The Urban Environment and Children's Health: Soils as an Integrator of Lead, Zinc, and Cadmium in New Orleans, Louisiana, U.S.A.

Environmental Research Section A 81, 117^129 (1999) Article ID enrs.1999.3966, available online at http://www/idealibrary.com on

The Urban Environment and Children's Health: Soils as an Integrator of Lead, Zinc, and Cadmium in New Orleans, Louisiana, U.S.A.1 H.W. Mielke,* C. R. Gonzales,* M. K. Smith,* and P.W. Mielke{ *Institute of Bioenvironmental Toxicology (CBR), College of Pharmacy, Xavier University, New Orleans, Louisiana 70125; and {Department of Statistics, Colorado State University, Fort Collins, Colorado; 80523 Received July 13, 1998

INTRODUCTION Soils are evaluated as a diagnostic tool of environmental conditions that in£uence health. The samples for this study are urban topsoil (0^2.5 cm depth) samples (n = 4026) analyzed for Pb, Zn, and Cd by inductively coupled plasma-atomic emission spectrometry (ICP-AES). The parent materials for New Orleans soils are derived from the Mississippi River, and alluvium from the Bonnet Carre¨ Spillway (n = 31) serve as control samples for this study. The urban samples were strati¢ed by census tract (n = 286). Blood Pb (BL) levels of children 6 years were also strati¢ed by census tract and paired with soil Pb (SL) (n = 175). A signi¢cant association (P = 1.26 610723) was found between median BL and median SL. The association was modeled by BL = 3.06+0.33 (SL)0.5 (correlation coe¤cient = 0.69 between the modeled BL and the observed BL and P = 3.56 610722). A median SL threshold (310 mg g71 and 5310 mg g71) for higher metal census tracts (HMCTs) and lower metal census tracts (LMCTs), respectively, represents median BL exposures above and below 9 mg dL71. HMCTs and LMCTs were characterized by demographic and socioeconomic data. HMCTs are more likely (P = 4.56 61076) inhabited by Blacks than by Whites. Of 13,803 children 6 years in HMCTs, 75% are Black and 22% are White, with other making up the remaining 3%. In LMCTs, the Black to White children ratio is 50:50. In HMCTs, socioeconomic indicators for Blacks are depressed compared to Whites. Zn and Cd are potentially phytotoxic in HMCTs. Children exhibit a steep rise in BL at SL 5100 mg g71, and empirically, a safe SL for most children is around 80 mg g71. SL is a useful diagnostic tool, and curtailing SL may complement primary Pb prevention for children. # 1999 Academic Press Key Words: lead poisoning; urban geochemistry; urban soil Pb map; Zn and Cd phytotoxicity; environmental justice.

Lead (Pb) poisoning as a health issue has been described as the silent epidemic.Young children are the most sensitive group to exposure (Mushak et al., 1989). This study focuses on the contribution of soil Pb (SL) to childhood lead exposure. Soils are an integrator and re£ect all of the activities that have taken place in the course of their existence. The study samples are from New Orleans, located on the Mississippi River Delta of southern Louisiana. The control samples for this study are alluvium, which are the parent materials for the soils of the delta. If there is a deposition of metal dust, regardless of the source of the metal (lead-based paint, incinerator fallout, gasoline, industry, etc.), the soils re£ect that deposition. At any instant, the accumulated results of the processes and activities that have occurred are recorded in the composition of the soil. A series of studies on urban SL began with gardens of Baltimore (Mielke et al., 1983).The inner city garden soils had disproportionately elevated Pb content compared to gardens located at increasing distance from the city (Mielke et al., 1983). Further investigations of residential SL in several cities of Minnesota and Louisiana con¢rmed the earlier results (Mielke et al., 1984, 1984/85; Mielke, 1991, 1993). The surveys in Minneapolis, St. Paul, and New Orleans resulted in SL maps of those cities (Mielke and Adams, 1989, Mielke, 1994). In general, the SL in large nonsmelter cities is highest in the inner city (with medians of 500 to 1000 mg g71) and declines steeply toward outlying communities. The suburbs and small cities contain substantially less SL (medians of about 50^60 mg g71 or less) than large cities. An understanding about the accumulation of Pb in urban soil was in£uential in the legislation to rapidly phase out Pb in gasoline (Rosner and Marowitz, 1985, Mielke, 1984).

1 This research was supported by MHPF/ATSDR cooperative agreement U50/ATU398948 to Xavier University.

117 0013 -9351/99 $30.00 Copyright # 1999 by Academic Press All rights of reproduction in any form reserved.

118

MIELKE ET AL.

A substantial national decline in blood Pb (BL) is attributed to the removal of Pb from gasoline as well as reducing Pb in the food canning process (Bolger et al., 1991, Brody et al., 1994, Pirkle et al., 1994). Currently, about 900,000 children under 6 years old in the United States have BL of at least 10 mg Pb per dL, a level high enough to adversely a¡ect their intelligence, behavior, and development (CDC, 1997). Higher Pb exposures are disproportional across the population and occur mainly among children living in poverty conditions in United States inner cities (Brody et al., 1994; Pirkle et al., 1994). This population consists primarily of African-American children. Some researchers suggest that Black race is a strong predictor for high BL (Lanphear et al., 1996, 1998; Sargent et al., 1995). However, suggesting that race is a predictor of high BL does not assist with understanding the underlying causes of the problem. In order to formulate policies and response actions, it is essential to understand the basis for the Pb-exposure problem. Previous evaluation of the association between children's Pb exposure, as measured by BL, and SL, paired by census tract, noted that SL is much better than age of housing as a predictor of BL (Mielke et al., 1997). The purpose of this study is to review the model of the association between SL and BL, to use the model to depict census tracts containing higher versus lower BL exposure risk in New Orleans, and to characterize demographic and socioeconomic indicators of these two groups of census tracts. MATERIALS AND METHODS

There are several components to the database and the research undertaken for this study. First, a SL (plus Zn and Cd) database was created. Second, a BL database was developed, and analysis of the association between the databases was conducted. Third, the ¢ndings were applied to contrasting and mapping the higher versus lower risk areas of the city. Finally, 1990 United States census data were used to characterize the demographic and socioeconomic indicators of higher and lower risk areas of the city.

water pressure and damage to the levee system that protects New Orleans from £ooding. The sediments carried by the Mississippi River were deposited as alluvium as the £oodwaters spread out and slowed in their £ow through the spillway to Lake Pontchartrain. Alluvium samples of various particle sizes, ¢ne sand to silt, were collected from the spillway.The alluvium samples represent the parent materials that compose the sedimentary soils of New Orleans and the entire lower Mississippi River Delta. Previous research on urban soils showed that SL occurs in community-sized (i.e., census tract) patterns in a city and that comparable quantities of SL occur in similar-sized communities of Minnesota and Louisiana (Mielke et al., 1989; Mielke, 1993). For this reason, census tracts were used to de¢ne communities. The protocol includes the following features: samples were collected from the top 2.5 cm of the surface within residential neighborhoods of census tracts and at least one block from the busiest streets and away from street corners. Early studies showed the following: that soils collected from the drip-line (i.e., next to the foundations) contained the largest and most variable quantity of Pb; that soil samples collected from mid-yards and open spaces (parks or playgrounds) contained the lowest and less variable amount of Pb; and that samples collected within a meter of the street contained quantities of Pb that were intermediate between the other two types of samples and also exhibited relatively low variability of Pb (Mielke et al., 1989; Mielke, 1991).Thus, for the New Orleans survey, we chose to collect 15 soil samples from each census tract; 9 samples were collected from within 1 m of the street, 3 samples from the drip-line, and 3 samples from vacant lots or mid-yard areas at least 3 m away from houses. All soil samples were brought to the laboratory and air-dried (24 h), sieved (2 mm screen), shaker extracted 5:1 (volume to weight) with 1 mol L71 nitric acid for 2 h at room temperature, centrifuged (1000g for 15 min), and ¢ltered (Fisherbrand Qualitative P4 Filter Paper) (re¢ned from Mielke et al., 1983). The extracts were analyzed, using inductively coupled plasma atomic emission spectroscopy (ICP-AES), for Pb, Zn, and Cd.

Soil Survey of Pb, Zn, and Cd in Control samples and Metropolitan New Orleans Samples

Association between Soil Lead and Blood Lead

The raw materials for this study were 31 control samples collected from Mississippi River alluvium in the Bonnet Carre¨ Spillway and 4026 urban soil samples collected from metropolitan New Orleans. The Bonnet Carre¨ Spillway alluvium was deposited between March 17 and April 18, 1997 when the gates of the spillway were most recently opened to relieve

Previous research in Minnesota indicated that the amount of SL of census tracts is connected with BL exposure for children in the same census tracts (Mielke et al., 1989). The BL data for Louisiana were from the State's O¤ce of Public Health Blood Lead Records.The BL data consists of records of 5765 children (6 years and younger) across the urban, suburban, and

119

URBAN SOILS AND CHILDREN'S HEALTH

rural spectrum of census tracts from Orleans and neighboring Lafourche Parishes (see Mielke et al., 1997). The research in Louisiana showed an association between SL and BL that was extremely strong, and Fisher's exact test yielded a P value of 3.2610724 (Mielke et al., 1997). The strength of this association exists without accounting for other factors known to in£uence exposure, such as the presence or absence of old housing, condition of neighborhood, seasonality, and ages of the children. The original ¢ndings relied on data obtained with atomic absorption spectrometry (AAS) methods (Mielke et al., 1997).The current review of the model relies on SL results analyzed with ICP-AES. Spearman's rho test is used to evaluate the association between environmental measures of central tendency (i.e., the median SL by census tracts) and the central tendency of children's exposure (i.e., the median BL by census tracts). Least sum of absolute deviations regression is used to model the SL and BL data (Gray et al., 1992). Higher Metal Census Tracts and Lower Metal Census Tracts A strong association between SL and BL justi¢es using the model to select a threshold that de¢nes the higher metal census tracts (HMCTs) and lower metal census tracts (LMCTs). In addition, a strong association justi¢es applying the threshold to all census tracts of New Orleans and mapping the HMCTs and LMCTs of the city. The threshold will be de¢ned as the median SL of a census tract that partitions the median BL of 4 9mg dL71 and 9 mg dL71 for children living in the same census tract (US DHHS, 1991). The HMCTs provide data about concentration of SL that is associated with higher BL exposures. Likewise, the LMCTs provide data about concentra-

tion of SL that is related to lower BL exposures and provides empirical data about an amount of SL that may be safe for most children. Census Tract Data The demographic and socioeconomic data for this study were obtained from the 1990 United States census survey. New Orleans consists of 286 census tracts (US Census Bureau, 1990). The numbers of census tracts populated by a simple majority of Blacks or Whites for both HMCTs and LMCTs were sorted by demographic and socioeconomic characteristics. The following demographic variables were evaluated: total population by race, total childhood population 6 years old by race. Socioeconomic indicators include median annual household income, median monthly rent, percentage of pre-1940 housing, percentage owner-occupied housing, percentage renter-occupied housing, percentage housing vacancy, and percentage of households headed by females with at least one child under 18. Geographic Information System (GIS) techniques were used to assemble soil data, demographic, and socioeconomic indicators for the HMCTs and LMCTs of New Orleans. RESULTS

Soil Pb, Zn, and Cd in Control samples and Metropolitan New Orleans Samples The upper part of Table 1 lists the percentiles of Pb, Zn, and Cd in alluvial control soils collected from Bonnet Carre¨ Spillway. The lower part of Table 1 lists percentiles of the results for Pb, Zn, and Cd for all census tracts of metropolitan New Orleans. Note the small Pb, Zn, and Cd content and their small range

TABLE 1

Description of Pb, Zn, and Cd in Fresh Mississippi River Alluvium and Metropolitan New Orleans Soils by Percentile Element

1 M HNO3 extractable metal (mg g71 soil) Mississippi River Alluvium N Min 10% 25% 50% 75%

90%

Lead (Pb) Zinc (Zn) Cadmium (Cd)

31 31 31

6.6 16.0 1.06

Element Lead (Pb) Zinc (Zn) Cadmium (Cd)

1.7 5.6 0.48

2.1 6.0 0.52

2.6 7.0 0.57

3.7 10.1 0.72

5.0 11.8 0.85

1 M HNO3 extractable metal (mg g71 soil) for all Metropolitan New Orleans samples N Min 10% 25% 50% 75% 4026 4026 4026

ND ND ND

12.4 25.4 0.9

37 54 1.7

120 131 3.2

330 312 5.2

Max 7.4 17.4 1.11

90%

Max

813 735 7.5

191,000 26,600 87

Note. ND is equal to 510 mg g71. The Mississippi River Alluvium samples were analyzed omitting sample dilution.

120

MIELKE ET AL.

in the alluvial soil samples compared to the high content and large range in urban soils. Association between Soil Pb and Blood Pb The association of SL and BL was evaluated from a database consisting of paired median SL and median BL for 175 census tracts. The Spearman's rho test of association yields a P value of 1.2610723 for the paired SL and BL data of 175 census tracts. The least sum of absolute deviations regression model equation is BL = 3.06 + 0.33 (SL)0.5 (correlation between the modeled BL and observed BL is 0.69 and the P value is 3.5610722). This model is illustrated in Fig. 1. In addition, the previous model from AAS data for SL is included in Fig. 1. Solving the equation for a median BL of 9 mg dL71 yields a median SL of 310 mg g71. HMCTs are de¢ned as median SL  310 mg g71 and LMCTs are de¢ned as median SL5310 mg g71. Accordingly, when the median concentration of SL is 310 mg g71, then an increasing number of children (about half or more of the children) exhibit BL concentrations of 10 mg dL71. When the median concentration of SL 5310 mg g71, then an increasing percentage of the children exhibit BL 510 mg dL71.

Pb, Zn, and Cd of HMCTs and LMCTs Table 2 lists groups of census tracts based on the median SL content (using 310 mg g71 as the partition for HMCTs vs 5310 mg g71 for LMCTs) along with the results for Zn and Cd. In the HMCTs, the median Pb is 481 mg g71 or 6 times larger than the median of 80 mg g71 for LMCTs. The HMCTs have a median soil Zn of 337 mg g71 or 3.5 times more and a median soil Cd of 5.0 mg g71 or 1.8 more than the LMCTs, respectively. With soil sample sizes of 677 and 3349, the Mann^Whitney test showed statistically signi¢cant di¡erences for Pb, Zn, and Cd between the two groups (P5 107100). Note that for SL, the lowest census tract median is 10.1 mg g71 and the maximum census tract median is 1130 mg g71.With regard to Zn and Cd, there is some overlap between HMCTs and LMCTs. Thus, while the association between Pb, Zn, and Cd is good, there are some census tracts where Zn and Cd do not follow Pb (Mielke and Smith, 1997). Figure 2, shows the relationship between maximum, mean, and minimum SL as a function of increasing median SL contents of the census tracts of metropolitan New Orleans. Drip-line soil samples next to the foundation often contain Pb-based paint

FIG. 1. Relation of median BL as a function of median SL derived from two di¡erent analytical methods, AAS and ICP-AES.The models of the relationship are from paired empirical data from 175 census tracts in the case of the ICP-AES database and 173 census tracts in the case of the AAS database (see text).

121

URBAN SOILS AND CHILDREN'S HEALTH

TABLE 2

Metal Concentrations (m mg g71) Associated with Median Soil Pb Subgroups Partitioned at 310 mg g71 (Higher Subgroup) and 5310 mg g71 (Lower Subgroup) for 286 Census Tracts in New Orleans Element

N

Lead (Pb) Zinc (Zn) Cadmium (Cd)

72 72 72

Element

N

Lead (Pb) Zinc (Zn) Cadmium (Cd)

214 214 214

Higher metal census tracts, median soil metal (mg g71) Min 10% 25% 50% 316 1141 2.0

334 199 3.5

378 256 4.2

481 337 5.0

Lower metal census tracts. median soil metal (mg g71) Min 10% 25% 50% 10.1 26 ND

30 53 1.7

chips that impart a Pb-spike to the soil sample. Pbspikes are less common in the mid-yard or streetside samples. Note that the maximum SL quantities strongly in£uences mean SL.The mean SL content increases logarithmically while the median SL content increases arithmetically, a trend especially striking at the lowest quantities. Figure 2 supports the use of the stable central tendency characteristic of the median for describing Pb, Zn, and Cd in soil for each census tract. Figure 3 is the HMCT and LMCT map of New Orleans consisting of three layers. The in£uence of median SL at a given point decreases as distance from that point increases, (i.e., Pb is inverse distance weighted). Layer 1 consists of median SL concentra-

47 65 2.1

80 97 2.8

75%

90%

Max

680 465 6.4

811 600 7.0

1130 910 15

75%

90%

Max

141 142 3.3

211 203 4.0

306 415 6.3

tion on centroids of a census tract grid generated by inverse distance weighted interpolation for the surface. Layer 2 is the 310 mg g71 SL isoline (contour line of equal SL concentration). Layer 3 is the 1990 United States boundary map of census tracts (ArcView, 1997; Hu, 1995).The 310 mg Pb g71 isoline in Fig. 3 de¢nes the areal extent of HMCTs and LMCTs. Demographic Data of HMCTs and LMCTs Table 3 displays the 1990 demographic data of the census of Metropolitan New Orleans. The ethnic composition of Metropolitan New Orleans was 57.0% White and 40.2% Black. Other (native Americans, Asians, etc.) make up the remaining 2.8 % of the

FIG. 2. Relation between maximum, mean, and minimum descriptive measures as a function of median SL for 286 census tracts of New Orleans.

122

MIELKE ET AL.

FIG. 3. New Orleans soil lead (Pb) map. The map consists of three layers: layer 1 is a £oating point grid generated by inverse distance weighted surface interpolation of census tract centroids with their assigned median soil lead concentration; layer 2 is the 310 mg g71 soil Pb isoline; and layer 3 is the census tract boundary map.

population. This study focuses on only the White and Black population of the city. In addition,Table 3 shows total Black and White population and total population of children 6 years old and younger in the census tracts of the HMCTs and LMCTs. Because these are total population data and not sample data, they

TABLE 3

Demographic Characteristics for Black and White Residents of Higher Metal and Lower Metal 1990 Census Tracts (n=286) of Metropolitan New Orleans Demographic characteristics Total population Black population White population Total children 6 years Black children 6 years White children 6 years

Higher metal census tracts (n=72)

Lower metal census tracts (n=214)

144,491 86,607 53,813

750,551 273,566 456,418

14,233

77,982

10,723

37,751

3,080

37,996

describe the actual situation of HMCTs and LMCTs. In 1990, 144,491 people lived in HMCTs and 750,551 people lived in LMCTs. Of the total population living in HMCTs, 60% are Black and 37% are White. Of the population living in LMCTs, 36% are Black and 61% are White. The 1990 census shows that 92,215 children 6 years old and younger lived in metropolitan New Orleans. Of this total, 14,233 (15%) of the children inhabit the HMCTs and 77,982 (85%) of the children live in LMCTs. Of the 14,233 children 6 years old and younger living in the HMCTs, 10,723 (75%) are Black, 3080 (22%) are White, and the remaining (3%) are other. For the 77,982 children 6 years and younger living in LMCTs, 37,751 (48%) are Black, 37,996 (49%) are White, and the remaining (3%) are other. Selected Socioeconomic Indicators of HMCTs and LMCTs Selected census tract data for HMCTs and LMCTs are given in Table 4. Note that sample data given in Table 4 also shows that Blacks are overrepresented

123

URBAN SOILS AND CHILDREN'S HEALTH

TABLE 4

Selected Socioeconomic Characteristics for Black and White Households of Higher Metal and Lower Metal 1990 Census Tracts (n = 286) of Metropolitan New Orleans Socioeconomic characteristic Simple Majority n = number of census tracts Ann. Income ($Med. of Med.) Mo Rent ($ Med. of Med.) Pre-`40 home % %Owner occupied %Renter occupied Vacancy (%) Female head (%)a

Higher metal census tracts (n = 72)

Lower metal census tracts (n = 214)

Black n = 47

White n = 25

Black n = 73

White n = 141

11,464

21,923

13,078

27,917

366

430

367

439

51.5 28.4 71.6 22.4 16.3

70.7 38.1 61.9 17.9 6.2

16.3 45.0 55.0 16.3 21.3

9.6 61.5 38.5 8.8 5.9

a

Female-headed household with at least one child under 18.

in the HMCTs.The Fisher's exact test of the number of census tracts consisting of simple majority Black and White in HMCTs compared with LMCTs yields a P value of 4.561076, indicating an overrepresentation of Blacks in HMCTs. In the following statements, the use of Black or White refers to census tracts (communities) with simple majority of one group or the other. Annual median household income. In the case of predominantly Black census tracts, the annual median household income (1990) was below the poverty level of $13,500 in both HMCTs and LMCTs.The median annual household incomes of Whites are almost twice that of Blacks in both HMCTs and LMCTs. Note that the median annual household incomes of both groups were slightly lower in the HMCTs than in the LMCTs. In the case of Blacks, the di¡erence is about $1500 while for Whites the di¡erence is about $6000.The annual median household income of White communities is about $10,500 higher than the predominantly Black communities in HMCTs and $14,800 higher in LMCTs. Monthly median rent costs. Monthly median rent costs are only slightly lower (*$64^$72 per month) for Black compared to White households living in HMCTs and LMCTs. For Blacks, the median monthly rent is *$366 and is the same for HMCTs and LMCTs. In addition, for Whites, the median monthly rent differences between $430 and $439 for HMCTs and LMCTs are small and essentially the same. For Black census tracts in HMCTs, the median annual rent is $4392 or 38% of median annual household income. For Black census tracts in the LMCTs, the median annual rent is $4344 or 34% of median annual household income. In contrast, the median annual rent, as a per-

cent of income for White census tracts is about 10^15% lower at 24% and 19%, within the HMCTs and LMCTs, respectively. Age of housing. Age of housing is recognized as a surrogate for Pb-based paint. The peak use of Pb in paint occurred in the mid-1920s, decreased rapidly by the late 1940's, and then declined slowly until its ban for residential use in 1978. The housing in the HMCTs is older than in the LMCTs. In the HMCTs that are predominantly Black, about 52% of the homes are older than 1940 in contrast to White census tracts where 71% of homes are pre-1940. In the LMCTs, White census tracts have newer homes, with 9.6% older than 1940, compared with Black census tracts where a 16.3% of the homes are pre-1940. HousingÐownership, rental, and vacancy. In HMCTs, home ownership is lower than in LMCTs. In predominantly Black communities, the owner occupied rate is 28% in HMCTs, compared to 45% owner occupied in LMCTs. In predominantly White communities, 38% of the homes are owner occupied in HMCTs compared to 62% owner occupied in LMCTs. In HMCTs, in both predominantly Black and White census tracts, rental property makes up the majority of the households. For Black census tracts, rental is 72% of the housing in HMCT and 55% in LMCTs. For White census tracts, rental makes up 62% of the housing in HMCTs and 39% in LMCTs. Census tracts with majority Black households have a higher vacancy rate (22.4% in HMCTs vs 16.3% in LMCTs) than census tracts with majority White households (17.9% in HMCTs vs 8.8% in LMCTs). Households headed by females with at least one child under 18 years old. Census tract data show that a high percentage of households in Black census tracts are headed by females with at least one child under 18 years old. In predominately Black census tracts of HMCTs there are 16% of the households headed by females with at least one child under 18 years old compared with 21% in the LMCTs. In predominantly White census tracts, about 6% of the households meet this condition. DISCUSSION

Parent Material and Urban Soils Alluvium derived from river sediment deposited in the Bonnet Carre¨ Spillway represents the parent material for the soils of the lower Mississippi River Delta. The results in Table 1 show that there are large di¡erences between Pb, Zn, and Cd of the alluvial

124

MIELKE ET AL.

parent materials compared with the residential soils of urban New Orleans. Note that alluvial parent material for the soils of the Mississippi River Delta contain 57.5 mg Pb g71 (see Table 1). Similarly low quantities of SL have been observed in soils collected from rural or small cities of Maryland, Minnesota, and Louisiana (Mielke et al., 1983, 1984/85, 1997; Mielke, 1993). Note also, that for Pb and Zn, there are no overlaps between the maximum of the alluvium and the 10th percentile results of the urban soils. In the case of the medians of the analytical results, there are over 30 times more Pb for all urban soil samples than in the alluvium control samples obtained from the Bonnet Carre¨ Spillway. There are also 13 times more Zn and 4 times more Cd in New Orleans soils than in the alluvium control samples. Comparing Table 1 and Table 2, a factor of 130 for Pb, 34 for Zn, and 7 for Cd exists between the medians of HMCT soils and alluvial control soils. Major di¡erences of SL between rural and urban soil samples have been reported by many researchers (see Mielke and Reagan, 1998). The comparison between fresh alluvium and urban soil samples shows that the source of the metals of urban soils is not from the parent materials of the soils of New Orleans. The Association between Soil Pb and Blood Pb The review between BL and the ICP data for SL shows that there is a consistent and signi¢cant association between paired median BL and median SL. The least sum of absolute deviations regression model equation is BL = 3.06 + 0.33 (SL)0.5 (correlation between the modeled BL and observed BL is 0.69 and the P value is 3.5610722) and is illustrated in Fig. 1. Also shown in Fig. 1 are the results based on AAS data (Mielke et al., 1997). The two SL^BL models are essentially the same and overlap across most of the curve. New Orleans provides empirical perspective on the association between SL and children's BL. Of particular interest is the steep rise in BL at SL concentrations 5100 mg Pb g71. Note that the steepest increase in BL occurs at median SL below 50 mg g71 and that the curve £exes between median SL 50 and 100 mg g71 and then slowly £attens as median SL rises. This is a critical ¢nding of this empirical study and supports the concept that children have particular sensitivity to environmental Pb (Mushak et al., 1989; Brody et al., 1994; CDC, 1997).The model also suggests the possibility that SL is a sensitive and robust environmental diagnostic tool for predicting where children may have elevated BL levels. The sensitivity of children to SL is probably linked to their distinctive behavior during childhood. Chil-

dren play on £oors and on the ground and they pick up dust and soil on their ¢ngers or their toys and place their ¢ngers and toys in their mouth. The amount of Pb picked up on hands has been investigated at New Orleans private daycare centers in inner city and outer city locations (Viverette et al., 1996). Private daycare centers are located in residential housing and have play yards that often contain bare soils. After playing outdoors compared to playing indoors, inner city children had over seven times more Pb on their hands while outer city children had only a fractional increase of lead on their hands (Viverette et al., 1996). Other researchers observe that child hand-tomouth behavior and pica for soil is a major link between environmental Pb and childhood Pb exposure (Sayer et al., 1974; Angle, 1982; Thorton et al., 1994; Calabrese et al., 1997). If children played in the alluvial parent materials that form delta soils (see Table 1), they would not be susceptible to the Pb exposure from normal hand-to-mouth behavior that they exhibit while playing in the HMCTs (see Table 2) of the urban environment. Higher Metal Census Tracts and Lower Metal Census Tracts The strength of the SL^BL model justi¢es using it to set a threshold for higher and lower exposed groups of children. Concerning follow-up activities for various classes of BL concentrations, CDC states that for Class IIA (10^14 mg dL71), ``Many children (or a large proportion of children) with blood lead levels in this range should trigger community-wide childhood lead poisoning prevention activities (US DHHS 1991).'' The strength of the SL^BL model justi¢es applying it to all of the New Orleans census tracts where there is comparable SL data. Table 2 shows that HMCTs contain a median SL of 481 and LMCTs contain a median SL of 80 mg g71. The HMCT and LMCT subgroups are mapped in Fig. 3. There are 72 HMCTs and 214 LMCTs. HMCTs are clustered within the inner city of New Orleans. For the HMCTs, the predicted median BL of children exceeds 9.0 mg dL71 and ranges from 9.1 to 13.9 mg dL71, with a median of 10.4 mg dL71. The LMCTs are generally distributed outside of the inner city. In the LMCTs, the predicted median BL is less than or equal to 9.0 mg Pb dL71 and ranges from 3.7 to 9.0 mg Pb dL71, with a median of 6 mg Pb dL71. The hypothesis that Pb exposure is directly associated with the degree of urbanization is supported by other studies. The higher inner city SL and lower outer city SL geographic pattern has been found in several cities, such as Baltimore, Minneapolis, St.

URBAN SOILS AND CHILDREN'S HEALTH

Paul, and New Orleans (Mielke et al., 1983, 1984; Mielke, 1994). In a database developed by the Minnesota Department of Health and the Minnesota Pollution Control Agency, SL was shown to occur in a pattern similar to BL (Mielke et al., 1989). Furthermore, a study in Duluth, Minnesota noted that location of residence is a reliable basis for conducting a childhood Pb poisoning screening program. Central Duluth is associated with three-quarters of all the children with blood lead 10 mg dL71 and nearly nine-tenths of the smaller group of more exposed children exhibiting BL 15 mg dL71 (Bronson and Renier, 1995). The National Health and Nutrition Evaluation Surveys (NHANESII and NHANESIII) show that degree of urbanization (i.e., rural, small city compared to inner city location) is an important factor in childhood Pb exposure (Brody et al., 1994). The processes that distribute Pb and create the pattern of SL appear to be a common feature of cities (Mielke and Reagan, 1998). Soil Pb Contamination of the Urban Environment Compared to the alluvium parent materials, there has been an enormous increase of SL in the city. The amount of Pb in urban soil is related to products and processes that transfer Pb into the city. Two major consumer products previously contained lead, Pbbased paint and Pb additives to gasoline. Both products are important contributors to the SL contamination problem, and, in addition to other sources, they are re£ected by the soils of the city. During United States production of these products, about the same amount of Pb (*6 million metric tons) were used to manufacture Pb-based paint as were used to manufacture additives to gasoline (Mielke, 1993). The two major sources exhibit di¡erent characteristics as to where they accumulate in the soil. Pb-based paint accounts for the locally intensive SL quantities of some samples as a result of chips of Pb paint that spike drip-line soils and less frequently other residential soil locations. In a given census tract, paint chips generally a¡ect the maximum and mean amounts of SL. In this survey, 12 of 15 or 80% of the soil samples in each census tract were collected within a meter of the street and away from the drip-line. Consequently, Pb-based paint chips usually associated with the narrow zone around houses do not usually in£uence the median SL of each census tract (see Fig. 2). In some parts of New Orleans, where the setback distances are small and houses are close together, lead-based paint is probably an important contaminant of the sample. The in£uence of lead-based paint as a contaminant of soil is brought into perspective when soil

125

collections from old communities of small towns with similar setbacks are compared with communities of large cities. Large cities exhibit signi¢cantly more SL than small towns (Mielke et al., 1984/85, 1989, 1997; Mielke, 1993). In addition to Pb-based paint applied on homes in the ¢rst half of the century, about 75,000 metric tons of Pb was consumed as gasoline additives in Louisiana since 1950 (Mielke, 1993). The massive quantity of Pb emitted by automobiles into the most tra¤c-congested parts of the city accounts for the geographically extensive area of Pb dust accumulation in urban soils. In the peak period of use of Pb-gasoline (late 1960's and early 1970's), an average (n = 8) of about 5.2 metric tons of Pb were emitted annually within a 1.5 km radius of major intersections of New Orleans (Mielke et al., 1997). At each busy intersection, the tonnage of Pb emitted by automobiles annually was equivalent to the annual stack emissions of a major Pb smelter.The consumer use of Pb-gasoline is a plausible explanation for the direct relationship between amount of SL in rural, small town, small city, large city, and inner city locations (i.e., degree of urbanization) (Mielke and Reagan, 1998). Demographic and Socioeconomic Indicators of HMCTs and LMCTs Race has been reported as a major factor in Pb exposure of children (Lanphear et al., 1996; Sargent et al., 1995; Weintraub, 1997). As the demographic data of Table 3 shows, the population of HMCTs is overrepresented by Blacks of all ages and especially children 6 years old and younger. For the total population of HMCTs, 60% of the people are Black. In the case of children 6 years and younger living in the HMCTs, 75% are Black. In New Orleans, race is a factor in children's Pb exposure because Black children are overrepresented in communities of the city that contain the highest amount of Pb in the environment. If White children live in HMCTs, they too belong to the highrisk group for Pb exposure. In a high SL community of Saint Paul, Minnesota, the highest BL exposures within a poor and mixed race group were exhibited by White, not Black children (Mielke et al., 1992). Race is not the cause of exposure. Lead in the environment is the cause of exposure. Unfortunately, in New Orleans, and apparently other cities, Black children are extremely overrepresented in excessively Pb-contaminated communities. SL appears to be an excellent diagnostic tool for identifying these communities. Evaluation of socioeconomic indicators of populations living in HMCTs and LMCTs assists with characterizing conditions that may impinge on children's

126

MIELKE ET AL.

exposure to Pb. The results of Table 4 show the di¡erences of income between Blacks and Whites in both HMCTs and LMCTs. Income is important because it relates to resources available for directly in£uencing the quality of life. Wealthier households can a¡ord to maintain their interior and exterior living areas, bring in new soil, landscape their properties with carpets of grass, construct a special Pb-safe play area for children, and even move to another community. Poorer households do not have economic opportunities to make the same choices. Poverty is recognized as a factor that is related to children's Pb exposure (Brody et al., 1994; CDC, 1997). Presumably, people living in owner-occupied homes have more direct control over the condition of their housing and the quality of play areas than do people living in rental housing. Rental housing is problematic because of the lack of incentive for occupants to invest time, money, and energy for changes that may directly protect children from Pb exposure. Vacancy is an indicator of the desirability of a community.Vacant property does not earn income, and there is a tendency for the absentee owners or landlords to let their unpro¢table real estate fall into disrepair. Disrepair of one property can result in depreciation of neighboring properties. For children, poorly maintained and deteriorating properties, especially within older residential homes, are known factors for increased Pb exposure (Mushak, 1989; CDC, 1997). Rental costs further re£ect on the socioeconomic disparity between Blacks and Whites of New Orleans. As shown in the results of Table 4, the monthly rental costs are similar for Blacks and Whites living in HMCTs and LMCTs. For Black communities the median annual rent is 38% of median annual household income in HMCTs and 34% of median household annual income in LMCTs. For Whites, median annual rental costs range from 19 to 24% of their median annual household income. Because of their potential in£uence on management of the environment of residential properties, the median income and disproportionately high rental costs re£ect on the economic di¤culty of Black communities compared to White communities in New Orleans. Age of housing is the primary factor used for targeting screening of young children for lead poisoning (CDC, 1997). The major argument used to focus on age of housing is that it is a surrogate measure for Pbbased paint. As stated above, the use of Pb-based paint peaked in the 1920's and declined until it was banned in 1978. Previous research conducted in New Orleans showed that newer housing (i.e., post-1940 housing) is an excellent predictor for low BL. Conversely, older housing (i.e. pre-1940 housing) was a poor

predictor of high BL (Mielke et al., 1997).The di¤culty of using age of housing is that while it is related to Pbbased paint, age of housing is not an indicator of other sources of Pb (i.e., incinerators, waste disposal, Pb-gasoline, etc.) that are important sources of exposure. Soils are an integrator of all sources of Pb and other metals that have been discharged as dust and deposited (including Pb-paint) in the environment. Furthermore, as this study shows, soils contribute diagnostic data that are associated with children's Pb exposure. Health Implications of Zn and Cd in Urban Soil Table 2 shows results of Pb, Zn, and Cd for the HMCTs and LMCTs. In the HMCTs, 50% of the soil samples are above 481 mg g71 or a ratio of six times more Pb than LMCTs. From Table 2, the HMCT: LMCT ratios are 3.5 for Zn, and 1.8 for Cd. Other research has shown that there is a strong linear association between Pb, Zn, and Cd, with a correlation coe¤cient that is 0.65 or higher and P values that are 510715 for the New Orleans census tracts (Mielke and Smith, 1997). A critical ¢nding is that a map of either Zn or Cd generally conforms to the map of SL (Mielke and Smith, 1997). Both Zn and Cd are toxic to plants, and for this reason, the combination of soil metals may increase the potential for childhood exposure. A SL content of 481 mg g71 is above the current 400 mg g71 EPA guideline of concern for residential sites where children have access to bare soil (US EPA, 1996). By itself, Pb is not strongly phytotoxic (Adriano, 1986). In the case of Zn, phytotoxicity is reported in the range of 500 to 700 mg g71, and for Cd, phytotoxicity is reported at 2^ 9 mg g71 (Adriano, 1986; Rump and Krist, 1992). As shown in Table 2, soils that contain a median SL of 800 up to 1100 mg g71 are associated with Zn in the range of 600 to 900 mg g71 and Cd in the range of 7 to 15 mg g71. The major problem is that Zn and Cd may reduce the amount of plant growth and ground cover where soil Pb concentrations are highest in the city. Given the empirical association between BL and SL, it is likely that children are also being exposed to a chemical mixture that includes Zn and Cd. Some clinicians suggest that excessive exposure to industrial metals is associated to varying degrees of central nervous system disorders ranging from learning disorders, mild forms of dementia, and the severest form of mental deterioration, Alzheimer's disease (Casdorph and Walker, 1995). A Safe Amount of Pb in Soil Determining the amount of Pb in soil that is safe for children is a critical practical concern. Table 2

URBAN SOILS AND CHILDREN'S HEALTH

provides an empirical indication of the quantity of Pb that is likely to be safe for most children. From Fig. 1, assigning a safe level is complicated by the fact that at 5100 mg g71 content of SL there is a steep rise in BL. According to this information, in order to address the especial sensitivity of children, a safe amount of SL should be 5100 mg g71. In the LMCTs, the median Pb content of soil is 80 mg g71. According to the model (see Fig. 1) this quantity relates to a median BL level of 6.0 mg dL71. The concentration of 80 mg Pb g71 in soil apparently provides a margin of safety for most children. But even at this amount of exposure, some children may be at such high risk due to their occasionally high pica behavior, that 80 mg Pb g71 is also too contaminated (Calabrese, 1997). However, using 80 mg Pb g71 soil content as a safe upper limit for children is in agreement with the health-based soil standard for residential soils in Sweden (Jones et al., 1998). CONCLUSIONS

This study reviews the health response of children to environmental contamination and concludes that a strong association exists between SL and BL of children 6 years old and younger. The results predict that median SL collected from residential census tracts is directly associated with median BL of children in the same census tracts. BL 10 mg dL71 is associated with reduced learning disabilities, behavior disorders, and long-term health problems.This study predicts that in over 70 high metal census tracts (HMCTs) of New Orleans, children face a high risk (50%) of having exposures of 10 mg dL71. The medical, learning, and social costs of Pb poisoning have broad and long-term implications. Fortunately, soils, collected systematically on a community by community basis, show strong potential as a diagnostic tool for ¢nding unfavorable environmental conditions. Soil integrates all sources of contamination of the environment. The study showed that soil Pb, Zn, and Cd contamination is clustered in census tracts in the inner city of New Orleans. The combination of Zn and Cd in association with Pb in New Orleans soils may intensify the potential for metal exposure because their phytotoxic properties can produce bare soil conditions. Demographic data shows that in HMCTs the population ratio of Blacks to Whites is 60:37. For children, the ratio of Blacks to Whites is 75:22. In the Black HMCTs, households have a median (1990) income of $11,464. About 72% of the Black HMCT households are rental property, and about 52% of the homes are older than 1940.Vacancy rates of HMCTs that are predominantly Black exceed 20%. There are many

127

characteristics that present unfavorable conditions for Black households inhabiting HMCTs. Policies and actions to improve conditions and reduce the Pb exposure must be implemented to address children's health concerns. The challenge is to protect children from Pb exposure during their ordinary play habits and hand-to-mouth behavior. Traditional Pb prevention e¡orts have focused on the control of Pb-based paint in old housing. This study suggests that a primary Pb prevention program would be complimented by taking into account Pb-dust from all sources that have accumulated in urban communities. It predicts that substantially reducing SL in HMCTs would signi¢cantly reduce the BL of children.There are numerous possibilities for reducing children's contact with contaminated soil. These include paving, constructing arti¢cial surfaces (decking to outdoor carpeting), and building sandboxes that are maintained with clean sand. Perhaps most promising in New Orleans is the possibility of transporting and distributing uncontaminated soil (such as identi¢ed at the Bonnet Carre¨ Spillway) to HMCTs of New Orleans and planting a vigorous ground cover (usually grass) on the clean soil. With leadership, careful design, and planning, an informed citizenry could be mobilized to undertake this action. These actions are relatively nontechnical and require logistic support for delivery of materials, clean soil, and a supply of seeds. Labor for soil spreading and planting could be arranged as an urban improvement project or on a volunteer basis, and such activity could be done at reasonable public expense. To protect the future urban environment, the accumulation of additional quantities of metals and other toxins must be averted. Public decision makers should be encouraged to create long-term plans for providing safer urban environments that nurture the growth and well-being of all young children. Soils as an integrator of metals provide a diagnostic tool about the quality of the environment, and corrective actions directed at contaminated soils should complement primary Pb exposure prevention. ACKNOWLEDGMENTS The content of this study is the responsibility of the authors. Special thanks are given to Xavier students Ayanna Buckner, Nitasha Burney, Pamela Daniels, Daphne Richardson, Ti¡any Pennick, and Gregory Washington for laboratory support. This study has bene¢ted from many reviewers including, Dianne Dugas and Kabrina Smith, of the LA O¤ce of Public Health, Environmental Epidemiology and Toxicology, and Georgia Bryant and Dr. Terry Fontham. Sarah Rountree provided editorial assistance. This research was supported by the Environmental Health and Toxicology Program sponsored by the MHPF/ATSDR cooperative agreement U50/ ATU398948 to Xavier University.

128

MIELKE ET AL.

REFERENCES Adriano, D. C. (1986). ``Trace Elements in the Terrestrial Environment,'' (Cd) p. 127, (Pb) p. 237, (Zn) p. 441. Springer-Verlag, New York. Angle, C. R., and McIntire, M. S. (1982). Children, the barometer of environmental lead. Adv. Pediatr. 29, 2^31. ArcView 3.0a. (1997). The software for ¢gure 3 was ArcView 3.0 with the Spatial Analyst, 3D Analyst (ESRI, Redlands, CA, USA), Xtools (Oregon Department of Forestry, OR, USA), and Spatial Tools extensions (USGS). Bolger, P. M., Carrington, C. D., Capar, S. G., and Adams, M. A. (1991). Reductions in dietary lead exposure in the United States. Chem. Spec. Bioavail. 3, 31^36. Brody, D. J., Pirkle, J. L., Kramer, R. A., Flegal, K. M., Matte, T. D., Gunter, E. W., and Paschal, D. C. (1994). Blood lead levels in the U.S. Population: Phase 1 of the Third National Health and Nutrition Examination Survey (NHANES III, 1988 to 1991). J. Am. Med. Assoc. 272, 277^283. Bronson, M. A., and Renier, C. M. (1995).The location of residence as a basis for childhood lead poisoning screening programs. Am. J. Public Health 85, 589^590. CDC. (1997). ``Screening Young Children for Lead Poisoning: Guidance for State and Local Public Health O¤cials.'' Centers for Disease Control and Prevention, Atlanta, GA. Calabrese, E. J., Stanek, E. J., James, R. C., and Roberts, S. M. (1997). Soil ingestion: A concern for acute toxicity in children. Environ. Health Perspect. 105, 1354^1358. Casdorph, H. R., and Walker, M. (1995) ``Toxic Metal Syndrome.'' Avery, New York. Gray,W. M., Landsey, C.W., Mielke, P.W., and Berry, K. J. (1992). Predicting Atlantic seasonal hurricane activity 6^11 months in advance.Weather Forecasting 7, 440^455. Hu, Jun. (1995). Methods of Generating Surfaces in Environmental GIS Applications, International User Conference May 22^26. ESRI (Environmental Systems Research Institute). 7 pp. Jones, C., Elert, M., and Christiansson, A. (1998). FÎrdjupad miljÎoch hÌlsoriskbedÎmning avsende Falu TÌtort. Kemakta Konsult AB, Falu kommun. 51 pp. Lanphear, B. P., Weitzman, M., and Eberly, S. (1996). Racial di¡erences in urban children's environmental exposures to lead. Am. J. Public Health 86, 1460^1463. Lanphear, B. P., Weitzman, M., Winter, N. L., Eberly, S., Yakir, B., Tanner, M., Emond, M., and Matte, T. D. (1996). Lead-contaminated house dust and urban children's blood lead levels. Am. J. Public Health 86, 1416^1421. Mielke, H.W. (1984).``Leaded Gasoline as a Factor in the Urban Lead Problem. Testimony before the Committee on Environment and PublicWorks on S.2609, (June 22, 1984).'' US Senate, Ninety-Eighth Congress (second session),Washington DC. Mielke, H. W. (1991). Lead in residential soils: Background and preliminary results of New Orleans. Int. J.Water Air Soil Poll. 57^58, 111^119. Mielke, H. W. (1993). Lead dust contaminated U.S.A. Communities: comparison of Louisiana and Minnesota. Appl. Geochem. Suppl. Issue No 2, 257^261. Mielke, H. W. (1994). Lead in New Orleans soils: New images of an urban environment. Environ. Geochem. Health 16, 123^128. Mielke, H.W., and Adams J. (1989).``Environmental Lead Risk in the Twin Cities.'' 22 p. Center for Urban and Regional A¡airs, H H Humphrey Center, University of Minnesota, Minneapolis, Mn.

Mielke, H. W., Adams, J. L., Reagan, P. L., and Mielke, P. W. (1989). Soil-dust lead and childhood lead exposure as a function of city size and community tra¤c £ow: The case for lead contaminated soil abatement in Minnesota. In ``Lead in Soil: Issues and Guidelines'' (B. E. Davies and B. G. Wixson, Eds.), pp. 253^271. Science Reviews Limited, Northwood, UK. Mielke, H. W., Adams, J. E., Hu¡, B., Pepersack, J., Reagan, P. L., Stoppel, D., and Mielke, P. W., Jr. (1992). Dust control as a means of reducing inner-city childhood Pb exposure. Trace Substances Environ. Health 25, 121^128. Mielke, H.W., Anderson, J. C., Berry, K. J., Mielke, P.W. Jr, and Chaney, R. L. (1983). Lead concentrations in inner city soils as a factor in the child lead problem. Am. J. Public Health 73, 1366^1369. Mielke, H.W., Blake, B., Burroughs, S., and Hassinger, N. (1984). Urban lead levels in Minneapolis: The case of the Hmong children. Environ. Res. 34, 64^76. Mielke, H. W., Burroughs, S., Wade, R., Yarrow, T., and Mielke, P. W. Jr. (1984/85). Urban lead in Minnesota: Soil transect results of four cities. Minn. Acad. of Sci. 50, 19^24. Mielke, H.W., Dugas, D., Mielke, P.W., Smith, K. S., Smith, S. L., and Gonzales, C. R. (1997). Associations between soil lead concentrations and childhood blood lead in urban New Orleans and rural Lafourche Parish, Louisiana. Environ. Health Perspect. 105, 950^954. Mielke, H.W., and Smith, M. K. (1997). Soil Pb, Zn, and Cd in Metropolitan New OrleansöTheir geochemical characterization and association. 4th International Symposium on Environmental Geochemistry. US Geological Survey Open File Report No. OF97-496. P 64. October 5^10,Vail, CO. Mielke, H.W., and Reagan, P. L. (1998). Soil is an important pathway of human lead exposure. Environ Health Perspect. 106, 217^229. Mushak, P., Davis, J. M., Crocetti, A. F., and Grant, L. D. (1989). Prenatal and postnatal e¡ects of low-level lead exposure: Integrated summary of a report to the U.S. Congress on childhood lead poisoning. Environ. Res. 50, 11^36. Pirkle, J. L., Brody, D. J., Gunter, E.W., Kramer, R. A., Paschal, D. C., Flegal, K. M., and Matte, T. D. (1994). The decline in blood lead levels in the United States: The National Health and Nutrition Examination Surveys (NHANES). J. Am. Med. Assoc. 272, 284^291. Rosner, D., and Markowitz, G.A. (1985). `Gift of God'?: The public health controversy over leaded gasoline during the 1920s. Am. J. Pub. Health 75, 344±352. Rump, H. H., and Krist, H. (1992).``Laboratory Manual for the Examination of Water, Waste Water and Soil,'' 2nd Edition, p. 45. VCH, Weinheim, Germany. Sargent, J. D., Brown, M. J., Freeman, J. L., Bailey, A., Goodman, D., and Greeman, D. H., Jr. (1995). Childhood lead poisoning in Massachusetts Communities: Its association with sociodemographic and housing characteristics. Am. J. Public Health 85, 528^534. Sayer, J. W., Charney, E.,Vostal, J., and Pless, I. B. (1974). House and hand dust as a potential source of childhood lead exposure. Am. J. Diseases Children 127, 167^170. Thorton, I.,Watt, J. M., Davies, D. J. A., Hunt, A., Cotter-Howells, J., and Johnson, D. L. (1994). Lead contamination of UK dusts and soils and implications for childhood exposure: An overview of the work of the environmental geochemistry research group, Imperial College, London, England 1981-1992. Environ. Geochem. Health 16, 113^122.

URBAN SOILS AND CHILDREN'S HEALTH US Census Bureau. (1990).Tracts and Block Numbering Areas, New Orleans, LA MSA,Table 32 Selected Structural Characteristics of Housing Units: and Summary Tape File 3A, Louisiana 040, Lafourche Parish 050, Census tracts 140.

129

Environmental Assessment O¤ce of Research and Development, US Environmental Protection Agency, Research Triangle Park, NC.

US DHHS. 1991. Preventing Lead Poisoning in Young Children. A statement by the Centers for Disease Control p 3. US Department of Health and Human Services. Public Health Service. CDC, Atlanta, GA.

Viverette, L., Mielke, H. W., Brisco, M., Dixon, A., Schaefer, J., and Pierre, K. (1996). Environmental health in minority and other underserved populations: Benign methods for identifying lead hazards at day care centres of New Orleans. Environ. Geochem. Health 18, 41±45.

US EPA. (1996). Urban soil lead abatement demonstration project. Volume 1: EPA Integrated Report. National Center for

Weintraub, M. (1997). Racism and lead poisoning. Am. J. Public Health 87, 1871.