Article

Journal of Exposure Science and Environmental Epidemiology (2011) 21, 395–407; doi:10.1038/jes.2010.31; published online 23 June 2010

Estimating perchlorate exposure from food and tap water based on US biomonitoring and occurrence data

David R Hubera, Benjamin C Blountb, David T Magec, Frank J Letkiewiczd, Amit Kumard and Ruth H Allene

  1. aUS EPA, Office of Ground Water and Drinking Water, Washington, DC, USA
  2. bCenters for Disease Control and Prevention, Atlanta, Georgia, USA
  3. cDanya International, Silver Spring, Maryland, USA
  4. dThe Cadmus Group, Arlington, Virginia, USA
  5. eUS EPA, Office of Chemical Safety and Pollution Prevention, Washington, DC, USA

Correspondence: Dr. David Huber, US Environmental Protection Agency, Office of Ground Water and Drinking Water, Standards and Risk Management Division 4607M, 1200 Pennsylvania Avenue, Washington, DC 20460-0001, USA. Tel.: +202 564 4878. Fax: +202 564 3760. E-mail: huber.david@epa.gov

Received 8 October 2009; Accepted 5 April 2010; Published online 23 June 2010.

Top

Abstract

Human biomonitoring data show that exposure to perchlorate is widespread in the United States. The predominant source of intake is food, whereas drinking water is a less frequent and far smaller contributor. We used spot urine samples for over 2700 subjects and estimated 24h intake using new creatinine adjustment equations. Merging data from surveys of national health (NHANES) with drinking water monitoring (UCMR), we categorized survey participants according to their potential exposure through drinking water or food. By subtracting daily food doses of perchlorate from the oral reference dose (RfD), we derive an allowances for perchlorate in tap water for several populations. The calculated mean food perchlorate dose in the United States was 0.081μg/kg/day compared to 0.101μg/kg/day for those who also had a potential drinking water component. The calculated 95th percentile doses, typically falling between 0.2 and 0.4μg/kg/day, were well below the RfD (0.7μg/kg/day) in all populations analyzed. Children aged 6–11 years had the highest mean perchlorate doses in food (0.147μg/kg/day), with an additional water contribution of only 0.003μg/kg/day representing just 2% of exposure. Pregnant women had a mean food dose of 0.093 vs 0.071μg/kg/day for all women of reproductive age. At the 95th percentile intake for both the total population and women of child-bearing age (15–44), the perchlorate contribution from food was 86% and from drinking water 14% (respectively, 30% and 5% of the RfD). At the mean for the same groups, the food to water contribution ratio is approximately 80:20. We calculate that an average 66kg pregnant woman consuming a 90th percentile food dose (0.198μg/kg/day) could also drink the 90th percentile of community water for pregnant women (0.033l/kg/day) containing 15μg/l perchlorate without exceeding the 0.7μg/kg/day reference dose.

Keywords:

perchlorate; drinking water; food; occurrence; NHANES; biomonitoring; UCMR

Top

Introduction

Perchlorate is an inorganic anion that occurs naturally and is synthesized for a variety of industrial applications. It is chiefly manufactured as ammonium perchlorate (AP), an oxidizer used in making solid propellants for rocket motors. It is used commercially as perchloric acid and as a salt of ammonia, potassium, magnesium, sodium, lithium and other cations. Perchlorate readily dissociates into its anionic form, ClO4, which has been found in some ground and surface waters. Chilean saltpeter contains small amounts of perchlorate, and that product's use in the United States as nitrate fertilizer may contribute to the presence of perchlorate in ground water and food in some areas (Urbansky et al., 2001).

Some atmospheric processes may also produce perchlorate naturally, and Dasgupta et al. (2006) have discussed the relative sources and quantities of perchlorate attributed to those processes and anthropogenic sources. Perchlorate can also form spontaneously in hypochlorite solutions used for drinking water disinfection (MA DEP, 2005). Perchlorate occurrence in water can result from both natural deposits and from its various uses, notably in fireworks, flares and explosives.

Plant uptake of perchlorate from soil, fertilizer or irrigation water can result in it being present at low levels in many food items (Sanchez et al., 2005). For example, water from Lake Mead and the Lower Colorado River contains perchlorate at parts-per-billion concentrations, the origins of which were two manufacturing facilities in Henderson, Nevada. These waters are used for agricultural irrigation in the Imperial and Coachella Valleys of southern California and southwestern Arizona, and supply drinking water in the tri-state area. Perchlorate in the irrigation water has led to its presence in a variety of forage and food crops distributed nationally (Hogue, 2003; Sanchez et al., 2009). Other studies have found trace levels of perchlorate in a wide variety of food items grown in many different regions around the world (El Aribi et al., 2006). The Total Diet Study (TDS) of the Food and Drug Administration (FDA) found that 74% of samples of 285 foods contained detectable levels of perchlorate. From these data, FDA developed estimates of average dietary perchlorate intake for 14 age–gender groups, which range from 0.08 to 0.39μg/kg/day (Murray et al., 2008).

The perchlorate anion is biologically significant specifically with respect to the functioning of the thyroid system. Because its charge, shape and size are similar to the iodide ion (I), perchlorate can bind and therefore compete with iodide at the sodium iodide symporter (NIS). NIS is a transmembrane protein that actively transports iodine into the thyroid follicles for organification and synthesis into the prohormone tetraiodothyronine (thyroxine or T4), and the active form triiodothyronine or T3. The thyroid hormones T3 and T4 have an important role in the regulation of metabolic processes in cells throughout the body and are also critical in the developing fetus, especially in brain development.

The National Research Council (NRC) of the National Academy of Sciences (NAS) evaluated all available health data and recommended a reference dose (RfD) for perchlorate on the basis of the Greer et al. (2002) human study finding a statistical no-observed-effect level (NOEL) based on the absence of iodine uptake inhibition at 0.007mg/kg/day. The NAS recommended an additional 10-fold uncertainty factor to account for differences between healthy adults in the study and the most sensitive life stage, the fetus of the pregnant woman who might have hypothyroidism or iodide deficiency (National Research Council (NRC), 2005). EPA defines a reference dose as an estimate, with uncertainty spanning perhaps an order of magnitude of a daily oral exposure to the human population (including sensitive subgroups) that is likely to be without an appreciable risk of deleterious effects during a lifetime. EPA established its RfD of 0.0007mg/kg/day (0.7μg/kg/day) consistent with the NAS report (US EPA, 2005).

NAS considered iodide uptake inhibition to be “the key biochemical event and not an adverse effect…that precedes all thyroid-mediated effects of perchlorate exposure.” Because the developing fetus depends on an adequate supply of maternal thyroid hormone for its central nervous system development during the first trimester of pregnancy, iodide uptake inhibition from low-level perchlorate exposure has been identified as a concern in connection with increasing the risk of neurodevelopmental impairment in fetuses of high-risk mothers. We therefore pay particular attention to women of child-bearing age or those who are pregnant in our study, along with children for whom data were available (ages ≥6 years). Males are included in our analysis, although the study by Blount et al. (2006) notes that for men, the presence of perchlorate did not serve as a predictor of either T4 or log TSH as it did for women.

Two main sources of perchlorate occurrence and exposure information have been used for the analyses presented in this paper that are aimed at understanding the relative contributions of food and drinking water to the daily intakes of perchlorate in the US population. These are the National Health and Nutrition Examination Survey (NHANES) urinary biomonitoring data and the EPA Unregulated Contaminant Monitoring Regulation (UCMR) data.

The NHANES data are from an ongoing stratified probability sample survey of the US non-institutionalized population that is designed to assess health and nutrition status (CDC, 2004). A complex multistage probability sampling design is used so that the ~5000 people sampled annually are representative of the US population based on age, sex, race/ethnicity and income. Urinary biomonitoring results from NHANES, which includes participants of ages ≥6 years, indicate that perchlorate exposure is ubiquitous in the US population (Blount et al., 2007). These data also associated exposure with non-adverse changes in thyroid hormone levels in susceptible populations, especially women with low iodine who are also tobacco smokers (Blount et al., 2006; Steinmaus et al., 2007).

In 1999, EPA developed the first UCMR rule (UCMR 1) program. EPA collected and analyzed drinking water occurrence data for perchlorate from a near census of 3086 large public water systems (PWSs) serving >10,000 people each and 797 representative small PWSs serving ≤10,000 people each. All UCMR 1 sampling was performed between 2000 and 2005, with the majority of sampling occurring between 2001 and 2003. Systems collected samples over a 12-month period, quarterly for surface water sources and biannually for ground water sources. These samples were collected at all entry points into their distribution systems. EPA found that 160 (approximately 4.1%) of all 3865 PWSs that sampled and reported had at least one analytical detection (at one or more sampling points) of perchlorate at levels ≥4μg/l (the method reporting limit). These 160 systems are located in 26 states and 2 territories: 8 were small systems and 152 were large systems. They reported 637 detections of perchlorate at levels ≥4μg/l in 387 sampling points. This represents approximately 11.3% of 5629 samples collected in systems having a positive sample somewhere, and approximately 1.9% of the 34,331 samples collected by all of the 3865 systems. In all, 39 systems had an average at or above the MRL, in which non-detects were counted as 2μg/l (half the MRL). The maximum reported concentration of perchlorate was 420μg/l, from a single surface water sample from a public water system in Puerto Rico, and the second and third highest were 200 and 70μg/l in two systems in Florida. The average concentration of perchlorate for the samples with positive detections for perchlorate was 9.85μg/l and the median concentration was 6.40μg/l.

In this study we use urinary perchlorate data from a random subset of NHANES 2001–2002 to estimate the daily perchlorate dose from food after categorizing study participants with respect to their potential for additional exposure from drinking water based on EPA UCMR tap water perchlorate data. We identified those NHANES study participants who are unlikely to have perchlorate exposure from drinking water based on the UCMR data and considered them to be representative of individuals whose exposure was likely to be from food sources alone. The perchlorate intake estimates for this “food exposure only” group are then compared with the RfD for toxicological perspective. The estimated perchlorate intakes from food can then be subtracted from the established RfD, yielding the additional oral intake that could also be consumed by those individuals through drinking water without exceeding the RfD. Figure 1 illustrates the difference between the relative contributions of food and water sources to the total exposure in comparison to the RfD, and depicts a method for determining a permissible dose through drinking water. The relative exposure contribution of the two sources has a bearing on the question of which of either source might provide a better opportunity to impact possible health outcomes over the other. Subtracting the national background contribution of food for various populations from the RfD reveals the remaining drinking water dose (converted to concentration) that, if found in any particular water system, would not exceed the RfD.

Figure 1.
Figure 1 - Unfortunately we are unable to provide accessible alternative text for this. If you require assistance to access this image, please contact help@nature.com or the author

Reference dose relative to exposure: calculating an allowance for water.

Full figure and legend (37K)

Top

Methods

EPA and CDC collaborated in this study to characterize the relative importance of food and drinking water as sources of perchlorate exposure. EPA contributed the national survey with monitoring data on perchlorate in public drinking water supplies, and CDC provided the national survey data on individual exposure measures including concentrations of urinary perchlorate. Tap water perchlorate was not measured as part of the 2001–2002 NHANES, and hence specific information about participants' tap water cannot be known directly, and only their county of residence was known. The researchers merged the NHANES 2001–2002 data with the UCMR public drinking water system database of EPA on perchlorate from a similar time frame (i.e., primarily 2001–2003). This NHANES/UCMR data merge was conducted at our request by the National Center for Health Statistics (NCHS) that retained participant confidentiality and matched results of co-location to a county (not zip code or water system) level.

The primary aim of merging the two data sets was to differentiate participants who may have had both water and food and water exposure to perchlorate from those whose exposure was likely from food alone. The estimated food-only dose of perchlorate for a specific sub-population was subtracted from the RfD to yield the remaining intake dose available through drinking water without exceeding the RfD. That is, a person consuming this dose of perchlorate from their tap water, in combination with the typical food amount, would not have a total exposure in excess of the RfD. In addition, we estimated the relative magnitudes of food vs drinking water perchlorate exposure in the US population.

To merge the data from UCMR and NHANES, we created three “Bins” for categorizing the NHANES study participants for whom we had perchlorate exposure data.

Bin I: Those subjects residing in counties where a drinking water system reported detectable perchlorate (≥4μg/l) at least once at any sample point were placed in Bin I. Because the merging of data was only possible at the county level, more specific location data (e.g., home addresses), which might have been useful to confirm that they were customers of that water system, could not be used because of privacy concerns. In addition, we could not delineate the distribution system boundaries served by an entry point to the distribution system that had a positive perchlorate result; therefore, we could not say conclusively that perchlorate was detected in tap water at the study participant's residence. In a few cases, more than one UCMR water system was located in the same county. If any of the systems in such counties had a detected level of perchlorate, then all NHANES individuals residing in that county were placed in Bin I.

Bin II: Those subjects for whom we did not have information about perchlorate in their water because there were no sampled locations from the UCMR in their county of residence were placed in Bin II.

Bin III: Those subjects who were hypothesized as unlikely to have been exposed to perchlorate from tap water by virtue of residing in counties where the drinking water system had been sampled for perchlorate but it was not found to be ≥4μg/l minimum reporting limit were placed in Bin III.

We deemed that the subjects in Bin III were, therefore, those whose urinary perchlorate levels resulted principally from food intake (it was also determined that none of the subjects worked in an industry using perchlorate). Assuming that the daily intakes calculated for Bin III are from food only, those intakes could be used as the basis for estimating an allowable drinking water intake that remained by subtracting the food portion of dose from the reference dose as described previously.

It should be noted that we are also assuming in this analysis that all subjects are customers of the public water systems in their area of residence. Approximately 15% of households in the United States use private wells; therefore, not all people in the NHANES cohort were necessarily served by public water systems monitored by the UCMR survey, and some may have been on private well water. The systems in Bins I and III are about evenly split between surface and ground (well) water sources, and hence private well use and the presence of perchlorate cannot be ruled out, although likely of negligible effect as discussed in a later section.

In addition, two other adjustments were made to the binning of individuals to better reflect those not expected to have a drinking water contribution. The first adjustment was adding to Bin III all individuals in Bins I or II who reported not consuming tap water in the day before providing the urine sample. This was justified on the basis of the relatively short half-life of perchlorate of approximately 8h (Crump and Gibbs, 2005).

NHANES also collected information on whether a subject used any form of home water treatment. Those devices considered included a pitcher water filter, ceramic or charcoal filter, water softener, aerator and reverse osmosis (RO). Only RO was considered to be an effective treatment for removing perchlorate from drinking water. Any individuals in Bins I or II who reported using RO water treatment device in their homes were moved into Bin III. Approximately 300 individuals were moved from Bins I and II to Bin III based on these two considerations. These additions did not result in any significant change to the mean perchlorate doses estimated for the Bin III group.

NHANES collected a single-spot urine sample for each of the 2001–2002 study participants. Because perchlorate is not metabolized by humans and, except for lactating women, is almost entirely excreted through the kidneys, urinary perchlorate levels allow an estimation of the amount of perchlorate actually ingested by the person. The amount of perchlorate ingested over a 24-h time frame would be approximated in a 24-h urine sample. However, having only spot urine samples, we used a creatinine adjustment calculation to as a surrogate for a 24-h urinary excretion value (Cockroft and Gault, 1976). Creatinine, which is a metabolic byproduct of creatine that is produced by muscle tissue, is excreted by the kidneys at an approximately constant daily rate proportional to an individual's age, sex, weight, height, race and lean body mass. The creatinine-adjusted estimates of intake used in this study were developed by one of the authors (David Mage) concurrent with performing the other aspects of this analysis (Mage et al., 2004, 2008). Urinary perchlorate in a single urination event (a spot sample) can be expressed as a fraction relative to creatinine: μg of perchlorate per g of creatinine excreted. Knowing the total constant daily amount of creatinine produced for an individual's physical characteristics, we can estimate his/her daily exposure to perchlorate. Using the fractional amount of daily creatinine in a single sample, we can use the same relative proportion to calculate the daily perchlorate excretion from the perchlorate amount in the sample. Therefore, assuming an omnivorous diet and normal renal function, we estimated 24-h creatinine excretion and corresponding perchlorate excretion, which for reasons noted above is assumed to be equivalent to that individual's 24-h intake. The daily intake is then converted to a dose in units of μg/kg/day using the subjects' body weight as further described in the online Supplementary Information.

We were able to generate estimates of daily perchlorate intake using this creatinine adjustment procedure for 2708 of the 2820 NHANES subjects with perchlorate urinary data. Data on height, weight and/or creatinine levels were missing for the remaining 112 subjects. Note that the mean urinary perchlorate levels for those 112 subjects were similar to 2708 subjects for whom data were available.

We recognize that there is inherent uncertainty in using the estimates of daily perchlorate dose as indicative of long-term averages for the population based on a single-spot urine sample of perchlorate and creatinine in the NHANES subjects. This is due to the timing of when the samples were taken relative to each individual's recent consumption of food and water, as well as the expected variability in intake and excretion related to an individual's dietary and water consumption patterns. This uncertainty due to such factors that affect short-term variability is expected to operate in both directions, that is, both underestimating long-term averages for some individuals, and overestimating them for others. Mendez et al. (2010) discussed the short-term variability aspect as a possible explanation for higher estimates of the intakes in the upper percentile values of the distribution obtained using the urinary excretion data as done here, vs their estimates based on a dietary simulation approach.

In recognition of the potential effects of short-term variability on our estimates, particularly with respect to the potential impact of unusually high values on the estimates of the mean and upper percentiles, we performed an outlier analysis on the 2708 estimated perchlorate intakes using the Grubbs' test for outliers (Grubbs, 1969). Outlier tests were performed both on the reported perchlorate urine concentrations and on the calculated intakes using the urine data and the creatinine adjustment. We decided that the intake values, which capture potential influences of both the observed urine levels and the various factors that go into the creatinine adjustment, were more appropriate for the outlier tests.

First, it was determined that the intake estimates for the overall NHANES subjects as well as those for various subsets were lognormally distributed. Therefore, all estimates of intakes presented in this study are obtained from fitting the estimates for individuals to lognormal distributions, using the NHANES weights in performing those fits, and conducting goodness-of-fit tests using Kolmogorov–Smirnov and Cramer-von-Mises methods. Note that the use of “mean” values throughout this paper refers to the expected values derived from the parameters for these fitted distributions.

Using the lognormal distribution for all subjects combined, four subjects (2 adult males and 2 adult females, all in Bin III) with calculated intakes ranging from 1.6 to 3.2μg/kg/day (i.e., approximately 20–50 times greater than the estimated mean intake) were identified as outliers at the α=0.05 level and were excluded from subsequent analyses. Excluding the estimates for these four data had a slight effect on the estimated means for Bin III adults, but no meaningful effect on the estimated quantile values (note that two of these same four individuals were determined to be outliers based on the urine data alone).

The distribution fitting and outlier tests were performed subsequent to an earlier presentation of the results of this study that appeared in EPA's 2008 preliminary regulatory determination notice (US EPA, 2008). Therefore, the results presented and discussed in this study differ slightly from the preliminary results presented in that Federal Register notice.

For data analysis, we used SAS (ver. 9.1.3, SAS Institute Inc., Cary, NC, USA) and SUDAAN (ver. 9.0.1, Research Triangle Institute, Research Triangle Park, North Carolina, USA), with supplemental analyses performed using Mathematica (ver. 7.0, Wolfram Research, Champaign, Illinois, USA).

Top

Results

Overall Estimates of Perchlorate Daily Intakes

Estimated daily perchlorate intakes for all subjects and for major subgroups based on age and sex are presented in Table 1. The overall mean value for all subjects taken together is 0.061μg/kg/day. Age seems to be a factor influencing intakes, with higher mean intakes observed more in younger than older groups. The mean intake was found to be 0.148μg/kg/day for ages 6–11; 0.082μg/kg/day for ages 12–19 and 0.076μg/kg/day for those ≥20 years old.


Sex also seems to be a factor, with higher mean values observed in males than females, with the age effect observed within the two sexes as well (higher mean values in younger subjects). It is noteworthy that the mean intake value of 0.088μg/kg/day for pregnant women in the 15–44 age group was higher than the mean value of 0.068μg/kg/day for all non-pregnant women in the 15–44 age group, and this difference was statistically significant at the 0.05 level. As shown in Table 1, intakes by pregnant women were also higher than that of non-pregnant women in the upper quintiles of intakes.

Estimates of Perchlorate Daily Intakes for Different Drinking Water Occurrence Groups

On the basis of the merged biomarker and tap water data, we estimated perchlorate doses for study participants in each of the three Bins. As stated, we included in Bin III subjects who reported that they had not consumed tap water in the 24h before their urine sample and those who treated their home tap water with RO systems. The effects of these adjustments were small for the reasons discussed in the next section. For example, the overall mean intake for 1765 individuals in Bin III was 0.080μg/kg/day before the adjustments and was 0.081μg/kg/day for 2059 individuals after the adjustments. We also used the creatinine adjustment equations (Mage et al., 2008) for children and adolescents and accounting for the lean body mass (non-water content) in relation to body mass index. After also correcting for adiposity, the creatinine-adjusted result is somewhat lower, proportionally more so in women than men.

Table 2 shows that the mean perchlorate intake values and various quantile values for Bin I (with food and drinking water sources) exceed those for Bin III (food sources only). The intake values for Bin I subjects at the mean and median levels are typically 15–30% higher than the values for corresponding Bin III subjects. These higher Bin III values when compared with Bin I is consistent with the presumed contribution of drinking water to intake in Bin I that is not present in Bin III subjects.


The Bin II intakes (unknown drinking water conditions) are similar to, although generally less than, those for Bin III. It was expected that Bin II results would be more similar to Bin III as the great majority of water systems, per the UCMR data, do not have perchlorate present at ≥4μg/l MRL. Because of the low number of subjects, we do not show the results of pregnant women in Bin 1 (highest value) n=8, or Bin II (lowest value) n=11, but the lower Bin II results caused the total intake results in Table 1 to be slightly lower that the food alone results in Table 3.


Additional consideration was given to the effect of pregnancy in the calculation of normal creatinine excretion and perchlorate intake. An increase in the perchlorate-to-creatinine ratio was observed in the pregnant NHANES subjects. Our calculations showed that an increase in the perchlorate dose was proportional to the increase in caloric intake typical in the second and third trimesters. The mean total caloric intake for the 24-h recall period for the 116 pregnant woman was 2184Kcal (95% CI 2031.8–2336.6) whereas the non-pregnant group of 524 women of child-bearing age averaged 1995.3 (95% CI 1924.4–2066.3). These data indicate that the pregnant women consumed 9.5% more calories compared with non-pregnant women of reproductive age (P=0.02).

Perchlorate is actively transported into breast milk during lactation, thus leading to less perchlorate excretion in urine (Kirk et al., 2005, 2007; Pearce et al., 2007). Bin III of our study population contained a small group of 19 lactating women. On the basis of urinary perchlorate excretion, these lactating women had a mean perchlorate dose of 0.079μg/kg/day compared with non-lactating women of reproductive age whose mean levels were0.083μg/kg/day. The lower estimated perchlorate doses are consistent with secretion of perchlorate in breast milk. Thus, perchlorate exposure estimates based on urinary excretion will likely underestimate exposure in lactating women.

Comparison of Estimated Food and Water Intake Estimates with the Reference Dose

Table 3 provides a detailed summary of comparisons made between the estimated food and water intakes for various age/sex populations in the three different bins reflecting expected drinking water exposure to perchlorate. It also provides estimates of the portion of the RfD that would still be available for intake through drinking water when counting the intake from food obtained from Bin III.

The difference between estimated intakes for NHANES subjects located in counties where a UCMR drinking water sample was positive for perchlorate (Bin I) and where no positives were found in the samples analyzed (Bin III) may allow an estimation of the portions of the daily intake that might typically come from the two main sources: namely, food and drinking water. For example, analyzing the overall mean for the total population category in Table 3, the difference between Bin I (total) and Bin III (food) is approximately 0.020μg/kg/day, which represents about 20% of overall exposure due to water. At the 90th percentile of the distribution, the difference between Bins I and III is 0.031, which indicates that water exposure contributes approximately 16% of total exposure, whereas food constituted the other 84% of exposure.

The differences between the means of Bins I and III were found to be statistically significant at the 0.05 level for all groups shown in Table 3, except for children of age 6–11 and women 15–44 years. However, the differences between the means of Bins I and III for these two groups (which we note had much smaller N counts in Bin I than the other groups) were statistically significant at the 0.10 level.

The data in Table 3 indicate that for the overall population, and for most of the individual groups shown, drinking water represents approximately 15–25% of the total intake at the mean. The notable exception to this is young children aged 6–11 years in which it represents only approximately 2% of total intake. At the 90th and 95th percentile values, drinking water represents approximately 10–30% of total intake for all groups except children aged 6–11 years in which the intakes for the Bin III (food only) subjects are similar to, although slightly higher, than those for the Bin I (food plus water) subjects. Again, this may be due in part to the relatively small number of subjects in Bin I for this age group.

The data in Table 3 also indicate that perchlorate intake from water at the mean (and for most percentile groups) typically comprises approximately 2–5% of the RfD value of 0.7μg/kg/day. Perchlorate intake from food at the mean is generally between 10% and 15% of the RfD. The notable exception to this is, again, children in the 6–11 age group in which drinking water perchlorate contributes <1% of the RfD and food perchlorate contributes >20%. The food choices in this age group, especially dairy, likely contribute to the high relative food contribution (Murray et al., 2008).

Relating the data in Table 3 to Figure 1, the relative source contribution of drinking water to total perchlorate intake for the national average is approximately 20% with food comprising the remaining 80% of total exposure. At the national average values, water and food represent perchlorate intakes of approximately 3% and 12% of the RfD, respectively.

The principal use of Bin III data is to calculate the portion of the perchlorate RfD remaining — after the food contribution is accounted for — that could be “allowed” in drinking water such that total intakes would not exceed the RfD. These are observed in Table 3 for the various groups, and for the means as well as the 90th and 95th percentile values these drinking water allowances generally fall in the range of 0.5–0.6μg/kg/day (71–86% of the RfD).

The conversion of the remaining water-available dose into an equivalent drinking water concentration value is obtained by multiplying dose by body weight and dividing by liters the tap water consumed per day. The pregnant woman's developing fetus has been identified as a sensitive life stage by the NAS. The first trimester may be particularly sensitive for brain development (National Research Council (NRC), 2005). On the basis of our estimates of food-only perchlorate dose at the 90th percentile value (for conservatism), a pregnant woman could ingest up to 0.502μg/kg/day perchlorate from drinking water (Table 3) without exceeding the reference dose. Using the average pregnancy weight, again for conservatism, rather than a first-trimester pregnancy weight, a 66kg woman and a body-weight-adjusted estimate of direct and indirect tap water ingestion of 33ml/kg/day by pregnant women at the 90th percentile (US EPA, 2004), it is estimated that she would drink approximately 2.2l per day. Dividing the product of the 90th percentile water allowance dose (0.502μg/kg/day) and the 66kg weight by the 2.2l per day yields 15μg/l drinking water concentration before exceeding the RfD.

Top

Discussion

Our analysis of the NHANES and UCMR data suggest that food typically contributes substantially more to perchlorate exposure than does tap water for the US population. The relative amounts will vary by locale depending on perchlorate levels in local tap water and food, but seem to fall in the range of approximately 4:1 food to water exposure ratio for the overall population and most age/sex subgroups. An exception seems to be young children aged 6–11 years in which drinking water seems to make a much smaller contribution to the total intake.

Knowing the level of perchlorate occurring in various water systems does not by itself allow a calculation of potential limits on drinking water concentrations. By focusing on our Bin III, that is, those NHANES subjects not drinking tap water or living in locations in which perchlorate was monitored for in tap water but not detected, we attempted to isolate the range of intakes from food alone. We were able to then derive a perchlorate concentration in drinking water that did not exceed the reference dose for various sub-populations using the 50th, 75th, 90th and 95th percentiles of food-only dose.

The TDS conducted by FDA (Murray et al., 2008) has provided useful estimates of perchlorate intake from food ingestion, whereas we sought to isolate the food intake data through biomonitoring data from urinary output to provide additional results. In fact, both study results compare well: in our study, mean perchlorate dose through food for women of reproductive age (15–44 years) was 0.071μg/kg/day whereas the mean result for the TDS for women of ages 14–16, 25–30 and 40–45 years was 0.09–0.10μg/kg/day. Average dose for males from food in TDS ranged from 0.08 to 0.14μg/kg/day. For males in the NHANES data, we estimated an average perchlorate dose of 0.088μg/kg/day. Thus, similar results were derived for food-based perchlorate dose, despite approaching the question from opposite starting points (excreting vs eating).

Exposure estimates had a degree of imprecision because of the varied and episodic nature of perchlorate exposure and excretion rates of perchlorate from the body. Mendez et al. (2010) observed that perchlorate dose estimates based on measuring perchlorate and creatinine levels in a single-spot urine introduce imprecision that increases uncertainty at the extremes of the distributions. Their recently published estimates of daily perchlorate intakes used a dietary simulation approach focused mainly on women of reproductive age (15–44 years old) and predicted perchlorate dose distributions with similar means, but lower 90th percentiles compared with our biomonitoring-based estimates presented in this study.

For women aged 15–44 years with drinking water as well as food contributions of perchlorate (our Bin I), Mendez et al. (2010) estimated the mean daily intake values of 0.067–0.079μg/kg/day compared with this paper's estimated mean of 0.083μg/kg/day; their 90th percentile was 0.10–0.12μg/kg/day compared with our 90th percentile daily intake value of 0.164μg/kg/day.

For women aged 15–44 years with no drinking water contributions (our Bin III), Mendez et al. (2010) estimated mean daily intake values of 0.064μg/kg/day and 90th percentile of 0.097μg/kg/day. For our current estimates, the mean and 90th percentile daily intake values are 0.071 and 0.141μg/kg/day, with our mean value lying between the estimates of Mendez et al. (2010) and Murray et al. (2008). As noted in the Methods section, our outlier analysis resulted in dropping two very high values out of the original 505 subjects for this group. This had a marked effect on the estimated mean value, but little effect on the 90th percentile value compared with our preliminary results.

To complete a relative exposure picture, the illustration in Figure 1 could be placed within a larger context to include other known iodine inhibitors consumed on a daily basis. The many goitrogenic compounds in the diet include thiocyanate from cruciferous (Brassica) vegetables and nitrate from fertilizer, both of which are inhibitors of iodine transport at the NIS, and soy isoflavones that affect thyroid peroxidase-mediated reactions necessary to organify the iodine (Divi et al., 1997). The perchlorate dose-equivalent (its dose considering its potency relative to other inhibitors) plus dose-equivalent exposures to all other inhibitors could be summed and compared with an aggregate iodine-inhibitor limit (DeGroef et al., 2006). Unfortunately, this relative exposure and risk picture is incomplete and no aggregate limit currently exists. Several papers have been published regarding iodine inhibitors in the diet and their relative impact, as well as the importance of adequate intake of iodine with which the inhibitors must compete (Tonacchera et al., 2004; Dasgupta et al., 2006; DeGroef et al., 2006; Tran et al., 2008). The inference is that when the body is replete with iodine, there is a decreased likelihood of a given dose of any iodine inhibitor having a toxicological effect.

The World Health Organization, along with the UN Children's Fund and the International Council for the Control of Iodine Deficiency Disease in 1994 recommended the median urinary iodide concentration be in the 100–200μg/l range to indicate adequate iodine intake and nutrition (100μg/l corresponds to an intake of approximately 150μg/day). The recommended daily allowance for pregnancy is 220 and is 290μg/day during lactation; the tolerable upper intake level (UL) for adults, the highest level likely to pose no risk of adverse health effects in almost all individuals, is 1100μg/day (Russell et al. 2001). The American Thyroid Association, citing a WHO recommendation, concludes 225 to 375 μg/day is optimal for pregnancy and lactation (Becker et al., 2006).

Although Caldwell et al. (2005) highlighted a stabilization of iodine intake after a downward trend between 1971–1974 NHANES I and 1988–1994 NHANES III, their data nevertheless show that 38% of females of ≥6 years and 37.7% of pregnant women have urinary iodine <100μg/l, and 7.31% of the latter are <50μg/l.

Significantly, in a study by Leung et al. cited by Dasgupta (2009), 31% of non-prescription and 72% of prescription multivitamins for pregnant women in the United States did not contain iodine. In those that listed iodine from KI the actual amount was 24% lower than the labeled amount, and those in which iodine was derived from kelp varied by 50%. Concern for hypertension has cut salt consumption, and only one-fifth of salt that is consumed contains iodine (Dasgupta et al., 2008). The NRC emphasized the importance of all pregnant women receiving adequate iodine, and recommended ensuring that by giving consideration to adding iodide to all prenatal vitamins (National Research Council (NRC), 2005).

Further research is needed to investigate the magnitude of exposure to iodine inhibitors, the possible thyroid consequences during times of stress on the thyroid, such as pregnancy, and for susceptible life stages, including the fetus, premature neonates, bottle- and breast-fed infants and children, as well as remedial measures addressing those with low iodine intake.

The Three-Bin Approach

Before conducting the data merge, we attempted to derive a food dose by using urinary data from individuals who did not consume tap water. The results revealed very little difference between food alone and food plus water in that population of subjects. We attribute this result to the fact that for the average consumer, the occurrence of perchlorate in tap water is rare enough that for most of the population it does not add to total urinary perchlorate, although it could in those specific locales identified as positive for perchlorate in UCMR. During the merge portion of the study, we analyzed a larger data set of those potentially exposed to perchlorate through drinking water and those who are likely to be only exposed through food.

The assignment of individuals to bins based on possible tap water perchlorate exposure was spatially and temporally imprecise; matching residence location and utility data by county alone for privacy does not allow for accurate categorization of individuals in counties with multiple utilities. Perchlorate levels in tap water may vary with time because of differences in source water, possibly leading to misclassification because of differing collection times for UCMR and NHANES samples. Bin I was populated by individuals residing in counties where UCMR provides some indication of perchlorate occurrence in public drinking water by having at least one positive perchlorate detection in the county. This positive result indicated that these subjects lived in the same county in which an entry point(s) in a system was positive at least once during UCMR monitoring (four required surface water sampling times and two for ground water). However, counties can be served by more than one water system, and individual systems can have multiple distribution system entry points. Therefore, the presence of perchlorate in one or more samples in a county does not indicate that the NHANES-tested individual residing in that county was drinking tap water containing perchlorate, that is, they were served water from the same water system or entry point in a water system that had tested positive. Exposure is relevant only to people residing in the service area of that affected entry point at the time it is affected, and not the whole population of the service area of the entire water treatment system at all times. Only 4.1% of systems detected perchlorate of ≥4μg/l. At the distribution entry points, 1.9% of water samples were positive. For many water utilities, perchlorate was only detected once. It is not surprising then that the mean results for Bin II (unknown) would tend to be more similar to Bin III (food alone) than Bin I, as Bin I was sorted to contain subjects from counties with a UCMR-positive system who possibly were ingesting perchlorate in tap water. Our data did not include temporal stratification because of the lack of available sampling dates to correlate with the NHANES data set.

Bin I, which is food plus tap water, is useful for subtracting the food results in Bin III to derive a sense of water exposure on a national basis for comparison with exposure from food alone. It represents a relative contribution from the two sources of exposure for the whole population. The greater the food contribution is to total exposure, the less significant the water contribution is to that exposure. As discussed in the Results section, at the 90th percentile for the total population, water exposure represented 16% of total exposure, with the remaining 84% of the exposure coming from food sources, whereas the total intake from both sources at the 90th percentile represents approximately 27% of the RfD. For systems having a more significant drinking water component of exposure, Bin III is the most important category for the purpose of this study: to investigate the dose of perchlorate in food and to derive the portion of the reference dose (converted to concentration) that remains for water exposure against which to compare individual public water system monitoring results.

Winter-grown lettuce and other produce and forage crops from southern California's Imperial Valley and Arizona's Coachella Valley irrigated with Colorado River water containing perchlorate have contributed to the presence of urinary perchlorate throughout the country regardless of local tap water quality. Ongoing remediation efforts have been successful in curtailing perchlorate from entering Las Vegas Wash leading to Lake Mead. A report compiled by EPA Region 9 of monitoring data from Nevada, California, and Tronox Corporation reports significant decreases in perchlorate releases and water concentrations in Lake Mead and the Colorado River. Results from a seep leading to the Las Vegas Wash decreased 95% from 2002 to 2005, and concentration in the Wash dropped 75–80%. The Nevada Division of Environmental Protection found that by 2005, the annual average perchlorate concentration at Willow Beach on the Colorado River below Hoover Dam declined 63% from 6.5p.p.b. in 2000 to 2.4p.p.b. in 2005 (US EPA, 2005). In a recent report, CDC reported on human exposure to environmental chemicals including perchlorate that compared urinary perchlorate data of 2001–2002 NHANES data with that of 2003–2004 (CDC, 2009). In all age groups, race/ethnicity, gender and percentiles, the urinary perchlorate decreased in the later survey by approximately 20%, although projecting a longer-term temporal trend may be premature (Mendez et al., 2010). It is possible that a decrease in concentration of perchlorate used to irrigate food crops consumed nationwide coincided with the national decrease in urinary perchlorate observed, and that effect might be expected to be more influential through the national exposure to foods than in the more localized intake of drinking water. However, the influence of foreign food imports, the sampling locations for NHANES and the seasonality of foods also factor into the results. The impact of the more recent NHANES data on our analysis would have been a somewhat lower dose in food to subtract from the RfD, leading to a correspondingly greater allowance in the drinking water.

The data merge did not distinguish the source of drinking water, either public or private, of the people. Nationally, approximately 15% of the population is served by private wells. The UCMR surveyed public water systems, governed by the Safe Drinking Water Act (SDWA). Whatever action might be contemplated with respect to perchlorate under the SDWA would only affect public water system users and the derivation of a water allowance considers the data for this group. If some of the private wells unknowingly contained some perchlorate and were placed into Bin III, the derived food dose might be slightly higher as a result, and the food and water dose (Bin I) slightly lower. The population of Bin I represents approximately 15% of the NHANES population of Bins I and III combined. If contamination is proportional in the 15% of private wells, then ~2% (15% of the 15% of wells) may have perchlorate in the water. This is close to 1.9% of positive samples in UCMR. Correcting for possible private well water contribution to urinary perchlorate data reveals an insignificant difference in results.

A minimum reporting limit of 4μg/l represents an uncertainty regarding the occurrence and concentration of perchlorate in tap water below this limit. Bin III includes only those people whose public water has been tested revealing no perchlorate at the reporting limit, or those using RO or only bottled water. But study participants may have been misclassified because of perchlorate in tap water at levels below the UCMR detection limit. For example, a 70-kg person drinking 2liters of water containing 3.5μg/l perchlorate daily would have a perchlorate dose of 0.11μg/kg/day (16% of the RfD). However, the fact that there was little discernable difference between urinary perchlorate levels among tap and non-tap water drinkers underscores that there is likely very little perchlorate in non-detected samples. Nevertheless, to account for such uncertainty we used a conservatively large food intake at the 90th percentile to subtract from the reference dose to derive a water limit that is smaller than it would be using the mean.

Creatinine Adjustment for Perchlorate Intake Estimates

The key assumption for use of the creatinine adjustment approach—that virtually 100% of ingested perchlorate is excreted in urine (with the exception of lactating women)—is supported by the data of Yu et al. (2002) who observed 99.5% excretion in 48h by male rats. Anbar et al. (1959) conducted a double-labeled isotope study confirming absence of perchlorate metabolism in humans Paulus et al. (2007) found that perchlorate is excreted 75–90% in 24h. However, several recent perchlorate exposure studies report that approximately 70% of a perchlorate dose is excreted in urine (Lawrence et al., 2000; Braverman et al., 2006). Lawrence et al. (2000) reported 100% excretion in four subjects in 24h and 50% excretion in three others. Greer et al. (2002) speculated that their results differ because of a more even temporal distribution of dosing compared with the study of Lawrence et al. (2000), in which ad libitum dosing by subjects may have resulted in disproportionate amounts ingested before clinic visits, causing uneven results. Braverman et al. (2006) reported 65–70% excretion of perchlorate doses within 24h, and no perchlorate detected in blood serum of any of their subjects 1 month after perchlorate dosing was discontinued.

Patterns of intake of food and water that contain perchlorate and the timing of sampling affect the amount of excretion detected in an individual's spot urine sample. There were some observed differences with respect to fasting times and urinary perchlorate. Generally, higher perchlorate values were observed in individuals with shorter fasting times before sampling as documented by Blount et al. (2006). This is expected because of the short half-life of perchlorate in the body. However, it was noted that the average fasting time in our data exceeded what would be expected overnight, indicating that some participants were perhaps not accurate in their recall, or preferred to leave the section blank rather than to recall particulars. Overall, the meal times, foods chosen and likely perchlorate content of food or water would sufficiently vary for the study population so that individual temporal differences in eating habits would not be significant with respect to the overall study objectives of calculating a daily mean or upper percentile food intake of perchlorate.

As previously stated, the NAS identified the maternal–fetal dyad as a sensitive population. Because of changes in body weight, water fraction, uterine muscle development (and creatine intake and creatinine excretion) and additional urine volume, estimates of 24-h creatinine excretion are less precise for pregnant women. Accordingly, some caution should be applied when interpreting perchlorate exposure estimates for this group. Profound changes occur in a woman's body throughout her pregnancy, and neither urine volume nor time from last void is measured in NHANES. The literature is mixed with regard to the deceptively simple question of typical urine volumes during pregnancy. Thorp et al. (1999) cite both increases and decreases in other studies, and their pregnant urine volumes are smaller than reported non-pregnant averages from Boeniger et al. (1993). Urination frequency increases during pregnancy, yet there is no consensus whether the 24-h urine volume increases or decreases throughout the pregnancy.

Strengths, Limitations and Uncertainties of Analysis

The strength of this analysis is that it estimates perchlorate doses based on perchlorate excretion levels from a large number of individuals rather than estimating perchlorate ingestion based on assumed food consumption. Our estimates are based on a diverse population and include questionnaire data to further evaluate potential exposure sources. The methodology provides a novel opportunity to use public water system occurrence and human biomonitoring data to directly inform decisionmakers. Subtracting the selected percentile food dose for a given population from the RfD provides an evaluative measure against which water system monitoring results can be compared.

An underlying assumption is that food and water are the only two primary sources of perchlorate exposure. Dermal absorption seems negligible, and because of its low vapor pressure, inhalation is only a factor for those exposed to particulates in industrial settings (Gibbs et al., 1998; Lamm et al., 1999), and none of the NHANES participants in our sub-sample worked in such a setting.

Summary

The study estimates the background dose of perchlorate from food for different populations, and uses the remainder of the reference dose as an allowance in drinking water. The authors used recent refinements in creatinine adjustments to more precisely estimate dose across different demographic groups. Methodologically, the use of NHANES urinary biomonitoring data, that is, output data, seems to be a viable and valuable tool for estimating the intake dose for a chemical such as perchlorate that is not metabolized and is excreted almost exclusively through the kidneys. It can augment the more traditional estimates of food-based doses such as FDA's TDS. Such biomonitoring analysis provides a tool for quantitatively comparing different exposure sources and routes of exposure to contaminants in the environment.

Top

Conflict of interest

The authors declare no conflict of interest.

Top

References

  1. Anbar M., Guttman S., and Lewitus Z. The mode of action of perchlorate ions on the iodine uptake of the thyroid gland. Int J Appl Radiat Isot 1959: 7: 87–96. | Article | PubMed | ChemPort |
  2. Becker D.V., Braverman L.E., Delange F., Dunn J.T., Franklyn J.A., Hollowell J.G., Lamm S.H., Mitchell M.L., Pearee E., Robbins J., and Rovet J.F. Iodine supplementation for pregnancy and lactation — United States and Canada: recommendations of the American thyroid association. Thyroid 2006: 10: 949–951. | Article |
  3. Blount B.C., Pirkle J.L., Osterloh J.D., Valentin-Blasini L., and Caldwell K.L. Urinary perchlorate and thyroid hormone levels in adolescent and adult men and women living in the United States. Environ Health Perspect 2006: 114(12): 1865–1871. | PubMed | ISI | ChemPort |
  4. Blount B.C., Valentin-Blasini L., Osterloh J.D., Mauldin J.P., and Pirkle J.L. Perchlorate exposure of the US population, 2001–2002. J Expo Sci Environ Epidemiol 2007: 17: 400–407 . | Article | PubMed | ISI | ChemPort |
  5. Boeniger M.F., Lowey L.K., and Rosenberg J. Interpretation of urine results used to assess chemical exposure with emphasis on creatinine adjustments: a review. Am Ind Hyg Assoc J 1993: 54(10): 615–627. | Article | PubMed | ISI | ChemPort |
  6. Braverman L.E., Pearce E.N., Xuenmei H., Pine S., Seeley M., Bech B., Mangani B., Blount B., and Firek A. Effects of six months of daily low-dose perchlorate exposure on thyroid function in healthy volunteers. J Clin Endocrinol Metab 2006: 91(7): 2721–2724. | Article | PubMed | ISI | ChemPort |
  7. Caldwell K.L., Jones R.T., and Hollowell J.G. Urinary iodine concentration: United States National Health and Nutrition Examination Survey 2001–2002. Thyroid 2005: 15(7): 692–699. | Article | PubMed | ISI | ChemPort |
  8. CDC. Fourth National Report on Human Exposure to Environmental Chemicals. Centers for Disease Control and Prevention, 2009, pp 243–246. Available at http://www.cdc.gov/exposurereport/pdf/FourthReport.pdf.
  9. CDC. National Health and Nutrition Examination Survey 2004, 2004 Available at: http://www.cdc.gov/nchs/nhanes.htm.
  10. Cockroft D.W., and Gault M.H. Prediction of creatinine clearance from serum creatinine. Nephron 1976: 16: 31–41. | Article | PubMed | ISI | ChemPort |
  11. Crump K.S., and Gibbs J.P. Benchmark calculations for perchlorate from three human cohorts. Environ Health Perspect 2005: 113(8): 1001–1008. | Article | PubMed | ISI | ChemPort |
  12. Dasgupta P.K., Dyke J.V., Kirk A.B., and Jackson W.A. Perchlorate in the United States. Analysis of relative source contributions to the food chain. Environ Sci Technol 2006: 40(21): 6608–6614. | Article | PubMed | ISI | ChemPort |
  13. Dasgupta P.K., Yining L., and Dyke J.V. Iodine nutrition: iodine content of iodized salt in the United States. Environ Sci Technol 2008: 42(4): 1315–1323. | Article | PubMed | ISI | ChemPort |
  14. Dasgupta P.K., Perchlorate: a couse for iodine deficiency? Environ Chem 2009: 6: 7–9. | Article | ISI | ChemPort |
  15. DeGroef B., Decallonne B.R., Van der Geyten S., Darras V.M., and Bouillon R. Perchlorate versus other environmental sodium/iodide symporter inhibitors: potential thyroid-related health effects. Eur J Endocrinol 2006: 155: 17–25. | Article | PubMed | ChemPort |
  16. Divi R.L., Chang H.C., and Doerge D.R. Anti-thyroid isoflavones from soybean: isolation, characterization, and mechanisms of action. Biochem Pharmacol 1997: 54(10): 1087–1096. | Article | PubMed | ISI | ChemPort |
  17. El Aribi H., Le Blanc Y.J.C., Antonsen S., and Sakuma T. Analysis of perchlorate in foods and beverages by ion chromatography couple with tandem mass spectrometry (IC-ESI-MS/MS). Anal Chim Acta 2006: 567(1): 39–47. | Article | PubMed | ISI | ChemPort |
  18. Gibbs J.P., Ahmad R., Crump K.S., Houck D.P., Leveille T.S., Findley J.E., and Francis M. Evaluation of a population with occupational exposure to airborne ammonium perchlorate for possible acute or chronic effects on thyroid function. J Occup Environ Med 1998: 40(12): 1072–1082. | Article | PubMed | ISI | ChemPort |
  19. Greer M.A., Goodman G., Pleus R.C., and Greer S.E. Health effects assessment for environmental perchlorate contamination: the dose response for inhibition of thyroidal radioiodine uptake in humans. Environ Health Perspect 2002: 110(9): 927–937. | Article | PubMed | ISI | ChemPort |
  20. Grubbs F. Procedures for detecting outlying observations in samples. Technometrics 1969: 11: 1–21. | Article | ISI |
  21. Hogue C. Rocket fueled river. Chem Eng News 2003: 81(33): 37–46. | Article | ISI |
  22. Kirk A.B., Dyke J.V., Martin C.F., and Dasgupta P.K. Temporal patterns in perchlorate thiocyanate, and iodide excretion in human milk. Environ Health Perspect 2007: 115: 182–186. | Article | PubMed | ISI | ChemPort |
  23. Kirk A.B., Martinelango P.K., Tian K., Dutta A., Smith E.E., and Dasgupta P.K. Perchlorate and iodide in dairy and breast milk. Environ Sci Technol 2005: 39: 2011–2017. | Article | PubMed | ISI | ChemPort |
  24. Lamm S.H., Braverman L.E., Li F.X., Richman K., Pino S., and Howearth G. Thyroid health status of ammonium perchlorate workers: a cross-sectional occupational health study. J Occup Environ Med 1999: 41(4): 248–260. | Article | PubMed | ISI | ChemPort |
  25. Lawrence J.E., Lamm S.H., Pino S., Richman K., and Braverman L.E. The effect of short-term low-dose perchlorate on various aspects of thyroid function. Thyroid 2000: 10(8): 659–663. | Article | PubMed | ISI | ChemPort |
  26. MA DEP. (2005) The occurrence and sources of perchlorate in Massachusetts. Draft Report. Massachusetts Department of Environmental Protection. Available at: http://www.mass.gov/dep/.
  27. Mage D.T., Allen R.H., Gondy G., Smith W., Barr D.B., and Needham L.L. Estimating pesticide dose from urinary pesticide concentration data by creatinine correction in the Third National Health and Nutrition Examination Survey (NHANES-III). J Expo Anal Environ Epidemiol 2004: 14: 457–465. | Article | PubMed | ISI | ChemPort |
  28. Mage D.T., Allen R.H., and Kodali A. Creatinine corrections for estimating children's and adult's pesticide intake doses in equilibrium with urinary pesticide and creatinine concentrations. J Expo Anal Environ Epidemiol 2008: 18(4): 360–368. | Article | ChemPort |
  29. Mendez W., Dederick E., and Cohen J. Drinking water contribution to aggregate perchlorate intake of reproductive-age women in the United States estimated by dietary intake simulation and analysis of urinary excretion data. J Expo Sci Environ Epidemiol 2010: 20(3): 288–297. | Article | PubMed | ISI | ChemPort |
  30. Murray C.W., Egan S.K., Kim H., Beru N., and Bolger P.M. US Food and Drug Administration's Total Diet Study: dietary intake of perchlorate and iodine. J Expo Sci Environ Epidemiol 2008, 1–10. | Article |
  31. National Research Council (NRC). Health Implications of Perchlorate Ingestion. National Academies Press, Board on Environmental Studies and Toxicology: Washington, DC 2005.
  32. Paulus B., Bazar M., Salice C., Mattie D., and Major M. Perchlorate inhibition of iodide uptake in normal and iodine-deficient rats. J Toxicol Environ Health A 2007. 70: 1142: 1149.
  33. Pearce E.N., Leung A.M., Blount B.C., Bazrafshan H.R., He X., Pino S., Valentin-Blasini L., and Braverman L.E. Breast milk iodine and perchlorate concentrations in lactating Boston-area women. J Clin Endocrinol Metab 2007: 92(5): 1673–1677. | Article | PubMed | ISI | ChemPort |
  34. Russell R.M., Beard J.L., Cousins R.J., Dunn J.T., Ferland G., Hambidge K.M., Lynch S., Penland J.G., Ross A.C., Stoecker B.J., Suttie J.W., Turnlund J.R., West K.P., Zlotkin S.H., Munro I.C., Murphy S.P., Young V.R., Trumbo P.R., and Yates A.A. Dietary reference intakes for vitamin A, vitamin K, arsenic, boron, chromium, copper, iodine, iron, manganese, molybdenum, nickel, silicon, vanadium, and zinc. National Academies Press, 2001.
  35. Sanchez C.A., Barraj L.M., Blount B.C., Scrafford C.G., Valentin-Blasini L., Smith K.M., and Krieger R.I. Perchlorate exposure from food crops produced in the lower Colorado River region. J Expo Sci Environ Epidemiol 2009: 19(4): 359–368. | Article | PubMed | ISI | ChemPort |
  36. Sanchez C.A., Crump K.S., Krieger R.I., Khandaker N.R., and Gibbs J.P. Perchlorate and nitrate in leafy vegetables of North America. Environ Sci Technol 2005: 59(24): 9391–9397. | Article | ChemPort |
  37. Steinmaus C., Miller M.D., and Howd R. Impact of smoking and thiocyanate on perchlorate and thyroid hormone associations in the 2001–2002 National Health and Nutrition Examination Survey. Environ Health Perspect 2007: 115(9): 1333–1338. | Article | PubMed | ISI | ChemPort |
  38. Thorp J.M., Norton P.A., Wall S.S., Kuller J.A., Eucker B., and Wells E. Urinary incontinence in pregnancy and the puerperium: a prospective study. Am J Obstet Gynecol 1999: 181(2): 266–273. | Article | PubMed | ISI |
  39. Tonacchera M., Pinchera A., Dimida A., Ferrarini E., Agretti P., Vitti P., Santini F., Crump K., and Gibbs J. Relative potencies and additivity of perchlorate thiocynate, nitrate and iodide on the inhibition of radioactive iodide uptake by the human sodium iodide symporter. Thyroid 2004: 14: 1012–1019. | Article | PubMed | ISI | ChemPort |
  40. Tran N., Valentín-Blasini L., Blount B.C., McCuistion C.G., Fenton M.S., Gin E., Salem A., and Hershman J.S. Thyroid-stimulating hormone increases active transport of perchlorate into thyroid cells. Am J Physiol Endocrinol Metab 2008: 294: E802–E806 . | Article | PubMed | ISI | ChemPort |
  41. Urbansky E., Collette T., Robarge W., Hall W., Skillen J., and Kane P. Survey of fertilizers and related materials for perchlorate (ClO4), 2001; EPA/600/R-01/tba.
  42. US EPA. Estimated per capita water ingestion and body weight in the United States–an update. EPA-822-R-00-001, 2004 Office of Water, Office of Science and Technology, US EPA.
  43. US EPA. “Integrated risk information system (IRIS), perchlorate and perchlorate salts.” February 2005. Available at: http://www.epa.gov/iris/subst/1007.htm.
  44. US EPA. Preliminary regulatory determination for perchlorate. Federal Register 2008: 73: 60262–60282.
  45. US EPA Region 9. Perchlorate Monitoring Results. Henderson, Nevada to the Lower Colorado River. December 2005. http://ndep.nv.gov/BCA/file/perchlorateeighthmonrpt123105.pdf.
  46. Yu K.O., Narayanan L., Mattie D.R., Godfrey R.J., Todd P.N., Sterner T.R., Mahle D.A., Lumpkin M.H., and Fisher J.W. The pharmacokinetics of perchlorate and its effect on the hypothalamus-pituitary-thyroid axis in the male rat. Toxic Appl Pharmacol 2002: 182: 148–159. | Article | PubMed | ISI | ChemPort |
Top

Acknowledgements

We express our appreciation to Peter Meyer and Deborah Rose at the National Center for Health Statistics (NCHS) for their outstanding cooperation in approving and conducting the data merge.

Supplementary Information accompanies the paper on the Journal of Exposure Science and Environmental Epidemiology website