Several phthalates, particularly diethyl phthalate (DEP) and di-n-butyl phthalate, can be used in personal care products (PCPs) to fix fragrance and hold color. We investigated associations between women’s reported use of PCPs within the 24 h before urine collection and concentrations of several urinary phthalate metabolites. Between 2002 and 2005, 337 women provided spot urine samples and answered questions regarding their use of 13 PCPs at a follow-up visit 3–36 months after pregnancy. We examined associations between urinary concentrations of several phthalate metabolites and use of PCPs using linear regression. Use of individual PCPs ranged from 7% (nail polish) to 91% (deodorant). After adjusting for age, education, and urinary creatinine, women reporting use of perfume had 2.92 times higher (95% CI: 2.20–3.89) concentration of monoethyl phthalate (MEP; the primary metabolite of DEP) than other women. Other PCPs that were significantly associated with MEP concentrations included: hair spray, nail polish, and deodorant. MEP concentrations increased with the number of PCPs used. PCP use was widespread in this group of recently pregnant women. Women’s use of PCPs, particularly of perfumes and fragranced products, was positively associated with urinary concentration of multiple phthalate metabolites.
Phthalates, diesters of phthalic acid, are used in numerous products, including pharmaceuticals, personal care products (PCPs), adhesives, paints, toys, medical devices, and building supplies.1 Because phthalates are not chemically bound additives in these products, phthalates may leach, migrate, and evaporate from these products, resulting in human exposure via multiple routes from direct or indirect contact. Due to their widespread use in common, everyday products, exposure to phthalates is nearly ubiquitous in westernized societies.
Phthalates can enter the body through many routes, including: ingestion of phthalate-containing foods, dermal application of products containing phthalates, and inhalation of air containing phthalate particulates,2 with route of exposure differing by compound. For example, diethyl phthalate (DEP), used in multiple PCPs, is more likely to enter the body through dermal or inhalational exposure whereas di(2-ethylhexyl) phthalate (DEHP), a commonly used plasticizer, enters the body through dietary ingestion more often than other routes of exposure.
Researchers have been increasingly concerned about the health effects from exposure to phthalates, especially during critical times of gestational and infant development. For instance, animal and human models have shown that exposure to certain phthalates, particularly di-n-butyl phthalate (DnBP) and DEHP, can disrupt the development of male reproductive organs.3, 4, 5, 6, 7, 8 Postnatal exposure to phthalates has also been associated with health effects. Phthalate metabolites in breast milk have been associated with neonatal hormone levels.9 Additionally, adult exposures have been associated with adverse impacts on male reproductive health.10, 11, 12, 13
The few studies that have investigated health consequences of phthalate exposure in non-pregnant women have focused primarily on women’s reproductive health. For example, several studies have linked plasma levels of certain phthalates and phthalate metabolites with endometriosis.14, 15, 16 Noteworthy, although measuring phthalates and phthalate metabolites in plasma is analytically possible, these measures are not likely to be good exposure biomarkers.17, 18 A study of Mexican women showed both positive and negative preliminary associations between urinary concentrations of some phthalate metabolites and breast cancer risk.19 In that same cohort of women, an analysis of the healthy controls has suggested associations between exposure to some phthalates and self-reported diabetes.20 It is worth noting that common studies of phthalate effects on pregnant women have focused on pregnancy/birth outcomes and offspring effects rather than effects on the woman herself; therefore, the potential health effects of phthalate exposure on women (whether pregnant or not) are still rather unknown.
Some phthalates are included in PCPs because of their ability to hold color, denature alcohol, and fix fragrance. DEP and, to a lesser extent, DnBP have been detected more often and at higher concentrations than other phthalates, particularly in perfumes and other heavily fragranced products.21, 22, 23, 24, 25 However, phthalates are seldom listed on product labels because current United States regulations do not require listing individual fragrance components.26 Additionally, no premarket approval is required before selling PCPs, though there is some industry self-regulation through the Cosmetics Ingredient Review panel.27 Previous studies have reported that use of certain PCPs were positively associated with urinary concentrations of the metabolites of DEP and other phthalates (e.g., DnBP, di-isobutyl phthalate (DiBP), dimethyl phthalate (DMP)).1, 21, 22, 28, 29, 30, 31, 32, 33
The goal of this study was to examine the associations between concentrations of urinary phthalate metabolites, including metabolites of DEP, DnBP, DiBP, and DMP, and reported use of PCPs within the 24 h before urine collection in a cohort of women who brought their children to a postnatal visit (within 3 years of birth).
The women of this study were recruited into the Study for Future Families (SFF), which was originally designed to assess the geographic variability of semen quality of their partners. SFF is a multicenter, pregnancy cohort study that recruited at prenatal clinics in Los Angeles, California (Harbor-UCLA and Cedars-Sinai), Minneapolis, Minnesota (University of Minnesota Health Center), Iowa City, Iowa (University of Iowa) and Columbia, Missouri (University Physicians), from September 1999 through January 2005. Methods are described at length elsewhere.34
Eighty-five percent of the participants in SFF agreed to be recontacted and eligible women were invited back for a second phase (SFFII) where the child of the pregnancy would undergo a physical examination. Eligibility criteria for the second phase of the study were: the pregnancy ended in a live birth, the infant was <28 months old (later expanded to 36 months old), the mother lived within 50 miles of a clinic, and the mother agreed to at least one study visit. Those visits occurred between April 2002 and December 2005. The analyses for this paper include women who completed a questionnaire and provided a spot urine sample at that study visit (n=337). There were no significant differences in demographics between eligible women who participated versus eligible women who did not participate. Human subject protection committees approved of and the participants signed informed consents for both phases of this study. The involvement of the Centers for Disease Control and Prevention (CDC) laboratory was determined not to constitute engagement in human subject research.
Demographic and Exposure Data
Information on the demographic variables used as covariates in this analysis were obtained from a questionnaire administered to the women during the first phase of the study (SFF). The women’s use of PCPs in the 24 h preceding the collection of their urine sample was obtained in a postnatal self-administered questionnaire completed during the second phase of the study (SFFII). Women indicated “yes”, “no”, or “not sure” about the use of any of the following: hair spray or hair gel, crème rinse/conditioner, shampoo, other hair care products, makeup (powder or liquid foundation), lipstick (not clear), rouge and blusher, eye makeup (mascara, liner, shadow), nail polish or nail polish remover, perfume/cologne, bar soap, liquid soap/body wash, lotion/mist (hand, body, or sun screen), and deodorant. No woman indicated using hair dye, hair bleach, hair permanent, or hair straightener/relaxer within the preceding 24-h time period, so those products were not included in the analysis. Other possible exposures of phthalates (such as exposure through drinking container plastics or contact with polyvinyl chloride) were not controlled for in these analyses.
Phthalate Metabolites Measurements
Women provided a spot urine sample on the same day that they completed the questionnaire that asked about PCPs use. After collection, the urine was transferred to cryovials, and stored at −20°C until it was shipped to the Division of Laboratory Sciences, National Center for Environmental Health, CDC for analysis. Briefly, the analytical method started with an enzymatic deconjugation and solid-phase extraction. Phthalate metabolites were separated using high-performance liquid chromatography and quantitatively detected using isotope-dilution tandem mass spectrometry, as described elsewhere in greater detail.35, 36 The CDC laboratory received no additional participant information. In addition to the study samples, quality control materials and blanks were analyzed to monitor method performance. The limits of detection (LOD) varied slightly by metabolite, but all were in the low nanogram per milliliter range. The nine phthalate metabolites presented in this paper are monoethyl phthalate (MEP), mono(2-ethylhexyl) phthalate (MEHP), mono(2-ethyl-5-hydroxyhexyl) phthalate (MEHHP), mono(2-ethyl-5-oxohexyl) phthalate (MEOHP), monobenzyl phthalate (MzBP), mono(3-carboxypropyl) phthalate (MCPP), monomethyl phthalate (MMP), mono-isobutyl phthalate (MiBP), and mono-n-butyl phthalate (MnBP). We applied correction factors of 0.66 and 0.72 to the MEP and MBzP concentrations, respectively, because the analytic standards used were of inadequate purity.37 Urinary creatinine was also measured using enzymatic reactions for each sample to assess urinary dilution.
Phthalate metabolites concentrations below the LOD were given a value of the LOD divided by the square root of two.38 All phthalate metabolites concentrations were log10-transformed for multivariable analysis to normalize the right-skewed distributions.
For univariate and bivariate analyses, phthalate metabolites concentrations were divided by creatinine. For multivariable analyses modeling log10-transformed phthalate metabolites concentrations, the square root of creatinine was included only as an independent covariate because this transformation provided the best linearity with the dependent variables.
In addition to the nine phthalate metabolites concentrations, one additional measure was examined: molar sum of three metabolites of DEHP. The summary measure for DEHP combines the DEHP primary and secondary metabolites measured (MEHP, MEHHP, MEOHP).
We analyzed several variables for possible confounding including: age at sample collection, employment, race, smoking, ethnicity, and education. Age was modeled as a continuous variable. Race, employment, smoking, ethnicity, and education were made dichotomous variables. These covariates were chosen a priori based on previous studies linking those variables with phthalate exposure and/or use of PCPs.29, 39, 40 After analyzing bivariate correlations and distributions, only age and education remained as possible confounders and were included in all multivariable models.
Factor analysis was performed on the use of PCPs to understand patterns of reported use. Using the Kaiser criterion of retaining factors with eigenvalues greater than one,41 two factors were obtained from the analysis. Makeup foundation, lipstick, and eye makeup defined Factor 1; shampoo and conditioner defined Factor 2. This analysis prompted the creation of two additional summary variables by combining the basic makeup and the basic hair care variables as proxies for these two factors.
Two-sided Wilcoxon signed rank tests were used to determine significant differences in median urinary phthalate metabolite concentrations in women who did and did not use individual PCPs (correlations between urinary phthalate metabolites and PCP use can be found in Supplementary Table S1). Multivariable linear regression adjusted the associations for age in years at study visit, education, and square root of creatinine. Neither pregnancy status nor time since last pregnancy changed effect estimates, so they were not retained in the models. Results of the multivariable regression are presented as the phthalate metabolite concentration ratio and confidence limits for women who used the product compared with those who did not use it. The ratio was calculated by taking the antilog of the multiple linear regression β coefficients. Statistical significance was set at P<0.05. All analyses were performed in SAS 9.2 (SAS Institute, Cary, NC, USA).
The characteristics of our sample of 337 women are shown in Table 1. Our sample contains predominantly white, educated women in their mid-20s to mid-30s. Most women reported having private insurance (77%) and having attended at least some college (91%). They were 30.1±5.1 years old and the median time since giving birth was 15 months (range: 1–37 months). More women came from Minnesota (37%) than any of the other participating study centers. Some women (8%) reported being pregnant at their postnatal visit.
Urinary Phthalate Metabolites
Spot urine samples from 337 women were analyzed for nine phthalate metabolites (98 women did not have MMP concentration quantified) (Table 2). For most of the phthalates, metabolite concentrations were detectable for most women. The most commonly detected metabolites were MEP (99.4%) and MEOHP (99.1%) and the least frequently detected metabolites were MMP (75.7%) and MCPP (78.6%). All urine samples contained at least two metabolites above the limit of detection and eight or nine metabolites in >87% of the women (Figure 1).
The reported use of PCPs varied widely by product type. As shown in Table 3, the most commonly used PCP was deodorant with 309 of the 337 (91%) of women reporting use within the past 24 h. Shampoo and lotion were the next most frequently reported products used by 270 women (80%) and 234 women (70%), respectively. Nail polish or nail polish remover had the fewest women reporting use at 7%, followed by other unspecified hair care products at 9%. As shown in Figure 2, all women reported using at least one PCP within the past 24 h and 25% of the women reported using nine or more PCPs. The median number of reported products used was 7 (data not shown). Some PCPs were often reported together (correlation>0.3; P<0.05) such as: makeup, eye makeup, and lipstick; shampoo and conditioner; and eye makeup and hairspray (data not shown).
Bivariate analysis of PCP use and phthalate metabolites urinary concentrations is presented in Table 3. The strongest and most significant differences were seen between PCPs and urinary MEP and, to a lesser extent, MnBP and MiBP, the metabolites of DnBP and DiBP, respectively. Women who used perfume, deodorant, hairspray, other hair products, or crème rinse/conditioner had significantly higher median urinary levels of MEP compared with women who did not use those products. Women who reported using foundation makeup had a median MCPP urinary concentration of 3.69 g/μg creatinine compared with the median concentration of 4.52 g/μg creatinine in women who did not use foundation makeup. Perfume use was associated with higher levels of MnBP and MiBP and lotion use was associated with a higher concentration of MMP, a metabolite of DMP.
No significant differences were found between any of the PCPs and the urinary concentrations of MBzP, MEHHP, MEHP, MEOHP, or ΣDEHP; therefore, the remainder of the analyses only includes MEP, MnBP, MiBP, MCPP, and MMP.
As the total number of PCPs reportedly used in the past 24 h increased, so did the median creatinine-adjusted, log10-transformed MEP concentrations (Figure 3). Adjusted median MiBP, MnBP, MCPP, and MMP urinary concentrations did not increase with increasing number of products used.
Table 4 shows the results of multivariable linear regression analyses of phthalate metabolites concentration as a function of PCP use after adjusting for education, creatinine, and age.
MEP urinary concentration was significantly associated with use of many products, from bar soap (ratio: 1.37; 95% CI: 1.02–1.83) to perfume/cologne (ratio: 2.92; 95% CI: 2.20–3.89). Foundation makeup had an inverse relationship with MiBP concentrations (ratio: 0.79; 95% CI: 0.66–0.96). MMP concentration was marginally associated with both lipstick and eye makeup (ratios of 1.23 and 1.22, respectively). MnBP concentration was significantly associated only with perfume or cologne use (ratio: 1.38; 95% CI: 1.14–1.66, adjusted R2=0.50). MCPP urinary concentration was not significantly associated with any PCP.
Multivariable regression of total number of PCPs used on log10 urinary MEP concentration (adjusting for age, creatinine, and graduate school education) showed a β coefficient of 0.08 (95% CI: 0.06–0.11). Other multivariable regressions of urinary phthalate metabolite concentrations were not significant (data not shown).
Basic Makeup and Basic Hair Care
To account for the correlation in product use, a factor analysis was performed, resulting in two oblique factors; only Basic Hair Care products (Factor 1, including shampoo and conditioner) and Basic Makeup (Factor 2 including eye makeup, foundation, and lipstick). The PCPs associated with the two factors were then included in regression models, together with perfume, and examined in association with the urinary concentrations of MiBP, MnBP, MEP, MCPP, and MMP. The multivariable regression coefficients (adjusting for age, education, and creatinine) are presented in Table 5. Perfume use was significantly positively associated with the concentrations of all metabolites except MMP that was associated with only Basic Hair Care products, and MCPP that was not associated with any product. Of note is how the crude and combined associations of Basic Hair Care and Basic Makeup with MEP concentrations are significant, but the inclusion of perfume in a multivariable model reduces those effect estimates. Using the factor scores from the factor analysis showed similar associations and confidence intervals for these regressions (data not shown).
Nearly all of the women in our study reported using multiple PCPs in the 24 h before urine collection, with over a quarter of them using more than nine different products. This widespread use of PCPs aligns with previous research on the usage patterns of PCPs.40, 42, 43, 44, 45 In our analyses, use of these products was positively associated with the urinary concentration of the metabolites of several phthalates, particularly MEP, the primary metabolite of DEP. This finding was expected given that previous research has connected DEP to PCPs and cosmetics.1, 21, 22, 25, 28, 29, 30, 31, 32, 33
Exposure to perfume resulted in a 2.92-fold difference in MEP concentration compared with not using perfume. These are relative numbers, but to put it in more concrete terms, the regression equations predict that a 31-year-old woman without a graduate school education and with an average creatinine concentration would have an MEP concentration of 64 ng/ml if she did not use perfume and 187 ng/ml if she did use perfume. Similarly, our data showed that increasing use of PCPs was significantly associated with increased urinary MEP concentrations. So, if the same 31-year-old woman used any three products, her expected urinary MEP concentration would be 44 ng/ml; however, if she used eight products, that expected concentration would be 112 ng/ml.
Our results are consistent with those of other studies that have examined associations between use of PCPs and urinary concentration of phthalate metabolites. Other studies have focused on men, children, and pregnant women, whereas our study examined a population of recently pregnant women (in fact, the same families in whom baby product use and infant phthalate metabolite concentrations were examined by Sathyanarayana et al.30). Like the study of minority pregnant women in New York City, we found that perfume use was significantly related to MEP urinary concentrations.32 Additionally, a study of men and PCPs found the same association with men’s cologne.29 Estimates showed similar effect magnitudes of perfume/cologne use on MEP urinary concentrations such that after adjusting for urinary dilution and covariates, someone exposed to perfume or cologne, on average, had twice the urinary concentration of MEP compared with someone not exposed. Similar to Berman et al., Duty et al., and Sathyanarayana et al., we found a dose-response effect of increasing use of PCPs and urinary concentrations of MEP.1, 21, 22, 28, 29, 30, 31
PCPs are often not used one at a time and often products are used simultaneously or in quick succession. Our analysis used factor analysis to account for such correlations in product use. We found two factors (i.e., personal care use patterns) among our population of women: basic hair care (shampoo and conditioner) and basic makeup (foundation, lipstick, and eye makeup). Using summary scores for these factor variables, we performed additional regressions to assess how using multiple products affected urinary phthalate metabolite concentrations. As shown in Table 6, while there were some associations between the basic hair and basic makeup groups and phthalate metabolites concentrations, the inclusion of perfume as a covariate dominated the regression models.
Phthalate metabolite concentrations in this population were comparable to those reported in other studies with similar populations of women (see Table 6). Interestingly, there are some relatively large differences in MiBP and MnBP concentrations across studies. For example, the study of healthy Germans had a median of 45 μg/g MiBP compared with the median of 3.8 μg/g MiBP among lactating women from North Carolina.
Urinary excretion gives little indication of the initial dose of a topical application. Janjua’s experimental study of whole-body topical phthalate application recovered <6% of the initial phthalate parent compound dose after 1 day.46 Given the small amount of recovered dose for products spread on the skin, no reliable conclusions can be drawn concerning the extent of exposure to our study population despite the finding of strong associations between topical product use and phthalate metabolite urinary concentrations.
For DEP, dermal irritation is also of concern because of the widespread use of this phthalate in dermally applied products, but there are no reports of primary dermal irritation with undiluted DEP28 and there are no reports of oral or inhalation toxicity of DEP.47 While there is little evidence from animal studies that DEP is reproductively toxic, several human studies have reported reproductive changes in male infants in association with concentration of MEP in prenatal urine samples9, 48 and in adult males.10, 12, 49 The disparity between these results in rodent and human studies has not been resolved, but may reflect the fact that reproductive toxicity testing in animals is via oral administration, rather than dermal or inhalation, the routes by which humans are primarily exposed to DEP.
Fragrance has emerged as the strongest predictor among PCPs of urinary concentrations of certain phthalate metabolites. In our initial analyses, we attempted to study the relationship between “fragrance” and urinary concentrations of phthalate metabolites. Unfortunately, self-reported data on fragranced PCPs is of questionable reliability. First, products marketed as fragrance-free may have phthalate-containing masking agents added to cover their chemical odors.50, 51 Second, products marketed as “natural” may also contain phthalates, even though the consumer believes them to be chemical-free. In both cases, subjects’ responses would be misclassified, potentially biasing results.
One limitation of this study is that we performed analysis by product category because brand names were not collected in our questionnaire. Our analysis, therefore, treated all products in a category as interchangeable. The Houlihan cosmetics study reported that phthalate content of products in the same category ranged considerably from undetectable to relatively high levels.21 However, even brand name information would not have remedied this limitation because products of the same category produced by the same company had a range of phthalate concentrations and not all brands have been tested for the phthalate content of their products, nor did we analyze any of the products used for their phthalate content. By combining products by category, we likely increased misclassification of exposure but non-differentially, so any bias in the effect estimates is likely to be toward the null.
Phthalate metabolites concentration varies within subject, even within a day52 so that assessment of phthalate exposure based on multiple samples is less variable than that based on a single spot sample. This is of particular concern when one is attempting to assess exposure over an extended period (e.g., the first trimester of pregnancy), and relating this exposure to a developmental outcome. However, in this paper we are attempting to relate reported product use in a brief (24-h) window to phthalate metabolite concentrations in urine collected at the end of that window. For these purposes, the single spot sample should provide a fairly adequate exposure measure.
Our questionnaire also did not ask about the frequency or amount of product used by the participants. The questionnaire’s recall period was relatively short—a span of 24 h—because the half-lives of phthalate compounds are on the order of hours, not days. We expected that some of our participants used products more often, or in greater quantity, than others but no single product in any excessive quantity.43 Frequency and amount were sources of variability that we could not explain with our models, which had adjusted R2 values in the 30–50% range.
Finally, this analysis focused solely on PCPs as the sources of exposure for phthalates. As previously stated, depending on the phthalate compound, other sources of exposure may be more prevalent and/or relevant. We do not expect these alternate exposures to vary according to PCP use, so the misclassification of phthalate exposure due to other sources is expected to be non-differential. Non-differential exposure misclassification may pull effects estimates toward the null, but that is not guaranteed.53
Women use multiple PCPs each day, which exposes them to a number of chemicals, including phthalates. Our study showed the strongest positive associations between PCPs and urinary concentrations of MEP; however, MiBP and MnBP also exhibited correlations with PCP use. After accounting for product use patterns, perfume emerged as the strongest, most significant predictor of MEP urinary phthalate metabolite concentrations.
This study was supported by grants from the US Environmental Protection Agency; National Institutes of Health Grants R01-ES09916 to the University of Missouri, MO1-RR00400 to the University of Minnesota, and MO1-RR0425 to Harbor-UCLA Medical Center; Grant 18018278 from the State of Iowa to the University of Iowa and 1RC2ES018736-02 to the University of Rochester. We gratefully acknowledge the technical assistance of Manori Silva, Jack Reidy, Ella Samandar, Tao Jia, and Jim Preau (Centers for Disease Control and Prevention, Atlanta, GA, USA) in measuring the urinary concentrations of phthalate metabolites. The findings and conclusions in this report are those of the authors and do not necessarily represent the views of the CDC.
personal care product
Study for Future Families
Study for Future Families, phase II
About this article
Supplementary Information accompanies the paper on the Journal of Exposure Science and Environmental Epidemiology website (http://www.nature.com/jes)
Environmental Science and Pollution Research (2017)