June 22, 2004

Rocket Fuel in Cows' Milk - Perchlorate: Milk Consumption Not Safe?

Infants and children may consume more perchlorate in milk than the EPA considers safe

Drinking milk from cows raised in some parts of California may expose infants and children to more of a toxic rocket fuel chemical than is currently considered safe by the U.S. Environmental Protection Agency (EPA) and the State of Massachusetts, according to unreleased tests by state agriculture officials and independent laboratory tests commissioned by Environmental Working Group (EWG).

In the first study to look for perchlorate in California supermarket milk, EWG found perchlorate in almost every sample tested - 31 out of 32 samples purchased from groceries in Los Angeles and Orange counties. EWG also filed a state Public Records Act request to obtain results of tests for perchlorate in milk by the California Department of Food and Agriculture (CDFA), which the agency has neither released publicly nor provided to state health officials. CDFA's tests found perchlorate in all 32 samples of milk collected from unspecified sources in Alameda, Sacramento, and San Joaquin counties. ‡1

EWG's computer-assisted analysis of federal dietary data shows that by drinking milk contaminated with the average levels ‡2 of perchlorate found in the two studies, 7 percent of women of childbearing age would get a daily dose of rocket fuel larger than the level currently considered safe by the EPA. But children are by far more at risk: Half of all children 1 to 5 would exceed EPA's safe dose just by drinking milk, and more than a third would get twice that dose. One-third of children 6 to 11 would get a larger dose than EPA says is safe, with one-fifth consuming twice as much.

Last month, California health officials adopted a public health goal for perchlorate of 6 parts per billion (ppb) in a liter of drinking water. (A public health goal, or PHG, is the level used to set an enforceable state drinking water standard, expected later this year.) But the EPA's provisional daily safe dose (RfD, for reference dose) is equivalent to 1 ppb - the same level as a risk assessment released last month by Massachusetts state scientists. [1, 2] Using the California health goal as a benchmark, only about 1 percent of children under 12 would get an unsafe dose of perchlorate by drinking milk.

The average perchlorate level in milk tested by the state, 5.8 ppb, was essentially the same as the maximum safe level for drinking water under the state's new PHG. The average level found in the EWG tests was lower, at 1.3 ppb, but could still present health risks, particularly for small children, who drink large amounts of milk relative to their size.

But both the EPA and Massachusetts have criticized the California goal as too weak in light of the available data on health effects at very low levels. According to the report from the Massachusetts Department of Environmental Protection, "the currently available data support a lower value in order to be sufficiently protective of sensitive individuals, including pregnant women and infants." [3] The report said that an even lower standard would be supported by the toxicity data, but that detection methodologies are not yet good enough to reliably detect perchlorate below 1 ppb.

What's more, EWG's dietary risk analysis does not include additional perchlorate exposure through drinking water or foods other than milk. In 2003, EWG and the Riverside Press-Enterprise found perchlorate contamination, at levels exceeding the EPA's provisional reference dose for drinking water, in samples of Southern California lettuce grown with contaminated water. [4, 5] EWG's study found that if women ate lettuce with perchlorate concentrations at the average level found in our contaminated samples, 57 percent would get a dose greater than the EPA's daily safe dose. These findings have since been confirmed by tests on lettuce and other foods conducted by the U.S. Army Corps of Engineers and academic researchers in Arizona and Texas. [6, 7, 8] Therefore, for many Californians, perchlorate exposure can't be assessed by looking at milk, or any other commodity, in isolation.

Fifty percent of children aged one to five would be getting more perchlorate per day than the EPA thinks is safe just by drinking milk graph

Perchlorate, the explosive main ingredient of solid rocket and missile fuel, can affect the thyroid gland's ability to take up the essential nutrient iodide and make thyroid hormones. These hormones regulate metabolism in adults and play important roles in the development of organ systems in fetuses, infants and children. Small disruptions in thyroid hormones in utero or during early development can cause lowered IQ; larger disruptions can cause mental retardation, loss of hearing and speech, or deficits in motor skills. [1] EWG's analysis of the latest scientific studies, showing harmful health effects from very low doses, argues that a national perchlorate safety standard should be no higher than one-tenth the EPA's currently recommended level of 1 ppb. [9]

There are no enforceable safety standards for perchlorate in drinking water at either the state or federal level. Attempts at setting them have been highly contentious and drawn out for more than a decade. Last year, in what was widely seen as another stalling tactic, the Bush Administration directed the National Academy of Sciences to review the EPA's 2002 study, and results are expected later this year. But evidence that at least two members of the review panel have financial ties to perchlorate polluters has raised what Sens. Barbara Boxer and Dianne Feinstein, in a letter to the NAS, call "serious issues about NAS review of conflict of interest, bias, and balance for this panel." [10] In June 2004, one of the two NAS panel members with ties to Lockheed was forced to resign.

During the weeks of December 8, 2003 and January 12, 2004, EWG and Physicians for Social Responsibility purchased 32 containers of milk from stores in Irvine, Pacific Palisades, Santa Fe Springs, Whittier and three different locations in Los Angeles. The samples included low-fat and whole milk from four well-known brands. Scientists at Texas Tech University, a leading center of perchlorate-related research, conducted the analyses. All but one sample contained measurable amounts of perchlorate, with the samples averaging 1.3 ppb.

Our study was the first to look for perchlorate in California milk. But in 2003, Texas Tech researchers were the first to find that milk could be contaminated. In that study, researchers found perchlorate in 7 out of 7 samples of whole milk purchased from local supermarkets, with levels as high as 6.3 ppb. The milk was purchased from grocery stores in Lubbock, where some water supplies are known to be contaminated with low levels of perchlorate. The samples represented six different brands of milk - essentially all of the brands of milk available in the area. [11] They also found perchlorate in a single sample of human breast milk taken from a woman living near Lubbock. The scientists were extremely thorough in validating their results: Each sample was tested three different times tested by two different laboratories, using two different methods. With these findings, plus the EWG and CDFA studies, there is no question that perchlorate is being passed from contaminated water into the U.S. food supply.

31 out of 32 samples of milk EWG tested had detectable perchlorate

  Brand City Purchased Fat Content Perchlorate level (ppb)
1 Brand D Los Angeles whole 3.62
2 Brand A Los Angeles 2 percent 2.66
3 Brand C Santa Fe Springs whole 2.06
4 Brand A Los Angeles whole 1.92
5 Brand D Santa Fe Springs whole 1.89
6 Brand A Los Angeles 2 percent 1.87
7 Brand B Pacific Palisades 2 percent 1.85
8 Brand C Santa Fe Springs whole 1.63
9 Brand B Los Angeles 2 percent 1.59
10 Brand A Whittier 2 percent 1.56
11 Brand C Santa Fe Springs whole 1.45
12 Brand C Santa Fe Springs "reduced fat" 1.38
13 Brand D Los Angeles whole 1.32
14 Brand C Santa Fe Springs "reduced fat" 1.31
15 Brand A Whittier whole 1.29
16 Brand A Whittier whole 1.08
17 Brand C Santa Fe Springs "reduced fat" 1.04
18 Brand C Santa Fe Springs whole 1.01
19 Brand B Pacific Palisades 2 percent 0.98
20 Brand B Whittier whole 0.95
21 Brand A Whittier whole 0.90
22 Brand A Los Angeles whole 0.78
23 Brand B Irvine whole 0.76
24 Brand B Irvine whole 0.74
25 Brand B Los Angeles whole 0.72
26 Brand A Whittier 2 percent 0.69
27 Brand B Irvine 2 percent 0.68
28 Brand B Irvine 2 percent 0.66
29 Brand A Los Angeles 2 percent 0.65
30 Brand B Irvine whole 0.60
31 Brand B Irvine whole 0.58
32 Brand B Los Angeles 2 percent ND
Average perchlorate concentration in all EWG milk samples: 1.3 ppb

After we completed analysis of our samples, EWG heard reports that the California Department of Food and Agriculture (CDFA) had also tested milk for perchlorate, but had not shared its findings with health officials. EWG requested CDFA's findings under the state Public Records Act. The state records showed CDFA collected samples of milk in Alameda, Sacramento, and San Joaquin counties, and found perchlorate in every one of them - in concentrations that were often higher than EWG's samples. The perchlorate levels in the CDFA samples ranged from 1.5 ppb to 10.6 ppb. The average level found, 5.8 ppb, is essentially the same concentration the state has said should be the limit for drinking water. [12]

Every milk sample CDFA tested had detectable perchlorate ‡1

County Perchlorate level (ppb) County Perchlorate level (ppb)
Sacramento 10.6 Sacramento 5.9
Sacramento 9.0 San Joaquin 5.6
Sacramento 8.8 Sacramento 5.5
Sacramento 8.8 Alameda 5.1
Sacramento 8.7 Sacramento 4.6
Sacramento 8.3 Sacramento 4.0
Sacramento 8.3 Sacramento 4.0
Sacramento 7.7 Alameda 4.0
Sacramento 7.6 Sacramento 3.9
Sacramento 7.4 San Joaquin 3.8
Sacramento 7.1 Sacramento 3.6
Alameda 6.7 Sacramento 3.4
Sacramento 6.7 Sacramento 3.1
Sacramento 6.4 Sacramento 2.1
Sacramento 6.1 San Joaquin 1.6
Sacramento 6.1 Alameda 1.5
Average perchlorate concentration in all CDFA milk samples: 5.8 ppb

Almost as disturbing as the concentrations of perchlorate found by CDFA is the fact that the agency kept the data secret. The state Office of Environmental Health Hazard Assessment, which was in charge of studies to set the recent public health goal, was not even informed that CDFA was doing a study. At a time when state officials up to and including Gov. Arnold Schwarzenegger were working to reach a final decision on a public health goal, and under intense lobbying by perchlorate polluters, CDFA apparently didn't think finding rocket fuel in supermarket milk was important enough to tell anyone. [13]

The documents received EWG from CDFA leave many questions about the state's testing program. They show that the samples were analyzed in April 2004 by the California Animal Health and Food Safety Laboratory at UC Davis. But they do not include information about how the study originated, how the samples were collected, what analytical method was used or whether the agency planned more tests.

EWG requested details that would put the results in context, but CDFA refused to supply them. The agency's chief lawyer replied to our request with a one-sentence statement: "These [documents] are our public records." Sources within the agency say the samples EWG obtained documentation of may be part of a much larger sampling project, but CDFA has released no additional information to the public or to state health officials.

Although we can't be sure of the source of the perchlorate found in any of CDFA's or EWG's samples, the chemical clearly is coming either from contaminated food or drinking water provided to livestock. Testing of public water supply wells around the state has turned up more than 350 drinking water sources in ten counties, with many sources still untested. [14] Although there has been little testing of agricultural wells, given the extent of perchlorate pollution it is virtually certain that at least some California dairy cows are drinking contaminated water.

But a more significant source of the perchlorate in milk is likely to be alfalfa grown with contaminated water. Scientists have found that alfalfa concentrates perchlorate to levels up to 950 times higher than found in the water it was irrigated with. [6] In 2002, California's Imperial County alone harvested more than 1.4 million tons of alfalfa for livestock feed - virtually all of it grown with water from the Colorado River. [15] This water is tainted with perchlorate from a now-closed manufacturing plant near Las Vegas, once owned by the Department of Defense and is now owned by defense contractor Kerr-McGee. [4]

EWG's and CDFA's findings of perchlorate in milk, and the agency's attempt to keep them quiet, call into question how rigorously the agency is upholding the state Food and Agriculture Code. A primary legislative charge in California's law regulating commercial livestock feed is to "to ensure in every way possible a clean and wholesome supply of meat, milk, and eggs for the benefit of the consumer." [16] CDFA claims that its inspection, sampling, testing, education, and "voluntary quality assurance" programs are "leading the nation in ensuring a safe and wholesome supply of commercial feed to the benefit of both the agricultural community and the consumer." [17] Ensuring that feed is uncontaminated is particularly important since, according to CDFA, California produces more livestock feed than any other state, and Imperial County is the state's leading feed producer. [17]

Although there are no safety standards for perchlorate in food, EPA has been working for more than a decade toward establishing a safe drinking water standard. In 2002, EPA proposed a reference dose (RfD) for perchlorate of 0.00003 milligrams per kilogram of body weight per day, which is the dose at which no adverse effect would be expected. [1] To assess the health risks of perchlorate in milk, EWG analyzed USDA's 1994-1996 Continuing Survey of Food Intakes by Individuals database, which monitors the food consumption of thousands of ordinary Americans. [18] The database contained information on more than 14,000 children (ages 1 to 11) and over 4,000 women of childbearing age (ages 15-44) whose milk consumption was measured over two days. (The database also includes information on other milk products such as cheese, yogurt and ice cream, but we only considered fluid milk consumption in our analysis.)

Using the USDA consumption data and adjusting for body weight, we used a probability-based model to calculate the percentage of individuals who would exceed EPA's provisional RfD if they drank milk with perchlorate concentrations found in EWG's and CDFA's studies. Our analysis showed that 51 percent of children aged 1 to 5 would get a perchlorate dose higher than the rfd, as would 35 percent of children aged 6 to 11 just by drinking milk. Thirty-five percent of children aged 1 to 5 and 20 percent of children aged 6 to 11 - would get a dose twice as high as the EPA thinks is safe. And 7 percent of women of childbearing age (15-44) would exceed EPA's RfD just by drinking milk.

Population of Concern Percent of individuals exceeding EPA's proposed reference dose
Analysis using EWG data only Analysis using EWG and CDFA data Analysis using CDFA data only
Infants and young children
(ages 1-5)
24% 51% 71%
(ages 6-11)
7% 35% 57%
Women of childbearing age
(ages 15-44)
0.2% 7% 14%

Source: [18]

It is clear from the USDA data that children are the most at risk for perchlorate in milk. Children not only tend to drink milk more often than adults, they also drink milk in greater quantities. For example, according to the USDA database, about 85 percent of children aged 1 to 5 drank milk on a given day, compared to only about 45 percent of women aged 15 to 44. And of those who drank milk, children between the ages of 1 and 2 drank an average of 1.8 cups of milk per day, while women in their twenties drank just a half-cup per day. EWG's analysis shows how much more likely kids are to exceed the EPA's proposed safety standard for perchlorate.

Milk appears to be a less significant exposure route for adults - unsurprising since adults typically drink much less milk in proportion to their body weight than kids. Yet, according to EWG analysis, 7 percent of women of childbearing age (15-44) would still exceed the EPA's proposed safe perchlorate level just by drinking milk. These calculations used milk concentrations from both EWG and CDFA's study. The percentages of individuals found to exceed EPA's reference dose are somewhat different if the calculations use only EWG's or CDFA's data. But no matter which subset of data is used in the calculations, it is clear that California milk is an important perchlorate exposure route for many children and some adults.

Again, none of our calculations account for water consumption or food consumption other than milk. With more than 350 drinking water sources in California contaminated with perchlorate, and the contaminated lower Colorado River irrigating 1.4 million acres of some of the nation's most productive farmland, it is very likely that many people are being exposed to perchlorate via multiple pathways. [14] EWG's 2003 study of winter lettuce found that if women ate lettuce with perchlorate at the average level found in our contaminated samples, 57 percent would get a dose greater than the EPA's rfd. [4]

Our findings have broad implications not just for the safety of the U.S. food supply, but also for current efforts to set perchlorate safety standards. Perchlorate has now been found to be taken up and concentrated by many foods besides lettuce and milk. Perchlorate exposure clearly is not just a risk for people in areas where the water is contaminated, but for anyone who consumes food produced with contaminated water.

‡ Footnotes

‡1 - CDFA documents number their samples from 1 to 34, but numbers 19 and 23 are missing.

‡2 - EWG employed a probability-based model to estimate the percentage of individuals in the USDA's CSFII database that would exceed the EPA's reference dose, based on their body weight and milk consumption patterns, if they drank milk contaminated with perchlorate at the levels found in EWG's and CDFA's samples. The perchlorate concentrations of individual EWG and CDFA samples (not average concentrations) were used in the analysis. A perchlorate concentration of zero was assumed for those samples where perchlorate was not detected.