Diet is a primary source of exposure for high-molecular-weight phthalates and bisphenol A (BPA), but little is known about the efficacy of various interventions to reduce exposures. We conducted a randomized trial with 10 families to test the efficacy of a 5-day complete dietary replacement (Arm 1; n=21) versus written recommendations to reduce phthalate and BPA exposures (Arm 2; n=19). We measured phthalate and BPA concentrations in urine samples at baseline, intervention, and post-intervention periods. We used Wilcoxon paired signed-rank tests to assess change in concentrations across time and multi-level mixed effects regression models to assess differences between Arms 1 and 2. Urinary di(2-ethylhexyl) phthalate (DEHP) metabolite concentrations increased unexpectedly from a median of 283.7 nmol/g at baseline to 7027.5 nmol/g during the intervention (P<0.0001) among Arm 1 participants, and no significant changes were observed for Arm 2 participants. We observed a statistically significant increase in total BPA concentration between baseline and intervention periods in Arm 1 but no significant changes in Arm 2. Arm 1 food ingredient testing for DEHP revealed concentrations of 21,400 ng/g in ground coriander and 673 ng/g in milk. Food contamination with DEHP led to unexpected increases in urinary phthalate concentrations in a trial intended to minimize exposure. In the absence of regulation to reduce phthalate and BPA concentrations in food production, it may be difficult to develop effective interventions that are feasible in the general population. An estimate of DEHP daily intake for children in the dietary replacement Arm was above the US Environmental Protection Agency oral reference dose and the European Food Safety Authority’s tolerable daily intake, suggesting that food contamination can be a major source of DEHP exposure.
Phthalates and bisphenol A (BPA) are synthetic endocrine disrupting chemicals (EDCs) that are used in a variety of common products and industrial processes.1, 2 The majority of humans in industrialized settings are exposed to these chemicals through diet, and other consumer products.3, 4, 5, 6, 7 High-molecular-weight phthalates such as di(2-ethylhexyl) phthalate (DEHP), which are used to add flexibility to plastics, are anti-androgenic.3 In experimental studies, in utero exposure to some phthalates causes a variety of male reproductive tract abnormalities including hypospadias, cryptorchidism, and reduced anogenital distance in offspring.8 In humans, prenatal exposures have been associated with reductions in anogenital distance as well as sex-specific changes in orientation and motor development in children, but prospective studies examining development of hypospadias and cryptorchidism have not been conducted.9, 10 Childhood exposure to DEHP and benzylbutyl phthalate (BBzP) has been associated with increased risk of allergic diseases including asthma and eczema.11, 12 BPA is an estrogen that has been associated with endocrine, reproductive, and neurodevelopmental abnormalities in animal studies.13, 14 In human studies, investigators report associations between prenatal BPA exposure and hypotonicity at birth, hyperactivity, anxiety, and depressive symptoms in girls, as well as increased wheezing in young children.15, 16, 17
Several scientific groups including the American Academy of Pediatrics and the Endocrine Society have concluded that EDCs may be harmful during developmentally susceptible time periods and recommend that pregnant women reduce exposures to these chemicals.18, 19, 20 Although recommendations to reduce exposures have been put forth by a variety of scientific and governmental bodies, there is little evidence that following these recommendations leads to decreased exposure as measured by urinary biomarkers.
Individuals can be exposed to phthalates and BPA through air, dust, water, and consumer products, but diet is a significant source of exposure for toxic phthalates (DEHP and BBzP) and BPA.4, 7, 21, 22, 23 How and when phthalates and BPA enter the food chain is not always clear, but the most likely source is plastic used in the manufacturing, processing, storage, and transport of food. These chemicals can also be part of food packaging including plastic containers, lids, can linings, dishware and utensils. A recent study of five families showed that a full dietary replacement of foods that were fresh, un-canned, and/or packaged and prepared without plastics led to significant reductions in urinary phthalate and BPA concentrations.22 Because complete dietary replacement is not feasible at the general population level, we investigated whether providing households with written recommendations produced by environmental health professionals aimed to decrease plastic use in food procurement, storage, and preparation would lead to similar reductions in exposure compared with dietary replacement. We hypothesized that written materials would not lead to reductions in exposures while dietary replacement would lead to significant reductions in urinary phthalate and BPA concentrations.
We conducted a two-arm, randomized study of 10 families to compare the efficacy of complete dietary replacement with fresh and organic, catered foods prepared without plastics (Arm 1) versus education using handouts describing best practice recommendations to reduce phthalate and BPA exposures (Arm 2). The study was conducted over 16-days: baseline (days 1–5), intervention (days 6–10), and post-intervention (days 11–16) time periods (Figure 1). During the intervention period, Arm 1 participants were provided and asked to eat catered foods and beverages that were fresh, local, and organic (when possible), prepared and packaged without plastics. We asked Arm 2 participants to follow written guidelines to reduce phthalate and BPA exposures developed by the national Pediatric Environmental Health Specialty Units. The handouts included descriptions of phthalates and BPA, sources of exposure with a focus on plastics, and how to reduce exposures to phthalates and BPA in daily activities (see Supplementary Materials).24
Eligibility criteria for study participants included: (1) having at least two 4- to 8-year-old children living in the home and (2) two parents living 100% of time in the same household as children. Families were first screened over the phone for eligibility. The second stage of screening included asking families to fill out a dietary questionnaire documenting each family member’s food preparation and consumption patterns over the past 24 h. We used these questionnaires to determine families at highest risk of dietary phthalate and BPA exposure from packaged/processed foods and foods prepared using plastics. Forty-one families in the Seattle area responded to flyers and emails on listservs advertising the study. We did not target any specific population but instead placed flyers in public libraries, the University of Washington, Seattle Children’s Research Institute, and listservs for local community mothers’ groups. Of these families, 17 were ineligible, and 10 were at low risk for dietary exposures to phthalate and/or BPA based on initial questionnaire responses. Of the remaining 14 families, we invited 10 families with the highest potential phthalate and BPA exposures based on proportion of foods that were processed or prepared/packaged in plastics to participate in the trial. All 10 families agreed to participate and were consented. Each family was randomized into one of two arms; (Arm 1) dietary replacement, n=21, and (Arm 2) written materials intervention, n=19. We collected data on gender of all participants, and age of all children. The Seattle Children’s Research Institute Human Subjects Review Board approved this research, and individual test results or urinary biomarkers were reported back to families at the end of the study.
Questionnaires were completed, and urine was collected at three time points during the trial: baseline (days 1–5), intervention (days 6–10), and post intervention (days 11–16) (see Figure 1). Each family received three study visits: (1) baseline, (2) end of intervention period, and (3) post intervention by the study coordinator. At the first visit, the study coordinator educated families about study procedures and provided a study calendar, a checklist for procedures to complete each day, questionnaires, glass food storage containers, and urine sample collection kits. Arm 1 participants were asked to eat only catered foods during the trial that was delivered 1 day before the intervention period. The study team worked closely with the local caterer to ensure that foods were fresh, local and, whenever possible, organic. Foods were also stored, prepared, and transported without plastics. We also asked families to use filtered water and consume drinks from non-plastic containers when possible. Arm 2 participants received educational materials at the first study visit for them to have adequate time to plan, buy, and prepare foods according to the written recommendations in the handouts. All families in Arms 1 and 2 were asked to eat foods using non-plastic utensils and dishware and encouraged to use the glass food storage containers provided. All families were surveyed to record any deviations from study protocol during the trial. We did not attempt to record quantities of foods consumed by participants.
All families in both study arms were asked to fill out detailed dietary questionnaires for each family member on days 1–5 and days 11–16. Arm 2 also filled out questionnaires during the intervention period (days 6–10). All families collected one urine sample per family member after dinner or just before bedtime on days 5 (baseline), 10 and 11 (intervention period), and 16 (post-intervention period) using pre-labeled 125 ml polypropylene urine collection containers and stored them in the freezer (see Figure 1). The study coordinator collected urine samples from the homes at study visit #2 and transported them on dry ice to the University of Washington Environmental Health Laboratory where they were stored in a −80 ° freezer until analysis. At study visit #3 (day 17 or 18), the study coordinator collected all remaining questionnaires and remaining urine samples.
Phthalate and BPA Analysis
We composited the two intervention period samples to improve reliability of measurement of urinary EDC during the intervention period. Therefore, each participant had a total of three urine samples (baseline, intervention composite, and post intervention) analyzed for phthalates and BPA. Phthalate metabolite analyses were based on the isotope dilution high-performance liquid chromatography negative-ion electrospray ionization-tandem mass spectrometry (HPLC-MS/MS) method 6306.03.25 Samples (0.5 ml) were analyzed for BPA before and after deconjugation, so that BPA contamination (as aglycone) during collection and sample processing could be distinguished from urinary eliminated BPA; we assume that aglycone BPA in urine before deconjugation is contamination. Direct injection of 50 μl of urine was used, and sample preparation was scaled to 0.5 ml.25 Procedure blanks were run with each batch of samples, and all were either non-detects or <10% of low calibrant. Therefore, samples were not blank corrected. All laboratory personnel were blinded to the identity of (e.g., study arm) samples.
Post-Intervention Food DEHP Detection
Owing to unexpectedly high urinary DEHP metabolite concentrations (results reported below), a separate analysis was performed on food samples used in the dietary replacement arm for DEHP to identify the source of contamination. We tested the exact same spice batches used in the study. We tested food ingredients from the same suppliers but were unable to test the exact same food ingredients used in the intervention because testing was performed 3 weeks after the study was completed. We first tested ingredients used in the largest quantity in the foods prepared by the caterer including: butter, canola oil, heavy cream, beef, chicken, and salt and pepper. We then tested a spice mixture, peanut butter, cane sugar, milk, honey, egg, oats, cheese, pork, and lamb. When we found that the spice mixture had high DEHP concentrations, we tested the individual spices: ground coriander, ground cinnamon, cayenne pepper, cumin, and star anise. Samples were screened for DEHP by extraction with acetonitrile and analysis of the extract by isotope dilution (HPLC-MS/MS). Before extraction, foods amenable to pulverization (e.g., seeds) were ground with a mortar and pestle. Meats were minced with scissors. Liquids and semi-solids (e.g., peanut butter) were extracted without processing. Butter was heated to liquefaction before extraction. Approximately 1 g of food was extracted once with 4 ml acetonitrile (Optima LC-MS grade, Fisher Scientific, Pittsburgh, PA, USA) containing ISTD (100 ng/ml) by vortexing for 1 min, hand shaking for 2 min, and sonication at 40 °C for 60 min. This single extraction was not considered a complete extraction but served the purpose of screening the food samples for DEHP. The extract was stored at −20 °C overnight to freeze out lipids and water. A 1 ml aliquot of the extract was transferred to an autosampler vial after centrifugation. Reverse-phase chromatography (Gemini, 3 μm, 150 mm, 2.0 mm; Phenomenex, Torrance, CA, USA) was performed with an Agilent 1200 HPLC using a short gradient at 0.4 ml/min. Mobile phase A was 20 mM ammonium acetate. Mobile phase B was acetonitrile. From initial conditions of 90% B, 98% B was reached in 3 min, and then held for 7 min. Initial conditions were achieved with a 5-min reverse gradient. The injection volume was 5 μl.
We examined the distribution of age and gender between arms using Student’s t-tests and χ2 tests, respectively. We then examined concentration and distribution of phthalate metabolites in each urine sample. Most metabolite concentrations were above the LOD, which was between 0.95 and 1.07 μg/l depending on the analyte. Concentrations below the LOD were assigned a value equal to the LOD divided by the square root of two.26 All phthalate and metabolite and BPA concentrations were logarithmically transformed to normalize distributions. Mono-ethyl-hexyl phthalate (MEHP), mono-2-ethyl-5-hydroxyhexyl phthalate (MEHHP), mono-2-ethyl-5-oxohexyl phthalate (MEOHP), and mono 2-ethyl-5-carboxypentyl phthalate (MCEPP) are all metabolites of a single parent compound (DEHP). Therefore, we used the sum of these metabolites (divided by molecular weight) to reflect total DEHP exposure. We conducted bivariate analyses to determine if urinary chemical concentrations differed by sex and age at baseline. We used a Wilcoxon matched-pairs signed-ranks test to assess if urinary concentrations between time periods were different within each arm. Outcomes did not differ with adjustment for urinary creatinine measurements and therefore unadjusted results are presented. We used a multi-level mixed effects regression model to determine if the change between baseline and intervention concentrations differed between Arms 1 and 2 (with adjustment for urinary creatinine). Because repeated measurements were collected on subjects and subjects were nested within families, we fit a mixed effects model with random intercepts at both the family level and subject-within-family level to account for the correlation structure. We also considered gender as a fixed effect in the regression. The addition of gender did not change results significantly while urinary creatinine was significantly associated with outcomes. Therefore, the final multilevel mixed effects model included subject and family as random effects with intervention arms, time periods, intervention-by-period interaction, and urinary creatinine as fixed effects. Analyses were conducted using Stata version 12 (Stata Statistical Software, College Station, TX).
We present results from five families (10 adults and 11 children) in Arm 1 and five families (9 adults and 10 children) in Arm 2. The distribution of gender and urinary creatinine concentrations did not differ between arms (Table 1). Urinary phthalate metabolites were above the limit of detection for over 95% of samples. Total BPA was above the limit of detection in 87% of all samples. The participant baseline values were lower for mono-ethyl phthalate (MEP) and total BPA compared with those reported in the National Health Nutrition and Examination Survey (NHANES) 2007–2008 (Table 2).27 Arms 1 and 2 baseline concentrations were similar for all phthalate metabolites. Urinary total BPA concentrations differed significantly between Arms 1 and 2 at baseline (Table 2). Phthalate metabolite and BPA concentrations in adults and children were similar at baseline.
We observed an unexpected and statistically significant increase in urinary DEHP metabolite concentrations (MEHP, MEHHP, MEOHP, and MCEPP), from 283.7 nmol/g at baseline to 7027.5 nmol/g during the intervention period (P<0.0001) among Arm 1 participants (Table 2; Figure 2). For the individual DEHP metabolites, we observed a 1670%, 2524%, 2297%, 2470% increase in MEHP, MEHHP, MEOHP, and MCEPP concentrations, respectively, between baseline to intervention time periods. The geometric mean (GM) sum of DEHP metabolite concentrations was higher for children (9763.7 nmol/g) compared with adults (4894.4 nmol/g) during the intervention period in Arm 1 but were similar in the post-intervention period (children, 207.6 nmol/g, and adult, 182.9 nmol/g). Change in individual and sum DEHP metabolite concentrations from baseline to intervention periods was significantly higher in Arm 1 compared with Arm 2 (P<0.001; Table 3). All other phthalate metabolite concentrations, among Arm 1 participants, increased between baseline and intervention time periods, but the change in monobenzyl phthalate did not reach statistical significance (Table 2). In contrast, we observed no significant changes in urinary DEHP or other phthalate metabolite concentrations from baseline to intervention periods in Arm 2 participants. We also observed that all urinary phthalate concentrations decreased from the intervention to the post-intervention period with a statistically significant decrease in concentrations for all DEHP metabolites among Arm 1 participants. We did not observe significant differences in change in concentration from baseline to intervention or from intervention to post-intervention time periods between Arms 1 and 2 for other phthalate metabolites. For the majority of urinary phthalates, post-intervention concentrations were similar to baseline concentrations except for MEP where post-intervention concentrations were significantly higher than baseline concentrations (results not shown).
The intervention total BPA GM concentration (1.6 μg/l) was significantly higher than the baseline value (0.8 μg/l) for dietary replacement participants (Table 2). We observed no significant changes in total BPA concentration between baseline and intervention periods for Arm 2. Total BPA post-intervention concentration was significantly higher than baseline concentrations (results not shown). We did not observe significant differences in change in concentration for total BPA from baseline to intervention or during intervention to post-intervention time periods between Arms 1 and 2 (Table 3).
Owing to the unexpected increase in DEHP concentrations in Arm 1, we tested food ingredients used in the dietary replacement intervention for DEHP contamination. Dairy products including butter, cream, milk, and cheese had concentrations above 440 ng/g. We observed that ground cinnamon and cayenne pepper had concentrations above 700 ng/g, and ground coriander had a concentration of 21,400 ng/g (Table 4).
The results of our randomized trial support the hypothesis that written recommendations to reduce dietary ingestion of plastics was insufficient to reduce exposures to high-molecular-weight phthalates and BPA. In the dietary replacement arm, we also observed a dramatic and unexpected increase in phthalate metabolites of one of the most toxic parent compounds, DEHP, despite strong efforts to develop a controlled dietary replacement focused on fresh, organic, and local foods without the use of plastics. We conclude that currently accepted methods to reduce phthalate and BPA exposures (both dietary replacement and written recommendations) may not lead to anticipated changes in urinary phthalate and BPA concentrations. This is the first trial to measure food DEHP concentrations concurrently with urinary phthalate concentrations, and it highlights how contaminated foods can contribute to excessive high-molecular-weight phthalate exposure.
We did not anticipate that our intervention would increase DEHP concentrations during the trial. Our study team undertook several measures to ensure that our dietary replacement would consist of fresh, local, organic food prepared, stored, and transported without plastics. For example, the caterer called local farms and asked that fresh foods be delivered in wood crates instead of plastic cartons. All dairy was delivered in glass (milk/cream) or paper except for one delivery of butter in plastic. In the kitchen, the cooks prepared dishes without the use of plastic utensils, appliances, or storage containers. Families were instructed to eat using ceramic dishes and metal utensils. We provided glass containers for food storage and transport. Despite these measures, DEHP metabolite concentrations increased significantly in the replacement group and then returned to concentrations similar to baseline after the dietary replacement was terminated. The DEHP metabolite concentrations increased for all but one participant, thus we are confident that the DEHP contamination was from the catered foods. We assessed adherence with food deviation questionnaires during the intervention period. Although families in the dietary replacement group reported some deviations from the catered foods (e.g., consuming tea, a piece of chocolate cake, chewing gum, drinking canned energy drinks), they were minor and never constituted a high percentage of overall food consumption. All the foods we measured for DEHP contamination, with the exception of ground coriander, milk, and cream were within the range of what has been reported in the scientific literature.7, 28, 29, 30 One study reported DEHP concentrations of 2598 ng/g in a generic grouping of spices, but this is a magnitude lower than that reported in our study.7 Wormuth et al.7 reports maximum DEHP concentrations in milk and cream to be 40 and 224 ng/g, and our concentrations were 673 and 488 ng/g, respectively. It may be that our findings reflect an isolated rare contamination event because of unusual processing or a packaging abnormality. It also could be the case that the food supply is systematically contaminated with high phthalate concentrations, which are difficult to identify.
Baseline and post-intervention phthalate concentrations for participants in both Arms were similar to values reported in NHANES, but urinary DEHP metabolite concentrations for Arm 1 participants during the intervention were 100-fold higher than those in the 95% percentile within the general population. The ground coriander was used in two dishes for study participants, a main entrée chicken dish and a black bean dip. Based on the recipes and ingredient lists from the caterer, we calculated that approximately 0.7 g of coriander was in each serving of dip and 0.1 g of coriander was in each serving of chicken. The dietary replacement took place over 5 days and therefore, there were approximately 15 servings of chicken and 30 servings of dip provided for each family over this time period. Children had much higher DEHP concentrations as compared with adults during the intervention period, which may reflect increased intake of foods high in DEHP (such as milk, cream, and foods with spices) and increased burden/unit body weight. In order to estimate the total daily intake of DEHP for the average child in the food replacement arm during the intervention period, we used the children-specific creatinine based formula put forth by Koch et al.,31 which accounts for weight (kg), DEHP excretion fraction, and urinary creatinine excretion concentrations:31
UEsum(μmol/g creat) is the excretion of sum of DEHP metabolites. We calculated the sum of MEHP+MEHHP+MEOHP+MCEPP, and the median urinary phthalate concentration from Arm 1 child data during the intervention period was 16.2 μmol/g creat. We used the EPA exposure factors handbook estimate for body weight of an average child age 3–6 years to be 18.6 kg.32 The CEsmoothed is the estimated value from Remer et al.33 for urinary creatinine excretion for a healthy Caucasian child with a weight of 18.6 kg (0.32 g creat/day).33 The FUE is the excretion/oral intake ratio as indicated by Koch et al.34 for the sum of MEHP, MEOHP, MEHHP, and MCEPP, which is estimated to be 0.594.34 The molecular weight for DEHP is 390 μg/μmol. Using these values, the estimated total daily intake of DEHP for a child in the intervention Arm is 183 μg/kg/day. This value is higher than the EPA oral reference dose of 20 μg/kg/day and the European Food Safety Authority total daily intake of 50 μg/kg/day for DEHP, suggesting that food can contribute to oral DEHP exposure that is above values determined to be safe by regulatory agencies.
This was a short-term exposure, and concentrations returned back to baseline after the intervention ended. We are unsure whether these exposures would be associated with adverse health outcomes, but the short-term nature of the exposures and the rapid decrease in the post-intervention period was reassuring. We reported all individual level results back to families stating that the results were unexpected. We also gave them a phone number to call and talk directly with the primary investigator if they had further concerns. No families reported health concerns, anecdotally, during the intervention period, but we did not collect specific data on these outcomes during the study.
Baseline total BPA concentrations were slightly lower than those observed in NHANES for similar populations. Although we observed an increase in total BPA between baseline and intervention periods in Arm 1 participants, the change in concentration was modest compared with the change in phthalate concentrations, suggesting that BPA was not a significant contaminant in the Arm 1 foods provided. The lower than expected baseline concentration could have led to difficulties in detecting larger changes in concentrations across time points or it could have been that food replacement and written recommendations do not aid in reducing exposure concentrations. We believe the former to be the case because other studies have reported measurable changes in BPA concentrations from dietary change.21, 22 In addition, obtaining single pre- and post-intervention urine samples may have introduced increased variability and therefore, decreasing the ability to detect a change during the intervention period. This study was conducted in the summer of 2011 in Seattle, WA, USA and only three families total reported eating canned foods on the screening questionnaire (high concentrations of BPA have been reported in canned foods), and these families were among the 10 chosen for the study. In future studies, we will recruit and screen more participants to identify those that would be at highest risk for BPA exposure through diet. As few families reported consuming canned foods during the recruitment phase, we did not expect to see high baseline BPA concentrations. These low baseline concentrations makes the lack of decrease observed during the intervention period difficult to interpret because we are unsure if the dietary replacement would lead to lower BPA concentrations in a family with higher BPA exposures.
Phthalate and BPA exposure concentrations did not decrease from baseline to the intervention period for participants in the education group of the trial. It is possible that participants may not have adhered to the written recommendations. We assessed adherence with follow-up questionnaires, and several families identified moderate or extreme barriers to following the written guidelines including: not enough time to make different decisions on food choices and storage as well as lack of education on appropriate food choices. Other factors that were reported to be a barrier included, “heaviness and cost of glass storage containers for food transport” compared with plastic containers. We believe that providing families with written recommendations and guidelines to reduce exposures is insufficient to affect change. They may need more explanation or help in interpreting the information and examples of how to implement the recommendations. In primary care practice, it has been shown that written guidelines/recommendations are ineffective in changing health-related behaviors.35, 36, 37 In addition, the handouts only focus on plastic use but not personal care product use or air and dust exposures. These other sources and routes of exposure may be contributing to overall phthalate exposures, but are not specifically addressed by written the recommendations used in this study. The last possibility is that following the guidelines appropriately does not lead to a reduction in exposure. Given that the Rudel et al.22 study showed that a dietary replacement can reduce exposure concentrations and the recommendations follow those used in the study, we do not believe this to be the case. Federal or industry wide regulation aimed at reducing phthalate and BPA concentrations in foods may be the only effective mechanism to ensure the food supply is safe from contamination. Until this kind of regulatory action occurs, it will be important to test improved methods for education and skill building to reduce phthalate and BPA exposures in order to appropriately educate patients and the public.
We thank the University of Washington Center for Ecogenetics and Environmental Health (grant number ES007033, NIEHS) who provided funding for this study. We thank the families that participated in this study.
endocrine disrupting chemicals
environmental health laboratory
high-performance liquid chromatography electrospray ionization-tandem mass spectrometry
National Health and Nutrition Examination Survey
Supplementary Information accompanies the paper on the Journal of Exposure Science and Environmental Epidemiology website (http://www.nature.com/jes)