Exposure assessment of adult intake of bisphenol A (BPA) with emphasis on canned food dietary exposures
Associated Data
Abstract
Bisphenol A (BPA) is a high-volume, synthetic compound found in epoxy resins and plastics used in food packaging. Food is believed to be a major source of BPA intake. In this study, we measured the concentration of BPA in convenience samplings of foodstuffs purchased in Dallas, Texas. Sampling entailed collection of 204 samples of fresh, frozen, and canned foods in two rounds in 2010. BPA was positive in 73% of the canned food samples, while it was found in only 7% of non-canned foods at low concentrations. The results of this food sampling program were used to calculate adult dietary intakes of BPA. A pathway approach combined food intakes, a “canned fraction” parameter which described what portion of total intake of that food came from canned products, and measured food concentrations. Dietary intakes were calculated as 12.6 ng/kg-day, of which 12.4 ng/kg-day was from canned foods. Canned vegetable intakes alone were 11.9 ng/kg-day. This dietary intake was compared to total intakes of BPA estimated from urine measurements of the National Health and Nutrition Examination Survey (NHANES). Total adult central tendency intakes ranged from 30 to 70 ng/kg-day for NHANES cycles between 2005 and 2010. Three possibilities were explored to explain the difference between these two approaches for intake estimation. Not all foods which may have been canned, particularly canned beverages such as soft drinks, were sampled in our food sampling program. Second, non-food pathways of exposure may be important for adults, including thermal paper exposures, and dust and air exposures. Finally, our canned food concentrations may not be adequately representative of canned foods in the United States; they were found to be generally lower compared to canned food concentrations measured in six other worldwide food surveys including three in North America. Our finding that canned food concentrations greatly exceeded non-canned concentrations was consistent with other studies, and underscores the importance of canned foods in the overall exposure of adults of BPA.
1. Introduction
Bisphenol A (BPA) is a high volume chemical with over six billion pounds produced annually worldwide (Vandenberg et al., 2007). BPA is used to make polycarbonate plastic and epoxy resins for linings in cans and pipes (Vandenberg et al., 2007). It has been commonly found in baby bottles, plastic drinking bottles, microwaveable food products, canned drinks and foods, toys, and medical devices (Vandenberg et al., 2007). The Food and Drug Administration (FDA) regulations no longer provide for the use of BPA-based polycarbonate resins in baby bottles and sippy cups and the use of BPA-based epoxy resins as coatings in packaging for infant formula (FDA, 2012, 2013). BPA is also used in dental sealants (Fleisch et al., 2010) and thermal paper used for receipts (Biedermann et al., 2010). BPA has been found in a variety of environmental media including food (Cao et al., 2011; EFSA, 2013; Geens et al., 2010; Liao and Kannan, 2013; Noonan et al., 2011; Schecter et al., 2010; Thomson and Grounds, 2005), dust (Geens et al., 2009; Loganathan and Kannan, 2011), paper money (Liao et al., 2012), and soil (Xu et al., 2008). Biological matrices sampled include blood (Cobellis et al., 2009; He et al., 2009), urine (CDC, 2009; EFSA, 2014; Zhang et al., 2011), saliva (Zimmerman-Downs et al., 2010), and breast milk (Otaka et al., 2003; Sun et al., 2004). Food is believed to be the main exposure source of BPA in humans (Kang et al., 2006; Lakind and Naiman, 2011; Vandenberg et al., 2007). Air, dust, thermal paper, and water also contribute to human exposure to a lesser extent (Biedermann et al., 2010; EFSA, 2013; Loganathan and Kannan, 2011; Wilson et al., 2007).
The usual biological matrix used to characterize exposure to BPA is urine (EFSA, 2014; CDC, 2009). The United States National Health and Nutrition Examination Survey (NHANES) has quantified BPA in urine for every sampling cycle starting in the 2003/4 cycle (see http://www.cdc.gov/nchs/nhanes.htm). The nearly 100% occurrence in every sampling cycle demonstrates the ubiquitous exposure of Americans to BPA (CDC, 2009). Simple dose reconstruction methods have been used to infer daily intakes of BPA based on urine measurements from NHANES (Lakind and Naiman, 2008, 2011; Lakind et al., 2012). These provide a very useful distribution of total daily intakes for the general population of United States. However, for the most part, they are not informative about the pathways of exposure. Exposure assessments focusing on exposure media concentrations in combination with exposure contact rates have typically been used in “pathway-based” assessments.
BPA exposure is of concern because animal and human studies have identified health effects associated with BPA exposure (NTP-CERHR, 2007; Rochester, 2013). Many of these focused on adverse effects on neurodevelopment, male and female reproductive systems alterations, and metabolic diseases (see Rochester, 2013 review).
We previously measured BPA in canned, fresh, and frozen foods purchased from supermarkets in Dallas, Texas, United States (Schecter et al., 2010). BPA was detected in 63 of 105 total samples, which included 88 food samples, 9 samples of infant formula, and 8 samples of pet foods. Only one positive result was from a non-canned sample, and that was of meat. The current study described here builds upon this earlier BPA food study by adding a second collection of a larger and more varied assortment of common foods from the same geographic region. Then, we combine the two rounds of food sampling, and estimate general adult population exposures to BPA from food. We compare these estimates to those made by back calculation from NHANES urine BPA measurements. This approach of comparing a “forward” or “pathway-based” intake estimation with a “backward” or “dose reconstruction” method has been previously used to assess children’s exposure to BPA (Christensen and Lorber, 2014). They found a good concurrence between the two methods showing the importance of dietary intakes, and specifically, the importance of consumption of canned food, in characterizing childhood exposure to BPA. We seek to similarly assess BPA exposure to adults in this study. We conclude the analysis by comparing measurements found in our combined food data set with measurements found in other surveys, and suggest future consideration of other food and non-food pathways that may be of importance to adult BPA exposures.
2. Materials and methods
2.1. Sample collection
Our previous study provided results from 105 samples (Schecter et al., 2010); we included results from 88 of these samples in this study (17 samples of infant formula and pet foods were not included). Samples were collected from supermarkets in Dallas, Texas, in March 2010. These 88 original samples included 84 samples from 28 specific food products sampled three times (such as three cans of a particular brand of tuna fish) and four individual samples of fresh meats and fish. The additional 116 individual food samples reported here for the first time were also purchased from supermarkets in Dallas in August 2010. We expanded the selection to include a variety of fresh, frozen, and canned foods. Unlike the first sampling, we did not obtain three samples of a specific food type; all 116 samples were unique samples (different containers and different manufacturers). This paper uses the combined set of 204 samples (116 plus 88) to assess dietary exposures. Samples were initially stored in their original containers and all except canned foods frozen at −80 °C and then shipped on dry ice to the analytic laboratory.
2.2. Analytical methods
For BPA measurement in food, the method followed the isotope dilution procedure. The samples, approximately 10 to 200 g wet weight per sample, were freeze-dried, internal standard 13C-BPA (Cambridge Isotope Laboratory, Andover, MA, US) was added, then homogenized. Extraction of the sample was performed by applying acetonitrile (Merck Darmstadt, Germany) in an ultrasonic bath at 40 °C. Liquid/liquid extraction was then performed with hexane (Merck Darmstadt, Germany), as well as purification on ENVI-Carb (Sigma-Aldrich/Supelco, St. Louis) using hexane as solvent.
The purified extract containing BPA was derivatized with bis(trimethylsilyl)-trifluoroacetamide (BSTFA) (Macherey-Nagel, Düren, Germany) forming the BPA-TMS-derivate. Further purification on silica column (Bakerbond SPE, Mallinckrodt Baker B.V., Deventer, Holland) was followed by concentration and addition of a recovery standard (13C-PCB #52; CIL, Andover, MA, USA).
Measurements were performed using High Resolution Gas Chromatography/Low Resolution Mass Spectrometry (HRGC/LRMS; GC 6890 Hewlett Packard and MS 5973 Hewlett Packard). The GC column was a DB5 MS with a length of 30 m, an inner diameter of 0.25 mm, and a film thickness of 0.25 μm (J&W Scientific). A splitless injector was used at 280 °C. The oven was set at 120 °C for 1 min, 10 °C/min to 240 °C, and 30 °C/min to 300 °C, then held 10 min. The flow rate for helium was 2.0 mL/min; interface was at 300 °C. The identification of the internal standard is performed by using the absolute retention time (RT) in the recorded ion chromatograms as well as the relationship of ion intensity of molecule ion and the relevant fragment ion. For determination of BPA, the relative retention time (RTT) referred to the internal Standard is used for additional identification criteria. The correct retention times are given by the analysis of the calibration standards under identical analysis conditions given for the sample analysis. The RTT for BPA needs to be in a window of ± 0,5% of the nominal value (given by the calibration measurements) based on the retention time of the internal standard. The absolute retention time for the internal standard needs to be in a window of ± 5% based on the nominal value measured at the calibration measurements. For each native and labeled component two mass traces (amu) were recorded. GC/MS-SIM settings were 357.2 and 372.2 (BPA-TMS) (amu) and 369.2 and 384.2 (13C12 BPA-TMS) (amu). Quantification ions and identification ions criteria for native and 13C labeled BPA are as follows:
|
| ||||
| Components | Quantification ion (amu) | Intensity (%) | Identification Ion (amu) | Intensity (%) |
|
| ||||
| BPA | 352.2 | 100 | 372.2 | 13 |
| 13C12-BPA | 369.2 | 100 | 384.2 | 13 |
|
| ||||
Recoveries (13C12 BPA-TMS) were between 65 and 112%.
QC/QA measures included a five point calibration curve, recalibration within each sequence of analysis (with a minimum of one blank in each batch of a maximum of 10 samples), and duplicate analyses of >50% of positive samples. The duplicate results matched in >90% of cases. When duplicates did not match, a third analysis was performed. The mean laboratory blank level for BPA was 0.43 ng absolute (n:15) with a standard deviation (SD) of 0.12 ng absolute. The pre-study method evaluation with test samples resulted in limit of detection (LOD) for BPA of 0.05 ng/g ww with a limit of quantification (LOQ) of 0.11 ng/g ww based on a 15 g ww sample. The actual study sample LOQs were 0.20 ng/g ww for food and 20 ng/L for liquids. Reported concentration measurements are the average values of the matched results.
Further information on the Eurofins BPA method for food analysis can be found at Eurofins (2014).
2.3. Exposure assessment
A “forward” approach was used to estimate a daily intake of BPA from food ingestion. This forward approach combined information on BPA concentrations in food with daily consumption rates to arrive at BPA intake. Intakes of BPA calculated in this manner were compared with intakes determined using a dose reconstruction, or “backward”, approach which started with a measurement of BPA in urine and back-calculated the intake necessary to have resulted in this measurement. BPA is an analyte routinely measured in urine in NHANES. For this study, we used NHANES data from the 2005/6, 2007/8, and 2009/10 cycles. The primary benefit to dose reconstruction estimates is that they integrate all pathways of exposure. Thus, by comparing the forward-based estimates with backward-based estimates, one can get an understanding of the validity and importance of the pathways being assessed in the forward-based approach.
The food samples from both rounds of sampling were combined into one data set. All samples were categorized into broad food types: vegetables; fruit; dairy; meat (including pork, beef, and poultry); fruit juices; and fish. For each category, samples were separated into “canned” samples and “all other” samples (which included fresh and frozen food products). Then, for each category, a “fraction canned” was determined. This is defined as the fraction of total consumption that is from a canned product. The fraction of all other consumption is simply, 1 — “fraction canned”. With this framework, total intake exposure from canned foods and all other foods can be estimated as:
Where
| It | = Total BPA intake from all foods, ng/kg-day |
| CRf | = consumption rate for food type f, g/kg-day |
| FCf | = fraction of CR of food type f that comes from canned food, unitless |
| CONCf,c/o | = concentration of food type f from canned “c” or from “o” other sources, ng/g |
Table 1 provides the values used in this assessment. Intake calculations were done only for adults in this assessment. Intakes as well as concentrations in the foods were mean values, and concentration means were determined with non-detects (ND) set equal to zero. Arithmetic mean (not geometric mean) concentrations were used instead of median concentrations because the consumption rates (amount of foods eaten) were provided as arithmetic mean values. An alternate intake using median concentrations with these mean consumption rates was derived for comparison. Discussions below examine the assumption for non-detects and also for pathways of exposure not considered here, such as dust ingestion or thermal paper exposure. Other food pathways are noted, including for foods sampled here but without detections, such as dairy, and for foods not sampled here, such as canned soft drinks and alcoholic beverages.
Table 1
Parameters used in the forward calculation of adult BPA intakes by food ingestion.
| Parameters | Value | Citation |
|---|---|---|
| I. Consumption rate, mean | ||
| Vegetable, g/kg-day | 2.5 | Per capita, ages 21–50, EPA (2011) |
| Fruit, g/kg-day | 0.9 | Per capita, ages 21–50, EPA (2011) |
| Meat, g/kg-day | 1.8 | Per capita, ages 21–50, EPA (2011) |
| Fish, g/kg-day | 0.23 | Per capita, finfish/shellfish, ages 20–50, EPA (2011) |
| Fruit juice, g/kg-day | 0.12 | ERS food availability data basea |
| II. Fraction canned | ||
| Vegetable | 0.24 | ERS food availability data basea |
| Fruit | 0.06 | ERS food availability data basea |
| Meat | 0.05 | Estimate, using information from Daniel et al. (2010) |
| Fish | 0.24 | ERS food availability data basea |
| Fruit juice | 0.50 | Estimate |
| III. Average BPA concentration, ng/g wwb
| ||
| Canned | Other | |
|
| ||
| Vegetables | 19.83 | 0.04 |
| Fruit | 0.26 | 0.02 |
| Meat | 2.01 | 0.02 |
| Fish | 1.08 | 0 |
| Fruit juice | 0.61 | 0.03 |
The consumption rates, CR, for each food type were determined for adults from EPA’s Exposure Factors Handbook (EFH) (EPA, 2011), with the exception of fruit juice, which was obtained from the United State Department of Agriculture (USDA) Economic Research Service’s (ERS) Food Availability data base (http://www.ers.usda.gov/data-products/food-availability-(per-capita)-data-system.aspx#.UidVCtKshNc). The EFH data used here were “recommended” values suggested by EPA for adult exposures, for general exposure assessment purposes. They were developed as arithmetic mean per capita intakes. “Per capita” is defined as all individuals, both consumers and non-consumers. The selected food consumption rates pertain to individuals aged 20–50, and were derived from food intake surveys performed in the early 2000s. The ERS food availability database provides per capita mean consumption rates for all ages for several food categories. While this is not ideal for adult exposures, a fruit juice consumption rate is not available in the EFH. Without providing an excess of detail, this database is, in essence, a nationwide mass balance of over 200 specific food commodities such as dairy products, nuts, meat, poultry, and seafood. Considering beginning and ending stocks, and known losses such as exports, this database calculates the total amount of the commodity “available” for consumption annually. By dividing by the US population, one gets an annual per capita consumption rate, which is easily converted to a daily consumption rate. Note this procedure does not consider food discarded or not otherwise consumed, so the tendency will be for consumption rates to be overestimated. The fruit juice consumption rate determined using this database was 0.12 g/kg-day. Very little information was available to assess exposure to “canned” foods versus “non-canned” (fresh, frozen, packaged other than in cans, and so on). Christensen and Lorber (2014) determined “canned fractions” for children’s food consumption rates using a database titled, “What we Eat in America” (see EPA, 2013). We developed a “fraction canned”, or FC, uniquely for this assessment for adult exposures. The ERS database noted above was the source of information for the FC for fruit, vegetables, and fish. For certain foods, the ERS database provides per capita availability of food commodities not only on a “total” basis, but also on a “canned” basis (i.e., two results are reported — the total per capita availability and the per capita availability for that portion packaged in cans). The FC is then determined as canned availability divided by total availability. The ERS-derived FC values were 0.24, 0.06, and 0.24 for vegetables, fruits, and fish, respectively, averaged over the years 2006–2010. Similar data are not available for meat and fruit juice in the ERS database. For meat, the most comparable information was provided in Daniel et al. (2010) who reported on trends in meat consumption in the US over time, including percent of meat that was “processed” including frankfurters, sausage, and luncheon meats, but not cured meats such as ham or bacon. Some processed meats are packaged in plastic and some in cans, including corned beef, chunk white chicken, and Spam®. All other meats were defined as “fresh”. Their analysis reported that “processed” meat was 22% of all meat consumed. With no other information found, we assumed that the FC for meats is half of what Daniel et al. (2010) found for processed meats, or 0.11 (11%). No information was found for the fraction of fruit juice that is canned. Canned juices might include frozen concentrated orange juice or other canned juices. Juices are also marketed in plastic or cardboard containers. For this assessment, we assumed that the FC for fruit juice was 0.50 (50%).
3. Results
Table 2 provides a listing of the samples with measurable levels (i.e., samples above the quantitation limit, QL, of 0.2 ng/g ww) uniquely reported here; recall that a first round of sampling had been reported in Schecter et al. (2010). Of 116 samples in the second round of sampling, a total of 31, or 27%, were measurable for BPA. Twenty-six of 37 canned samples, or 70%, were measurable, while only five of 79 non-canned foods, or 6%, were measurable for BPA. Quantified concentrations ranged from 0.24 ng/g ww in a fresh peach to 149.0 ng/g ww in canned cut green beans. The second and third highest measurements were 121.0 ng/g ww in canned corn and 80.6 ng/g ww in canned mixed vegetables.
Table 2
BPA levels in fresh, canned and frozen foods that were above the quantitation level of 0.2 ng/g ww.
| Food | Type | Food category | BPA level (ng/g ww) |
|---|---|---|---|
| Cut green beans no salt added | Canned | Vegetables | 149.00 |
| Golden sweet corn | Canned | Vegetables | 121.00 |
| Original mixed vegetables | Canned | Vegetables | 80.60 |
| Sweet peas | Canned | Vegetables | 66.90 |
| Whole peeled tomatoes no salt added | Canned | Vegetables | 26.60 |
| Small sweet peas | Canned | Vegetables | 20.50 |
| Cut green beans | Canned | Vegetables | 5.60 |
| Peeled tomatoes | Canned | Vegetables | 5.19 |
| Whole kernel sweet corn | Canned | Vegetables | 3.19 |
| Cut green beans | Canned | Vegetables | 2.85 |
| Whole kernel corn | Canned | Vegetables | 2.41 |
| Premium mixed vegetables | Canned | Vegetables | 1.69 |
| Whole tomatoes | Canned | Vegetables | 0.56 |
| Sweet peas chicharos | Canned | Vegetables | 0.35 |
| Mixed vegetables | Canned | Vegetables | 0.34 |
| 98% fat free chicken breast | Canned | Meat | 5.70 |
| Premium quality corned beef | Canned | Meat | 3.48 |
| Premium quality deviled ham spread | Canned | Meat | 2.36 |
| Corned beef | Canned | Meat | 1.73 |
| Corned beef | Canned | Meat | 1.64 |
| Spam classic | Canned | Meat | 0.78 |
| Lean smoked ham water added | Canned | Meat | 0.33 |
| Chunk white chicken | Canned | Meat | 0.26 |
| Canned light tuna | Canned | Fish | 1.07 |
| Mandarin oranges whole segments | Canned | Fruit | 2.03 |
| Yellow cling peach slices | Canned | Fruit | 0.31 |
| Fresh green beans | Non-canned | Vegetables | 0.38 |
| Fresh green beans | Non-canned | Vegetables | 0.28 |
| Frozen mixed vegetables | Non-canned | Vegetables | 0.31 |
| Fresh peach | Non-canned | Fruit | 0.24 |
| 100% vegetable juice cocktail | Non-canned | Fruit/veg juice | 0.41 |
Table 3 provides a summary of concentrations by category for the samples in the second round of sampling compared to the earlier first round of sampling. As seen in Table 3, data from both rounds of sampling are consistent, which is not unexpected. Samples from both rounds were obtained in the same year — the first round occurred in March 2010 and the second in August 2010. The same laboratory conducted the analysis using the same protocol. Medians and percent positives were consistent between the two rounds. The mean of canned vegetables for the second round, 32.5 ng/g, was higher than the mean from the first round, 11.9 ng/g. However, the medians were essentially the same, 5.7 (2nd round) and 5.2 (1st round) ng/g. The mean from the second round was driven by four samples that were >66 ng/g ww. Because of this comparability, we combined data from both surveys for purposes of the exposure assessment (See Supplemental Material, Table S1 for the combined sampling results). The combined data set of 204 samples included 92 samples of frozen and fresh foods and 112 samples of canned food. Overall, BPA was detected in 7% of fresh and frozen foods, and in 73% of canned foods.
Table 3
Summary of food concentrations from the current and the previous studya used to derive mean concentrations for exposure calculations.
| Food type | N | Percent positive | Meanb (ng/g ww) | Medianb (ng/g ww) | Range of positives |
|---|---|---|---|---|---|
| I. Current study — canned | |||||
| Vegetables | 15 | 100 | 32.5 | 5.2 | 0.3–149 |
| Fruit | 6 | 33 | 0.4 | ND | 0.3–2 |
| Meat | 11 | 73 | 1.5 | 0.8 | 0.3–6 |
| Fish | 3 | 33 | – | – | 1 |
| Dairy | 2 | 0 | – | – | – |
| II. Current study — all other packaging | |||||
| Vegetables | 24 | 13 | 0.04 | ND | 0.3–0.4 |
| Fruit | 9 | 11 | 0.02 | ND | – |
| Meat | 12 | 0 | – | – | – |
| Fish | 3 | 0 | – | – | – |
| Dairy | 15 | 0 | – | – | – |
| Fruit & vegetable juice | 16 | 6 | – | – | 0.4 |
| III. Previous study — canned | |||||
| Vegetables | 24 | 88 | 11.9 | 5.7 | 0.4–65 |
| Fruit | 3 | 0 | – | – | – |
| Meat | 27 | 85 | 2.2 | 1.2 | 0.3–7 |
| Fish | 9 | 66 | 1.3 | 0.9 | 0.8–4 |
| Dairy | 6 | 0 | – | – | – |
| Fruit & vegetable juice | 6 | 100 | 0.6 | 0.6 | 0.5–0.8 |
| IV. Previous study — all other packaging | |||||
| Fruit | 3 | 0 | – | – | – |
| Meat | 3 | 33 | – | – | 0.1 |
| Fish | 1 | 0 | – | – | – |
| Dairy/grain (mac and cheese) | 3 | 0 | – | – | – |
| Meat/grain (spaghetti & meat) | 3 | 0 | – | – | – |
| V. Previous study — other | |||||
| Infant formula — canned | 9 | 30 | 0.4 | ND | 1–1.2 |
| Pet food — canned | 8 | 38 | 0.1 | ND | 0.2–0.3 |
The final adult food ingestion intake was determined from consumption rates, fractions canned, and the concentrations shown in Table 1. Overall calculated intake was 12.6 ng/kg-day, with canned food accounting for 12.4 ng/kg-day, or 98%, of the calculated food intakes. Canned vegetables accounted for 11.9 ng/kg-day, with canned meat second at 0.4 ng/kg-day. The substitution of ND with zero did not greatly affect this outcome. If instead a value of half the QL, or 0.1 ng/g ww, was substituted, total intake would rise from 12.6 to 13.3 ng/kg-day. The change is almost entirely due to the categories that had all or mostly non-detects. At ND = 0, the food group averages were 0.0 (no detects in the group) or very low, 0.02 ng/g ww (which might occur, for example, with 1 detection near the QL of 0.2 ng/g ww). At ND = ½ QL, the food group average became nearly half QL at 0.1 ng/g ww.
This intake is lower than total intakes determined from urine measurements of BPA from NHANES. Lakind and Naiman (2011) used simple dose reconstruction methods to estimate geometric mean intakes of 39 and 29 ng/kg-day for ages of 20–39 and 40–59 years, respectively, from NHANES 2005/6. For the next cycle of NHANES, 2007/8, Lakind et al. (2012) estimated geometric mean intakes of 44 and 32 ng/kg-day for these same age ranges. Using similar procedures as used in these two studies, we determined geometric mean intakes of 34 and 30 ng/kg-day for the same two age ranges for the 2009/10 NHANES cycle (analysis not shown). However, the intakes calculated in this study from food concentrations and food consumption rates used means for both parameters, so it may be more appropriate to compare the overall 12.6 ng/kg-day mean intake to an analogous mean dose-reconstructed intake from the NHANES data base, rather than the geometric mean [medians were also provided in the Lakind and Naiman (2011) and Lakind et al. (2012) articles that were slightly lower than the geometric means]. Using similar techniques for dose reconstruction, we calculated a mean intake of BPA for adults ages 20–59 of 68 ng/kg-day for the NHANES 2009/10 sampling cycle (analysis not shown). Similarly, one could also calculate a “median” dietary intake rather than a “mean” dietary intake as done in this paper. This might provide a better direct comparison to the median or geometric mean intakes noted above. Although median exposure consumption rates were not provided in the EFH (EPA, 2011), concentrations could be assigned median instead of mean values. Looking only at canned vegetables (which dominated the intakes), the median of the combined dataset was 5.2 ng/kg ww, compared to the mean of 19.8 ng/kg ww. This median is about one quarter of the mean, so a quick estimate of the “median” intake might be one-fourth that calculated using means, or about 3.2 ng/kg-day.
Overall, we found that the central tendency of dietary intakes were in the range of 3 to 13 ng/kg-day. This is lower than the central tendency of intakes derived from urine measurements, which were found to be in the range of 30 to 70 ng/kg-day.
High-end intakes can result from exposure to foods with the highest levels of BPA, and also from other pathways, discussed below. For example, if one were to consume an average amount of canned vegetables of 0.6 g/kg-day (calculated as 2.5 g/kg-day total vegetables × 0.24 fraction canned, Table 1) at a concentration of 150 ng/g (the highest found in our survey was 149 ng/g), the exposure would be 90 ng/kg-day. If all of this individual’s average consumption was canned instead of only 24% of it, than his/her exposure could exceed 300 ng/kg-day NHANES data also show high end exposures in the general population that could be on the order of 5 times higher than central tendency exposures. The 95th percentile exposure determined from urine data from NHANES 2009/10 for adults age 20 to 59 is about 170 ng/kg-day (analysis not shown), compared to a geometric mean of about 32 ng/kg-day. There is uncertainty in these calculations aimed at understanding the high end of BPA exposures, but clearly the possibility exists for individuals to have occasional days where their exposures could be several times higher than what we have calculated as the central tendency of exposure.
4. Discussion
This paper reports on a second round of sampling of foods for BPA contamination, and combines these levels with an earlier sampling to also conduct an exposure assessment of BPA from the same geographical location. An overall intake of 12.6 ng/kg-day for adults was calculated, dominated by intakes from consumption of canned vegetables. A second measure of intake can be derived from biomonitoring, specifically in this case from urine concentrations of BPA. This intake from biomonitoring represents all pathways of exposure, and the total intake derived from NHANES was substantially higher than the food intake of 12.6 ng/kg-day. Specifically, the total intake representing a central tendency (median as well as mean, and for different years including the 2009/10 cycle which would correspond to our food sampling time) was in the range of 30 to 70 ng/kg-day.
Clearly, there is a discrepancy between BPA intakes derived from our food data and urine measurements. One possibility is that our concentrations of BPA are low and not representative of typical US food concentrations although our samples were purchased from nationwide supermarket chains, which do not typically sell local foods; but more representative food concentrations in this assessment might have led to intake estimations nearer to those found using NHANES. Table 4 provides a comparison of canned food concentrations from this study with canned food concentrations from six other studies in the literature. Noonan et al. (2011) and Liao and Kannan (2013), reported on canned food sampled in the United States, while the other four studies were from Canada (Cao et al., 2011), Europe (EFSA, 2013), New Zealand (Thomson and Grounds, 2005), and Belgium (Geens et al., 2010). Generally, these were market basket surveys, reasonably comprehensive in scope and intent. All had provided information specifically on canned food concentration, allowing for this table and comparison.
Table 4
Canned food concentrations of BPA from our study compared with findings from other studies.
| Food category | Data from this study
| Data from other studies
| Location and specific foods | ||
|---|---|---|---|---|---|
| % Pos (n) | Meana ng/g ww | % Pos (n) | Meana ng/g ww | ||
| Vegetables | 92 (39) | 20 | 100 (6) | 23 | Canadianb; beans, beets, peas, tomatoes |
| 82 (73) | 22 | Europec; unspecified | |||
| 65 (17) | 12 | New Zealandd; tomatoes, corn, beetroot, peas | |||
| 96 (25) | 88 | USAe; green beans, corn, tomatoes, peas, misc. veg (wax beans, spinach, stir fry, oyster mushrooms, almond jelly) | |||
| 92 (12) | 31 | USAf; green beans, carrots, tomatoes, tomato paste, mushrooms, kidney beans, potatoes, olives | |||
| 100 (10) | 37 | Belgiumg; carrots, corn, tomatoes, bamboo, mushrooms, olives; 2 veg soups included | |||
| Fruit | 22 (9) | 0.3 | 50 (4) | 1 | Canadianb; cherries, pineapple |
| 79 (14) | 12 | Europec; unspecified | |||
| 0 (16) | ND | New Zealandd; apricots, peaches, pineapple, fruit salad | |||
| 57 (14) | 5 | USAe; fruit cocktail, pineapple, sliced peaches | |||
| 100 (3) | 3 | USAf; mandarin orange, pear, pineapple | |||
| 100 (4) | 12 | Belgiumg; fruitmix, peaches, pears, applesauce | |||
| Meat | 82 (38) | 2 | 100 (1) | 10 | Canadianb; luncheon meats |
| 62 (16) | 64 | Europec; unspecified | |||
| 33 (6) | 21 | New Zealandd; meat unspecified | |||
| 100 (17) | 58 | USAe; pork & beans, chili | |||
| 100 (5) | 3 | USAf; chicken (breast, sausages), ham | |||
| 100 (2) | 26 | Belgiumg; chicken soup, sausages, ravioli which may have been partially meat not included | |||
| Fish | 58 (12) | 1 | 100 (1) | 106 | Canadianb; unspecified |
| 66 (67) | 33 | Europec; unspecified | |||
| 50 (8) | 23 | New Zealandd; salmon, tuna | |||
| 100 (6) | 12 | USAe; tuna, albacore, mackerel | |||
| 88 (8) | 7 | USAf; sardines, clams, tuna, unspecified | |||
| 100 (4) | 75 | Belgiumg; 2 tuna samples same brand at 169 and 126 ng/g, salmon and anchovy low at 3.6 and 0.9 ng/g | |||
| Dairy | 0 (8) | ND | 100 (1) | 15 | Canadianb; evaporated milk |
| 100 (3) | 20 | Europec; unspecified | |||
| Fruit/vegetable juice | 100 (6) | 0.6 | 50 (4) | 0.2 | Canadianb; vegetable, apple, citrus |
| 0 (2) | ND | USAf; orange, grape | |||
| 100 (4) | 3 | Belgiumg; apple, orange, vegetable, tropical juice | |||
Canned foods from our current and previous surveys are mostly lower in BPA content than canned foods in other surveys, with the exception of vegetables. The mean of 39 canned vegetable samples over the two rounds of sampling reported here is about 20 ng/g ww with a high detection frequency of 92%, comparable to mean concentrations of 12, 22, 23, 31, and 37 ng/g ww, and detection frequencies of 65, 82, 100, 92, and 100%, from five of the six other studies. The data from the remaining study, which was from the US, showed an average of 88 ng/g ww, with 24 of 25 samples showing a positive detection (96%) (Noonan et al., 2011). Six of 25 samples in this other US study had concentrations higher than 100 ng/g ww while only 2 of the 39 canned vegetable samples in our study had concentrations over 100 ng/g ww. BPA concentrations in canned meat were higher by over an order of magnitude in all other studies except for Liao and Kannan (2013) where a mean concentration of 3 ng/g ww was found, similar to the 2 ng/g ww found here. For the other categories of fruit, fish, and dairy, the literature reported mean concentrations that were between 1 and 2 orders of magnitude higher than levels found in this survey. Interestingly, the three studies from the US are lower in canned fish BPA concentration as compared to four other studies outside of the US. The canned fish in the three US studies are 1 (this study), 7, and 12 ng/g, while it was 23, 33, 75, and 106 in four other studies. The 106 ng/g average was from only one sample tested in Canada (Cao et al., 2011) so it likely was not representative, but two samples of tuna (from different cans of the same brand) were above 130 ng/g in the study originating from Belgium (Geens et al., 2010).
It is unclear why the foods in our survey are lower than those in other surveys. Quality assurance and control procedures described earlier would suggest validity for the analysis of the samples in this study. It is worth noting that the data reported here includes two sampling cycles and hence, two laboratory analysis cycles, and as described earlier, concentrations found were consistent between sampling cycles. It is possible that differences in analytical methodologies between the various studies might explain the difference. Noonan et al. (2011) also notes that concentrations in our first round of sampling are lower than found in their survey of foods in the Washington DC metropolitan area, and identifies differences in analytical protocols that might explain the difference. For example, they note our use of a freeze-drying step prior to the addition of internal standard, which they did not use and was not used in the Canadian total diet study (Cao et al., 2011). In addition to differences in analytical protocols, possibly an in-depth examination of the specific sources of food, whether national or local brands, could explain the differences in at least the three studies from the United States. Assuming that the food concentrations in this study are indeed lower than typical for US foods, alternate concentrations can be substituted for the final concentrations used in the exposure assessment shown in Table 1. For example, if one to average the study averages from the three US studies — this one, Noonan et al. (2011) and Liao and Kannan (2013) — higher concentrations for use in the exposure assessment would result. For example, the canned vegetable concentration would be 46.3 ng/g instead of 20 ng/g. Using 3-study averages for all foods, a recalculated total food intake would be 32.5 ng/kg-day, now within the range of 30 to 70 ng/kg-day surmised from NHANES data.
Another potential source for the difference between the forward-based intake estimate and the biomonitored intake estimate comes from the fact that the survey did not include some canned products which could be impacted by BPA, such as canned beverages other than juices. The EFSA report (EFSA, 2013) contains a comprehensive set of data on the occurrence of BPA in canned beverages, including carbonated and non-carbonated, alcoholic and non-alcoholic, and from several European countries. EFSA found BPA in over 50% of samples in the various studies of canned beverages they included in their compilation, with study-specific high concentrations in the 5 to greater than 20 μg/L range, and central tendency concentrations in the 0.5 to 1.0 μg/L range. Cao et al. (2009) analyzed 69 canned soft drink products in Canada, and BPA was detected in all samples (detection limit = 0.045 μg/L), with 85% of the samples less than 1 μg/L and a high of 4.5 μg/L. If an individual weighing 60 kg consumed one 12 oz. (0.35 L) can of a beverage containing 1 μg/L of BPA, his/her intake would be about 6 ng/kg-day, which is almost 50% of the 12.6 ng/kg-day calculated from all foods otherwise.
Finally, there is the possibility that there are non-food or certain food pathways not considered in this exposure assessment. EFSA (EFSA, 2013) considered several foods and beverages for European exposures which were not considered here, including grains/grain-based products, legumes/nuts/oilseeds, dairy, alcoholic and non-alcoholic beverages, and drinking water impacted by polycarbonate (PC) sources (piping or water containers produced using PC). They also quantified important non-food pathways. Using data from around the world, and making estimates of transfers in the case of dermal contact, they estimated these mean exposures for adults for non-dietary pathways: thermal paper (shopping receipts, credit card receipts, etc.) −18 ng/kg-day, cosmetics −1.2 ng/kg-day, indoor air −0.7 ng/kg-day, and dust (mostly ingestion) −0.1 ng/kg-day (EFSA, 2013). From all pathways, EFSA (EFSA, 2013) determined a mean forward-based intake of about 45 ng/kg-day for adults. This EFSA (EFSA, 2013) analysis of about 20 ng/kg-day from non-food pathways suggests about half of all typical exposures for adults may come from these non-food pathways.
5. Conclusion
Our approach to assess dietary intakes of BPA for adults is unique in that it separately considered canned foods and non-canned foods, necessitating the development of a “canned fraction” to adjust overall food consumption. To our knowledge, this had been done only once before for assessing children’s exposure to BPA (Christensen and Lorber, 2014) using a different data source to assigned “canned fractions”. There was already an understanding that BPA concentrations were much higher in canned as compared to non-canned foods, and our survey and others separated out canned foods for analysis. Our analysis suggests that further work is necessary to fully understand the extent of adult exposures to BPA by different canned foods and by different pathways, particularly by thermal paper. More samples of representative US foods are indicated and a standard analytical approach should lead to more refined estimates of BPA exposure routes and amounts. Following changes in levels over time and from various geographical regions is indicated.
At present, the intakes calculated from biomonitoring and food are both well below current government benchmarks. The EPA’s Reference Dose, RfD (IRIS, 2014) is set at 50 μg/kg-day, and as of January 2014, the European Union’s Tolerable Daily Intake, TDI (EFSA, 2014) was reduced from 50 to 5 μg/kg-day. These are about 2 to 3 orders of magnitude higher than the central tendency ranges of intakes found in this study and elsewhere, which are in the ng/kg-day range. As noted above, high end intakes (at the 95th percentile) are only about 5 times higher than the median levels, so one can surmise that total intakes at the high end of the general population are still substantially less than these benchmark levels. However, caution should be exercised in drawing conclusions based on this comparison. The current EPA RfD was published in 1993, and hundreds of studies have been conducted since then, better characterizing the potential for health impacts from exposure to BPA including the potential for health effects at low doses (NTP, 2007). A recent survey article reviews the literature on low dose studies and identifies uncertainties with their use and interpretation (Teeguarden and Hanson-Drury, 2013). The Food and Drug Administration (FDA) of the US also conducted an assessment of the recent toxicity data of BPA, including low dose studies (see http://www.fda.gov/food/ingredientspackaginglabeling/foodadditivesingredients/ucm064437.htm). They identified uncertainties with low dose toxicity studies, and concluded that recent toxicity data does not change their decision to use the existing NOAEL (No Observed Adverse Effect Level) of 5 mg/kg-day for risk assessment purposes. It should be noted that BPA is an endocrine disruptor with the ability to interact not only with multiple estrogen receptor forms, but also several other nuclear and membrane-bound receptors, and neither EPA’s RfD nor the European TDI take into account the cumulative exposures to multiple environmental endocrine disruptors to which humans are exposed.
In any case, widespread exposure to BPA has prompted the search for alternatives for use of BPA in polycarbonate baby bottles (C&EN News, 2009) and for food companies to remove BPA from canned foods. For example, Campbell’s announced the removal of BPA in its soup products in 2009 (http://www.greenbiz.com/blog/2012/03/06/campbells-stop-using-bpa-soup-cans). Lakind (2013) discusses the issues with can coatings for food with a focus on the use of BPA, and Lakind and Birnbaum (2010) caution that replacement chemicals for BPA in canned foods could pose issues all their own. For example, Kinch et al. (2015) found that low doses of both BPA and Bisphenol S (BPS), a common analog used in BPA-free products, induced precocious hypothalamic neurogenesis in zebrafish. They cautioned that BPA-free products are not necessarily safer.
In conclusion, this study, along with Christensen and Lorber (2014), underscores the important role of canned foods in children and adult exposures to BPA.
Acknowledgments
This study reflects the views of the authors and not the policies of the U.S. Environmental Protection Agency. This study was funded by the intramural research program of the National Cancer Institute/National Institutes of Health, and also the Pfeiffer Research Foundation. The authors appreciate the assistance of Susan Euling of the Environmental Protection Agency who provided input regarding impacts from exposure to BPA.
Appendix A. Supplementary data
Supplementary data to this article can be found online at http://dx.doi.org/10.1016/j.envint.2015.01.008.
Footnotes
Disclaimer
The views expressed in this article are those of the authors and do not necessarily reflect the views or policies of the U.S. Environmental Protection Agency. Mention of trade names or commercial products does not constitute endorsement or recommendation for use.
Competing financial interest declaration
All authors declare they have no actual or potential competing financial interest.
References
- Biedermann S, Tschudin P, Grob K. Transfer of bisphenol A from thermal printer paper to the skin. Anal Bioanal Chem. 2010;398(1):571–576. [PubMed] [Google Scholar]
- C&EN News. Babies on board. Chemical and Engineering News. 2009 Aug 31;:20. [Google Scholar]
- Cao XL, Corriveau J, Popovic S. Levels of bisphenol A in canned soft drink products in Canadian markets. J Agric Food Chem. 2009;57:1307–1311. [PubMed] [Google Scholar]
- Cao XL, Perez-Locas C, Dufresne G, Clement G, Popovic S, Beraldin F, et al. Concentrations of bisphenol A in the composite food samples from the 2008 Canadian total diet study in Quebec City and dietary intake estimates. Food Addit Contam, Part A. 2011;28(6):791–798. [PMC free article] [PubMed] [Google Scholar]
- CDC. Fourth National Report on Human Exposure to Environmental Chemicals. Department of Health and Human Services: Centers for Disease Control and Prevention, National Center for Environmental Health; Atlanta, Georgia: 2009. (Available at, http://www.cdc.gov/exposurereport/pdf/FourthReport.pdf. Accessed 7/15/2013) [Google Scholar]
- Christensen KLY, Lorber M. Exposure to BPA in children – media-based and biomonitoring-based approaches. Toxics. 2014;2:124–157. [Google Scholar]
- Cobellis L, Colacurci N, Trabucco E, Carpentiero C, Grumetto L. Measurement of bisphenol A and bisphenol B levels in human blood sera from healthy and endometriotic women. Biomed Chromatogr. 2009;23(11):1186–1190. [PubMed] [Google Scholar]
- Daniel CR, Cross AJ, Koebnick C, Sinha R. Trends in meat consumption in the USA. Public Health Nutr. 2010;14(4):575–583. [PMC free article] [PubMed] [Google Scholar]
- EFSA. EFSA Panel on Food Contact Materials, Enzymes, Flavourings and Processing Aids (CEF) European Food Safety Authority; 2013. DRAFT Scientific Opinion on the risks to public health related to the presence of bisphenol A (BPA) in foodstuffs — part: exposure assessment. ( http://www.efsa.europa.eu/en/consultations/call/130725.pdf Accessed 7/15/2013) [Google Scholar]
- EFSA. EFSA Topic: Bisphenol A. 2014 http://www.efsa.europa.eu/en/topics/topic/bisphenol.htm (Accessed 12/15/2014)
- Environmental Protection Agency (EPA) Exposure Factors Handbook 2011 Edition (Final) 2011 (EPA/600/R-09/052F; Available at, http://cfpub.epa.gov/ncea/risk/recordisplay.cfm?deid=236252 Accessed 7/15/2013)
- Environmental Protection Agency (EPA) Food commodity intake database. What We Eat In America. 2013 (Available online: http://fcid.foodrisk.org/ Accessed 8/31//2013)
- Eurofins. Bisphenols in food. Eurofins Implements Innovative UHPLC–ESI-MS/MS Technique. 2014 (Available online: http://www.eurofins.de/media/9981108/bisphenol_a_eng.pdf Accessed 12/17/2014)
- FDA. Final Rule 77 FR 41899. Food and Drug Administration; Washington, D.C: 2012. Indirect food additives: polymers. (July 17, 2012; http://www.gpo.gov/fdsys/pkg/FR-2012-07-17/html/2012-17366.htm. Accessed 11/1/2013) [Google Scholar]
- FDA. Indirect Food Additives: Adhesives and Components of Coatings. Final Rule Food and Drug Administration; Washington, D.C: 2013. (78 FR 41840, July 12, 2013; https://www.federalregister.gov/articles/2013/07/12/2013-16684/indirect-food-additives-adhesives-and-components-of-coatings. Accessed 11/1/2013) [Google Scholar]
- Fleisch AF, Sheffield PE, Chinn C, Edelstein BL, Landrigan PJ. Bisphenol A and related compounds in dental materials. Pediatrics. 2010;126(4):760–768. [PMC free article] [PubMed] [Google Scholar]
- Geens T, Roosens L, Neels H, Covaci A. Assessment of human exposure to bisphenol-A, triclosan and tetrabromobisphenol-A through indoor dust intake in Belgium. Chemosphere. 2009;76(6):755–760. [PubMed] [Google Scholar]
- Geens T, Apelbaum TZ, Goeyens L, Neels H, Covaci A. Intake of bisphenol A from canned beverages and foods on the Belgian market. Food Addit Contam. 2010;27:1627–1637. [PubMed] [Google Scholar]
- He Y, Miao M, Herrinton LJ, Wu C, Yuan W, Zhou Z, Li DK. Bisphenol A levels in blood and urine in a Chinese population and the personal factors affecting the levels. Environ Res. 2009;109(5):629–633. [PubMed] [Google Scholar]
- IRIS. Bisphenol A. (CASRN 80-05-7) IRIS, US EPA; 2014. ( http://www.epa.gov/iris/subst/0356.htm. Accessed 11/5/2014) [Google Scholar]
- Kang JH, Kondo F, Katayama Y. Human exposure to bisphenol A. Toxicology. 2006;226(2–3):79–89. [PubMed] [Google Scholar]
- Kinch CD, Ibhazehiebo K, Jeong JH, Habibi HR, Durrasch DM. Low-dose exposure to bisphenol A and replacement bisphenol S induces precocious hypothalamic neurogenesis in embryonic zebrafish. Proc Natl Acad Sci. 2015 http://dx.doi.org/10.1073/pnas.1417731112 (Published ahead of print 1/12/2014) [PMC free article] [PubMed]
- Lakind JS. Can coatings for food and beverages: issues and options. Int J Technol Policy Manage. 2013;13:80–95. [Google Scholar]
- Lakind JS, Birnbaum LS. Out of the frying pan and out of the fire: the indispensable role of exposure science in avoiding risks from replacement chemicals. J Expo Sci Environ Epidemiol. 2010;20:115–116. [PubMed] [Google Scholar]
- Lakind JS, Naiman DQ. Bisphenol A (BPA) daily intakes in the United States: estimates from the 2003–2004 NHANES urinary BPA data. J Expo Sci Environ Epidemiol. 2008;18:608–615. [PubMed] [Google Scholar]
- Lakind JS, Naiman DQ. Daily intake of bisphenol A and potential sources of exposure: 2005–2006 National Health and Nutrition Examination Survey. J Expo Sci Environ Epidemiol. 2011;21:272–279. [PMC free article] [PubMed] [Google Scholar]
- Lakind JS, Levesque J, Dumas P, Bryan S, Clarke J, Naiman DQ. Comparing United States and Canadian population exposures from National Biomonitoring Surveys: bisphenol A intake as a case study. J Expo Sci Environ Epidemiol. 2012;22:219–226. [PMC free article] [PubMed] [Google Scholar]
- Liao C, Kannan K. Concentrations and profiles of bisphenol A and other bisphenol analogues in foodstuffs from the United States and their implications for human exposure. J Agric Food Chem. 2013;61:4655–4662. [PubMed] [Google Scholar]
- Liao C, Liu F, Kannan K. Bisphenol S, a new bisphenol analogue, in paper products and currency bills and its association with bisphenol A residues. Environ Sci Technol. 2012;46:6515–6522. [PubMed] [Google Scholar]
- Loganathan SN, Kannan K. Occurrence of bisphenol A in indoor dust from two locations in the eastern United States and implications for human exposures. Arch Environ Contam Toxicol. 2011;61:68–73. [PubMed] [Google Scholar]
- Noonan GO, Ackerman LK, Begley TH. Concentration of bisphenol A in highly consumed canned foods on the U.S. market. J Agric Food Chem. 2011;59:7178–7185. [PubMed] [Google Scholar]
- NTP. NTP-CERHR Expert Panel Report on the Reproductive and Developmental Toxicity of Bisphenol A; Report No. NTP-CERHR-BPA-07. National Toxicology Program, U.S. Department of Health and Human Services, Center for the Evaluation of Risks to Human Reproduction; 2007. (Published Nov 26, 2007; Available at, http://wf1-ext-vip.niehs.nih.gov/evals/bisphenol/BPAFinalEPVF112607.pdf Accessed 7/15/2013) [Google Scholar]
- Otaka H, Yasuhara A, Morita M. Determination of bisphenol A and 4-nonylphenol in human milk using alkaline digestion and cleanup by solid-phase extraction. Anal Sci. 2003;19(12):1663–1666. [PubMed] [Google Scholar]
- Rochester JR. Bisphenol A and human health: a review of the literature. Reprod Toxicol. 2013;42:132–155. [PubMed] [Google Scholar]
- Schecter A, Malik N, Haffner D, Smith S, Harris TR, Paepke O, et al. Bisphenol A (BPA) in U.S. Environ Sci Technol. 2010;44(24):9425–9430. [PubMed] [Google Scholar]
- Sun Y, Irie M, Kishikawa N, Wada M, Kuroda N, Nakashima K. Determination of bisphenol A in human breast milk by HPLC with column-switching and fluorescence detection. Biomed Chromatogr. 2004;18(8):501–507. [PubMed] [Google Scholar]
- Teeguarden JG, Hanson-Drury S. A systematic review of bisphenol A “low dose” studies in the context of human exposure: a case for establishing standards for reporting “low-dose” effects of chemicals. Food Chem Toxicol. 2013;62:935–948. [PubMed] [Google Scholar]
- Thomson BM, Grounds PR. Bisphenol A in canned foods in New Zealand: an exposure assessment. Food Addit Contam. 2005;22:65–72. [PubMed] [Google Scholar]
- Vandenberg LN, Hauser R, Marcus M, Olea N, Welshons WV. Human exposure to bisphenol A (BPA) Reprod Toxicol. 2007;24(2):139–177. [PubMed] [Google Scholar]
- Wilson NK, Chuang JC, Morgan MK, Lordo RA, Sheldon LS. An observational study of the potential exposures of preschool children to pentachlorophenol, bisphenol-A, and nonylphenol at home and daycare. Environ Res. 2007;103(1):9–20. [PubMed] [Google Scholar]
- Xu J, Wu L, Chen W, Chang AC. Simultaneous determination of pharmaceuticals, endocrine disrupting compounds and hormone in soils by gas chromatography–mass spectrometry. J Chromatogr A. 2008;1202(2):189–195. [PubMed] [Google Scholar]
- Zhang Z, Alomirah S, Cho H, Li Liao YC, Minh TB, Mohd MA, et al. Urinary bisphenol A concentrations and their implications for human exposure in several Asian countries. Environ Sci Technol. 2011;45:7044–7050. [PubMed] [Google Scholar]
- Zimmerman-Downs J, Shuman D, Stull S, Ratzlaff R. Bisphenol A blood and saliva levels prior to and after dental sealant placement in adults. J Dent Hyg. 2010;84(3):145–150. [PubMed] [Google Scholar]