Volume II Chapter 4.0 Pages 3 of 4 page next page 4

4.6.2. U.S. Food 4-25

4.6.3 European Food 4-38

4.6.4 Canadian Food 4-41

4.6.2. U.S. Food

The published data on measured levels of CDDs, CDFs, and dioxin-like compounds in U.S. food products have generally come from studies of a specific food product(s) in a specific location(s) rather than from large survey studies designed to allow estimation of daily intake of the chemicals for a population. For example, CDD/Fs are not routinely monitored in the U.S. Food and Drug Administration's (FDA) Surveillance Monitoring Program for domestic and imported foods (conversation between Dr. S. Page, FDA, and G. Huse, Versar, Inc., February 8, 1993) nor are they routinely monitored by the U.S. Department of Agriculture (USDA) in the National Meat and Poultry Residue Monitoring Program (conversation between Dr. E. A. Brown, USDA-FSIS, and G. Schweer, Versar, Inc., February 8, 1993).

However, USDA has developed some site-specific, though dated (late 1970s), CDD monitoring data. These efforts were in response to a decline in general health noted by inspectors in several cattle herds in Michigan. Wood products in the local barns and other cattle holding premises, presumed to be treated with pentachlorophenol (PCP), were suspected as the cause of this health decline (Buttrill et al., no date; Tiernan et al., 1978). PCP was suspected to contain trace CDD and CDF levels as manufacturing contaminants at that time. In response to this incident, two national investigations were performed by USDA.

The first study involved the analysis of peritoneal adipose and liver samples collected from beef cattle in 23 States (Tiernan et al., 1978), while the second study involved the analysis of adipose tissue samples (body region not specified) collected from dairy cattle in 30 States--neither study specified the cattle breeds for any sample. HxCDD, HpCDD, and OCDD were screened for in the analyses of samples from each study. In the beef cattle study (Tiernan et al., 1978), 220 samples were analyzed: 189 peritoneal adipose samples and 31 liver samples.

No residues were detected in any liver samples. A total of 19 (i.e., 10 percent) of the 189 adipose samples were found to positively contain HxCDD, HpCDD, or OCDD levels >0.10 ppb, while 56 (i.e., 30 percent) contained levels <0.10 ppb that were detectable based on the signal-to-noise ratio of the analytical instrumentation. OCDD accounted for the majority of the samples that positively contained CDDs (i.e., 17 or 9.0 percent) while only 3 samples contained HxCDD and 2 samples contained HpCDD residues, respectively. A total of 358 adipose samples were analyzed in the dairy cattle study (Buttrill et al., no date).

Nine samples (i.e., 2.5 percent) positively contained CDD levels >0.19 ppb or the "level of reliable measurement", while another 30 samples (i.e., 8.4 percent) contained CDD levels that were identifiable yet below the "level of reliable measurement" (i.e., not positively identified due to low concentration levels). As with the beef cattle study results, OCDD accounted for the majority (eight) of positive samples. HpCDD was identified in only a single sample that also contained OCDD. HxCDD was identified as well in only a single sample.

The data from the USDA studies are not useful for estimating CDD/F exposure for two reasons. First, the samples were analyzed for only 3 of the 17 CDD/F congeners with dioxin-like toxicity, and these were reported on a homolog basis rather than a congener-specific basis. Second, the limit of detection was at or above 0.1 ppb or 100 ppt. Background levels for individual congeners appear to be much less than 100 ppt. For example, the highest congener levels in beef fat analyzed by Fürst et al. (1990) were 5.4 ppt for OCDD.

FDA has also conducted some limited analyses for the higher-chlorinated dioxins in market basket samples collected under FDA's Total Diet Program (Firestone et al. 1986). Food samples found to contain PCP residues >0.05 m g/g were analyzed for 1,2,3,4,6,7,8-HpCDD and OCDD. Also, selected samples of ground beef, chicken, pork, and eggs from the market basket survey were analyzed for these dioxin congeners, regardless of the results of PCP residue analysis.

A total of 16 ground beef samples, 18 pork samples, 16 chicken samples, and 17 eggs samples with no PCP contamination were collected between 1979 and 1984 at various locations throughout the United States and analyzed for 1,2,3,4,6,7,8-HpCDD and OCDD.

No dioxin residues were detected in any of the ground beef or egg samples. OCDD was observed at detectable concentrations in only 2 of the 18 pork samples (27 ppt 53 ppt) and 2 of the 16 chicken samples (29 ppt, 76 ppt). One chicken sample with PCP residues >0.05 m g/g had detectable residues of both 1,2,3,4,6,7,8-HpCDD (28 ppt) and OCDD (252 ppt). Egg samples from Houston, Texas and Mesa, Arizona with PCP residues >0.05 µg/g had detectable 1,2,3,4,6,7,8-HpCDD levels ranging from 21 ppt to 588 ppt, and OCDD levels ranging 80 ppt to 1610 ppt.

These levels were attributed to local PCP contamination (Firestone et al., 1986). Milk samples contaminated with PCP at levels ranging from 0.01 m g/g to 0.05 m g/g PCP contained no detectable dioxins. It should be noted that these food residue data were not used in this assessment of dioxin exposures in the United States because the reported limits of detection (10 to 40 ppt) for the FDA analyses were considerably higher than the levels of dioxins observed in foods from more recent studies.

Also, the study only analyzed for residues of 2 of the 17 toxic CDD/CDF congeners. Finally, the study focussed on samples with PCP contamination and, therefore, was not generally representative of background exposures.

The primary sources of information on background levels of CDD/Fs in U.S. foods are studies conducted by the California Air Resources Board (CARB), the results of background analysis from the NCASI study (Stanley and Bauer, 1989; LaFleur et al., 1990) and Schecter et al. (1993). Each of these three studies is summarized below.

CARB collected multiple samples of seven types of foods from commercial food sources in two urban areas of California (Stanley and Bauer, 1989). Foods were collected randomly, but an emphasis was placed on food stuffs of California origin (Stanley and Bauer, 1989). The types of food stuffs included saltwater fish, freshwater fish, beef, chicken, pork, milk, and eggs.

A total of 210 samples were collected in Los Angeles (30 individual samples of each of the 7 types of foods), and 140 samples were collected in San Francisco (20 individual samples of each of the 7 types of foods). Food items were composited before chemical analysis to obtain a sample that was representative of average levels of PCDDs and PCDFs in the food stuffs, increase the probability of detection, and reduce the cost of chemical analysis.

Samples were composited separately for each type of food stuff, within each geographical area. Each composite sample contained 6 to 10 individual food samples, and 5 to 8 composite samples were analyzed for each food type. Beef (ground beef), pork (bacon), and chicken samples were analyzed on a lipid weight basis, but were subsequently converted to a wet weight basis, for the purposes of this report, by multiplying the lipid weight concentration of CDD/CDFs by the fraction of fat contained in the food product of interest. Milk and fish samples were also analyzed on a lipid weight basis. Egg samples were analyzed for CDD/CDFs on a wet weight basis. The CARB data are summarized in Table 4-2.

The NCASI study (as described by LaFleur et al., 1990; and Henry et al., 1992) collected random food samples directly from the shelves of grocery stores located in the southern, midwestern and northwestern regions of the United States. The samples were analyzed for 2,3,7,8-TCDD and 2,3,7,8-TCDF. These data are summarized in Table 4-3.

Schecter et al. (1993) conducted a complete congener analyses of 18 food samples collected directly from a supermarket in Binghamton, New York in early 1990. The samples included five fish, three types of beef (ground beef, beef sirloin tip, and beef rib steak), one chicken drumstick, one porkchop, one lamb, one ham, one bologna, one heavy cream, and four types of cheese. The following ranges of TEQ levels on a whole weight basis were found: fish: 0.01 - 0.13 ppt; meat: 0.03 - 1.5 ppt; and dairy products: 0.04 - 0.7 ppt. These data are summarized in Table 4-4.

Beef and Pork
Background TEQ concentrations of CDD/Fs in beef/veal and pork were estimated using data from eight CARB samples (Stanley and Bauer, 1989), three NCASI background samples (LaFleur et al., 1990), and three samples from Schecter et al. (1993). The CARB and Schecter et al. samples were analyzed for 16 2,3,7,8-substituted CDD/F congeners; the NCASI samples were analyzed for 2,3,7,8-TCDD and 2,3,7,8-TCDF only. At least one congener was detected in 13 of the 14 composite beef samples.

One sample had no detectable congeners. The congeners most frequently detected in beef/veal were 1,2,3,4,6,7,8-HpCDD and OCDD, and only one congener was not detected in any of the samples. For the purposes of this report, the total whole weight TEQ for beef was calculated by assuming that the lipid content of beef was 19 percent and by using one-half the detection limits to represent the concentration of nondetectable CDD/F congeners in the samples.

table Table 4-2 Summary of Dioxin/Furan Data Collected in the California State Air Resources Board Study. table Table 4-3 Summary of U.S. Food Data from NCASI Study.
expand table Table V2 4-2 expand table Table V2 4-3
table Table 4-4 Summary of Schecter et al. (1993) Data on U.S. Foods.

Using this methodology, the total background TEQ was estimated to be 0.48 ppt for beef on a wet weight basis. If nondetectable concentrations are assumed to be zero; the estimated total TEQ for beef is estimated to be 0.29 ppt.

All of the pork samples analyzed by CARB, NCASI, and Schecter et al. (1993) had at least one 2,3,7,8-substituted CDD/F at detectable concentrations.

The hepta- and octa-chlorinated dioxins and furans were detected most frequently, and only one congener was not detectable in any of the samples analyzed.

expand table Table V2 4-4
Using an assumed lipid content of 15 percent and one-half the detection limit to represent nondetectable concentrations, the estimated total whole weight TEQ for pork is 0.26 ppt. If nondetectable concentrations are assumed to be zero, the estimated TEQ for pork is 0.10 ppt. Therefore, the TEQ concentration for pork is expected to be between 0.10 ppt and 0.26 ppt.

Chicken and Eggs
Background TEQ concentrations for chicken are based on data from CARB (Stanley and Bauer, 1989) and Schecter et al. (1993). Nine composite chicken samples were analyzed for 16 2,3,7,8-substituted CDD/F congeners.

All chicken samples contained detectable concentrations of at least one CDD/F congener. The hepta-chlorinated CDD/Fs and OCDD were detected most frequently, and 5 of the 16 congeners were not detected in any of the 9 chicken samples. The total background whole weight TEQ for chicken is estimated to be 0.19 ppt using an assumed lipid content of 15 percent and one-half the detection limit to represent the concentration of nondetectable congeners. Using zeros to represent nondetectable concentrations, the estimated total whole weight TEQ for chicken is 0.07 ppt.

Therefore, the TEQ for chicken is expected to be between 0.07 ppt and 0.19 ppt. Background TEQs for eggs are based on data from CARB (Stanley and Bauer, 1989). For eggs, seven out of eight composite samples had no detectable concentrations of 2,3,7,8-substituted CDD/F congeners.

Only one sample contained detectable concentrations of OCDD, 2,3,7,8-TCDF, and 1,2,3,4,6,7,8-HpCDF. Using one-half the detection limit for nondetectable concentrations, the estimated total TEQ for eggs is 0.14 ppt; but using zero for the nondetectable concentrations, the estimated TEQ for eggs is only 0.0004.

Milk and Milk Products
Background levels of CDD/Fs in U.S. milk are based on a very limited data set. CARB (Stanley and Bauer, 1989) analyzed eight packaged milk samples and found an average of 0.06 ppt TEQ. Becuase these samples may have been impacted by leaching of CDD/Fs from packaging materials, they were eliminated from the analysis of background concentrations conducted for this report.

LaFleur et al. (1990) analyzed a single background milk sample for 2,3,7,8-TCDD and 2,3,7,8-TCDF. The sample contained 2,3,7,8-TCDD at a concentration of 0.0018 ppt and nondetectable concentrations of 2,3,7,8-TCDF. Based on the LaFleur et al. (1990) data, the TEQ for these two congeners is estimated to be 0.0018 ppt whether one-half the detection limit or zero is used to represent the nondetectable concentration of 2,3,7,8-TCDF.

EPA (1991b) collected milk samples from several sites in the vicinity of a municipal waste incinerator in Rutland, Vermont, and two background samples from a dairy farm 123 kilometers from the incinerator where no obvious industrial sources of CDD/F were present. All samples were taken from bulk storage tanks at the farms.

The report indicated that facility emissions could not be correlated with the levels of CDD/F and other contaminants measured in various environmental media. For all milk samples, the majority of the congeners were not detected. It was reported that only OCDD was consistently detected at levels from 0.2 to 2.4 pg/g in the farms near the facility.

The levels in milk from the three farms near the facility ranged from about 0.2 to 0.4 pg of TEQ/g whole milk, and the TEQ for the background samples collected from the distant farm was 0.12 pg/g. The TEQs were calculated by EPA (1991b) by setting the nondetects equal to the detection limit. The 0.12 ppt TEQ background value estimated by EPA is nearly 2 orders of magnitude higher than the TEQ for milk based on the NCASI data. (This is probably due largely to the incomplete congener analysis conducted by LaFleur et al.)

Examination of the raw data supporting this study indicated that all of the CDD/F congeners in the background sample were nondetectable. Consequently, if nondetects are set to zero, the total background TEQ for milk would be zero. If half the detection limits are used to calculate the total TEQ level, the estimated value is 0.07. Therefore, the total background TEQ level for milk is expected to be between zero and 0.07 pg/g.

Some idea of the total TEQ level of CDD/F in milk samples can be gained by assuming that levels in beef fat are similar to levels in milk fat. This assumption implies that the differences in feeding/raising practices of dairy cattle vs. beef cattle do not cause substantial differences in CDD/F exposure. Beef contains approximately 20 percent fat, and whole milk is about 4 percent fat.

Thus, on a whole food basis, CDD/F levels in beef should be about five times higher than in milk. Support for this concept can be seen in the German data presented in Table 4-2. This table shows that the TEQ level in milk fat is 1.35 ppt and in beef fat is 1.08 ppt. On this basis, the North American data for beef (0.48 ppt of TEQ) suggest that milk would be about 0.1 ppt of TEQ. This lends support to the background level reported in EPA (1991b) of 0.12 ppt, based on only two samples.

Schecter et al. (1992) reported on the analysis of 2,3,7,8-substituted CDD/Fs in U.S. dairy products. Cottage cheese, soft cream cheese, and American cheese samples were selected randomly from New York supermarkets and analyzed on a wet-weight basis.

All of the dairy products sampled had at least 13 detectable congeners out of the 17 evaluated. Only one congener (1,2,3,7,8,9-HxCDF) was not detectable in any of the five dairy products. Based on these data, the total background TEQ concentration of dairy products is estimated to be 0.36 ppt if one-half the detection limit is used to represent nondetectable concentrations and 0.35 if zero is used to represent nondetectable concentrations.

Fruits and Vegetables
Data on CDDs and CDFs in U.S. fruit and vegetable products are extremely limited. The Ministry of the Environment, Ontario, conducted a study of CDDs and CDFs in locally produced and imported fruits and vegetables, some of which originated in the United States (Ministry of the Environment, 1988; Birmingham et al., 1989).

Samples of fresh apples, peaches, potatoes, tomatoes, and wheat products were analyzed. In general, the minimum detection limits for these analyses were less than 1 ppt. The report indicated that "fruit and vegetable samples were substantially free of PCDD and PCDF residues, especially the more toxic tetra, penta, and hexachlorinated forms" (Ministry of the Environment, 1988). OCDD was the only congener detected in any of the samples.

One apple and one peach sample contained detectable OCDD concentrations (8 ppt and 0.6 ppt, respectively). Detectable OCDD concentrations were found at concentrations ranging from 1 to 3 ppt in potatoes and 0.6 to 0.7 ppt in wheat samples. None of the tomato samples contained detectable levels of any CDD or CDF congeners. Based on these results, Birmingham et al. (1989) estimated the TEQs for fruits, vegetables, and wheat products to be 0.004 ppt, 0.002 ppt, and 0.0007 ppt, respectively.

As discussed in Volume III, dioxin contamination of fruits and vegetables is thought to occur primarily via particle deposition or vapor adsorption onto outer layers with little penetration to inner portions. Plant uptake from the soil via the roots is generally considered negligible. However, the work of Hülster and Marschner (1993) indicates that zucchini and pumpkins were exceptions.

For these plant species, it appears that root uptake occurs and leads to a uniform concentration within the fruit. The concentration of CDDs and CDFs in zucchini squash grown on "uncontaminated" soil (0.4 ppt TEQ soil concentration) ranged from 0.5 to 0.7 ppt TEQ dry weight. These reported values may be converted to whole weight TEQ concentrations by using an assumed moisture content of 93.7 percent (USDA, 1979-1984). The resulting range of whole weight concentrations for zucchini is 0.03 to 0.04 ppt TEQ. Müller et al. (1993) also evaluated CDDs and CDFs in vegetables (carrots, lettuce, and peas) grown at both contaminated plots and control plots.

For the control plots, the highest levels of CDDs and CDFs were observed in carrot peels: 0.55 ppt TEQ dry weight, or 0.07 ppt TEQ whole weight, assuming a moisture content for carrots of 87.8 percent (USDA, 1979-1984). Lower concentrations were observed in samples from the cortex of the carrots, indicating that the "contamination source for the peel of carrots is the soil" (Müller et al., 1993). Lettuce concentrations ranged from 0.1 to 0.4 ppt TEQ dry weight.

This is equivalent to a whole weight concentration range of 0.005 to 0.018 ppt TEQ, assuming a moisture content of 95.4 percent for lettuce (USDA, 1979-1984). Concentrations in peas from contaminated plots ranged from 0.04 to 0.12 ppt TEQ dry weight (0.004 to 0.013 ppt TEQ whole weight, assuming a moisture content of 88.9 percent). Lower concentrations in peas (i.e., close to the detection limit; exact value not given) were reported for control plots. Similar data for vegetables grown in the United States were not available.

Vegetable Oil
The high fat levels in vegetable oil suggest that it may be important to consider as a source of human exposure. Vegetable oils can be made from a variety of plants including corn, olives, peanuts, sunflower seeds, safflower seeds, linseed, and cotton seed. Many of these items are protected from atmospheric deposition, which implies that their CDD/F levels would be low.

However, Theleen (1991) estimated that vegetable oil could contribute about 10 percent of a person's total daily intake in the Netherlands (14 of 120 pg TEQ/d). This estimate was based on the Fürst et al. (1990) study that found nondetects for most congeners except some of the higher chlorinated congeners of CDD and CDF (detection limit = 0.5 ppt). Half the detection limit was used for the nondetects, and most of the congeners were not detected. Consequently, the actual value could be much lower. No data could be found on CDD/F levels in vegetable oil in North America.

U.S. Food Summary
The U.S. food data are summarized in Table 4-5. The background TEQ estimates are presented first assuming that nondetects equal half the detection limits and second assuming that nondetects equal zero. For food groups such as eggs, a wide range of TEQ estimates are seen indicating a high percent of nondetects among individual congeners.

The upper mean TEQ estimates are generally comparable to the TEQ estimates derived from studies conducted in Germany and Canada (as discussed below). ...

table Table 4-5 Summary of CDD/CDF Levels in U.S. Food (pg/g Fresh Weight).
... These studies did not report TEQs based on assuming nondetects equal to zero but did report many nondetects in some food groups. In summary, the limited number of U.S. food samples and the high incidence of nondetects make an uncertain basis for estimating national background levels. However, the general agreement with food level estimates reported for Canada and Germany provides some reassurance that these U.S. values are reasonable.

It is clear, however, that a large survey is needed to confirm residue levels of CDD/F in the U.S. food supply.

For the purposes of calculating background exposures to CDD/Fs via dietary intake, the upper-range background TEQs (i.e., those calculated using one-half the detection limit for the nondetects) were used (See Chapter 5.)
expand table Table V2 4-5

4.6.3 European Food

Relatively extensive multiyear surveys of the levels of CDDs and CDFs in food are being undertaken in Sweden and the United Kingdom (de Wit et al., 1990 and Startin et al., 1990). The most extensive investigations reported to date that involve testing of a variety of randomly selected food samples collected within the framework of official food control have been performed in the Federal Republic of Germany (Beck et al. 1989; Fürst et al., 1990).

The detailed results of these studies are included in Appendix B. Fürst et al. (1990) analyzed 107 food samples collected in Germany. The results of this study are presented in Table 4-6. All samples, except some of the milk, were randomly collected during official food monitoring programs. The authors speculated that a source may have been near the areas where the milk samples were collected because they appeared higher than other milk tested in Germany which showed levels around 1 ppt TEQ.

In a later report, Fürst et al. (1991) reported that a much larger survey of dairies in Germany had been completed. This survey analyzed 168 samples of milk and milk products collected at dairies prior to bottling. They found an arithmetic mean of 1.35 pg of TEQ/g of fat. TEQs in these studies were estimated by assuming that nondetects equalled half the detection limits.

The percent detected was not reported. Fürst et al. (1991) provided a summary of the results of several European studies. The data summaries relevant to background levels in meat and dairy products from Fürst et al. (1991) are presented in Table 4-7. Fürst et al. (1991) report that information on CDD and CDF levels in vegetables and fruits is scarce and that the available data indicate a background of below 1 ppt.

Beck et al. (1989) analyzed 12 food samples collected randomly from food markets in West Berlin, Germany. Chicken, eggs, butter, pork, ocean perch, cod, herring, vegetable oil, cauliflower, lettuce, cherries, and apples were analyzed for CDD/Fs. CDD/Fs were detected in samples of animal origin in the ppq to ppt range (fat weight basis). No CDD/F congeners were detected at a detection limit of 0.01 ppt (whole weight basis) in samples of plant origin.

table Table 4-6 CDD/CDF Levels in German Food. table Table 4-7 CDD/CDF Background Levels in Some European, Canadian, and New Zealand Food.
expand table Table V2 4-6 expand table Table V2 4-7

Theelen et al. (1993) collected food products from various locations in the Netherlands and analyzed them for 2,3,7,8-chlorine substituted dioxins, furans, and planar PCBs. Meat samples were collected from slaughter houses throughout the Netherlands. Fish, mixed meats, and cheeses were gathered at various grocery stores.

Mixtures of foods in these categories were prepared based on the proportion of the average annual consumption rate that different food items in these categories represented. The food industry provided purified oils and fats. Mixtures of these items were also prepared in proportion to their annual use in the Netherlands. The concentration of CDD/Fs in these food products are presented in Table 4-8.

4.6.4 Canadian Food

Birmingham et al. (1989) analyzed CDD/F residues in food collected in Ontario, Canada. Most of the food was grown in Canada, although some was from the United States. They reported analyzing 25 composite samples from 10 food groups.

The precise number of samples in each food group was not reported. No TeCDD, PeCDD, HxCDD, TeCDF, or PeCDF were found at detection limits of 0.1 to 7 ppt. Low ppt levels of some of the higher chlorinated CDD/Fs were detected in some foods. TEQ levels were also estimated for the major food groups.

However, as shown in Table 4-9, these data were reported on a homolog basis. It is unclear what procedure was used to convert the homolog data to TEQ. The text implies that nondetects were treated as zero for purposes of estimating TEQ.

In addition to the animal food data shown in Table 4-9, measurements were also made in potatoes, apples, tomatoes, peaches, and wheat. Only OCDD was detected at levels ranging from 0.6 to 8 pg/g fresh weight. The TEQ totals for vegetables were reported as 0.004 ppt for fruit, 0.002 ppt for vegetables, and 0.0007 ppt for wheat- based products. The procedure used to develop these TEQ estimates was not clear.