: update of experimental and human evidence, 2007-2013. Environ Health Perspect; 122: 775–86. [DOI] [PMC free article] [PubMed] [Google Scholar]
  30. Pham HQ, Marks MJ (2004)Epoxy resins. Kirk-Othmer encyclopedia of chemical technology. John Wiley & Sons; pp. 1–125. Available at http://onlinelibrary.wiley.com/doi/10.1002/0471238961.0516152407011414.a01.pub2/abstract. Accessed 31 May 2018. [Google Scholar]
  31. Ren D-S, Wu C-H, Chang X-L et al. (2012)Exposure levels of bisphenol A among workers in a resin factory. J Environ Occup Med; 29: 269–73 [in Chinese]. [Google Scholar]
  32. Snyder RW, Maness SC, Gaido KW et al. (2000)Metabolism and disposition of bisphenol A in female rats. Toxicol Appl Pharmacol; 168: 225–34. [DOI] [PubMed] [Google Scholar]
  33. Stahlhut RW, Welshons WV, Swan SH (2009)Bisphenol A data in NHANES suggest longer than expected half-life, substantial nonfood exposure, or both. Environ Health Perspect; 117: 784–9. [DOI] [PMC free article] [PubMed] [Google Scholar]
  34. Thayer KA, Doerge DR, Hunt D et al. (2015)Pharmacokinetics of bisphenol A in humans following a single oral administration. Environ Int; 83: 107–15. [DOI] [PMC free article] [PubMed] [Google Scholar]
  35. Thayer KA, Taylor KW, Garantziotis S et al. (2016)Bisphenol A, Bisphenol S, and 4-Hydroxyphenyl 4-Isoprooxyphenylsulfone (BPSIP) in urine and blood of cashiers. Environ Health Perspect; 124: 437–44. [DOI] [PMC free article] [PubMed] [Google Scholar]
  36. Toner F, Allan G, Dimond SS et al. (2018)In vitro percutaneous absorption and metabolism of Bisphenol A (BPA) through fresh human skin. Toxicol In Vitro; 47: 147–55. [DOI] [PubMed] [Google Scholar]
  37. USEPA, United States Environmental Protection Agency (2014)Bisphenol A alternatives in thermal paper. Final report Available at http://www2.epa.gov/sites/production/files/2014-05/documents/bpa_final.pdf. Accessed 31 May 2018.
  38. USEPA, United States Environmental Protection Agency (2018)Toxic Release Inventory, TRI Explorer. Release reports Available at https://iaspub.epa.gov/triexplorer/tri_release.chemical. Accessed 31 May 2018.
  39. Völkel W, Colnot T, Csanády GA et al. (2002)Metabolism and kinetics of bisphenol a in humans at low doses following oral administration. Chem Res Toxicol; 15: 1281–7. [DOI] [PubMed] [Google Scholar]
  40. Wang F, Hua J, Chen M et al. (2012)High urinary bisphenol A concentrations in workers and possible laboratory abnormalities. Occup Environ Med; 69: 679–84. [DOI] [PubMed] [Google Scholar]
  41. WHO, World Health Organization (2011)Toxicological and Health Aspects of Bisphenol A. Report of Joint FAO/WHO Expert Meeting, 2–5 November 2010 and Report of Stake-holder Meeting on Bisphenol A, 1 November 2010, Ottawa, Canada. Section 6.2 Epidemiological Studies, pp. 31–34. Geneva, Switzerland: WHO; Available at http://apps.who.int/iris/bitstream/10665/44624/1/97892141564274_eng.pdf?ua=1. Accessed 31 May 2018. [Google Scholar]
  42. Xiao G-B, Shi J-L, He G-H et al. (2005)Investigation into serum BPA and sex hormone level of workers in epoxy resin manufacture. J Environ Occup Med; 22: 295–8 [in Chinese]. [Google Scholar]
  43. Zalko D, Jacques C, Duplan H et al. (2011)Viable skin efficiently absorbs and metabolizes bisphenol A. Chemosphere; 82: 424–30. [DOI] [PubMed] [Google Scholar]
  44. Zhou X, Kramer JP, Calafat AM et al. (2014)Automated on-line column-switching high performance liquid chromatography isotope dilution tandem mass spectrometry method for the quantification of bisphenol A, bisphenol F, bisphenol S, and 11 other phenols in urine. J Chromatogr B Analyt Technol Biomed Life Sci; 944: 152–6. [DOI] [PubMed] [Google Scholar]
  45. Zhou Q, Miao M, Ran M et al. (2013)Serum bisphenol-A concentration and sex hormone levels in men. Fertil Steril; 100: 478–82. [DOI] [PubMed] [Google Scholar]
  46. Zhuang W, Wu K, Wang Y et al. (2015)Association of serum bisphenol-A concentration and male reproductive function among exposed workers. Arch Environ Contam Toxicol; 68: 38–45. [DOI] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

wxy042_suppl_Supplementary_Material
Click here for additional data file. (566.8KB, pdf)

Articles from Annals of Work Exposures and Health are provided here courtesy of Oxford University Press

RESOURCES