United States Environmental Protection Agency Region 5 77 W. Jackson Blvd. Chicago, II 60604-3507 Illinois, Indiana Michigan, Ohio , Wisccnsn Environmental Sciences 7 November 1992 905R92002 &EPA Project LEAP— Phase 1 Spatial And Numerical Dimensions of Young Minority Children Exposed to Low-Level Environmental Sources of Lead ------- SPATIAL AND NUMERICAL DIMENSIONS OF YOUNG MINORITY C}IILDREN EXPOSED TO LOW-LEVEL ENVIRONMENTAL SOURCES OF LEAD BY WILLIAM H. SANDERS III Director, Environmental Sciences Division U.S.E.P.A. Region 5 United States Environmental Protection Agency Region 5, Chicago Environmental Sciences Division Project LEAP (Lead Education and Abatement Program) 77 West Jackson Blvd, Chicago, Illinois 60604-3507 ------- ABSTRACT SPATIAL AND NUMERICAL DIMENSIONS OF YOUNG MINORITY CHILDREN EXPOSED TO LOW-LEVEL ENVIRONMENTAL SOURCES OF LEAD BY William H. Sanders III, Director Environmental Sciences Division Region 5 United States Environmental Protection Agency Chicago, Illinois A population comparative risk algorithm was developed as a pilot study for the Agency’s Lead Strategy, as a Region 5 Comparative Risk initiative, and as an environmental equity project. All known environmental sources of lead in 83 cities in the Midwest were assessed to develop a population comparative risk analysis for childhood exposure to leatL A secondary objective was to discern the association of proximity of transportation corridors, to elevations in blood-lead levels. The selected at risk population were African-American and Hispanic children under seven years of age. Measured and postulated values were derived to approximate lead concentrations in air, drinking water, soi1 and dust. Sources included in the analysis were major point sources of lead and lead compounds (from the Toxic Release Inventory national data base), ambient air concentrations, reported drinking water concentrations, municipal waste combusters, abandoned hazardous waste sites, and operating hazardous waste facilities. Using concentrations specific to census tracts within each city, the EPA Uptake Biolcinetic Model was used to estimate the pmbabiliiy distribution of blood-lead levels for each area, and to estimate the percent of children expected to exceed a criterion value of 10 p.g/dL blood-leaS Although considered to be conservative, the analysis concluded that in 1988 a total childhood population of 154,000 Midwest children were expected to have blood-lead levels exceeding 10 g/dL, including 55,000 Afri can-American and 12,000 Hispanic children. No association was found for proximity of transportation corridors to elevated lead-blood levels, for the Minneapolis/St. Pau Minnesota, study area. This report constitutes Phase 1 of a three phase projecL The purpose of this phase is to screen a large number of cities for future lead reduction efforts through the use of a comparative risk analysis. Phase 2 will include testing in a small number of communities, as well as public education and outreach. Phase 3 will be remediation of environmental sources of lead in one or more communities. 1 ------- ACKNOWLEDGMENTS I am deeply grateful to the U.S. Environmental Protection Agency, particularly the Region 5 Regional Administrator Vakias V. Adamkus and Deputy Regional Administrator Ralph Bauer, who provided tremendous support in this endeavor, culminating in a sabbatical that provided the time to complete a complex study. l’his study relies heavily upon previously gathered data, by design, and I want to thank those too numerous to mention that provided assistance and data for the analysis. Three individuals and organizations are, however, particularly noteworthy: Mr. Douglas M. Benson, Coordinator, Lead Program, Division of Environmental Health, Minnesota Department of Health, for graciously providing the blood- lead database for the Minneapolis/St. Paul Blood-lead Survey; the Minnesota Pollution Control Agency for providing the counterpart soil-lead data for the Twin Cities; and the U.S. Department of Housing and Urban Development, for providing the raw database used for the National Housing Survey. Finally, my deepest appreciation to a husband and wife team in the U.S. Environmental Protection Agency, Environmental Sciences Division, Mr. Larry Lehrman and Mrs. Loretta Lehrman. The computer support, both hardware and software, and technical/programming assistance proved invaluable for efficacy in conducting the study. But, as important, their personal support and encouragement proved indispensable to the spirit, as I struggled through to completion. U ------- TABLE OF CONTENTS CHAPTER PAGE EXECUTIVE SUMMARY 1. INTRODUCI’ION 1 2. UTERATURE REVIEW 4 2.1 Toxicological Profile 4 2.1.1 Internal Exposure 4 2.1.2 Encephalopathy/Lethality 7 2.1.3 Neurological Impairment 8 2.1.4 Developmental Toxicity 9 2.1.5 Aggregated Studies Analysis 14 2.1.6 Growth 14 2.1.7 Toxicological Summary 14 2.2 Adequacy of Studies 16 23 Biological Monitoring Techniques 17 2.4 Typically Encountered Environmental Levels 18 2.5 At-Risk Population 2.5.1 Spatial/Numerical Estimates of At Risk Population 24 2.5.2 Lead Screening Programs 2.6 At Risk Population Estimates by Sources/Routes of Exposure 29 2.6.1 Lead-Based Paint 29 2.6.2 Leaded Gasoline 34 2.6.3 Stationary Sources 35 2.6.4 Dust and Soils 36 2.6.5 Drinking Water 41 2.6.6 Lead in Food 43 2.7 Special Concern For Exposure of the Fetus 43 2.8 Special Emphasis: Ethnicity 2.9 Research Needs 47 3. STUDY OBJECI1VES 49 4. METHODOLOGY 50 4.1 Study Scope and Methodology Overview 50 4.2 Study Area 52 43 Contribution to Childhood Lead Levels Fmm Air Emissions 57 43.1. Industrial Source Complex Long Term Model 57 4.3.2. ISCLT Sensitivity Analysis 59 4.3.3. Ambient Air Data 4.3.4. Air Emissions 63 4.33. Municipal Waste Combusters 4.4 Drinking Water Data 64 4.5 Soil and Dust Contributions to Elevated Blood-lead Levels 64 4.5.1. RCRA and Operating Landfills 64 U’ ------- TABLE OF CONTENTS (CONTINUED) 4.5.2. Abandoned Hazardous Waste Site Data .65 4.5.3. Derivation of Soil and Dust Values 65 4.6 Lead Uptake Biokinetic Model 67 4.6.1. UBK Sensitivity Analysis 72 4.7 Selected Area for Verification of Lead Screening Approach: MinneapolisISLP.aul 81 4.7.1 Minneapolis/St. Paul Demographic, Biological, and Soils Data 81 4.7.2 Minneapolis/St. Paul Statistical Analyses 82 4.8 Derivation of City ExceedanCe Estimates 83 5. RESULTS 85 5.1 Overview/Introduction to Results 85 5.2 Environmental Data Categorical Assessments 85 5.2.1 Ambient Air 85 5.2.2 Air Emissions 85 5.2.2.1. ISCLT Modeling Results 5.2.3 Municipal Waste Combusters 91 5.2.4 Drinking Water 94 5.2.5 RCRA and Operating Landfills 95 5.2.6 Abandoned Hazardous Waste Sites (Superfund) 99 5.2.7 Environmental Data Qualitative Summary 102 5.4 Chosen Cities 107 5.4.1 Minneapolis/St. Paul Environmental Sources of Lead 107 5.4.2 Blood-Lead DatalDefllOgraphics 109 5.4.3 Minneapolis/St. Paul Correlation Analysis 112 5.4.4 Minneapolis/St. Paul Regression Analysis 117 5.5 UBK City Results 121 6. DISCUSSION 131 6.1 Demographics 131 6.2 Environmental Data 133 6.3 Correlation Analysis 134 6.4 Regression Analysis 135 6.5 City Estimates of ExceedanCe 136 6.6 Uncertainties 137 7. CONCLUSIONS 140 8. RECOMMENDATIONS 143 iv ------- TABLE OF CONTENTS (CONTINUED) CITED LITERATURE. 145 BIBLIOGRAPHY 153 COMPANION REPORTS: •PROJECr LEAP—PHASE I APPENDIX Appendix A Air Monitoring Summary for 1988 Appendix B Soil and Dust Values Based Upon DHUD National Housing Survey Data Appendix C Values for Uptake Biokinezic Model Air Concentrations Appendix D 1988 Toxic Release Inventory Environmental Emissions in the Midwest Appendix E 1988 Toxic Release Inventory Environmental Emissions in MSA Cities Appendix F 1988 TRI Sources in the Midwest MSA Areas Exceeding 4000 Pounds/year Appendix G Industrial Source Complex Model Results Appendix H Municipal Waste Combuster Inventory in the Midwest Appendix I Drinking Water Supply Data Summary for 1988 MSA Area Cities Appendix J Resource Conservation and Recovery Act Facilities in Midwest MSA Cities Appendix K Toxic Release Inventory Reported On-Site Disposal in Midwest MSA Cities Appendix L Final and Proposed Sites in the Midwest with Lead, November 1989 Appendix M Final and Proposed Sites in the Midwest with Lead, November 1989 in MSA Cities Appendix N Minneapolis/St. Paul Soil Lead Concentrations by Census Tract Appendix 0 Common Log of Blood-lead Values by Ethnicity Appendix P Common Log of Blood-lead Levels by Census Tract Appendix 0 Regression Results for Full Model Appendix R Regression Results for Final Model Appendix S Regression Results for Revised Final Model Appendix T UBK Community Area Exceedance Results •PROJECT LEAP— PHASE 1 GIS APPENDIX .PROJECr LEAP— PHASE 1 SUMMARY REPORT •FROYECTO “LW” RESUMER DEL INFORME V ------- LIST OF TABLES TABLE PAGE I HEALTH EFFECTS SUMMARY 15 II BLOOD-LEAD SCREENING PROGRAM RESULTS FOR CHILDREN IN 16 MIDWEST CITIES IN 1981 III CHILDREN UNDER 7 YEARS OF AGE IN THE MIDWEST RESIDING IN UNSOUND LEAD-PAINTED HOUSING 32 IV METROPOLITAN STATISTICAL AREA CENTRAL CITY DEMOGRAPHICS ....53 V PARTICLE SIZE DISTRIBUTION INPUT TO INDUSTRIAL SOURCE COMPLEX MODEL 58 VI UPPER AIR DATA ANALYSIS 61 VII INDUSTRIAL SOURCE COMPLEX LONG TERM MODEL RUN COMPARATIVE ANALYSIS 61 VIII SOIL AND DUST CONCENTRATIONS FOR PB BASED UPON DHUD NATIONAL HOUSING SURVEY DATA 66 IX UPTAKE BIOKINETIC MODEL DEFAULT VALUES 69 X MODEL DEFAULT BLOOD-LEAD AND LEAD UPTAKE 70 XI UPTAKE BIOKINETIC MODEL SENSITivITY ANALYSIS 75 XII SOURCES WITH TRI REPORTED TOTAL AIR EMISSIONS EXCEEDING 4,000 POUNDS/YEAR IN 1988 87 XIII MAXIMUM CONCENTRATIONS OF LEAD FOR MODELED SOURCES 90 XIV MUNICIPAL WASTE COMBUSTER INVENTORY FOR METROPOLITAN STATISTICAL AREA CITIES IN REGION.5 93 XV TOXIC RELEASE INVENTORY REPORTED ON-SITE DISPOSAL IN 1988 FOR MSA CITIES 98 XVI NATIONAL PRIORITY LIST FACILITIES IN METROPOLITAN STATISTICAL AREA CITIES WITH LEAD AS OF NOVEMBER 1989 101 XVII QUALITATIVE SUMMARY OF ENVIRONMENTAL EXPOSURES FOR MSA CITIES IN 1988 103 XVIII CHILDREN UNDER 6 YEARS OF AGE WITH BLOOD-LEAD LEVELS EXCEEDING 10 tGIDL BASED UPON 1986-1987 BLOOD-LEAD SURVEY FOR MINNEAPOLIS AND ST. PAUL 110 XIX BLOOD-LEAD VALUES ( LGIDL) BY ETHNICITY FOR MINNEAPOLIS AND ST. PAUL, MINNESOTA BLOOD-LEAD SURVEY IN 1986-1987 111 xx CORRELATION ANALYSIS OF MINNEAPOLIS/ST. PAUL 113 XXI SELECTED CENSUS TRACF DATA FROM MINNEAPOLIS/ST..L4UL 115 XXII CORRELATION ANALYSIS— CENSUS TRACF LEVEL 116 XXIII SUMMARY OF STEP WISE PROCEDURE FOR DEPENDENT VARIABLE ACTUAL BLOOD-LEAD CONCENTRATION FOR MINNEAPOUSISE..PAIJL...119 xx iv NUMBERS OF CHILDREN UNDER 7 YEARS OF AGE IN THE MIDWEST EXPECTED TO EXCEED 10 iG/DL BLOOD-LEAD LEVEL IN 1988 122 xxv TOp NKEI) CITIES BY PERCENTILE OF CHILDREN EXCEEDING 10 &G/DL PB-B xxvi Top p &N}(E1) CITIES BY NUMBER OF CHILDREN EXCEEDING 10 1 &G/DL PB-B 129 vi ------- UST OF FIGURES FIGURE PAGE 1 Industrial Source Complex Long Term Point Source Grid 60 2 Uptake Biokinetic Model Default Concentration Curve 71 3 Select Drinking Water Concentrations 77 4 Select Soil and Dust Concentrations 78 5 Select Ambient Air Concentrations 79 6 Uptake Biokinetic Model Default Concentrations with Geometric Standard Deviation of 1.8 80 7 Total Air Emissions 1988 Toxic Release Inventory 86 8 Major Air Emission Facilities 88 9 Municipal Waste Combusters In U.S. EPA Region 5 94 10 Major On-Site Disposal Facilities 97 11 Superfund NPL Sites 100 12 Scatter Plot of Modeled Blood-lead Vaules Vs. Actual Blood-lead Values 117 v i i ------- EXECUTIVE SUMMARY This research considers the known environmental sources of lead in 83 cities in the Midwest, estimates the probability distribution of lead in African-American and Hispanic children (as well as the total childhood population) under seven years of age in each of the cities, and compares the numbers of children at risk. The approach thus developed is a population comparative risk screening methodology for ranking geographic areas as to potential lead toxicity. This data analysis report is the first phase of a three phase effort. Phase 2 will be to conduct sampling in a small number of communities, as well as to begin public outreach and education on the dangers of environmental exposure to lead. Phase 3 will be to conduct remediation of environmental sources of lead (e.g., soil and dust) in one or two communities. The objective of Phase 1 is to estimate relative blood-lead levels in childhood populations and to compare geographic areas to ascertain the severity. For each metropolitan statistical central city area, environmental data were obtained for the major sources of exposure. This included stationary source air facilities, municipal waste combusters, ambient air quality measurements, drinking water supplies, and operating as well as abandoned hazardous waste sites. Where available, actual concentrations were used. Default values were established for each environmental medium where actual measurements had not been taken. Major air emission sources were modeled to calculate associated air concentrations. The results were used in a qualitative assessment of environmental exposure. Demographic data were obtained from a geographic information systems application (provided by the Geographic Information Systems Management Office, Region 5, U.S. EPA). That office provided data at the census tract and community area (aggregation of census tracts) levels for each city. In general, a census tract has a population of about 4,000 people. Environmental data (air, drinking waler, soil and dust Prqfrct LEAP— Pha.. 1 viii ------- concentrations) associated with each tract were obtained in order to calculate blood-lead level distributions in affected populations. Based upon these environmental concentrations for each census tract/community area, the Uptake Biokinetic Model (described in Section 4.7) was run to calculate an expected percent exceedance for each area. The percentage, applied against the population data for the tract, provided an estimate of the number of children under seven yeais of age at risk of lead exposure. Further aggregations allowed for a city total, as well as a numerical ranking of cities. Data from a single geographical area, Minneapolis/St. Paul,Minnesota, was selected to test the methodology. That area had available measured blood-lead levels, along with pertinent demographic information. Two statistical procedures were performed. A simple correlation analysis was conducted to ascertain whether modeled blood-lead levels, based primarily upon the environmental data for the area, were associated with actual measured blood-lead levels. An association would indicate the viability of the approach in comparing cities. The correlation analysis indicates a correlation coefficient of 0.3. It is only statistically significant, however, at the 0.10 level. The second statistical procedure was conducted so further analyze the contribution of environmental pathways of exposure to elevations in blood-lead levels and, in particular, to ascertain whether mobile sources (i.e., proximity to a major transportation corridor) could account for a portion of the elevation in blood-lead levels. No association was found. An analysis of environmental data indicates that a tremendous quantity of lead is still being released into the environment, and that quite typically a small (relative) number of sources contribute most of the contaminant. For the six Midwest states, industry released nearly 450,000 pounds of lead and lead compounds into the air in 1988. Seventeen sources out of nearly 350 reporting facilities accounted for almost one-half of the total emissions. Nevertheless, air quality, based upon measurements of the ambient air, was excellent, with few exceedances of the primary air quality standard for lead. Point sources of Prqj.ct LEAP— Phi.. 1 ix ------- emissions, although many in number, generally do not cause concerns (a measurable increase in the ambient air-lead concentration). The notable exceptions are a few high emitting industries. For those industries, the increased ambient air-lead concentration, as modeled, is expected to occur near the source. Although there is a large amount of lead emitted into the air, only a few sources emit lead and lead compounds in sufficient amount to exceed the ambient air quality standard for air (refer to Section 5.5.2). Only two of 17 modeled stationary sources of air-lead emissions had calculated maximum point downwind air-lead concentration values projected to exceed the air quality standard of 1.5 tg/in 3 . Drinking water supplies are also typically safe, although exposure does continue in some communities. Violations of the drinking water standard are rare. Exposure to lead through soil and dust, associated with operating and abandoned hazardous waste sites, may occur in a few cities. The majority of sites, however, are located beyond the boundaries of the central cities assessed and, consequently, do not generally pose a threat. The research placed special emphasis on the risk posed by low-level environmental sources of lead to African-American and Hispanic children. These populations are thought to be at particular risk (refer to Section 2.8). For children residing in central cities of one million population or more, and annual family income less than $6,000, 68 percent of African-American children are projected to have blood-lead levels exceeding 15 t .g/dL For white children in the same socioeconomic strata, the percent projected so exceed that value is much smaller, at 36 percent. Seven cities in Region 5 are in the top 10 of the 83 cities assessed in the Midwest by virtue of having both the highest percentages of children as well as the greatest numbers of children that may exceed 10 g/dL blood-lead concentration. Those cities are Milwaukee, Wisconsin; Detroit, Michigan; Minneapolis and St. Paul, Minnesota; and Cincinnati, Akron, and Cleveland, Ohio. The analysis indicates that the States of Illinois and Michigan had the largest numbers of African- American and Hispanic children under seven years of age calculated so exceed 10 &g/dL blood-lead level. Pmj.ct LEAP— Pb 1 X ------- This includes 28,000 and 16,000 minority children, in the respective states, due to environmental sources of lead. Every Region 5 state has community areas where elevated blood-lead levels are of concern. For the six Region 5 states, all cities combined, the total childhood population under seven years of age was 1,359,000 in 1988. The findings indicate that 154,000 children, or 11 percent of the total, would have blood-lead levels exceeding 10 igfdL. The predominant environmental sources are lead contaminated soil and dust. This includes 55,000 African-American and 12,000 Hispanic children. The cities with the highest potential for sizable numbers of African-American and Hispanic children with blood-lead levels calculated as above 10 tg/dL are Chicago, illinois, 27,000; Detroit, Michigan, 13,000; Milwaukee, Wisconsin, 5,000; aeveland, Ohio, 4,000-, Cincinnati, Ohio, 2,000; and Indianapolis, Indiana, 2,000. It is important to note that this methodology is for population screening purposes. It expands upon the use of an Uptake Biokinetic Model for derivation of blood-lead levels. Such use of the model has not been attempted before. The Uptake Biokinetic Model was developed specifically for application at abandoned hazardous waste sites for which measured environmental lead concentrations are known. The Uptake Biokinetic Model has only been validated at that spatial scale. This methodology applies it at a much larger spatial scale. It includes both estimated and measured environmental concentrations, and uses the model as part of a population risk screening approach. Consequently, the results may have no practical value as a prediction of the actual number of children expected to have elevated blood-lead levels. Nor was that the intent of the methodology. The value of the approach is in the comparison between cities. It is specifically to locate areas within a city that may be expected to have higher rates of lead exposed children than other areas. The intent of the population screening methodology is to use the relative number to set priorities for intervention efforts within a city or region. The reader is particularly cautioned that the numbers of children cited in this research are as derived by the computerized methodology. The methodology is a screening tool. It is not a methodology to predict PrqJ.ct LEAP— Phase I xi ------- aetual nuthber,ofcliildrefl at risk., ------- 1. INTRODUCFION The insidious effects that lead causes on the health of children have received increased attention in recent years. Due to mouthing behavior, increased uptake of lead compared to adults, nutrition and other factors, children under seven years of age present a subpopulation at increased risk to the adverse effects of lead exposure. Within this population subgroup, it has been well demonstrated that African- Americans, particularly in lower socio-economic situations, are a subpopulation group at even greater risk. Hispanic children may also be at higher risk. The reasons for a dissimilarity between white and African- American children are unclear. It is clear, however, that the difference is seen at all socioeconomic levels. Measurements and projections of blood-lead levels for African-American children consistently reflect elevated blood-lead levels. Reports from the second National Health and Nutrition Examination Survey, based upon data from 1976 to 1980, illustrates the substantial difference in blood-lead prevalence levels based upon ethnicity. Among African-American children six months to five years of age, only 2.5 percent of African-American children, compared to 14.5 percent of white children, had blood-lead levels less than 10 .tg/dL (Lin-Fu, 1992). For families with an annual family income < $6,000, 18.5 percent of African-American children, contrasted to only 5.9 percent of white children, exceeded 30 tg/dL (children aged six months to five years). The percentage was 10.9 percent exceeding 30 tg/dL for all races. For that same age group, the geometric mean blood-lead level was 19.6 p.g/dL for African-American children, 14 p.tg/dL for white children, and 14.9 p.g/dL for all races. Although complete data are not available for children of Hispanic origin, the Agency for Toxic Substances Disease Registry (ATSDR, 1988) postulates that it is reasonable to assume that the association between high blood-lead levels and lower socioeconomic income Status would hold true for this population as well. Hispanic children, accordingly, may also be at elevated risk. As research continues, the level of blood-lead concentration of concern continues to be lowered. PrqJectLEAP—Pbase l 1 ------- More and more studies add to the weight of evidence for health effects in children at levels previously thought to be safe. The fact that lead is a transplacental contaminant is even more alarming because internal exposure can begin in the fetus. The exposure can continue to contribute to body tissue burden of the young child if the child is subsequently brought into a lead-contaminated environment. A significant evolving concern is that many of the effects of low-level lead exposure are not readily observable in the individual child, unlike physical manifestations caused by acute lead poisoning. Acute (observable) effects are usually associated with lead-based paint. Health effects are generally ascertained not through clinical diagnosis of the individual patient, but rather through epidemiologic study of large groups of children already suffering from the chronic effects of lead exposure. These chronic effects are generally not observed in the individual child. Effects may include lower intelligence and other neuropsychologic deficits, hearing impairment, stunted growth, reduction in attention span, and other reported health impacts. Some studies suggest the lack of a threshold. This is extremely problematic. Even though acute poisoning and exposure have been recognized, generally associated with lead-based paint contamination, chronic exposure and effects caused by low-level lead exposure in the environment are difficult to recognize. This nation has experienced a tremendous reduction in lead emitted into the environment by the phase down of lead in gasoline. The reduction has been paralleled by a significant concomitant reduction of the average blood-lead levels in this Country. Lead, however, remains pervasive in our environment. It is in the homes of tens of millions of families and serves as a continuous source of contamination and exposure via lead paint. Lead remains in some sources of drinking water in the home. It remains in soil and dust, caused potentially by both exterior and interior lead-based paint, as well as historical or ongoing deposition from mobile sources of nearby industry. Even the nation’s food supply still contains some lead, albeit in small quantity. The aggregate effect from multiple sources, in a specific geographic area, may be sufficient to cause concern. p J.ct LEAP— Phase 1 2 ------- The major objective of this research is to examine environmental sources of lead that may be linked to chronic health effects in young children. In particular, such effects may be exacerbated by an aggregation of low-level environmental exposures to lead and lead compounds that result from multiple pathways of exposure. The research effort does not account for the direct effects of lead-based paint consumption. The methodology does take into account the indirect contribution to exposure from lead- based paint via lead-contaminated soil and dust. It is recognized, nevertheless, that lead-based paint provides the largest contribution to elevated blood-lead levels. This is particularly the case for acute lead- poisoning events. This effort, however, is to assess the extent to which low-level environmental sources of lead may also contribute to elevated blood-lead levels. It constitutes the first phase of Project LEAP: analysis of existing environmental data pursuant so a comparative risk analysis of childhood exposure to lead for the study cities. The goal is to discern a logical direction for future lead reduction efforts. Phases 2 and 3 will follow, to address lead testing and remediation, respectively. This report documents the development of a management tool to identify and prioritize geographic areas having children with elevated environmental exposures to lead which may constitute a health risk to young children. The methodology explores the application of an Uptake Biokinetic Model, developed by the U.S. Environmental Protection Agency for site specific application, on a much larger scale than its original design and intent. Pruj.ct LEAP— Phase 1 3 ------- 2. LITERATURE REVIEW 2.1. Toxicological Profile “The EPA (1986a) and ATSDR (1988) are concerned that the emerging evidence of a consternation of effects, including inhibition of AL4-D activity and pyrimidine-5 ‘-nucleotidase activity and reductions in serum 1,25-dihydrooxyvitamin D levels, is indicative that low-level lead exposure has a far reaching impact on fundamental enzymatic, energy transfer, and calcium homeostatic mechanisms in the body, which are expressed through subtle effects on neurobehavioral indices, growth and blood pressure “(ATSDR, 1990). The evidence that low level lead exposure is a health concern, particularly for young children, has emerged from a host of studies concomitant with the recognition that today, such exposure has become pervasive in the United States. This is especially alarming as we gain a fuller understanding of the aggregate effects of the multitude of (external) environmental exposures that contribute to internal exposures. That internal exposure is typically assessed via ascertainment of the blood-lead (Pb-B) level, a measure historically associated not with low level, chronic exposure, but rather with the acutes effect caused by lead-in-paint poisoning. 2.1.1. Internal Exposure For decades, scientists have recognized that high exposure to lead results in encephafopathy, colic, anemia, nephropathy, and electrocardiographic abnormalities. High exposure can cause spontaneous abortions in females, and decreased fertility in men (AThDR, 1990). McMichael c i al. (1986) reported on miscarriage and still births among pregnant women. ATSDR (1990) notes that the primary source of lead (Pb) in children is via the gastrointestinal tract. It is distributed in blood, soft tissue, and bone. In human blood, 99 percent of the lead in blood in attached to erythrocytes (with over 50 percent of this pool bound to hemoglobin) (ATSDR, 1990). The balance is deposited in blood plasma, and can be transported to soft tissues. Lead in bones is found in two components, an inert pool with a half life of decades, and PrqJ.ct LEAP— Pha.. 1 4 ------- a labile pool having the ability to exchange readily between bone and blood or soft tissue (ATSDR, 1990). According to a model proposed by Rabinowitz et al. (1976), the blood component half life is 36 days, soft tissue is 40 days, and bone is iO days (about 27 years). A number of age related differences exists between lead distribution and body burden of children, in comparison to adults. In a controlled expenment, Griffin et al. (1975) found that blood levels returned to near-normal after about two months subsequent to termination of exposure to airborne lead. In contrast, the biological half life in two year old children has been measured to be about 10 months, (Succop et al., 1987). Further, in adults, about 95 percent of the total body burden is in bones, while in children, the percentage is approximately 73 percent (ATSDR, 1990). It is noted that lead accumulation in most soft tissues (the kidney, brain, and liver) is of much smaller proportion than lead which accumulates in bone. Blood-lead which is not retained in one of these compartments is excreted by the kidney, or is excreted through biliary clearance into the gastTointestinat tract (AThDR, 1990). It is also noted that the physiological stress of pregnancy can mobilize lead from maternal bone. This creates additional exposure for the developing fetus, resulting, consequently, in greater danger to the fetus. The transpiacental transfer of lead has been cited in a number of studies over the years. In a Glasgow, Scotland, study of 236 mothers and infants, the geometric mean blood-lead levels were found to be 14 ig/dL for mothers, and 12 p.g/dL for infants (Moore Ct al., 1982). According to the Public Health Service (ATSDR, 1988), there is no metabolic barrier to fetus uptake of lead; consequently, exposure of women during pregnancy results in lead uptake by the fetus (i.e., physiological stress results in increased exposure of the fetus). Differential internal exposure risk appears to continue after birth. Infants from birth to two years have been shown to retain 32 percent of the total amount of lead absorbed, according to a study by Ziegler ci al. (1978); whereas a study by Rabinowitz et al. (1977) discerned a one percent retention rate in adults of the absorbed amount of inspired lead, derived by ATSDR from the Rabinowitz et al. study data. The Rabinowitz et al. study itself found that the average respired lead intake of 14 pg/day, inhaling air containing 2 pg/rn 3 lead, Pr iject LEAP— Ph 1 5 ------- resulted in a calculated increase of 0.06 tgfgm in the blood-lead level. The interaction of lead with other chemicals in the body has also been extensively studied, and is a matter of concern. “In humans, the interactive behavior of lead and various nutritional factors is appropriately viewed as particularly significant for children, since this age group is not only particularly sensitive to the effects of lead, but also experiences the greatest changes in relative nutrient status” (ATSDR, 1990). Data supporting this conclusion is available from a number of sources. Studies have found that calcium intake is inversely correlated with increasing blood-lead levels (ATSDR, 1990). Watson et al. (1980) reported that iron deficient adults absorbed lead two to three times greater than lead- replete adults (thus 20 to 30 percent of dietary input, versus 10 percent). Studies have found increased lead absorption with low dietary calcium, increased lead absorption and toxicity with iron deficiency, and that low zinc in the diet increases lead absorption (ATSDR, 1990). Mahaffey (1990) found that lead absorption and toxicity increased for subjects with diets low in calcium. He also found that long term iron deficiency, as well, increased the absorption and retention of lead. Mahaffey concluded that longitudinal, prospective studies are needed to evaluate the effectiveness of nutrition as a preventive strategy for lead intoxication. Lower-level exposures affect the synthesis of heme, and decreases the circulating levels of the active form of vitamin D, 1,25-dihydroxyvitamin D, in children. “This form of vitamin D is largely responsible for the maintenance of calcium homeostasis in the body” (ATSDR, 1990). In a study by Rosen et al. (1980), the researchers found that lead-exposed children with Pb-B levels of 33 to 120 .tg/dL had notable reductions in serum levels for both 1,25-dihydroxyvitamin D and Pb-B over the entire range of blood-lead levels measured in the study. EPA (1986a) concludes that lead’s interference with heme synthesis may be the basis for the effects on vitamin D metabolism. Low-level lead exposures causes inhibition of erythrocyte ALA-D, down to the lowest observed blood-load levels of approximately three to five g/dL (ATSDR, 1990). This has been confirmed PrqJ.ct LEAP— Pba.. 1 6 ------- particularly for child studies by Secchi et al. (1974) (minimum subject Pb-B value of 16 .tWdL), Wada et al. (1973), Hernberg and Nikkanen (1970), Chisolm c i al. (1985a), and Rods et at. (1976) (mimimum subject Pb-B value of 4.7 .tg/dL). The lowest observed adverse effects level (LOAEL) for ALA-D and heme synthesis is thought to occur below 10 tg/dL (ATSDR, 1990). Based upon a review of studies by EPA (1986a), ATSDR (1988), and Grant and Davis (1989) the threshold for accumulation of erythrocyte protoporphyrin (EP) or zinc protoporphyrin is approximately 15 ,.tg/dL, the presumed Lowest Observed Adverse Effects Level (LOAEL) for children. EPA (1986a) concluded that inhibition of the enzyme erythrocyte pyrunidine-5’-nucleotidase may occur in workers at Pb-B levels at or exceeding 44 tg/dL, and in children that inhibition of the enzyme is seen down through the lowest blood-lead levels of approximately seven j.tg/dL, based upon data of Angle et at. (1978) and Angle et al. (1982). The LOAEL for children consequently appears to occur at less than 10 tg/dL under intermediate and chronic exposure scenarios. A study by Rosen et al. (1980) found strong indication of an inverse correlation between Pb-B and serum 1,25-dihydroxyvitamin D, that was observed in children over the blood-lead levels measured in the study, from 33 to 120 ptg/dL 2. 1.2.EncephalopathyfLethaljty For the oral route of exposure, the range of blood-lead levels associated with encephalopathy in children was about 90 to 700 or 800 .tg/dL, with a mean of approximately 330 .tg/dL (ATSDR, 1990). The range associated with death is approximately 125 to 750 tg/dL, with a mean of 327 tg/dL For the inhalation route of exposure in adults, lead encephalopathy is the most severe neurobehavioral effect (ATSDR, 1990). Early symptoms include dullness, irritability, poor attention span, headache, muscular tremor, loss of memory, and hallucinations. The condition can worsen to delirium, convulsions, paralysis, coma, and, ultimately, death. This is generally not observed in adults until levels exceed 120 p.g/dL Such studies of signs and symptoms indicate that the lowest observed-effect levels for overt signs and symptoms of neurotoxicity is in the range of 40 to 60 tg/dL (ATSDR, 1990). In children, acute lead poisoning other PrQJ.ct LEAP— Ph... 1 7 ------- than signs of encephalopathy have been observed at levels of approximately 60 to 450 tg/dL. Acute lead poisoning in children causes death from Pb-B levels equal to or exceeding 125 ig/dL, as reported by NAS (1972), based upon studies by Chisolm (1962) and Chisolm and Harrison (1956). Although the Chisoim studies did not address lethality directly, the latter study noted four deaths, and estimated the total lead in the soft tissues of the individuals to be 20 to 100 mg. Grant and Davis (1989) suggest that Pb-B levels that can produce death are basically the same as those associated with acute encephalopathy. Such effects are usually observed in children from approximately 100 tg/dL 2.1.3. Neurological Impairment AThDR (1990) paraphased an EPA (1986a) report, that concluded “that the consistent pattern of lower JO values and other neuropsychologic deficits among the higher lead exposure children in these studies indicate that cognitive deficits occur in apparently asymptomatic children with markedly elevated blood-lead levels (starting at 401060 lg/dL and ranging up to 70 to 80 p.g/dL).” EPA concluded that approximately five JO decrement points is a reasonable estimate of the extent of JO decrements associated with markedly elevated blood-lead levels (mean approximately 50 to 70 tg/dL) in children that do not exhibit signs and symptoms of lead poisoning (EPA, 1986a). 10 deficits of approximately four points are associated with blood-lead levels of 30 to 50 tg/dL (ATSDR, 1990). In studies reported in 1986 and 1987, Hawk et al. (1986) replicated the Study with a cohort of 75 African-American children, aged three to seven yeara old. All were of low socioeconomic status. Using a backward stepwise multivariate regression analysis statistical technique, they found a “highly significant linear relationship between the Stanford-Benet 10 scores and contemporary blood-lead levels, over the entire range of 6 to 47 p.tg/dL”. Hawk et a!. (1986) and Fulton et a!. (1987) reported a significantly inverse linear association between cognitive ability and blood-lead levels. There was no evident threshold down to the lowest Pb-B of approximately 6 tg/dL The LOAEL for JO effects is, consequently, thought to be less than 10 .tg/dL (ATSDR, 1990). PrqJ.ct LEAP— Ph ... 1 8 ------- A study by Robison ci al. (1985) of 75 African-American children aged three to seven years old. The study determined that hearing decreased for the study group, and that the severity of hearing loss increased linearly wIth historical blood-lead levels, in the range of 6.2 to 56.0 p.g/dL. Schwartz and Otto’s (1987) logistic regression analysis of NHANES II (the second National Health and Nutrition Examination Survey) data suggests the probability of elevated hearing thresholds with significant increases across the entire range of blood-lead levels of < four ig/dL to > 50 p.g/dL. The study involved 4,519 children aged four to 19 years, and was controlled for several confounding variables available from the data set. A study of the effects on peripheral nerve function suggests an increased susceptibility to lead neumpathy, among children with sickle cell disease (Erenberg et al., 1974). 2.1.4. Developmental Toxicity Developmental toxicity of lead has been assessed in several studies. A study in Boston by Needleman et al. (1984) found an association between lead exposure and congenital abnormalities, including undescended testicles (Hydrocele). Several epidemiological studies have also been conducted. A Port Pine, South Australia study of 595 children was reported by Vimpani ci a!. (1989) as well as Baghursi et al. (1987). These studies determined that the geometric mean values of Pb-B increased from approximately 14 p.g/dL at six months of age, to approximately 21 p.tg/dL at 15 and 24 months of age. Depressed Mental Development Index (MDI) scores were found to be significantly associated with higher post-natal blood-lead levels, as well as with six month blood-lead levels, although such an association was not found with pne-natal delivery or with cord Pb-B level. The study found a two point deficit in MDI at age 24 months, for every 10 tg/dL increase in Pb-B at age six months. In a Bellinger et a!. (1985a,b, 1986a,b,1987) prospective study of 249 middle-to-upper-middle-class children in Boston, Massachusetts, the researchers determined that the high-lead group (having a mean cord blood level of 14.6 jtg/dL) demonstrated an average deficit of 4.8 points on a “covariate-adjusted” MDI score, when compared to the low-lead group (having a mean of 1.8 tg/dL cord blood level). The difference was 5.8 points at six PrQJ.ct LEAP— Phi.. 1 9 ------- months, and 7.3 points at 12 months.This inverse relationship held for ages six, 12, 18, and 24 months of age. The Bellinger study (1985a) considered several variables, including demographics (race, parental age, education, marital status, occupation), medical/reproductive history, index pregnancy, labor and delivery, neonatal status (such as birth weight and infections), post natal status (such as hospitalizations and temperature), postnatal environment (including HOME, maternal JO, family stress, and feeding method), and cord-blood level as an ordinal categorical value (low, medium, or high). The HOME (Home Observation for Measurement of the Environment) assesses the quality of the rearing environment. The study found a statistically significant association of blood cord levels and MDI scores, when the MDI scores were adjusted for length of gestation and total HOME score. A further analysis by Bellinger et al. (1990) found that children with high (10 to 25 .tg/dL) umbilical cord-blood levels achieved significantly lower MDI scores through two years of age, than infants with low (c three g/dL) or medium (six to seven ptg/dL) cord blood levels. The cord blood level, however, was found not to be significantly related to performance (using the McCarthy Scales of Childrens’ Abilities) at age 57 months. The study found that delta Z a derived index for a child’s “developmental trajectory” between 24 and 57 months of age, to be significantly related to higher HOME scores, higher social class, and more intelligent, older mothers. It was not, however, significantly related to gender or ethnicity. According to the report, “The associations between performance trajectory between ages 24 and 57 months and several of these characteristics, including high social class, high HOME score, and high maternal 10, are consistent with the hypothesis that environmental enrichment facilitates the rate and extent of recovery on compensation” [ from lead associated cognitive deficit]. EPA (1986a), Davis and Svendsgaard (1987), Grant and Davis (1989), and ATSDR (1988) concluded from several studies of neurobehavioral effects of pre-natal lead exposure (Ernhar* ci al., 1985; Wolf et a!., 1985; Davis and Svendsgaaid, 1987; Winnede ci al.,1985a,b), that neurobehavioral effects, indeed, are associated with prenatal internal exposure levels. Maternal or cord blood-lead concentrations Project LFAP— Phi .. 1 10 ------- of 10 to 15 tg/dL, and possibly lower were found to be associated with such effects (ATSDR, 1990). (The Ernhart 1985 study, however, for which cord Pb-B levels ranged from 2.6 ig/dL to 14.7 .tg/dL with a mean of 5.8 JAg/dL, concluded that “... the results do not provide a reasonable level of support for the hypothesis of adverse effects due to intrauterine low-level Pb exposure”.) ATSDR (1990) notes the criticism of the flaws of the studies reviewed, that showed both positive and no effects at low blood-lead levels. The ATSDR further notes that a 2 to 8 point deficit for an individual child may not be clinically significant, but that a 4 point reduction in a normal distribution of MDI scores for a given population of children, would result in an increase of 50 percent of the children scoring below 80, which the report called “a grave consequence” (ATSDR, 1990). The Cincinnati Lead Program Project continues to follow study subjects into their early school years to discern whether early deficits (i.e., decrements in Bayley mental index scores) persist into later life (that is, do the observed effects of low level lead exposure continued at the same magnitude over time). Dietrich c i a!. (1990) reports on the relationship between prenatal and postnatal lead exposure and development status of two-year-old infants. Families for the study were recruited from “lead-hazardous areas” of Cincinnati, Ohio, based upon pediatric case histories of lead poisoning. A total of 297 infants, with a mean blood-lead level of 17.45 tg/dL, participated. The sample was 86.2 percent African- American. The mothers were predominantly from lower social classes, unmarried, and on some form of public assistance. Developmental assessments were conducted at ages three, six, 12, and 24 months. The three part Bayley Scales of Infant Development were used: Mental Development Index (MDI), Psychomotor Development Index (PD!), and Infant Behavior Record (IBR). The researchers collected social as well as medical background data to test as potential confoundeis, including race, maternal age and tobacco use. The study employed multiple regression analysis with backward elimination of nonsignificant covariates and confounders (in the reduced model, while all variables were included in the multiple regression Prq .ct I FAP- Ph... 1 11 ------- analysis). The lead variables were analyzed both in terms of tg/dL and of a transformation to their natural logarithms. The prenatal and neonatal blood-lead levels were found to be low, with a few subjects exceeding 25 tg/dL Most reached the highest blood-lead level during the second year. About 25 percent had at least one serial blood-lead of 25 p.g/dL during the second year. Prenatal blood-lead was found to be significantly related to six month MDI after statistical adjustment for 10 potential covariates and confounders, at six months, but only for males. It was insignificant for females at this age. The study did not provide reasons for the gender difference. The study also found, for Hollingshead socioeconomic status scores below the sample median of 17, a covariate adjusted reduction of 0.757 MDI points for each .tg/dL increase of neonatal blood-lead (p = 0.0316). This was statistically insignificant, however, for a status score above 17. A two year follow-up determined that there was no statistically significant relationships between prenatal or postnatal blood-lead level variables and Bayley MDI. The relationship with Bayley IBR factor scores also had statistically insignificant results. Dietrich ci al. conclude that the lack of inverse relationships suggests that those infants of mothers with higher prenatal blood-lead levels may have overcome their early developmental deficits. The authors note that these results are inconsistent with previous studies by Bellinger and the Port Pine Study of 1988, and cite as caveats the limitation of Bayley scales of measurement. The authors also noted that the two other studies did find continuing harmful effects at two years of age. The documented toxic effects of lead on the human fetus include a lowering of the gestational age, reduction in birth weight, and reduced mental development, all of which may occur at relatively low Pb-B levels (ATSDR, 1990). McMichael et al. (1986) found that the risk of pre-term delivery increases about four times as cord or maternal Pb-B increases from s eight so >14 tg/dL Dietrich et a!. (1986; 1987a) reported a significant inverse association between prenatal Pb-B levels in the mother, and birth weight, withtheeffectobscrvcddownto12to13 tg/dL. Prqj.etLEAP—Pbe l 12 ------- A Bellinger et al. (1987a) study reported significant deficits of 4.8 points in the Bayley MDI at ages six to 24 months of age, in children whose Pb-B at birth ranged from 10 to 25 ig/dL, contrasted to children whose Pb-B level at birth was less than three WdL. Dietrich et a!. (1987a) also reported an inverse correlation between prenatal or neonatal blood-lead levels and MDI, in the range of one to 25 .tg/dL PrQJ.ct LEAP— Pha.. 1 13 ------- 2.1.5. Aggregated Studies Analysis Needleman et a!. (1990a) performed what was termed a meta-analysis of 12 of 24 studies that used multiple regression analysis to study the effect of childhood exposure to lead on 10. They found overall evidence of a strong link between low-dose lead exposure and intellectual deficit in children. The analysis concluded that even though the studies had significant variation in their individual power to find an effect, 11 of 12 of the studies reviewed reported an association between adverse health effects and lead exposure. 2.1.6. Growth The effects of lead exposure upon growth in the young child have been recognized as far back as 1929, when Nye (1929) reported on runting (stunted growth) and chronic nephritis in overtly lead- poisoned children in Australia. (Nye, in turn, cites a report by A. Jefferis Turner of lead poisoned children in Brisbane, in the year 1892.) Schwartz et al. (1986), based upon data for 2,695 children under seven years of age from the second National Health and Nutrition Evaluation Survey (NHANES II) study, provides even stronger evidence of this effect. Through the use of a stepwise multiple regression analysis technique, the Schwartz group concluded that blood-lead levels for the range of five to 35 p.tg/dL, were a “statistically significant predictor of children’s height (p.c.0001), weight (p<.OO1), and chest circumference (p<.O 26 ), after controlling for age in months ... race, sex, and nutritional covariates.” The strongest relationship found was between Pb-B and height, with regression models indicating no threshold down to the lowest observed Pb-B of five jtg/dL There was no indication of a threshold within the study range. 2.1.7. Toxicological Summary These studies indicate that there are several effects of major concern regarding low-level exposure, including neurobehavioral effects Mental Development Index (MDI) and Intelligence Quotient (10) deficits, as well as elevated hearing thresholds, and growth retardation (for young children with pie-natal exposure as well as for children suffering from post-natal exposure) (ATSDR, 1990). There appears to Proj.ct LEAP— Phii. 1 14 ------- be no indication of a threshold down to the lowest level of internal exposure (Pb-B c 10 xg/dL) (ATSDR, 1990). Health impact are summarized in Table I. :$wdy : : TABLE I Health Effects Summary : Neahh Etfea 14B VSut Rosen et at., 1980 Interference with heme synthesis, decreased level of 1,25-dehydroxyvisamin D 33 to 120 Secchi ci al.,1974; Wada et al., 1973; Hernberg and Nikkenen, 1970; Chisolm et aL,1985a; and Roels et at., 1976 Inhibition of erythrocyte ALA-D <3 to 5 AThDR, 1990 Lowest observed adverse effect level (LOAEL) for ALA-D and heme synthesis <10 EPA, 1986a; Grant and Davis, 1989 Acrumulation of crythrocyte pyrinmidine (LOAEL) 15 Angle et at., 1978; Angle et at., 1982 Inhibition of enzyme erythrocyte pyrinmidine-5-nucteotidase 44 ATSDR, 1990 Encephalopachy in children 90 to 700 NAS, 1972; Chisolm, 1962; Chisolm and Harrison, 1956 Death . 125 to 750 Hawk et at., 1986; h ilton et at., 1987 Decreased IQ 6 to 47 AThDR, 1990; EPA 1986a Decreased 10 of 4 points; of 5 points 30 to * 50 to 70 Robison et at., 1985; Schwn and Otto, 1987 Decreased hearing acuity 6.2 to 56M <4 to >50 Needleman ci at., 1984 Hydrate (undescended testide) Not specified Ernharrt et at., 1985; Wolf c i at., 1985; Davis and Svendsgaard, 1987; Winnede ci at., 1985a,b Neurobehavjoraj effects <10 to 15 McMichael ci at., 1986 Pm-term delivery risk s8 to 14 Dietrich et at., 1986, 1987. Decreased birth weight asaociatcd with mother’a lb-B 12 to 13 Dietrich et at., 1987a Mental Development Index deficit 1 to 25 Nyc, 1929 Stunted growth Not specified ScbwartzetaE,1986 Rcducdonlnheightandweight c5to35 In a speech given on October 7, 1991, Health and Human Services Secretary Louis Sullivan cited Pn J.d LEAP —Phase! 15 ------- an announcement by the Centers for Disease Control, for a lower “threshold of concern” for blood-lead levels in children (Sullivan, 1991). The new threshold is 10 p.tg/dL, coupled with recommendations for “...levels of action for intervention.” Dr. Sullivan called lead poisoning “...the number one environmental threat to the health of children in the United States.” 2.2. Adequacy of Studies The United States Congress, in Section 110(3) of the Superfund Amendments and Reauthorization Act (SARA) of 1986, tasked the ATSDR with preparing a toxicological profile for each of 100 most significant hazardous substances found at the Comprehensive Environmental Response, Compensation, and Liability Act (CERCLA) National Priority List (NPL) sites, and for each profile to provide “An examination, summary, and interpretation of available toxicological information and epidemiologic evaluations on a hazardous substance in order to ascertain the levels of significant human exposure for the substance and the associated acute, subacute, and chronic health effects” (ATSDR, 1990). The result of that mandate has been the preparation of a very intensive and extensive review of the literature on the human health effects of lead. l’his was presented in pertinent part in the preceding section. Congress also required ATSDR to determine whether sufficient information existed to ascertain the levels of exposure for a given chemical that posed endangerment to human health. ATSDR categorized the sufficiency of data from human studies for specific health endpoints as sufficient, some, or no information available to make a definitive determination for each endpoint, for both cancer and noncancer (A1SDR, 1990). ATSDR judged, for combined oral and inhalation studies, that some information exists for lethality, acute systemic toxicity, reproductive toxicity, and carcinogenicity. Sufficient information is deemed to be available, based upon the extensive literature review, for intermediate systemic toxicity, chronic systemic toxicity, and developmental toxicity. No information (derived from human health studies) is available for the dermal route of exposure for any of the seven health endpoinss reviewed. PmJ.ct LEAP— Phas. 1 16 ------- Information is available, however, for animal data. There appears to be sufficient information available on reproductive toxicity (oral), along with some evidence of carcinogenicity via the dermal exposure route (ATSDR, 1990). It is noted that the Interagency Regulatory Assessment Group has rated lead as a Group B-2 possible human carcinogen (1987), and that EPA (1991) has determined lead to be a Class B-2 probable human carcinogen. Noting the difficulties in ascertaining the length of exposures and that distinctions are somewhat arbitrary, ATSDR recommends the joint consideration of intermediate and chronic systemic toxicity data together (AThDR, 1990). ATSDR has determined that the data do not clearly indicate NOAELs for humans, recognizing that the associations between blood-lead levels and neurobehavioral indices, blood pressure, growth, and heme synthesis, occur over a wide range of Pb-B concentrations. Further, there are no indications of threshold values through the lowest Pb-B levels. More than 100 ongoing federally sponsored projects involving lead toxicity have been identified, including several prospective studies on the effects of lead toxicity on neurobehavioral changes in childhood populations (ATSDR, 1990). The existence of a large data base relating Pb-B levels to measured lead concentrations in air, diet, drinking water, dust and soil, has also been noted (ATSDR, 1990). ATSDR, nonetheless, considers such measures to be an imperfect measure of body tissue burden. The better measure is the level of lead in teeth and bones, together with Pb-B, to better measure both past exposures and current body burden. ATSDR specifically cites the EPA lead uptake biokinetic model (EPA, 1991a) for estimating blood-lead levels. The model is based upon exposures, and has been validated by an investigation of young children living near industrial lead sources that contribute to lead concentrations in ambient air, soil, and dust. 2.3. Biological Monitoring TechniQues As noted, however, blood-lead measurement is the most common method of assessing exposure to lead. The half life of lead in human blood is 28 to 36 days (ATSDR, 1990). It is noted that the detection limit at most clinical laboratories is three to five &g/dL (ATSDR, 1990). PruJ.ct LEAP— Pha.. 1 17 ------- The use of erythrocyte protoporphyrin (EP) measurement for screening asymptomatic children for lead toxicity was recommended (CDC, 1985; American Academy of Pediatrics, 1987), recognizing that elevated EP was otie of the earliest and most reliable indications of impairment of heme biosynthesis. Further, EP is used because of the contamination problems in measuring blood-lead levels (ATSDR, 1990). In humans, it is noteworthy that Pb-B values are distributed in a log-normal distribution. Accordingly, researchers should use the geometric mean and the geometric standard deviation when analyzing the distribution data (ASTDR, 1988). 2.4. Ty ica1ly Encountered Environmental Levels EPA (1986a) found that the relationship of blood-lead levels to lead concentrations in air, food, and water is curvilinear, such that the increase in Pb-B is less at high levels than at low exposure levels. The clear implication is that concern is warranted when children, in particular, are subject to even low levels of environmental exposure. The range of normal air concentrations is 0.1 to 2.0 pg/rn 3 (A1SDR, 1990). The median blood-lead-level/inhalation-concentration for children is approximately 1.92 p g/dL blood per pg/rn 3 air, based upon three major studies (AThDR, 1990). Aggregate values, including indirect blood-lead contribution from dust and soil, range from three to five pg/dL per pg/rn 3 air (Brunekreff, 1984). Angle et al. (1984) defines the value at four to five p.g/dL for indirect exposure, additive to the direct inhalation contribution. The Centers For Disease Control (1985) has determined that concentrations of lead in soil of 500 to 1,000 p.g/g result in Pb-B in children exceeding background levels of Pb-B. EPA’s (1985) estimate of the contributions of blood-lead levels from various media have been provided based upon background levels, and the resultant levels from the addition of incremental concentrations of lead in air (EPA 1986, 1986a). Mean background contributions for non-air sources are 2.37 pg/dL from food, water, and beverages; 0.30 ig/dL from dust; and 1.65 pg/dL from air. The total background contribution is 4.32 pg/dL The background Pb-B range is 4.32 pg/dL to 16.72 pg/dL (including 9.40 pg/dL from ingested PrqJect LEAP— Ph e 1 18 ------- dust, and 3.00 .tWdL from ingested air, at the upper value). EPA found the estimates to be higher than predicted when compared to observations in children living in areas with measured lead in air concentrations. By comparison, Piomelli et al. (1980), during an expedition ascending the Marsyandi River in the Manang district of Nepal in the foothills of Annaparna and Dhaulagiri, sampled the blood of local inhabitants. The geometric mean Pb-B concentration was 3.4 p.g/dL. Only 10 of 103 individuals tested exceeded 10 p.gIdL. In 1986, the U.S. production of refined lead from primary sources totaled 808 million pounds, and from secondary sources totaled another 1,356 million pounds (ATSDR, 1990). It is noted that recycling of old scrap metal supplies 45 percent of U.S. demand. In that year, consumption in the nation was 2,480 million pounds, including 319 million pounds of lead imported to this country (ATSDR, 1990). Anthropogenic emissions constitute the primary source of lead in the environment, and as of 1984, gasoline combustion, in particular, was responsible for approximately 90 percent of all anthropogenic emissions (ATSDR, 1990). This percentage has been reduced dramatically, however, due to the phase down of lead in gasoline. Atmospheric deposition is the largest source of lead found in both soils and surface waters. The lead particles are removed from the atmosphere principally via wet and dry deposition. ATSDR (1990) stales that soil and sediments appear to be important sinks. The average residence time in the atmosphere is seven to 30 days, during which period lead can be transported up to thousands of kilometers. According to EPA (1986a), natural emissions of lead from volcanoes and windblown dust are thought to be of minor significance. EPA (1986a) has also estimated the anthropogenic lead emissions into the atmosphere for the year 1984. The 1984 estimate for gasoline production was 34,881 tons/year, or 89.4 percent of the total emissions of 39,016 tons/year. Based upon a current gasoline standard of 0.1 g Pb/gallon-gasoline, the estimated 1988 lead emission would be 1,100 tons/year. In the atmosphere, lead exists mostly in particulate form (ATSDR, 1990). Large size particles, 19 ------- particularly those with aerodynamic diameters exceeding 2 I.tm, settle out of the atmosphere fairly rapidly and are deposited relatively proximate to the emission source. Smaller particles may travel thousands of kilometers. In waters, at a pH > 5.4, the total solubilily of lead is about 30 ig/l in hard water, and 500 i.tgfl in soft water (ATSDR. 1990). The ratio of lead in suspended solids to lead in dissolved form has been found to vary from four to one in rural streams, to 27 to one in urban streams (EPA, 1986a). it is noted that lead does not appear to be biomagnified in the food chain, but may accumulate in flora and fauna. In aquatic organisms, the lead concentrations are typically highest in benthic organisms , such as algae, and are lowest in the upper-trophic predators (e.g., carnivorous fish) (ATSDR, 1990). Most lead is retained strongly in soil, and very little is transported into surface or groundwater, according to reports by EPA (1986a) and the Zimdahl and Hasseti (1977). In soils with a pH 5 having more than five percent organic content, atmospheric lead is retained in the upper two to five cm of soil, if it is left undisturbed. EPA (1986a) estimates that four to five million metric tons of lead from gasoline combustion remain in the dust, soils, and sediments of the U.S. Although lead does have a high degree of immobility in soil (Zimdahl and Hassett, 1977), wind action may induce mobilization to the atmosphere and thus downwind transport. This soil entrainment may be of significance to the atmospheric burden downwind from stationary sources, particularly smelters and superfund sites (ATSDR, 1990). As noted previously, preschool age children, pregnant women, and their fetuses constitute the population at highest risk. ATSDR also notes the increase risk to white males aged 40 to 59 years old (AThDR, 1990). The health endpoint of concern for this latter risk group is hypertension. ATSDR (1990) estimates the baseline intake for a two year old child to be 466 &g/day, and for an adult female to be 37.5 .ig/day. Additional exposure results from residing in an urban environment, proxunity to stationary lead sources, residences and other building structures containing lead-based paint, pica (eating disorder of some younger children), both primary and secondary occupational exposure, smoking (from tobacco products containing lead), and wine consumption (for wines containing lead) Prqjsct hAP— Phs 1 20 ------- (ATSDR, 1990). In addition, proximity to a superfund site is thought to increase risk of exposure. At the time of the A1SDR (1990) report, lead and lead compounds had been discovered at 635 of 1,177 National Priority List (NPL) sites. The levels of lead in the ambient air range from 0.000076 tg/m 3 in remote areas, to over 10 tg/m 3 near stationary lead sources (ATSDR, 1990). Confined places (e.g., parking garages, tunnels, and toll booths) may have unusually high concentrations of lead in air. In surface waters in the United States, EPA (1986a) has found typical levels of five to 30 .tg/l. Sediments contain a considerably elevated concentration of approximately 20 rig/kg (ATSDR, 1990). In ground water, the typical range is one to 100 p.gfl (EPA, 1986a). EPA (1988b) estimates that 99 percent of 219 million people in the United States that utilize public drinking water supplies, are exposed to water with levels of lead < 0.005 mg/i, and that about two million consumers are exposed to drinking water exceeding this value. The range is, on average, 10 to 30 i.g/l in households, schools, and office building drinking water supplies, although corrosive water, lead pipes, and lead solder joints can, singly or in combination, produce much higher concentrations (EPA, 1989b). Soils adjacent to roads traveled since 1930 may have as much as 10,000 .tg/g lead (EPA, 1986a), while soils near homes with exterior lead-based paint may have even higher soil-lead concentrations. ATSDR (1990) draws upon studies conducted in Baltimore, Maryland, and Minnesota, to conclude that the highest soil-lead levels generally occur in inner city areas, especially in areas where there has been an historically high amount of traffic. Lead is also found in dairy products, meat, fish, poultry, fruits, sugar, and beverages (EPA, 1986a). Canning processes, in particular, can increase the concentration of pre-canned foods from eight to 10-fold. According to the Food and Drug Administration (Gunderson, 1988), the baseline intake via food consumption for the years 1982 to 1984 was 23.0 tg/day for a two year old child, 29.6 pg/day for Proj.et LEAP— Pha.. 1 21 ------- an adult female, and 40.9 ig.dL for an adult male. Elias (1985), based upon an analysis of food residues and using the 1984 U.S. Food and Drug Administration’s Marketbasket Survey, postulates total food consumption to be 25.8 .tgfPb/day (including 2.8 tg/day from water) for a two year old child. For a male aged 25 to 30 years, Elias estimates a lead consumption of 54.7 jig/day. It is also noted (ATSDR, 1990) that additional exposure through dietary exposure, from atmospheric dust, is experienced by those living in an urban environment, at 91 .i day for children, and 28 jLg/day for adults. Lead content of dusts can be a significant source of exposure, particularly for young children (ATSDR, 1990). It is estimated that children ingest five times more dust particles than adults do (EPA 1986a). Lead-based paint confounds the problem for young children. EPA (1986a) has found concentrations of 1,000 to 5,000 pg/cm 2 for lead-based paint chips. Chisolm (1986) estimates that between 40 to 50 percent of the currently occupied housing in the United States may contain lead-based paint on exposed surfaces. Cigarette smoke is yet another source of lead exposure, with each cigarette containing approximately 23 to 12.2 pg lead (ATSDR, 1990). From two to six percent of the lead may be inhaled in the smoke. Consequently, given the greater propensity for lead uptake by children, secondary smoke poses yet another threat to children under seven years of age, as well as to the developing fetus. Additional exposure to children in the home, as well as to others, is also plausible via secondary occupational exposure from workers in lead processing industries. Workers may bring home lead dusts on their clothing. Other sources of lead exposure, such as housing renovation activity, are also now being more fully recognized. Marino et al. (1990) reported on an outbreak of severe lead-based paint poisoning in a family that was exposed to lead dust and fumes generated during the removal of lead-based paint in the family’s rural farm house. Multiple coats of lead-based paint were being removed over a 10 week period. The removal methods were sanding, torching, and the use of heat guns. These methods produced wood Proj.ct LW- Ph 1 22 ------- particles, fine dusts, and fumes that could be ingested or inhaled. Symptoms were first noticed in the family dog, found by the veterinarian upon examination to be weak, dehydrated, and depressed. The animal was determined to be lead-poisoned, and subsequently died. The mother of the family began feeling weak and tired. The daughter complained of stomach aches in the mornings. The father suffered severe nausea during weekends of renovation work. All were found to have elevated blood-lead levels. EPA (1986a) concluded, appropriately, that “... lead is a pervasive environmental contaminant that causes a wide variety of adverse health effects in humans. In short, lead is potentially toxic wherever it is found, and it is found everywhere”. 2.5. At Risk Population Section 118(t) of SARA requires the ATSDR to prepare a comprehensive study on lead poisoning in children. ATSDR (1988) noted that much of the data needed to prepare the report was not available in peer- reviewed literature, and, consequently, was developed specifically for the report to Congress. In a given year, A1SDR estimates that an estimated 400,000 fetuses are exposed to maternal Pb-B > 10 g/dL, within Standard Metropolitan Statistical Areas (SMSAs). For other exposures, the estimation problem is more problematic. AThDR (1988) found that “The actual number of children exposed to lead in dust and soil at concentrations adequate to elevate Pb-B levels cannot be estimated with the data now available.” The opinion expressed is that the regulatory actions of the 1970’s to address existing lead- based paint in housing “have been a clear failure” (ASTDR, 1988). The current average Pb-B levels in the United States today, in some segments of the population, are 15 to30 times higher than the theoretical mean value of 0.5 tg/dL, calculated for pre-industrial humans (ASTDR, 1988). ATSDR (1988) selected 1984 as the base year for estimating the number of children at or above selected Pb-B levels, because all of the required enumerations were available for that year. The findings from NHANES II were utilized to derive prevalences for demographic and socioeconomic strata within Pr J.ct LEAP- Pb e 1 23 ------- the childhood population, in order to judge the numbers of exposed children. The method was to allocate the total number in each Standard Metropolitan Statistical Area (SMSA) for selected strata of age, race, income, and (where possible) urbanization categories. The various strata were then added to obtain national totals for each strata. Each stratum population number was then multiplied by the prevalence for the three selected Pb-B levels (using national prevalence rates), adjusting prevalence from 1978 to 1984 levels, to account for the reduction of lead in gasoline. It is noted that NHANES II did not report Pb-B levels for specific geographic areas, but rather reported for socioeconomic, demographic, and ethnic strata for the nation as a whole. Consequently, due to the lack of geographic specificity of data, the ATSDR report considers SMSAs collectively, and not individually. The data (ATSDR, 1988) is further limited to young white and African-American children, because NHANES II did not include sufficient numbers of Hispanic and other-race children, in order to enumerate prevalences in those ethnic sub-populations. The strata analyzed were African-American and white; 0.5- to two-year-old children, three- to five-year-old children (although these age bands were subsequently merged to derive a 0.5- to five-year-old child age band); urban status (central city, outside of central city); and family income. The size of SMSA (< or> one million population) was also provided. It is noted with specificity that just as the ASTDR study adjusted the 1978 prevalence rates to account for reductions in lead-in-gasoline and lead-in-food from 1978 to 1984, so too are these prevalence rate estimates now overly conservative, due to the further reductions of lead-in-gasoline and food for today (recall in particular the significant reduction in leaded- gasoline emissions to 1,100 metric tons by the year 1990). 2.5.1. Spatial/Numerical Estimate of At Risk Population ASTDR (1988) Pb-B criteria values of 15, 20, and 25 1 tg/dL were as calculated from NHANES data by EPA’s Office of Policy, Planning, and Evaluation (ASTDR, 1988), using logistic regression analysis techniques to update prevalences to 1984. Tables are provided, separately for central cities and outside central cities, on the projected percentages of children 0.5 to five years old that are PmJ.ct [ ZAP— Ph. 1 24 ------- estimated to exceed selected Pb-B levels by family income, race, and urban status within SMSAs. The range is from 36 percent> 15 .tg/dL for white children in populations centers one million and income < $6,000, to 0.5 percent > 25 tg/dL for white children, with family income >$15,000. For African- American children, in the same categories, the range is from 67.8 percent> 15 i.tg/dL, to 2.2 percent >25 .tgfdL Similar patterns are presented for children residing outside central cities. Estimates of the numbers of children in the age band who are projected to exceed the three criteria levels of Pb-B are provided by family income and race. For central cities, with SMSAs < one million population, the projections are 301,100 children >15 p.g/dL, 93,800 > 20 .tg/dL, and 27,500 > 25 .tg/dL. For central cities with SMSAs > one million population, the numbers are even greater, 901,800 children > 15 tg/dL, 301,700 >20 p.g/dL, and 86,200 > 25 p.g/dL Overall, for the 1984 United States childhood population of 13,840,000, 2,381,000 are expected to have Pb-B values >25 g/dL (ASTDR, 1988). Specific concerns with the limitations on accuracy of these projections are noted in the ASTDR report. In particular, the Hispanic child population is not included, and that population segment is experiencing high birth and growth rates. Some of this population is associated with lower economic and central city strata and, consequently, are expected to have higher predicted prevalence rates across the criteria Pb-B levels. “The most important finding, however, is that no strata of these children are totally exempt from risk of Pb-B levels high enough to represent a potentially adverse health impact” (ASTDR, 1988). 2.5.2. Lead Screenin2 Programs Lead has a long history of use by man as well as harm to man, extending hundreds of years back in time. In this country, not much public concern was evident in the early part of this century. In the early 1930’s, however, the Baltimore Health Department became interested in lead poisoning (Lin-Fu, 1982). Not much attention was shown by health officials in other cities until the early 1950’s. At that time, New York, Chicago, and Philadelphia began case finding as well as public education efforts. During this period through the mid-1960’s, health officials found hundreds of cases of lead poisoning in several PrqJ.ct LFAP- Phi.. 1 25 ------- large older cities (Lin-Fu, 1980). These included Baltimore, New York, Philadelphia, and Chicago. Between 1959 and 1963, Cook County Hospital in Chicago treated 182 children for lead encephalopathy (Lin-Fu, 1982). Of the cases, 51 died. A mass of data on childhood lead poisoning was published in the 1950’s to early 1060’s (Lin-Fu, 1982). Most of the public was unaware of the problem. Many in the public health profession failed to react. According to Lin-Fu (1982), the turmoil and awakening of social conscience of the mid-1960’s brought a sudden realization of the magnitude of childhood lead poisoning in this country. It was during this period that it was discovered that lead poisoning was epidemic in the inner city slums (Lin-Fu, 1980). In 1966, Chicago began the first mass blood-lead screening program in the nation. New York and other cities did the same. It was also during this decade that health officials unexpectedly discovered asymptomatic children with elevated blood-lead levels. This discovery sounded an alarm that health care workers needed to recognize lead absorption in preventing lead poisoning disease, and that subclinical toxic effects of lead were a concern (Lin-Fu, 1980). The pervasive source of lead exposure in children, from lead in dust and soil, also became apparent (Lin-Fu, 1992). Childhood screening programs discovered a high prevalence of elevated blood-lead levels in children that could not be fully explained by ingestion of lead contaminated paint chips. The U.S. Surgeon General issued a statement in 1970 which effectively shifted the emphasis of health care workers from case finding to lead poisoning prevention. He advocated mass screening to find cases of elevated blood-lead levels (Lin-Fu, 1982). Shortly thereafter, the 1971 Lead-Based Paint Poisoning Prevention Act became law. The Act provided funds for mass screenings. Mass screening funded by the Act began in mid-1971. From January 1972 through December 1978, 2,485,320 children were screened by federally funded projects (Lin-Fu, 1980). Of these, 170,738 children were found to have elevated Pb-B or EP levels. In fiscal year 1982, these lead screening programs, along with other public health protection Pi LFAP- Ph . 1 26 ------- programs, were incorporated into the Maternal and Child Health (MCH) Block Grant Program (ASTDR, 1988). The screening programs are targeted primarily at case finding for children with Pb-B levels serious enough to warrant medical intervention. The classification schemes have changed over the years. Analysis for elevated EP has been the first step in screening, although it is recognized that some children having elevated Pb-B will pass the EP test; accordingly, the EP screening test does produce false negatives. Table II, based upon data presented in the ATSDR (1988) report, presents the results of screening programs in 16 cities in the Midwest region of the country, for fiscal year 1981, using the 1978 Centers for Disease Control classification of lead toxicity of 30 p.g/dL Pb-B and 50 p.g/dL EP. In 1988 Congress enacted the Lead Contamination Control Act. Among other provisions, the Act authorized the Centers For Disease Control to Provide grants to States and local health agencies to fund childhood lead poisoning prevention programs (DHHS, 1991c). The grants are to screen children for lead poisoning; to ensure environmental as well as medical follow up for lead-poisoned children; and to provide education about lead poisoning. PrqJ.et LEAP-Ph...! 27 ------- TABLE II Blood-Lead Screening Program Results for Children in 16 Midwest Cities in 19811 Piog ram tacation , Number Screened Number With Elevated ‘ 32,861 PbB Chicago, IL 2,070 -- Kankakee, IL 2,464 56 Madison County, IL 2,288 105 Rockford, IL 2,341 30 Waukegan-Lake Col, 1L 3,570 35 Illinois (other programs) 5,184 145 FT. Wayne, IN 532 19 Detroit, MI 19,281 926 Grand Rapids, M I 688 19 Wayne Co., MI 1,818 75 St. Paul, MN 2,107 15 Akron, OH 4,637 149 Qncinnati, OH 9,085 191 Qeveland, OH 14,151 921 Beloit, WI 779 15 Milwaukee, WI 6,640 316 ‘Elevated Blood-lead (Pb-B) is based upon the Caiters for Disease Control lead toxicity dasification of & 30 pgAlL Pb-B and t 50 pgklL erythrocyte protoporphyrin. Project LEAP— Phase 1 ------- ATSDR has also compiled the number of lead poisonings determined by screening programs for fiscal year 1983 (ASTDR, 1988). The numbers of children screened and cases of confirmed lead toxicity, by state are Illinois: 25,340 and 136; Indiana: 1,265 and 1; Michigan: 14,700 and 434; Minnesota: 1,816 and 18; Ohio: 19,543 and 416 (the number evaluated for lead toxicity did not necessarily include all those screened who may have been lead poisoned); and Wisconsin: 4,322 and 187 (some of the 187 cases are estimates by respondents, not necessarily the result of Pb-B testing) (ASTDR, 1988). The cases ranged from 0.1 to 4.3 percent of the children screened. It appears that the rate of chronic lead poisoning in young children is decreasing somewhat (ATSDR, 1988). The numbers of children with elevated blood- lead levels, as well as the percentages of screened children that have lead toxicity, indicates that lead poisoning is a continuing problem. That conclusion is supported by an analysis of lead-screening statistics for the Chicago Department of Health from 1981 to 1985 (ASTDR, 1988). That analysis suggests that there has been minimal change over these years in the percent of children screening positive for lead poisoning. The prevalence of Pb-B levels above 30 tg/dL in young children sampled by NHANES II was higher than that which would be predicted from the state and local screening data (ASTDR, 1988). It would appear, consequently, that screening programs may not be addressing the totality of the at-risk population. 2.6. At Risk Population Estimates By Sources/Routes Of Exposure 2.6.1. Lead-Based Paint Qearly the greatest amount of attention and data in recent yeais has been on the contamination and health problems caused by lead-based paint. A prospective study of inner-city children conducted by Qark c i al. (1985), that found that children who have the highest Pb-B levels lived in the worst housing. The housing-quality accounted for more than 50 percent of the Pb-B variability in 18 month-old children. The study also found that children in public (versus private) housing, near a heavily used interstate PrqJ.ct LEAP- Ph. 1 29 ------- highway, had the lowest Pb-B levels. This result indicates that, for the study, air-lead from highways had only a very limited impact on blood-lead in children. Further, the study found that although rehabilitated housing contained lower lead paint levels than public housing, children in rehabilitated housing had higher blood-lead levels than those in public housing, suggesting to the authors that lead sources in the immediate neighborhood of the rehabilitated housing may be a factor. A more plausible explanation, however, is the probability of the inadequacy of rehabilitation. Performed incorrectly, such units pose substantial risks of reexposure of children returned to the housing units. Chisobn et al. (1985b) found, in a prospective study of children in Baltimore, that children returned to homes subsequent to lead paint abatement/removal actions, experienced significantly higher Pb-B levels than children returned to public housing that was free of leaded paint. ATSDR (1988) noted a “great decline” in the number of very severe cases of lead poisoning in the U.S., but notes that “... the basic epidemiological picture characterizing paint-lead associated toxicity has not materially changed for chronic interaction.” To derive the number of children at risk via this route of exposure, ATSDR’s (1988) method was to use estimates of the ratio of children under seven years of age per 1,000 housing units, together with estimates of categories and numbers of lead-painted houses with problems such as peeling paint, broken or cracked plaster, or holes in walls. Problem dwellings were as defined by the American Housing Survey of the U.S. Bureau of the Census (ASTDR, 1988). The fraction of total housing units to that of such defined “problem” units was used to derive estimates of the total number of children in lead-painted homes, and the number of children in lead-painted homes categorized as problem dwellings. The ATSDR study used estimates and calculations by Pope (1986), whose efforts addressed four major areas of the country, including the Midwest. ATSDR (1988) also notes a comprehensive unit-by-unit study that was conducted in the city of aiicago in 1978, that assessed the Pb-B levels and the presence of leaded paint in 80,000 individual PruJsct LEAP— Pb 1 30 ------- housing units. In general, however, ATSDR found “a general dearth of nationwide studies that estimate the number of children living in paint-containing homes who have elevated Pb-B levels...” The study approach was to take the number of children living that the U.S. Census Bureau estimates to be living in deteriorated housing having 100 percent lead paint, and then to approximate “the most logical” prevalence for the stratum (as discussed earlier), that would be applicable to children in such housing (ASTDR, 1988). The stratum chosen was inner city, dense population, and lowest income, with the further assumption that many of the children in such areas would be African-Amencan. Paint with a lead concentration 0.7 mg/cm 2 was chosen as the criterion value for distribution, with an estimation that 99 percent of pre-1940 housing stock, 70 percent of houses built from 1940-59, and 20 percent of the housing stock built during 1959-74, would exceed this value. Thus, for the U.S. housing inventoly of 80, 390,000 (1983 Survey, U.S. Bureau of the Census), ATSDR estimates that 52 percent (41,964,000) of the units exceed the criterion value (ASTDR, 1988). Ii is further noted that the 0.7 mg/cm 2 criterion value is based upon a CDC (1985) statement. Pope’s method (Pope, 1986) was used to classify housing for age groupa by unsound housing (i.e., deteriorating paint). The study denved, via these considerations, a best national estimate of 1,772,000 children, and a national upper bound estimate of 1,996,000 children under seven years of age living in unsound lead- painted housing. For the Midwest region, the derived numbers are (derived from ATSDR, 1988): Prujict LEAP— Phase 1 31 ------- TABLE III Children Under 7 years of age in the Midwest Residing in Unsound Lead-Painted Housing 2 Age of Housing tnnsr usr — Housing with Peeling Paint La — Number of ———-- Pre-1940 264,000 - J4 ,000 - 1940-1959 159,000 47,000 - 1960-1974 47,000 14,000 Pre-1980 - 470,000 1 39,000 For comparison purposes, the numbers for all four regions of the nation and all housing, are 1,840,000 housing units, and 520,000 children, it is noted that these estimates (based on Pope’s work) include non- SMSA housing stock, and exclude potential exposures that may result from renovation of older urban housing (the so-called urban gentrification phenomenon, as discussed previously, for example, in the study by Marino et at. of a rural farm house renovation) (ASTDR, 1988), due to an inability to quantify such units. A recently released study, the Comprehensive and Workable Plan for the Abatement of Lead- Based Paint in Privately Owned Housing (HUD, 1990), determined there to be no correlation between lead-based paint and household income, and that more units have lead paint on the exterior walls than on interior walls. The national survey estimated that 38 percent of all homes occupied by families with young children have priority hazards, and notes that blood-lead screening programs reach only five percent of the young children in the nation. 2 San: The Nature end Extent of Lead Poisoning in Children in the United Statet A Report to Conga AISDR, 1988. PrqJ.et LEAF — the .. 1 32 ------- The objectives of the national housing survey were to determine the incidence of lead-in-dust in dwelling units, as well as lead-in-soil in and around residences; and to define the characteristics of housing with varying levels of potential lead hazard in order to determine priorities for abatement. The study population was the United States population residing in pre-1980 housing stock. A statistically based survey was derived for the nation. A sample size of 284 housing units was chosen to represent 77 million housing units. The survey assessed interior and exterior paint (concentration and condition) by year built, the type of housing, the threshold level of lead concentration, and the census region. The stratification was on type (privately owned single-family and privately owned multifamily) and construction date (before 1940, 1940 to 1959, and 1960 to 1979). The sample units were geographically clustered in 30 counties (of 3,000 in the nation). The researchers employed X-ray fluorescence (XRF) to test for lead paint concentration, and also collected and analyzed dust and soil samples. Is is noted with particularity that XRF does not distinguish between paint lead on the surface and lead beneath the surface (e.g., old paint under a fresh cover, or lead pipes). The national survey found that 57.4 million homes, representing 74 percent of the study population (residing in 77 million homes), contained lead-based paint (LBP 3 ). This included 9.9 million homes whose families have young children. The percent of LBP housing units by strata was determined to be 90 percent for pre-1940 housing, 80 percent for 1940-59 housing, and 62 percent for 1960-79 housing. The Midwest census region had 76 percent LBP housing, compared to the 74 percent national value. Distribution for the housing unit was determined to be 14 percent LBP interior only, 23 percent LBP exterior only, and 37 percent LBP on both interior and exterior walls (for a total of 74 percent). Thus lead-based paint was determined to be more common on the exterior of homes. Nonintact paint was estimated to exist in 13.8 million (of 57.3 million) units containing LBP. Of ‘LBP is defined as greater than of equal to 1.0 mgk m 2 , measured by XRF, in a crdance with the Federal Standard for LBP eatablisbed in Section 566, Houning and Community Development Act of 1987 (HUD, 1990). PrqJ.ct LEAP— Ph... 1 33 ------- the 13.8 million, 5 percent are interior only, 11 percent are exterior only, and 2 percent are estimated to be both interior and exterior surfaces. Thus, 18 percent of the total housing stock is estimated to contain nomntact LBP, defined as exceeding five ft 2 of LBP in a dwelling being defective (HUD, 1990). Further, the paint is estimated to be damaged in 21 percent of units with exterior LBP, and in 13 percent of the units with interior LBP. Citing the interim guidelines developed by the Department of Housing and Urban Development on clearance levels for dust, post abatement’, the authors note that fully 17 percent of the occupied homes that contain LBP exceed the guidelines. Only four percent of the homes free of LBP, in contrast, were determined to have excessive dust-lead. The chance of having excessive dust-lead if lead-based paint exists versus no lead-based paint was thus calculated to be 17:4. Surprisingly, the study found that the incidence of dust-lead is almost as low for homes with interior LBP only, as for homes with no LBP, while the incidence is approximately the same for units with interior or exterior LBP. The study concludes, consequently, that interior dust-lead-contamination is more likely generated by exterior LBP than by interior LBP. The incidence was found to be highest for units containing both interior and exterior LBP. Most dust was found to be located around windows. 2.6.2. Leaded Gasoline Recent consumption of leaded gasoline in the United States alone shows that in the 10-year period from 1975 to 1984, inclusive, this country consumed 654.6 X iO gallons of gasoline, resulting in the dispersal of 1,087.8 X io metric tons of lead in the U.S. (ASTDR, 1988). “Gasoline lead makes a sizable contribution (about 90 to 95 %) to the total atmospheric lead burden in developed countries such as the United States” (ASTI)R, 1988). From 1975 to 1984, however, U.S. gasoline lead consumption decreased by 73 percent. ASTDR states that studies and data indicate that past gasoline lead consumption resulted in airborne lead that “added significantly to atmospheric and soil/dust/food burdens, and that via both direct and indirect rouses, such input contributes 20 to 25% to Pb-B levels.” The pathway can be very 200 pgf& for floors, 500 g/ft 2 for window sills, and 800 ig/ft 2 for window wells (DHUD, 1990). PrqJ.ct LEAP— Pb 1 34 ------- significant as a route of exposure for children, with elevated blood-lead to airborne lead concentration ratios of five to six gIdL Pb-B rise for each p .g/m 3 increase in air-lead concentration. The NHANES II data supports the high correlation between reduction in leaded-gasoline and the decrease in Pb-B levels in the general population (ATSDR, 1988). The methodology utilized for estimating numbers exposed was to restrict the enumeration to the 100 largest cities, where the highest exposures, due to mobile sources (including soil/dust routes of exposure resulting from past deposition), were expected to occur. For the estimated 1984 population of 50,597,300 residents of these areas, 11 percent (5,565,700) are estimated to be children under seven years of age (ASTDR, 1988). ATSDR, estimating the numbers of children falling below criteria Pb-B levels, and then projecting to the year 1990, found that gasoline lead phase down alone, will not be sufficient to reduce all Pb-B levels down to levels considered to be acceptable. The agency determined that, in the year 1990, the numbers of children estimated to be below criteria Pb-B levels, as a result of lead in gasoline phaseout, are 25 gIdL-119,000; 20 jtg/dL-400,000; and 15 p.g/dL-1,252,000 (ASTDR, 1988). 2.6.3. Stationary Sources This nation has 11 lead mines, five primary smelters and refineries, 60 secondary smelters, and 132 plants, the latter for manufacture of lead-acid batteries (ASTDR, 1988). Soil and dust levels near these sources range from 500 to 5,000 ppm, with exponential decreases with distance from the source. A 1977 investigation by Yankel et al. (1977) of one- to nine-year-old children living near a smelter in Silver Valley, Idaho, clearly demonstrates an association of airborne lead concentration as well as (elevated) Pb-B levels with distances from the source. That study modeled the natural log of blood-lead, house dust, soil, age, occupational factors, and air concentration. The researchers determined that 99 percent of the children adjacent (within 1.6 km) to a smelter had Pb-B levels exceeding 40 p.g/dL. Air concentration alone accounted for 55 percent of the variance in Pb-B levels. A CDC study in two smelter communities in Montana (CDC, 1986a) and Idaho (CDC, 1986b) found that the only significant PmJ.ct LEAP— Ph . 1 35 ------- environmental source causing elevated blood-lead levels in children was lead in the soil and house dust, resulting from smelter operations. Thus previous fallout remains a main contributor to elevated Pb-B levels. It is also noted that blood-lead remains elevated even when airborne levels have been reduced to low levels. Consequently, ATSDR (1988) recommends that closed facilities be included in studies, to account for the impact of previous lead emissions and deposition. Based upon previous studies, A1SDR estimates that between 1 and 26 percent of children living near primary lead smelters would exceed the CDC criteria for lead toxicity of 25 p gIdL Pb-B and 35 p .g/dL EP. Four percent of children residing near secondary smelters would also exceed the criteria values (ASTDR, 1988). The report quotes an estimate by the EPA Office of Air Quality Planning and Standards, of 21,000 children exposed via primary lead smelters (within five km of the source), and 187,000 exposed via secondary lead smelters (within two km of the source). Some 25,000 children are estimated to be exposed from lead-acid battery plants (within a one km radius), for a total childhood exposure count of 233,000. 2.6.4. Dust and Soils Brunekruf et al. (1983), in a study conducted in the Netherlands, determined that household dust- lead concentration increases by 40010700 ppm for each pg/rn 3 rise in airborne lead. This was as reported by A1SDR, apparently based upon data presented in the Brunekruf study. The Bruenekruf study found a Pb-B to air-lead concentration of one to two p.g/dL per pg/m 3 . The outdoor measured air values ranged from 0.10 to 0.27 pg/rn 3 . Soil-lead generally was found to be less than 500 ppm. Dust-lead ranged from geometric mean values of 58 to 81 pg/rn 2 for two inner city areas studied, with a range of values from 22 to 740 ppm. EPA (1986a) also reviewed reports on the relationship of lead in soil and dust to Pb-B levels. Generally, the review found that lead in soil and dust of 500 to 1,000 ppm begins to affect Pb-B levels in children (Baker at al., 1977; Mielke et *1., 1984). The Mielke Twin Cities, Minnesota, study of inner city areas found that 50 percent of the individuals with lead poisoning lived in housing containing Prqj.ct LEAP— Pha.. 1 36 ------- soil-lead levels of 500 to 999 .tgfg, and that 40 percent lived in homes with values of >1,000 gIg. The authors cite both house paint and leaded gasoline as contributors. The study also found, from right-of-way soil samples, that lead levels low to high correspond to ligbt to heavy traffic (however, the report did not indicate a p-value or other indication of statistical significance for this finding). Over half the Minneapolis homes in the study had soil-lead levels exceeding 50 tg/g. The variability of measured soil-lead concentrations, sometimes a 100-fold order of magnitude difference in concentrations between the front and back entrance of the home, was noted as a precaution in interpretation of the soils data. Clark et aL (1987) determined an increase of Pb-B by 6.2 p.Lg/dL for each 1,000 ppm increase in soil-lead concentration. Studies (EPA, 1986a) show a range of values, from 0.6 to 6.8 1 tg/dL rise in Pb-B level for 1,000 ppm incremental increases in soil-lead concentration. Recognizing that soil/dust information was not available at the time of the report (beyond a limited number of site specific studies), the (ATSDR, 1988) report authors recommend the use of multiple linear regression analysis to account for different contributions to a child’s Pb-B level, preceded by a representative sampling of dusts and soils from the urban and rural areas of each of the nation’s four major regions. Because such a statistically based representative sampling program was not available at that time, the report used an admitted overestimate of exposure by combining the major routes: paint lead in pre- 1940 housing with the highest lead content - 5.9 million children; gasoline lead in the 100 largest cities - 5.6 million; and stationary sources - 0.2 million, for a total of 11.7 million exposed children (ASTDR, 1988). A reliable method to apportion Pb-B values to primary contributors was called for by ATSDR. That call was answered in part by the Comprehensive and Workable Plan (DHUD, 1990) that derived soil/paint correlations, and speculated about the contributions to elevated blood-lead levels. From multiple regression and pathway analyses, the HUD report determined that excessive dust-lead levels occur more often in houses with LBP (intact or not) than in housing without LBP. Elevated blood-lead levels were associated more often with housing with nonintact LBP on exterior walls, than with intact exterior PrQJ.ct LEAP— Ph 1 37 ------- LBP. HUD concludes that young children in homes having nonintact LBP, or excessive dust-lead, are at highest risk. Of 57 million occupied homes having LBP, less than 10 million are occupied by families with children under the age of seven. Of these homes, 3.8 million units have high dust-lead levels or nonintact paint. According to the survey report, soil-lead is within the guidelines 5 of 500 ppm, 79 percent of the time that LBP is present. The analysis estimated the numbers of occupied dwellings with soil-lead associated with the presence and condition of exterior LBP. The percentages of homes exceeding the guidelines by strata were estimated to be 6 percent for homes with no LBP, 21 percent for homes with intact LBP, 48 percent for nonintact LBP, and 27 percent with homes containing any exterior LBP. Overall, 18 percent of 63 million occupied housing units are estimated to have soil-lead exceeding 500 ppm. A strong statistical association was thus found between the presence of lead-based paint and lead- contaminated soil. The probability of excessive soil-lead was derived as 4:1 for LBP exterior compared to LBP-free exterior. The report analyzed hypothesized pathways from paint to dust, by determining the correlation coefficients between the natural logarithms of the pairs of survey measurements of lead associated with a pathway. The correlation coefficients determined were paint-on-wall:dust-on-window-sill——0.25; dust- on-window-sill :dust-in-window-well-—O.46; dust-in-window-wethsoil-at-drip-line-—O.42-O.45; and soil-at- drip-linesoil-at-remote-location-—O.68. All of the correlations were found to be statistically significant at or below 0.05, with some at the 0.001 level. Thus, if high lead concentrations were found at one location, values tended to be high everywhere. The regression of dust variables on paint variables support the conclusion that paint is one of the sources of lead in dust. Derived R 2 values to discern the fraction of the dependent variable explained by ‘The DHUD report refers so an EPA interim gwddine having a range of values for i l land uinccnsratron of 500 to 1000 ProJ.ct LEAP— Pb... 1 38 ------- the independent variables yielded values ranging from 0.11 to 0.30. From the regression analysis, the authors conclude that lead from exterior paint is brought inside the house, and that lead from interior paint contaminates the soil outside the house. When the age of housing is added to the regression analysis, that variable helped explain lead levels in most of the regressions. The older the dwelling, the higher the estimated lead levels. The regression values ranged from 0.13 to 0.43. HUD postulates that age of the home “merely proxies for lead-based paint”, and that age may measure other sources, such as auto emissions. HUD notes, however, the difficulty in estimating the percent of lead in dust and soil that can be attributed to LBP. From the regression analysis, approximately 20 to 25 percent of the variation in dust and soil-lead is explained by paint variables (HUD notes that this could be low). Consequently, the source of most of the lead in soil is not explained by the model. Thorton c i al. (1990) studied lead in garden soils and household dusts in England, Scotland, and Wales. They found that 10 percent of the floor dusts exceeded 2,000 .tg/g. The two-year-olds and their home environs were sampled for inside dust, soil, road soil, wipes, food and water, and venous blood. The intent of the study was to assess lead intake from dusts in relation to other sources. The study reported a geometric mean for lead in the surface (zero to five cm) garden soils to be 266 p.g/g and for house dust to be 561 tg/g. In London, the mean values were 654 Lg/g for soils, and 1010 j.i.g/g for dust. A highly significant correlation between household dust and garden soil was determined (r= 0.531, p= 0.001, n=4512). Overall, the geometric means were determined to be 11.7 p.g/dL blood-lead; playroom air 0.27 p.g/m 3 ; bedroom air 0.26 g/m 3 ; external air 0.43 p.g/m 3 ; dust soil 424 p.&g/g; soil 313 .tgfg; dust loading 60 Lg/m 2 ; handwipes 5.7 ptg; food and beverage 161 p.tg/weelq and water 19 .tg/l. The study found that the correlation of Pb-B levels with indoor air concentration to be virtually zero. The correlation of blood-lead levels with dust-lead was determined to be r= 0.34, with water to be r= 0.39, and with soil- lead to be r= 0.18. The association with dietary variables was found not to be statistically significant. The researchers used multiple linear regression to assess the relative importance of various sources, PrQJsct LEAP— Pha.. 1 39 ------- with the model log Pb-B ( .tg/100 dl) = 0.55 + 0.10 log xi + 0.14 log PbW ( .tg/l) + 0.07 S, where Pb-B = blood-lead concentration PbW = water lead concentration xi = dust loading x rate of hands touching all objects, and S = 0,1 depending upon whether parents smoked cigarettes or not. The analysis determined that adding air-lead concentrations, soil-lead concentrations, or dietary-lead intake gave nonsignificant regression coefficients, and only marginal improvements to the R 2 value. The study concluded that the Birmingham study for the first time demonstrated a relationship between levels of environmental lead within the home and blood-lead in a two-year-old child. In this country, a comprehensive study was concluded in 1987 by the Minnesota Pollution Control Agency (MPCA) and the Minnesota Department of Health (MPCA, 1987). The report “provides the results of soil testing throughout Minnesota and blood-lead screening of children residing near sites in Minneapolis and St. Paul identified by the MPCA as having at least 1,000 parts per million (ppm) of lead in soil.” The study sampled soils in five major cities and 27 counties in Minnesota, including census tracts in Minneapolis, St. Paul, Duluth, Rochester, and St. Cloud. A total of 2,485 soil samples were taken. Overall, 85.8 percent of the samples were found to be <500 ppm 6 . Only 7 percent of the samples were found to exceed 1,000 ppm. For areas designated as play areas, only five of 564 samples exceeded 500 ppm, and none exceeded 1,000 ppm. For foundation samples (defined as being within five ft of a structure), however, 53 percent (of 413 samples) exceeded 500 ppm, and 31 percent exceeded 1,000 ppm. Surprisingly, only 4 percent (22 samples) of the street side samples (generally, the parkway areas adjacent to the street) exceeded 500 ppm, and only one of 593 samples exceeded 1,000 ppm. As expected, the study found that samples taken from sites occupied by industrial point sources had very high lead ‘The study notes that sod lead levels om vary by >50 percent, depending upon the (laboratory) anal ydcal method used, and also that sod samplea from the same yard may vary by a factor of 100. Comequently, the mean soil concentration values are deemed to be of questionable value. Pr LEAP- Phi..! 40 ------- concentrations. The researchers performed a regression analysis and found a general tendency for streetside soil- lead concentrations to increase with increasing traffic, but judged the relationship to be weak, with 29 percent of variation in streetside soil-lead concentration attributable to average daily traffic count. The study also noted that the degree of contamination in the streetside samples, ostensively from vehicular traffic, is far less than soil-lead concentration along foundations. Maximum soil-lead concentration accounted for little variation (R 2 = 0.0541, p = 0.0060) of measured blood-lead. The average daily traffic count accounted for some of the variation in street side lead concentration (R 2 = 0.2888, p= 0.0001). The report concludes that the relationship between soil-lead and blood-lead appears to be weak. The Minnesota Department of Health (MDH) also conducted a lead screening program for youths aged 6 months to 6 years living near the sites with soil-lead concentrations exceeding 1,000 ppm. Of 743 children screened (742 EP test and 656 blood-lead tests), 13 were determined to have lead toxicity in accordance with the CDC criteria of> 25 j.tg/dL blood-lead and 35 .tg/dL EP. Another 24 had elevated blood-lead (>25 g/dL Pb-B but <35 &g/dL EP), and 65 were determined to have iron deficiency. Twenty percent (134) of the inner city children tested had Pb-B equal to or exceeding 15 .tgIdL. The Minnesota Department of Health noted that the children tested “live in older, poor, inner city neighborhoods dominated by lead painted housing, high traffic density, and the highest residential soil-lead concentrations found in the study.” The average blood-lead level of screened children was 10 p .g/dL, which the study compared to the NHANES 1980 national average Pb-B of 16 p.g/dL for children under five years old. 2.6.5. Drinkmnn Water Most contamination via this source results from domestic plumbing and plumbing in public buildings, including lead pipe connections, lead-based solder in copper plumbing, and corrosive water in plumbing (ASTDR, 1988). Water fountains and drinking water coolers in schools and other public PreJ.et LEAP— Pb ... 1 41 ------- buildings are potentially important sources, as well. EPA (1986a; A1’SDR 1988) and ATSDR (1988) note that lead is absorbed in the human body at 35 to 50 percent from water, compared to 10 to 15 percent from food; consequently, lead in water poses a three to five multiple risk compared to food having the same lead concentration. Lead absorption rates are even higher for children, resulting in even higher increased risks of exposure. EPA (1986b) estimates that 42 million people in the U.S. may be exposed to lead in drinking water exceeding 20 pg 1 at the lap. This is based upon 772 samples from a random grab sampling program conducted in 580 cities in 47 states. Data from this survey indicate that 16 percent of water from U.S. kitchen taps exceed 20 pg / I, noting also a problem of lead leaching from new water connections, that was not considered in the swvey. In addition, the survey did not consider drinking water from water coolers in séhools, another documented potential source of lead contamination. ATSDR (1988) assessed exposure by age of housing stock, considering the use of lead pipes for pre-1920 homes, iron pipes for homes built between 1920 and 1949, the use of lead solder during the period 1950-1984, and that fresh solder may have been used during the two most recent years preceding the ATSDR report, 1985-6. Using this approach, the estimated population at risk is set at 1.8 million children in new housing, and 4.89 million in older housing (assuming 1\3 of the housing built before 1939, or 10 percent of the housing stock) contained lead pipes. The effects of corrosivity are also noted. From the 42 million people estimated to be exposed above 20 p.g/dL (thought to result in an increase in Pb-B levels), 3,780,000 children (9 percent of 42 million) are estimated to be exposed. Levels in drinking water can be high (up to 1,000 p.i.g/l) due to leaching of lead from lead pipe and leaded solder Joints (EPA, 1991a). The concentration varies with the amount of lead in the plumbing and with the comsiveness of the water. Soft or acidic waters tend to be more corrosive, and consequently tend to contain higher concentrations of dissolved lead. An analysis performed for the Environmental Protection Agency, which included public water supply systems’ data for the States of Indiana, Michigan, and Minnesota, indicates that these states have only 1.6 percent of the public water suppliers delivering highly corrosive water (EPA PiuJsct LEAP— Phi.. 1 42 ------- 1988). In general, water with a pH of eight or above and high alkalinity is less corrosive than water with a pH < eight and low alkalinity (highly corrosive) (EPA, 1991b). EPA estimates the water from lead service lines to be 10 gfl for water systems with highly corrosive water, and five g/l for systems with moderately corrosive water (Memorandum, Cohan, 1991). In an EPA (1986b) analysis of the benefits of reducing the lead in drinking water standard, EPA estimated that 241,000 children had blood-lead levels exceeding 15 .tg/dL due to lead in drinking water (as a result of the action of corrosive water on aged piping), including 11,000 having Pb-B levels exceeding 30 .tg/dL. 2.6.6. Lead in Food Lead enters food processing mainly through lead-soldered cans, which practice was to be phased out beginning in the late 1970s (ASTDR, 1988). Studies have found varying levels of lead intake in children, based upon foods consumed. Recognizing the centralized food distribution in this country, that all children (indeed the entire population) are exposed via this route, ATSDR estimates that 9 percent of the 1985 population, or 21 million children, are exposed by food intake. By making a series of assumptions and relying on the results of previous surveys, the report estimates that a maximum of 5 percent of children five months to six years of age are “at or approaching a dietary lead exposure that pushes their body burden close to that associated with early toxicity if they are also exposed to other typical lead sources” (ASTDR, 1988). The continual decline of lead in food, however, is noted, along with a myriad of uncertainties associated with the five percent estimate. By excluding children zero to five months of age, the population estimate of 21,405,000 (citing the World Almanac 1987) is reduced to 19,474,000. A 5 percent exposure rate would then result in 973,000 children at risk, based upon an Pb-B increase of 10 g/dL 2.7. Special Concern For Exposure Řf The Fetus To estimate exposure of the yet-to-be-born, ATSDR considered women in SMSAS of childbearing age for the year 1984, with four strata: white and African-American women, and age ranges 15-19 and Pr J.ct LEAP— Phase 1 43 ------- 20-44. Using estimated prevalence and logistic regression to extend NHANES II Pb-B levels in the general population to the year 1984, the authors then applied the prevalence to the four strata for 1984. The resultant geometric mean Pb-B levels were 3.4 j.tg/dL for white females 15-19 yeais of age; 5.2 tg/dL for white females 20-44 years old; 5.1 xg/dL for African-American females 15-19 years old; and 7.3 ig/dL for African-American females 20-44 years of age. An estimated 41,300,000 females are thought to be in the four strata. ATSDR estimates that 3,595,000 could be pregnant (in a given year), with 403,200 (at risk annually, for fetuses of white and African-American women living in SMSAs) having Pb- B levels 10 .ig/dL; 69,400 15p.g/dL; 14,500 20 p.g/dL; and 3,800 25 g.tg/dL. The report acknowledges both overestimation and underestimation errors due to the limitations of methodology, data availability, and assumptions. Estimations were not calculated for individual SMSAs. 2.8. Special Emphasis: Ethnicity A crucial finding of the ATSDR (1988) study is the substantial difference in estimated prevalence of blood-lead levels based upon ethnicity. The Agency provided projected percentages of children 6 months to 5 years old expected to exceed 15, 20, and 25 pg/dL Pb-B, who live inside central cities of Standard Metropolitan Statistical Areas with populations greater than one million. The starkest difference is at the lower socioeconomic level with annual family incomes of less than $6,000. For African- American children, an astounding 68 percent are projected to exceed 15 .ig/dL Pb-B, compared to 36 percent for white children. A difference is indicated across all socioeconomic strata. For annual family income exceeding $15,000, 26.6 percent of African-American children are projected to exceed 15 p.tg/dL , contrasted to 7.1 percent for white children. This is compelling evidence of an increased exposure risk for African-American children. Because the projections rely upon NHANES II data, a similar comparison was not provided for Hispanic children. Data were not available from NHANES for such analyses. It is plausible, however, given similar socioeconomic circumstances of the African-American and Hispanic population, that Hispanic children could also be at increased risk of the harmful effects of low-level lead PruJ.et IL P- Phiae 1 44 ------- exposure. There is, moreover, city specific analyses, based upon blood-lead screening programs, to support this contention. Evidence that suggests that elevated blood-lead values are a significant concern in the Hispanic community as well as the African-American community is based upon screening programs, rather than upon epidemiological studies. Fernadez et a!. (1990) studied the demographic patterns of 485 lead- poisoned children in the City of Chicago. Ninety-four of the cases studied were minority. Their analysis indicated that African-American and Hispanic children are disproportionately affected by lead poisoning. The study found that 69 percent of the cases were African-American children, and that 25 percent of the children were Hispanic. In contrast, chicago’s African-American population is 41 percent, and the Hispanic population is 17 percent. The analysis found that even in community areas with a “fairly even racial/ethnic composition”, African-American and Hispanic children suffered disproportionately from lead poisoning. The authors noted limitations in the analysis. The sample data were not representative of the entire population of the city. True incidence could not be calculated. Further, the data was sometimes incomplete. The relationship to socioeconomic characteristics of the neighborhoods studied was also noted. Data from the Minnesota Department of Health 1986-87 Blood Lead Survey supports this conclusion (Memoranda, Benson, 1991). The survey was conducted for 451 children in Minneapolis and 584 children in St. Paul. Data for St. Paul indicated an average blood-lead value of 9 g/dL for African- American children, and 7 .i.g/dL for both Hispanic and white children. The population size for the latter, however, was quite small at 7 children. For Minneapolis, the average values were 9 tg/dL for African- American children also, but 8 .tg/dL for white children, and 12 p.g/dL for Hispanic children. Overall, for five Minnesota cities including Minneapolis and St. Paul, the analysis determined the percent of those screened exceeding 10. j tg/dL blood-lead. The percentages were 33.3 percent for African-American children, 25.7 percent for white children, and 43.9 percent for all others (including Hispanic, but excluding PraJ.ct LEAP— Phase 1 45 ------- American Indian children). The author notes also that the screened children are not necessarily representative of the entire population of the cities. The Public Health Service (PHS), in Healthy People 2000 (Health and Human Services, 1991a), has set an objective to “Reduce the prevalence of blood-lead levels exceeding 15 p.g/dL and 25 p.tg/dL among children aged six months through five years to no more than 500,000 and zero, respectively.” The baseline for the objective is an estimated three million children with Pb-B levels exceeding 15 .tg/dL, and 234,000 children with Pb-B levels exceeding 25 tg/dL, in the year 1984. The 1984 baseline of inner-city low-income African-American children (having an annual family income <$6,000 in 1984 dollars) was 234,900 exceeding 15 .tg/dL (with a year 2000 target objective of reduction to 75,000 children), and 36,000 children exceeding 25 g/dL (with the corresponding year 2000 target objective of a reduction to no children). The Public Health Service refers to this as a special population target. Such a special emphasis is supported by the findings of Danford et al. (1982a). Danford and her colleagues found, in a study population consisting of mentally retarded individuals, that 30 percent of African-American children aged one to six years have abnormal ingestion behavior, compared to 10-18 percent in the same age strata for white children. Danford (1982b) notes, however, that interpretation of survey results on the incidence of pica is complicated by several factors, including limitations on statistical methods used, inconsistent definitions of pica, and (statistically) small numbers of subjects. Danford also cites a cultural hypothesis for pica. In some African cultures, the consumption of soil during pregnancy is thought to suppress nausea. She asserts that, “given the deeply ingrained geophagy of the African cultures that supplied the bulk of slaves to the New World, it is not surprising that the practice persists in the black subculture of the United States.” Consumption of lead-contaminated soil, as a conscquence of such practices, would cause elevated blood-lead levels. PHS further, in its Strategic Plan for the Elimination of Childhood Lead Poisoning (HHS, 1991), asserts that “Poor, minority children in the inner cities, who arc already disadvantaged by inadequate PrnJ.ct hAP- Phi.. 1 46 ------- nutrition and other factors, are particularly vulnerable to this [ lead poisoning] disease.” The Strategic Plan focuses heavily upon lead poisoning because of its importance to public health protection. Needleman (1990) speculates about the social cost of exposure, based upon his ongoing study of a cohort of children followed into the 19th year of life. Needleman and David Bellinger had found, when the cohort was in grade five, that the incidence of grade retention at that time was significantly higher in the high (blood) lead group. Further, the attention span of the high lead group was disturbed. Needleman, in retesting 132 of the children, found the relative risk for not graduating from high school, associated with lead, to be 4.8. He asserts that the high lead group in adult years are clumsier, have poorer reading scores, more depression, and higher rates of hard drug use (no statistical presentation, however, was provided in the paper). Further study is to be done. Bellinger et al. (1990) add that “Children already stressed by sociodemographic disadvantages may be less able to weather the additional stress of high prenatal lead exposure.” The Needleman analysis also indicates that lead is associated with increased risk for attention deficit disorder (ADD) (attributable risk of 0.51), and that attention deficit disorder in turn is a risk factor for antisocial behavior. Needleman determined the attributable risk for antisocial behavior, given ADD, to be 0.58. Using these findings, he postulates a joint probability of delinquency, given lead exposure, to be that 20 percent of (juvenile) delinquency is lead-associated. He is currently examining this relationship. 2.9. Research Needs The Public Health Service (HHS, 1991b) calls for research studies to determine the relative contributions of various pathways of lead to children’s blood-lead levels, particularly from paint, dust, soil, air, food, water, parental occupations, and hobbies. HHS notes particularly that the dietary contribution of lead in calcium supplements, especially when consumed by pregnant women, should be assessed. The ATSDR (1988) report aptly deScribes the current situation on lead exposure: “A: the same P J.ct LEAP— Pha .. 1 47 ------- time that progress is being made to reduce some sources of lead toxicity, scientific determinations of what constitute safe’ levels of lead exposure are concurrently declining even further. Thus, increasing percentages of young children and pregnant women fall into the ‘at-risk’ category as permissible exposure limits are revised downward Accompanying these increases is the growing dilemma of how to deal effectively with such a widespread public health problem. Since hospitalization and medical treatment of individuals with Pb-B levels below approximately 25 &g/dL is neither appropriate nor even feasible, the only available option is to eliminate or reduce the lead in the environment” (emphasis added). In concluding its report to Congress, ATSDR (1988) cites the need for comprehensive studies, at the regional level, of the impact and geographic distribution of lead sources upon exposed populations. The need to eliminate low-level environmental sources of lead is clear. Far too many children are still exposed to concentrations of lead in dust and soil that cause unacceptable blood-lead levels. A lessor number are exposed to excessive air-lead and lead in drinking water. The relatively higher risk that confronts African-American and Hispanic children, compared to the general population, is also apparent. It is uncertain, however, where these children are located, and in what numbers, due to such environmental exposures. Gathering actual data for all environmental pathways of exposure for the entire population is neither practical nor feasible. Even creasing such a data base for the much smaller minority childhood population would be a daunting task. Consequently, as an alternative, a population screening methodology So guide public health officials to geographic areas where children are at high risk, is needed. Prqjsct LEAP— Pb 1 48 ------- 3. STUDY OBJECI1VES Children under seven years of age having low blood-lead levels, resulting from environmental exposures to lead from multiple pathways of exposure, experience a significant health threat. Further, the danger posed to specific communities within the Midwest region of the nation, is oftentimes not detected via either environmental monitoring of exposures to lead and lead compounds, or via biological measurements such as ascertainment of blood-lead levels. Consequently, large numbers of children at risk to low level exposure to lead are undetected and thus, unprotected. OBJECTIVE 1: Develop a population comparative risk approach for estimating the number and location of African-American and Hispanic children under seven years of age, at risk of exposure to lead with blood-lead levels exceeding 10 tg/dL Include a “hot spot” selection scheme that accounts for all known routes of environmental exposure to lead. OBJECTIVE 2: Conduct an analysis to ascertain the predictive ability of the approach for selecting “hot spot” areas, by comparing modeled blood-lead levels to measured blood-lead levels. OBJECTIVE 3: For a selected city, examine the association of elevated blood-lead levels with proximity of children to transportation corridors (lead exposure due to historical deposition of lead in gasoline and/or current emissions from mobile sources). FrqJsct LEAP— Pb. 1 ------- 4. METHODOLOGY 4.1. Study Scope and Methodology Overview In 1987 the USEPA published a document entitled ‘Unfinished Business”, which provided a best professional judgment review of agency programs and environmental problems from the perspective of comparative risk. Since that time, each individual medium program office at USEPA headquarters, as well as each of the 10 regional offices, were tasked with development of a comparative risk analysis pertinent to the program or geographic region of concern. The intent of the approach was to discern and prioritize environmental problems affecting human health and the environment, to determine whether Agency programs were adequately addressing the existing and emerging environmental concerns, and to assess whether resource shifts (generally at the margin) could impact priority environmental problems that otherwise would not be addressed. The Region 5 office’s comparative risk study was completed in the summer of 1990. Several cross-cutting concerns were identified. Lead was identified by several program areas as one of the multi-program pollutants of concern. The region selected lead as a priority area, and tasked the program managers, and a project director, with development of a comprehensive strategy and implementation plan to address and remediate lead contamination in the six state region. The group recognized that lead poisoning in children is now considered to be a national epidemic by many in the public health community. Lead exposures from exterior and interior residential paint, in particular, as well as exposures from contaminated soils and dust in and around structures present in most urban areas, drinking water, air emissions, food, occupational settings, and hobby activities, result in multiple pathways of exposure. These exposures are responsible for a number of adverse health effects in humans, especially in children. Because children are at elevated risk a targeted population has been chosen to be children under seven years of age. Within this population group, African-American and PrqJ.ct LEAP— Phi.. 1 50 ------- Hispanic children are particularly targeted in recognition of increased body burden susceptibility and thus vulnerability to the uptake and effects of lead exposure. Project LEAP is a multi-media and multi-program approach having four basic components: data analysis and targeting; pollution prevention; education and intervention activities; and abatement activities. The project is being implemented in three phases. It is a component of the Agency Lead Strategy. Project LEAP Phase 1 focuses on data analysis, air modeling of major sources, prioritization of sources and areas for targeting purposes, and selection of geographic areas for attention during the subsequent phases of the Project. Phase 2 will focus upon specific geographic areas of concern with an emphasis upon on-site measurement, e.g., of soil and dust concentrations. Phase 2 will also include continuation of pollution prevention efforts, and initiation of public education and outreach efforts in coordination with other agencies. Phase 3 is envisioned to be actual abatement activities for a selected communicity. Lead exposures from exterior and intenor residential paint, in particular, as well as from contaminated soils and dust in and around structures present in most urban areas, drinking water, air emissions, food, occupational settings, and hobby activities, result in multiple pathways of exposure. These exposures are responsible for a number of adverse health effects in humans, especially in children. Because children are at elevated risk, a targeted population has been chosen to be children under seven years of age, as well as the fetus. Within this population group, African-American and Hispanic children are particularly targeted in recognition of the vulnerability of this population to the uptake and effects of lead exposure. The approach of this effort was to estimate the probability distribution of blood-lead in childhood populations. Determination of severity for each city would then allow for comparisons of geographic areas. For each metropolitan statistical area central city, environmental data were obtained for the major sources/routes of exposure (i.e., point sources of air emissions, municipal waste combusters as a special case categorical source of air emissions, ambient air quality measurements, drinking water supplies, and PrQJ.ct LEAP— Ph... 1 51 ------- operating as well as abandoned hazardous waste sites). Where available, actual concentrations were used. Default values were established for each environmental medium where actual measurements had not been taken. Sensitivity analyses were conducted to assess the impact of assumed (default) values on the blood- lead uptake estimate. Demographic information was obtained from a geographic information systems application (derived and provided by the Geographic Information Systems Management Office, Region 5, EPA). Information was provided at the census tract or community area (aggregation of census tracts) levels for each city. Environmental data (i.e., media concentrations) associated with each tract were provided in order to calculate blood-lead level distributions in affected populations. A single geographical area, Minneapolis, t. Paul, Minnesota, was selected to test the viability of the approach. That area had measured blood-lead levels available, along with pertinent demographic information. A simple correlation analysis was conducted to ascertain whether modeled blood-lead levels were associated with actual measured blood-lead levels. An association would indicate the viability of the approach in comparing cities. Based upon environmental concentrations for each census tract/community area, the Uptake Biokinetic Model (described in Section 5.5) was run to calculate an expected percent lead exceedance for the pertinent area. The percentage, applied against the population data for the tract, provided an estimate of the number of children under seven years of age at risk. Further aggregations of geographic areas provided city totals. 4.2. Study Area The Study area includes 83 cities located in 60 metropolitan statistical areas in the Midwest. These cities represent the central cities in all of the metropolitan statistical areas in the States of Illinois, Indiana, Minnesota, Wisconsin, Michigan, and Ohio. Each city is shown, along with selected demographic information, in TABLE IV. PrQJ.ct LEAP—Ph...! 52 ------- TABLE IV Metropolitan Statistical Area Central City Demographi& — : I Whiic African. flispani Bittt Rock Island 43,720 82.48 - 15.18 3.36 F 695 15.3 II Moline 44,500 95.64 1.17 5.38 667 14.5 Chicago 3,005,072 49.59 39.83 14.05 53912 18.0 Kankakee 27,220 70.53 28.19 1.08 543 19.1 Peoria 110,290 81.49 16.69 1.39 1931 16.5 Bloomington 46,250 92.80 5.70 1.38 860 18 5 Normal 36,790 91.99 6.07 .79 368 9.8 Champaign 59,180 84.52 12.74 1.23 800 13.3 Urbana 35,770 84.08 9.99 1.76 560 16.4 Rantoul N/A 8 Springfield 100,290 88.04 10.79 .66 1817 17.9 E. St. Louis 49,470 4.16 95.56 .94 1474 28.7 Granite City 35,150 98.76 .20 1.61 560 15.7 Rockford 135,760 84.27 13.19 2.89 2294 16.8 Ibta1 $třte otiUb a ř < >. ij11.i .u Gary 136,790 25.16 jj;j .:•: .: . . 70.84 q :.:• . .. . .. .. ... . . :.:... . 7.10 2574 . 18.0 Hammond 86,380 89.48 6.40 8.30 1224 13.7 E. Chicago 36,950 47.85 29.66 42.27 617 16.6 South Bcnd 107,190 79.50 18.29 2.36 1862 17.4 Mishawaka 41,400 97.93 1.08 .71 602 14.6 44,180 86.02 12.56 1.28 866 20J Goshen N/A ‘Sour : County and City Data Book 1988 U.S. Department of Commen , Bureau of the Census. ‘N/A not available. Data w not attainable for theae cities. II I’ II Pi j.ct LFAP— Ph.. 1 53 ------- I Afr 1can ’ . . Hi sp an ic Bbth . Amencan 1%4 • rate per :1,000 .Po . .. • . Ft. Wayne 172,900 83.24 14.55 2.20 3166 19.1 La Fayette 44,240 97.14 1.63 1.14 848 19.2 Kotomo 45,610 90.57 —_85.65 8.11 1.41 843 18.6 Anderson 61,020 13.70 .63 828 13.4 Muncie Indianapolis 72,600 719,820 89.47 77.10 9.54 21.78 .80 .88 969 12812 13.1 18.0 Terre Haute 57,920 90.11 8.49 .77 891 15.2 Bloomington - 52,500 91.10 4.31 139 688 13.2 Evansville 129,480 90.36 8.83 .49 1954 15.0 New Albany 94.32 5.19 .61 566 14.9 Ibtait State N of Ind iana . 5504 ,000 > t1 US tSP $0084 t4 6 Saginaw 72,470 — 57.37 . 35.55 9.01 1557 21.1 Bay City 39,700 —__94.69 1.79 4.68 701 17.6 Midland 35,890 96.26 139 142 537 14.2 Muskegon Grand Rapids — 39,810 — 186,530 76.04 80.93 21.42 15.73 2.98 3.16 867 3937 21.9 213 Lanc ing 128,980 80.42 13.94 632 2566 20.1 East Lansing — 48,120 90.31 5.22 1.80 404 - 8.6 Flint — 145,590 56.17 —- 41.43 249 3129 21.0 Detroit 1,086,220 34.38 63.07 2.41 18523 17.0 Ann Arbor 107,810 85.10 - 9.33 2.08 1414 13.1 Battle Creek 54,080 75.00 22.79 1.90 948 17.4 Jackson 36,970 82.45 15.43 2.03 705 18.7 Kalamazoo — 77,230 81.40 15.60 1.87 1416 18.3 1 Prujiet LW— Phase 1 54 ------- City .i•.. • : •: i• I MOOrhead :f .. : . •• Population . : •• • :• • :: I I ; 28,360 e I 97.80 % Anie ican ---- - - - .46 % Daa • . — : 1.02 total Birth B 1 4• . i,ocx • 426 14.6 Duluth 82380 96.98 .83 .45 1298 15.2 St. Cloud 42,850 97.70 33 .44 718 17.1 Minneapolis 356,840 87.30 7.66 1.26 6301 17.6 St. Paul 263,680 90.01 4.92 2.91 5040 19.0 Rochester lbt*t State orMbrneaot - 58,130 4075,970 97.37 96J6 .65 131 .71 .79 1297 71 22.3 16 0 Toledo 340,680 80.06 17.41 3.01 5594 163 Cleveland 535,822 53.55 43.80 3.10 10162 18.6 Akion 222,060 76.78 22.23 .65 3451 15.2 Lorain 72,210 79.44 11.89 14.36 1114 15.3 Canton 87,110 83.06 15.99 1.31 1508 16.9 Steubenville 23,580 84.72 14.25 .72 319 13.1 Wheeling N/A Marietta N/A Youngstown 104,690 64.43 33.34 332 1641 15.2 Warren 52,900 8112 1&13 .66 934 17.3 MansfIeld 51,340 83.08 16.05 1.11 955 18.4 Lima 45,990 78.72 20.41 1.10 881 19.1 Dayton 178,920 62.05 36.89 .86 3535 19.5 SpringfIeld 69,500 81.87 17.24 .73 1158 16.5 Columbus 566,030 76.24 22.11 .82 10406 1&4 Hamilton 65,050 91.75 7.16 .69 1208 18.9 Middletown 46,090 8&00 11.57 .46 818 18.7 Cln nn 369,750 65.15 33.85 .78 7312 19.7 PrqJ.ct LEAP— Pb. 1 55 ------- Eau c laire 54 ,580 9&66 .25 .38 793 14 3 Wan 3Z24 0 98.80 .07 .30 519 1 6 3 Green Bay 93 ,470 97 .25 .25 .68 1542 17.1 Oshkosh 51 ,190 98.35 . 5 9 32 741 14.8 Neenab N/A M i lwaukee 605,090 73.34 r io 4.10 11800 19.0 Racinc 82 ,440 81.91 14.74 .34 1642 19.7 Kenosba 74,960 93.89 3.62 4.00 1232 16.3 Madi son 175,830 94 .33 2.70 1.31 2580 15.1 Jancsville 51,790 98.95 .22 .71 901 17.5 Beloit 33 ,760 86.99 1130 1.00 583 17.1 LaGossc 47,650 98.75 .29 .48 710 14.9 Sheboygan 47,410 98.28 .12 1.60 812 17.0 App le ton •_ . — - $%.) ipt i ; 98 .27 : 3 .C ____________ .08 .55 cL . a I 1 052 • 1&9 731$1 Png.ct LEAP— Pbs. 1 56 ------- 4.3. Conthbution to Childhood Lead Levels From Air Emissions 4.3.1. Industrial Source Complex Long Term Model Air-lead concentrations resulting from significant point sources were estimated using the Industrial Source Complex Long Term (ISCLT) Model, Personal Computer Version. The model is an advanced Gaussian plume model that uses the steady-state Gaussian plume equation for a continuous source to calculate concentrations for point sources. The model uses statistical wind summaries to calculate seasonal or annual concentration values, and a wind- profile exponent law to adjust the observed mean wind speed from the measurement height to the emission height for plume rise and other parameters. Plume rise is calculated due to momentum and buoyancy as a function of downwind distance for stack emissions. Pasquill’s method is used to account for buoyancy induced dispersion. The ISCLT requires input data arrays of the joint frequency of occurrence of wind speed and direction for each Pasquill stability category and season (when the season option is selected); an array of the mean ambient air temperatures as a function of stability category and season; and an array of the median mixing layer heights as a function of wind speed, stability category, and season. Source specific information needed includes emission release rate, stack height and diameter, gas exit velocity, and gas exit temperature. The “regulatory default” option of the model was selected for the analysis. The regulatory default option includes final plume rise at all receptor locations, stack-tip downwash, buoyancy induced dispersion, default wind profile coefficients, default vertical potential temperature gradients, and revised wake effect procedures. The particle size distribution was added to the model, in accordance with Agency recommendations (Rothblatt Memorandum, “Refined Metals Lead Modeling Analysis, December 8 1989), as shown in Table V. This particle size distribution provides a better estimate of the actual particle sizes expected, in comparison to the default particle size distribution In the ISCLT model Project LEAP— Phase! 57 ------- TABLE V Particle Size Distribution Input to Industrial Source Complex Model 1 . .. :. ..: Mean Mass Diameter ..:: .... .079 ScuIin Velocity .000129 Setthn .:. .237 Reflection Co 1.0 4.08 .00363 .157 1.0 11.1 .0262 .68 20.4 .0877 .20 .52 30.27 .194 .16 .26 40.19 .342 .12 0 50.15 .532 .08 — 0 60.11 .764 .04 0 Meteorological input arrays were obtained by the following process. Surface meteorological data files were obtained from the National Climatologic Data Center weather monitoring stations closest to the source to be modeled, along with upper air data files. A “STAR” (Stability Array) program was run on each set of meteorological data in order to convert the data into the format used by the !SCLT model. The STAR program converted the raw meteorological data into the proper format required by the ISCLT model for the joint frequency of occurrence of wind speed and direction for each Pasquill stability category A through F. A file of hourly temperatures was also created. The temperature and upper air data files were further processed using a statistical analysis program to derive the mean air temperatures and mixing layer heights, and to calculate the median mixing heights by stability and wind speed categories. The statistical analysis program was used as a convenient method for calculating mean and median values, and for sorting data for use by the ISCLT program. The three data arrays were incorporated into a single file specific to each source. PrqJ.ct LEAP— Ph. 1 58 ------- Emission rates are those reported in the Toxics Release Inventory data base for 1988 (raw data was retrieved from the U.S. EPA national computer center). Steady-state emissions were assumed for the source to calculate a gram per second emission rate from the annual loading. The Aeorometric Information and Retrieval Facility Subsystem (AIRS-FS) was utilized as the data source for facility specific information on physical properties of emission releases, that were required to model emissions at 17 selected facilities. As an approximation, multiple stacks were combined into a single stack by weight-averaging emissions. For modeling purposes, the derived stack height, stack diameter, temperature of gas at release, and gas exit velocity were used, together with the Toxic Release Inventory (TRJ) reported emission rate for the facility. The AIRS-FS also contained emission data for 532 sources of lead emissions. That information, however, was deemed inappropriate for use by the project. Much of the data had been estimated. The estimation factors are currently being updated. The use of the quantity of lead data in AIRS-FS, based upon the existing emission factors, would provide questionable results. Consequently, quantity emission from the TRI data base was used. Stack height, stack diameter, and stack gas exit temperature were obtained from the AIRS-FS data base for eight of the 17 sources: LaClede Steel Co., Alton, Illinois; Chemetco, Inc., Hartford, Illinois; Refined Metals Corp., Beech Grove, Illinois; Quemetco, Inc., Indianapolis, Indiana; Inland Steel Co., East Chicago, Indiana; Kohler Co., Kohier, Wisconsin; Gopher Smelting and Refining Co., Eagan, Minnesota; and North Star Steel, St. Paul, Minnesota. The ISCLT model was run for these sources, and for the additional seven sources using default values. 4.3.2. ISCLT Sensitivity Analysis Recognizing that upper air data (used to discern mixing heights in the model) are available for a very small number of weather stations (Flint, Michigan; Dayton, Ohio; Green Bay, Wisconsin; and St Cloud, Minnesota) for 1988, the ISCLT model was run using each, and the resulting lead concentrations compared at the four locations. The locations included x and y coordinates (-2000,2000), (0,2000), Project LEAP— Pha.. 1 59 ------- (200,0), and (2000,2000) in meters (Figure 1). The Flint, Green Bay, and St. Cloud upper air stations (resulting in three observations for each x,y coordinate location) provided extremely close air-lead concentrations, as shown in Table VI. Running the ISCLT model with data for the three upper air stations results in a mean value of 1.693 .tg/m 3 air-lead. The standard deviation of 0.004 is quite small. Consequently, the choice of upper air station for inclusion in the modeling of air-lead concentiations is basically irrelevant, especially given the other assumption made in order to run the model. Green Bay, which provided values close to the mean values, was selected. ISCLT Point Source Grid Oist.aioa From Point Sour.. 2 00C 1580 - (-2000.2000) - (200,2000) (2000,2000) - 1000 . ._..... . .. . . -2 -1 o (2OO,0) 2 3 Distance From Point Source (Thousands) Rscsptor Ps$t 1801.1 I.dus*. Sour.. G,.u$sn Long Tirm) Figure Industrial Soune Complex Long Term Point Source Grid PrqJsct LEAP— Ph e 1 60 ------- TABLE VI Upper Air Data Analysis Analysis Variable: Lead Concentration Locat ion of ;y No of Observations Mtnzmwn ( j ig/tn ’) Maximum ( j i g /rn 3 ) Mean 4tg1&) Standa id Deviation coordinate 1 3 0.043570 0 .050765 0.047196 0.003630 2 3 0.111866 0.130074 0.120823 0.009107 3 3 1.693079 1.700655 1.697299 0.003861 4 3 0.033252 0.041245 0.037243 0.003996 A similar series of model runs were conducted with varying source specific inputs of exit gas temperature, stack diameter, and stack height, recognizing that those parameters were not available for all sources. A model default of stack temperature of 432 degrees Kelvin, stack diameter of 2.4 meters, stack height of 35 meters, and stack gas exit velocity of 11.4 meters/second were compared to varying inputs for a source at the same emission rate. The modeled lead concentrations, for a selected point -200,0 meters west of the source, are shown in Table VII. TABLE VII industrial Source Complex Long Term Model Run Comparative Analysis I k 111k I Default 432 35 2.4 11.7 0.055 Runi 426 34 1.7 1L7 0.085 Run2 121 22 7.2 11.7 0.044 Run 3 432 35 2.4 35.1 0.023 Pn1frct LEAP—. Ph ... 1 61 ------- Decreasing the stack diameter from 2.4 to 1.7 metets, along with very minor changes to the exit temperature and stack height, results in an increase above the “default” run from 0.055 tgIm 3 to 0.085 p .g/zn 3 at selected grid point (-200,0). Both values indicate that associated quarterly values would be well below the ambient air quality standard of 1.5 .ig/rn quarterly average. A reduction in the stack gas temperature from 432 degrees K to 121 degrees K, along with a lower stack height (35 m to 22 in) and a larger stack diameter (2.4 in to 7.2 m) results in a slight decrease in the concentration from 0.055 p.glni 3 Pb to 0.044 p .g/m 3 . Tripling the gas exit velocity front 11.7 to 35.1 m/sec, while holding all other parameters constant, results in a decrease in the concentration to 0.023 iWm 3 . Consequently, choosing the default value for a source in place of a source specific exit velocity, where the actual exit velocity at the source is greater, would result in a conservative (i.e. higher) estimate of air-lead concentrations. 4.3.3. Ambient Air Data Lead concentrations in the ambient air are reported as part of the National Ambient Monitoring SysterniState and Local Monitoring System (NAMSISLkMS). The network of monitoring stations is administered by State and local agencies. Monitors are sited to ascertain compliance with criteria air pollutant standards, including lead. Although the monitors are not strategically placed to be statistically representative of a geographic area, the measured air quality at the stations do provide an indication of overall air quality in an area. Many arc sited near point sources or in locations expected to experience maximum spatial concentrations. By nature of the siting criteria, many of the Metropolitan Statistical Area cities had actual concentration data for lead. The data quality is excellent, because the EPA conducts a rigorous quality assurance program for the NAMS,SLAJvfS system. Flagged data indicates that the data is of questionable quality. The results of 1988 monitoring data were obtained for project analysis. Where monitored data were available the annual average concentration was used to characterize air quality for a city. The Prqj.ct LW- Phi.. 1 62 ------- program default of 0.20 iWm 3 was selected when actual monitoring data was not available. The ambient data was provided to the EPA Geographic Infonnation Systems Management Office to create a spatial data coverage for comparison to estimated concentrations derived from point source modeling. A summary of the ambient air concentrations is contained in Appendix A. 4.3.4. Air Emissions The Toxic Release Inventory (TRI) data base was utilized as a source of information for point sources of lead emissions, particularly for air emissions. The national computer center was queried for a listing of all releases in the Midwest of lead and lead compounds, and that data was subjected to further analysis. TRJ was chosen because it is the most comprehensive data base available on toxic releases into the environment. The data is provided to the EPA and the states as required by the Emergency Planning and Community Right-to-Know Act of 1986. According to the 1990 report, “Toxics in the Community National and Local Perspectives” (EPA, 1990), an analysis of data quality and completeness found the data to be quite accurate in the aggregate. On-site visits by Agency personnel determined that the total volume of reported releases were just 2 percent lower than corrected figures. The audit found that almost 80 percent of all release estimates were without error. Ii is noted, however, that only two-thirds of the companies nationwide that were required to report, did so. Further, not all manufacturing facilities must report; consequently, the TRI data base does not account for all toxic emissions. 4.3.5. Municipal Waste Combusters Municipal Waste Combusters (MWCs) were analyzed as a special category of potential air emissions of lead. The sources, which are not required to report by the Emergency Planning and Community R.iglit-to-Know Act of 1986, but could have substantially large emissions of lead due to the incineration of lead in the municipal waste stream. The EPA Region 5 Municipal Waste Combuster Coordinator provided a listing of MWCs in the six states. Further information on the facilities was PrqJict LEAP— Phi.. 1 ------- obtained by direct written and verbal communication with State agency personnel and facility operators. The sources were to be modeled to discern air concentrations resulting from operations, if a facility were deemed to be an important source (i.e., operations would be expected to result in a measurable increase in the ambient lead concentrations). 4.4. Drinking Water Data The Federal Data Reporting System (FRDS), operated at the EPA National Computer Center, was accessed for information on violations of the drinking water standard. FRDS tracks community water supplies’ compliance with monitoring requirements, maximum contaminant level (MCL) exceedances, variances, enforcement actions, and population. State agency records were also obtained. Community water supplies are required to participate in a quality assurance program, and to report the results of data analysis to the state agency. Data quality is consequently considered to be excellent. The test results for a city was used, when available. When a non-detect value is reported, one-half of that value was used, in accordance with EPA risk assessment guidelines (EPA, 1989). Otherwise, a UBK program default value of 4.0 i .g/l was used. Corrosivity of drinking water supplies was taken into account to recognize the contribution from lead pipe leads, by assuming a higher value for drinking water in homes built prior to 1949. Housing age data was provided only as prior to 1949. This was done as a substitution for homes built before 1920, which are more likely to have lead-pipe leads). 4.5. Soil and Dust Contributions to Elevated Blood-lead Levels 4.5.1. RCRA and O eratin Landfills There is no central data base to query for Resource Conservation and Recovery Act (RCRA) facilities that currently treat, store, or dispose of lead and lead-based compounds. Consequently, information on facilities operating in the MSA cities was obtained by contacting RCRA program personnel at EPA and the stale agencies. In many cases, facility operators were contacted to obtain more specific ProJ.ct LEAP— Pb... 1 64 ------- information. The TRI data base was accessed to find facilities disposing of lead and lead compounds on site (designated operating landfills). Each site was then characterized as to the potential for human exposure. 4.5.2. Abandoned Hazardous Waste Sites Data A November 1989 listing, Final and Proposed NPL (National Priority List) Sites With Lead, was used to identify NPL Sites in the six states that listed lead as a primary OT major constituent of concern. Specific information on sites located in the MSA cities was obtained from summary sheets and from the more comprehensive reports on file for each facility. 4.5.3. Derivation of Soil and Dust Values Data developed and utilized by the Department of Housing and Urban Development to prepare the Comprehensive And Workable Plan (DHUD, 1990) were obtained to derive values for soil and dust concentrations. Although data were coded by region, which would allow assessing data specific to the Midwest states, the national data base was selected to avoid weakening the representativeness of the statistically based sample. The data were generated under a rigorous quality control regimen, and is of good quality, except for dust concentrations reported in ppm. Problems with use of that information was flagged by DHUI) due to problems in the laboratory. The weight of the filter, upon which dust was collected, could not be accurately measured. Consequently, the values in the data base may not be accurate. The calculated values, however, although flawed, are ordinate indications of dust concentrations, in that lead-dust concentration increases with age of dwelling. The data were therefore judged to be adequate for use when categorized by housing age bands. The validity of this judgement is assessed in the UBK sensitivity section 4.7.1. In order to calculate mean soil and dust concentrations from data in the DHUD data base, several soil and dust sampling locations were selected for analysis. The locations were standard points of reference from which DHUD obtained samples at each home. For example, soil concentrations were Pr LEAP- Pb e 1 65 ------- determined at the front entrance and rear yard locations for each home. Selected locations included dust mass concentrations in ppm at six locations, and dust sample results at six locations in sgfff. A statistical program was run on these to obtain minimum, maximum, mean, and standard deviations of lead, by housing age category. The calculated values from two locations for each home, one for dust and one for soil, were used to characterize soil and dust concentration for each census tract. These values were used for inclusion in the Uptake Biokinetic (IJBK) Model. Results of a statistical analysis of selected variables from the National I-lousing Survey, including derived variables used to calculate soil arithmetic means and geometric means for dust concentration samples, are shown in Appendix B. Dust concentration in ppm at the common-entrance location (of all the houses surveyed), and soil concentration at the dwelling-entrance location, were selected for use in the study. Values are shown in Table VIII. These values, rounded and prorated to reflect actual housing counts in each area, were used in the UBK model as the soil and dust concentrations associated with the age of housing for each census area (see page 84). Table VIII Soil and Dust Concentrations For Pb Based Upon DHUD National Housing Survey Data .— te At t$ i J thse Prqj.et LEAP—. Pbs 1 66 ------- 4.6. Lead Uptake Biokinetic Model The Uptake Biokinetic Model was developed by the U.S. EPA and was validated using blood-lead and associated environmental concentrations for individual children (e.g., soil and dust values for the child’s home). The model has not been validated at the large scale as applied by the study methodology. It is specifically emphasized that this research explores the application of the model on a scale much different from the original design and intent of the model. The effort is to determine whether the UBK model can be used as an effective risk management tool to suggest which areas might have comparatively more children at risk to environmental sources of lead. Thus, previously the UBK model has been used to predict site-specific distributions of blood-lead leveLs in childhood populations in the vicinity of lead point sources. This alternative use is to compare modeled levels between cities and areas within cities. The model uses assumptions regarding behavioral and physiologic parameters that determine intake and absorption of lead from air, soil, dust, drinking water, and lead point sources. Behavioral and physiologic assumptions vary by age of child, and include time spent indoors and outdoors; time spent sleeping; breathing volume; deposition efficiency in the respiratory tract; diet (based upon a national food basket survey, and not specific to the Midwest or to individual cities); and absorption efficiency in the gastrointestinal and respiratory tracts. The Uptake Biokinetic Model PC Version 0.5 (EPA, 1991d) is thus a mathematical simplification of lead exposure-effect relationships. The model uses estimates of exposures to predict the distribution of blood-lead concentrations in populations, for user selectable age ranges of children. This analysis uses the full age range of the model, 0 to 84 months of age, and 10 .i.g/dL as the cutoff point for exceedances. Marcus’ study suggests that the default value Geometric Standard Deviation (GSD) of 1.42 used by the UBK model, based upon the nationwide NHANES II study, may be too small (Marcus, 1991). In U.S. communities having much lower blood-lead values, and where there are a diversity of lead sources, Projict LEAP— Ph ... 1 67 ------- for some children in smelter and mining towns indicated a range of unadjusted GSD values from 1.67 to 1.79, in three very disparate types of communities (Marcus, 1991). The communities assessed were Kellogg, Idaho; East Helena, Montana; Leadville, Colorado; Telluride, Colorado; and Midvale, Utah. Marcus analysis calculated both raw and adjusted GSDs for these communities. He determined, for purposes of his analysis, that a GSD value of “... 1.66 fits neatly between the maximal raw GSD and the minimal adjusted GSD in all cases ...“. Although the default value for the geometric standard deviation (GSD) of blood-lead values is 1.42, appropriate for point sources of lead, this analysis uses a standard deviation of 1.7. The wider value is more appropriate for area sources of lead. In addition, the wider GSD better reflects the uncertainty of the spread in blood-lead data values for a population. Table IX provides the UBK model default values for indoor air concentration, diet (based upon Food and Drug Administration National Food Basket Survey for 1988), soil and dust, and paint. The UBK model Calculated Blood Pb and Pb Uptakes shown in Table X are those derived from the default values shown. The associated Figure 2 displays the probability density function for the selected age range of children, along with the percentage of the population expected to exceed the cutoff value. The probability distribution function is a mathematical representation of how blood-lead levels would be distributed in a given population. The impact of various assumptions regarding lead concentration of environmental sources (air, water, soil, and dust), as well as selection of a GSD value, is assessed in the following section. Piidect LEAP-. Pha.. 1 68 ------- TABLE 1X 9 Uptake Biokinetic Model Default Values I I ABSORPT ION METhODOLOGY bocar Absoq*wn I - T r 4 9 I I1i CONCEWiBATiON 0 .200 &g Pb/ms f agj . . . . . ... . . . : I Indoor Air Pb conceu ttation: flU % of outdoor ; . . . : R ! . ‘ : I Ot licr Ak ?ar*incta$: . . . : : . . :. I s I . 1 U I I I - 0 -1 1.0 2.0 32 .0 1 -2 2.0 3.0 32.0 2 -3 3.0 5 .0 32.0 3 -4 4 .0 5 .0 32.0 4 -5 4.0 5.0 32.0 5 -6 4D 7.0 32.0 6 -7 4.0 7.0 32.0 r L L ‘J I jflY m I L— 0 - 1 200.0 200.0 1 -2 200.0 200.0 2 -3 200.0 200.0 3 -4 200.0 200.0 • 4 -5 200.0 200.0 5•6 200.0 200.0 — 6 -7 200.0 . 200.0 II II II II I I I H i i I I I I 11 I I H H I I I H 0 Based upon program default va!ua in the U Bicidnefic Model Piidnt LEAP— Pbs 1 69 ------- TABLE X Model Default Blood-lead and Lead Uptake: :..C fS . . :: . :: jSlO oO: ItWl :{jAg?4.W ..::.:.:. 1 . 1 i.. : I . .::. ;::ô W4a Y ).: •. •:; •. .. WnJtSUptJkt 03-1 3.30 9.38 — 6.00 1-2 3.01 10.03 - 6.00 2-3 2.98 10 .56 — 6.00 3-4 3.04 10.48 6.00 4 -5 3.12 10.41 — 6.00 54 3.15 10.72 — 6.00 6-7 —... 3.18 ....... 11.11 . . . . . . . . . . . . . . . . . 6.00 —. Uptake Y ear : . . .:: :: i i? DM Uptake 34 Water Uptake ;:ç. . .:::4 j !RM SJ ) Paw Uptake : : . :f .: ($j /4 y ) Air .:... 03-1 2.94 0.40 0.00 0.04 1-2 2.96 1.00 0.00 0.07 2-3 3.40 1.04 0.00 0.12 3-4 3.29 1.06 0.00 0.13 4-5 3.18 1.10 0.00 0.13 5-6 3.38 1.16 0.00 0.19 6-7 3.74 1.18 0.00 0.19 Based upon information pertinent to each medium pathway/source contribution, as discussed previously, the model was run for subareas within each of 83 cities, at the community area level (an aggregation of census tracts). The population age range for the model runs was zero to 84 months of age. The “percent above” percentages were then used to calculate numbets of children expected to exceed 10 ,LWdL Pb-B. PnJ.et LEAP— Pbs. 1 70 ------- ii IS 12 14 LI D OI4CV4TJIRT ION C ug.i’dL) S to 54 Hon FIGURE 2 Uptake Biokinetic Model Default Concentration Curve Prq .ct LEAP— Ph 1 71 ------- 4.6.1. UBK Sensitivity Analysis A sensitivity analysis was conducted in order to ascertain the impact of various assumptions made in conducting the study, and to provide a sense of how a range of environmental media concentrations affects the blood-lead level outputs from the model. The analysis considered various concentration levels for drinking water, soil and dust, and outdoor air, as well as different geometric standard deviation values. For each model run, except for the parameter of concern, all other values were held constant at the model default values. Table X displays how varying environmental media concentrations affect mean blood-lead concentrations. The model runs use a Geometric Standard Deviation (GSD) of 1.7. Associated Figures 3, 4, and 5 show the probability density functions for each medium. Table X I summarizes information from the figures. The blood-lead levels vary only slightly when drinking water concentration is increased from 0 pig/I to 4.0 pig/I. The latter is the model default value. Consequently, drinking water concentrations in these ranges would be expected to contribute only a small amount, as an environmental pathway, to the numbers of children expected to exceed 10 p&g/dL Soil and dust, however, can contribute significantly. Particularly for housing built prior to 1949, the concentrations of lead in soil and dust, due to historical deposition as well as lead-based-paint, can result in high numbers of potentially exposed people. To assess the impact of using a value for dust-lead concentration that may not be accurate, the model was run using the same soil concentration as dust concentration. This was in order to compare an assumption of soils concentration equals dust concentration, to the use of the calculated dust concentrations in combination with the calculated soils concentration. Using the same soil as dust value versus the calculated dust value of 565 ppm does not change the percent expected to exceed 10 ig/dL Pb-B. Therefore, for the oldest housing age category, there is no impact of using 565 ppm versus 555 ppm soil-dust, each together with a soil lead concentration of 555 ppm. For housing age category 1940 to 1959, using a 175 ppm dust value in place of the PrqJuct LEAP— Phi.. 1 fl ------- calculated dust-lead concentration of 55 ppm, results in an elevation in the percent expected to exceed 10 g/dL of less than 1 percent. For the newest housing age category, substituting a dust-lead concentration of 90 ppm for the calculated dust-lead value of 20 ppm, results in virtually no change in the percent estimate of exceedance. Thus the model is not sensitive to using the calculated dust-lead values. Running the model for dust alone (with model default values for air, drinking water, and diet) indicates that, except for the highest dust-lead concentration, there is minimal contribution to the percent expected to exceed the criterion Pb-B value. For housing age category prior to 1940, the modeled percent exceedance of 4.55 percent indicates an increase of slightly more than 3 percentage points associated with increasing dust concentration from 200 ppm to 565 ppm. Compared to the model derived percent exceedances when soil and dust values are both held at 0, essentially the full 4.55 percent is associated with the dust concentration of 565 ppm. Outdoor air, at low levels (generally expected, except where a significant point source is in the vicinity of a population), is not expected to contribute greatly to an increase in Pb-B levels. When the air quality standard is greatly exceeded, however, as may be caused by a point source or lead-contaminated dust, the Pb-B levels of nearby residents (within one to two km) are expected to rise significantly. As indicated in the table, an air concentration ten times the standard would result in a percent exceedance of almost 13 percent. The model is very sensitive to the choice of blood-lead GSD for the population. The GSD model default value of 1.42, thought to be applicable to point sources of lead, as mentioned, results in a 0.05 percent exceedance of 10 p.g/dL A GSD of 1.7, selected as discussed earlier, results in a 1.44 percent exceedance, while a GSD of 1.8 results in an even greater exceedance of 2.47 percent (Figure 6). At higher input levels of environmental concentrations, the spread would be even more dramatic. When applied to a population, for example of 1,000 children, the number of children expected to exceed 10 &g/dL for the model default concentrations would vary from 0 to 14 to 25 children at GSDs of 1.42, 1.70, Project LEAP— Ph.. 1 73 ------- and 1.80, respectively. Consequently, the choice of a GSD value for the model has a significant effect upon the estimated number of children at risk of elevated blood-lead. PrQJsct LEAP— Phi.. 1 74 ------- TABLE X I Uptake Biokinetic Model Sensitivity Analysis 1 ° l r • ! ‘ ‘ ! “ ? “ ““ 1 r rrTn h 1i • : — : • . • . . : : • : • • • : • : : • : : I I 1 T I Faramett i Gtomc t ic 4$ txceeduig tJBI( Remarks Canctntmtioa Mean 10 .0 tgtdL Run Value (sf1) Children aged No .......... rinking Water (pg/i) 0 2.92 0.98 1 23 3.09 1.29 2 For level of detection 5.0 pg/I . For level of detection 7.Opg,’l 44 ) 319 E44 4 DelauliSk 15.0 3.94 3.81 5 New drinking water standard. 50.0 6.33 18.69 6 Old drinking water standard. Soil& (Dust) (pg lg) 0(0) 1.47 0.01 1 L 21D(200) 319 244 DefaultSue 500 (500) 5.78 14.56 3 Superfund lower range value. 1,000 (1,000) 10.10 49.00 4 Superfund upper range value. 555 (565) 6.3 17.56 5 Housing age prior to 1940 175 (55) 2.41 0.34 6 Housing age 1940-1959 90(20) 1 91 0.00 7 Housing age 1960-1979 555 (555) 6.25 17.56 Assumes soil conczntration = dust concentration 175 (175) 3.02 1.15 90 (90) 2.29 0.25 0(565) — 4.17 — 435 . Dust concentration only I I I 11 Ii I I II I I It I ‘° Mo le that the level of detection for lead in drinking water varies by state, either 5.0 pgfl a 7.0 pg4. The significance of the soil and dust values is discussed in Sec t ion 44.6. Prqj.ct LEAP- Pus. 1 75 ------- [ 0(55) 1.79 035 0 (20) 1.62 0.02 Outdoor Air ( g/m 3 ) 0 3.16 0 1 0.20 1.50 319 3.41 144 2.13 2 3 Pth Quart u t erly average standard. 15.0 5.63 12.86 4 Ten times standard II II Prqj.ct LEAP—. Phe 1 76 ------- ii H S 3LOUD LEW ON 4TJ T lOll C uQ/ ) S to SI MonThs FIGURE 3 Select Drinking Water Concentrations Prq .ct LEAP— Phase 1 77 ------- I! Ii S S IS IS lICOD L HCD4TIIRT ION C ugldL) S to S4 Nosiths FIGURE 4 Select Soil and Dust Concentrations PrqJ.ct LEAP— Pba.e 1 78 ------- S B 1 5 12 ii 10 15 3L005 LSVED ONcE1TJVVT ION C u d’dL) S to S4 Month, FIGURE 5 Select Ambient Air Concentrations Prq .ct LEAP— Pha.. 1 79 ------- ii I i 31.000 L CWICW4TT ION C ugtdL 0 to 04 NDntIn FIGURE 6 UBK Default Concentrations with Geometric Standard Deviation of 1.8 PrqJ.ct LEAP— Ph... 1 80 ------- 4.7. Selected Area for Verification of Lead Screening Approach: Minneapolis/St. Paul 4.7.1. Minneapolis/St. Paul Demographic. Biological, and Soils Data Minneapolis/St. Paul MSA was selected for verification of the population screening approach (Objective 2) and for analyzing the association of blood-lead level to mobile sources (Objective 3) because its data were available in computerized format. The Minnesota Department of Health (MDH) conducted blood-lead testing in the area during the years 1986 and 1987. Reports of the results of the Minnesota Department of Health 1986-87 Blood-lead Survey were reported by memoranda from the Lead Program Coordinator, Division of Environmental Health, MDH, (Douglas Benson, Office Memorandum, October 11, 1991). A total of 1,410 children were surveyed, mostly in the Twin Cities, to ascertain blood-lead values and to find lead-poisoned children. The data collected in support of the survey were provided on computer disk. The data base contained 1,034 records for Minnesota and St. Paul. Data included blood- lead level, census tract of home, ethnicity, gender, birth date of child, years of education for the father and mother, and year blood-lead sample was taken. It is important to note that the blood-lead survey was conducted in geographic areas where soil-lead values had been previously determined to exceed 1,000 ppm. Environmental and demographic data were added to each record. Blood-lead-modeled values were calculated from the UBK model for the age and census tract location for each child/record. Environmental concentrations ascertained for the census tract of residency for each child were included to account for relevant routes of exposure. Most particularly, soils data for each census tract were obtained from the Minnesota Department of Pollution Control. The data base was that used by the MDH to prepare the Soil Lead Report to the Minnesota State Legislature (MPCA, 1987). Geometric mean soil-lead concentrations were calculated from the raw data, for each census tract. The geometric mean was selected in order to compare values with modeled blood-lead values. The UBK model assumes a lognormal distribution. The Project LEAP— Pba.e 1 81 ------- model calculates a geometric mean. A Geographic Information Systems applications was use to determine the distance from a transportation corridor for each census tract. That distance was then included in the data set for each child. The distance between the center of each census tract and the closest heavy duty transportation rn’ was determined for each census tract. 4.7.2 Minneapolis/St. Paul Statistical Analyses Two statistical procedures were conducted. To gauge the predictive ability of the comparative risk approach, the UBK model was run for each child’s data set, using the child’s age and the environmental data (concentrations) pertinent to the census tract of residency. The geometric mean estimated blood-lead levels were then compared to the measured blood-lead levels, using a simple correlation procedure. The correlation analysis was then repeated, grouping the children by census tract. The mean Pb-B values for each group was then compared to the UBK modeled Pb-B value for the census tract. A multivariable regression analysis procedure was employed to discern the contributions of various pathways of exposure. The procedure was limited by the lack of variation of some of the data. Neither drinking water nor air concentrations varied (sufficiently) across geographic areas and, consequently, the variables were not included in the regression analysis. The predominant housing age for each census tract was assigned to each record. Housing age was included in the model to account for historic deposition from mobile sources, both exterior and interior lead-based-paint, and deposition from point sources. Insufficient information is available, in general as well as for this analysis, to distinguish between and partition the contributions from these sources. For the remaining data/variables (full model on the following page), a regression analysis using stepwise comparison/replacement of independent variables was conducted. The decision point p-value for selecting a variable in the model was p = 0.05. U lnt 3 , 94, 494, and 694 Prnj.ct LEAP— Phi.. 1 ------- The full model is of the form: Log Pb-B, = + 1 Log Pb Bm + 2 HAC + I3 3 Dist + 4 AGE + 5 E 1 + + 7 E 1 + 8 GEN + p 9 FAT + ftOMOT + 11 Soil + 12 INC Where Pb-B 1 , Pb-B HAC = measured lead blood from survey (pg/dL) = model estimate of lead blood level ( gIdL) = 1 if house built before 1949 2 if house built 1950-59 3 if house built 1960-69 4 if house built 1970-79 = distance from centroid of census tract to nearest heavy duty highway (meters) = Age of child (years) = 1 if ethnicity is white, 0 otherwise = 1 if ethnicity is African-American, 0 otherwise = 1 if ethnicity is American Indian, 0 otherwise = 1 if gender is female, 0 otherwise = number of years of father’s education = number of years of mother’s education = soil lead concentration measured for census tract (ppm) = family income for census tract (dollars) DIST AGE El E2 E3 GEN FAT MOT SOIL INC 4.8. Derivation of City Exceedance Estimates To derive an estimate of exceedance of the 10 .i.g/dL blood-lead value for each group of census tracts, the UBK model was run with data pertinent to the area. Ambient air concentrations for the city (refer to Section 4.4.3.) and drinking water concentrations for the city (refer to Section 4.5) were used as input. No data from the air quality modeling efforts were used in the city computations. The results of the air quality modeling for the 17 sources were used in the qualitative analysis only. Soil and dust concentrations were calculated as a weighted average of the actual number of houses in each housing age category, for each census tract group, based upon the soil and dust concentration values derived from HUD data (refer to Section 4.63.). To illustrate, consider a census tract with 200 homes built before 1940, 300 homes built between 1940 and 1959, and 100 homes built from 1960 to 1979. Calculation of soil and dust values for the census tract, for input into the model, would be as follows: PrqJ.ct LEAP— Phi.. 1 83 ------- Housing Age Number Calculated Soil Value Calculated Dust Value of Homes ( ppm) ( ppm ) 1960-1979 100 (100/600)090= 015.0 (100/600)020= 003.3 1940-1959 300 (300/600)175= 087.5 (300/600)055= 027.5 pre- 1940 200 (200/600)555= 185.0 (200/600)565= 188.3 Total/Ave. 600 287.5 219.1 Thus, the soil and dust concentrations for the census tract, for input into the UBK model, would be 287.5 ppm and 219.1 ppm, respectively. The percentage of population expected to exceed 10 tg/dL, derived from the UBK model, was then multiplied by the total, African-American, and Hispanic childhood counts to derive the number of children expected to exceed the criterion Pb-B value for the census group area. The numbers for all census tract groups were then totaled to derive an exceedance number for each city. The number of new borne by ethnic categoly was calculated by applying the city specific birth rate to the total, African-American, and Hispanic populations. The UBK derived percentages were multiplied by those numbers to derive an estimate of fetuses that would exceed 10 p.g/dL Pb-B. Census tract groups were similarly aggregated to derive city totals. It is important to note that this methodology is for population screening purposes. The results may have no practical value as a prediction of the actual number of children expected to have elevated blood- lead values. Nor was that the intent of the methodology. The value of the approach is in the comparison between cities, and specifically to areas within a city that may be expected to have higher rates of lead exposed children than other areas. The intent of the population screening methodology is to use that indication to set priorities for intervention efforts within a city or region. The reader is particularly cautioned that the numbers are as derived by the computerized methodology and therefore appear to be precise. They are not. Project LEAP- Phase 1 84 ------- 5. RESULTS 5.1. Overview/Introduction to Results Results are presented for each environmental category, along with the modeling results for 17 air emission sources, soil and dust derivations, and a qualitative summary of the results for each city. An environmental profile and the results of statistical analyses pertinent to Minneapolis and St. Paul, Minnesota, are provided. Finally, the results of Pb-B modeling for each city, are presented. 5.2. Environmental Data Categorical Assessments 5.2.1. Ambient Air During 1988, there were few exceedances of the air quality standard for lead (three month average of 1.5 .tg.m 3 ). Most monitois reported in the tenths or hundreds of a Lg/m 3 The notable exception is Eagan, Minnesota, with a single fourth-quarter exceedance of 1.8 g/m 3 . The average lead concentrations utilized as the air concentration values in the UBK model are listed in Appendix C. Cities for which a program default values were used are also listed. 5.2.2. Air Emissions Tables showing total emissions in the six states and total emissions in the 83 cities are provided as Appendices D and E, respectively, based upon the Toxic Release Inventory data base. Figure 7 shows total air emissions of lead by state. The inventory contains 497 facility emission reports. Total air releases from 342 sources reporting release to air equals 449,304 pounds for calendar year 1988. Twenty-one sources, or 6.1 percent of the total number of sources, release 261,051 pounds annually, or 58.1 percent of the total air emissions. In the MSA areas, 226 sources account for 314,904 pounds annually. Seventeen sources with annual releases to the air greater than 4,000 pounds release 216,459 pounds annually. This accounts for 48.2 percent of total air emissions in the six states. These 17 sources are listed in Table X II. They constitute a mere 5 percent of the number of sources in the six states. Appendix F provides locational information. Prqjset LEAP- Phase 1 85 ------- Total Air Emissions 1988 Toxic Release Inventory 200 N U m b a r 150 P 0 U 100 0 f S 0 U r C a a 150 50 I 100 0 200 ( T h 0 U a a n d S ) R a a a a d No. — Lbs — 50 Ohio I Michigan Indiana Illinois 169 166.1 81 87 110 15 I 49.561 66.369 70.644 30.02 0 FIGURE 7 Total Air-lead Emissions 1988 Prq .ct LEAP— Pb 1 86 ------- TABLE XII Sources with TR.I Reported Total Air Emissions Exceeding 4,000 Pounds/Year in 1988 Trr .... B Location A ir Einisstcns (Pounds) Columbus, OH 61,300 Toledo, OH 6,711 Cleveland, OH 4,200 Warren, OH Steels, Canton Works Canton, OH 4,600 Mansfield, OH 7,231 Products Division Dayton, OH 4,250 Beech Grove, IN 9,870 Indianapolis, IN 5,485 Division Ecoise, MI 11,590 Saint Johns, MI 5,740 Co. Eagan, MN 13,812 Hanford, IL 11,570 East Chicago, IN 17,900 Koh ler, WI 29,200 Saint Paul, MN 12,480 Aiton. IL gr cjj 4J tA *44 s 4,677 t f l ! Prtdset LEAP— Pbs. 1 87 ------- Figure 8 shows the spatial distribution of the sources of lead and associated annual release amounts. Gopher Major Air Emission Facilities pounds/year FIGURE S Major Mr-lead Emission Facilities PrqJ.ct LEAP- Ph... 1 88 ------- 5.2.2.1. ISCLT Modeling Results Model run results for the seventeen major air emission facilities, included in Appendix G, were provided to the GISMO for spatial representation and to relate to census tracts. The resultant concentrations of air-lead were cross checked with ambient air data, where the latter was available. Table XIII summarizes the concentrations of lead determined for each of the 17 sources at a comparable grid point (200, 200) meters. This is generally the point of maximum concentration. Beyond 200 meters from the source, concentrations begin to decrease rapidly, generally to tens and hundredths of a pg/rn 3 . At the extremities of the model grid, 2,000 meters from the source, concentrations were in the thousandths of a pg/rn 3 , for all but the largest sources. All sources were less than hundredths of a at the extreme points. The maximum concentrations for the 17 sources varied from a low of 0.118 pg/rn 3 downwind from Acussar Dayton Thermal Products Division in Dayton, Ohio, to the two highest maximum concentrations values calculated at 1.792 pg/rn 3 for Kohler Co., Kohler, Wisconsin, and 1.693 pg/rn 3 for Ol-Neg TV Products, Inc., Columbus, Ohio. Except for these two sources, all sources had annual concentrations of less than unity. PrqJect LEAP- Ph.. 1 89 ------- TABLE XIII Maximum Concentrations of Lead for Modeled Sources Ftacskty Name Locat ion Estimated Concentration :;..::::: Ol-Neg TV Products, Inc. ....::..::.:::..:3 ..:: . .. Columbus, OH .. . . . .... 1 .693 DuPont Toledo Plant Toledo, OH 0.138 Oatey Co. Qeveland, OH 0.132 Copperweld Steel Co. Warren, OH 0.151 Republic Engineered Steels, Canton Works Canton, OH 0 .135 Empire Detroit Division Mansfield, OH 0.229 Acustar Dayton Thermal Products Division Dayton, OH 0.118 Refined Metals Corp. Beech Grove, IN 0.584 Quemetco, Inc. Indianapolis, IN 0.244 National Steel Great Lakes Division Ecorse, MI 0.306 Federal-Mogul Saint Johns, MI 0.136 Gopher Smelting & Refining Co. Eagan, MN 0.179 Chemetco, Inc. Hanford, IL 0.912 Inland Steel Co. East Chicago, IN 0.527 Koh ler Co. Kohler, WI 1.792 North Star Steel Minnesota Saint Paul, MN 0302 LaClede Steel Co. Alton, IL 0.146 P. J.et LEAP— Ph 1 90 ------- 5.2.3. Municipal Waste Combusters There are 32 municipal waste combusters in the Agency’s inventory, although three facilities are not currently operating. The three are all outside the study area. A significant finding is thai estimated emissions exceed actual emissions gathered by stack test emissions, generally by an order of magnitude or more. Appendix H provides information on each of the facilities, including design capacity, estimated emissions, stack test emissions, and comments pertinent to each facility. The 32 facilities, when all were operating, had annual emissions of 62,288 pounds of lead, based upon estimated emission factors and stack test emissions. Stack test emissions, available for 17 of the 32 sources, were utilized when available. Of the 32 sources, 15 are located in the project MSA cities. The facilities are listed in Table IV. Figure 9 shows the location of the facilities. An analysis of• the 17 sources with emissions data indicates the problem of using estimated emission factors. For those sources, estimated emissions total 349.13 pounds/day, while emissions based upon stack test information total only 46.78 pounds/day. Notably large differences for estimated and stack test emissions, respectively, include the Indianapolis facility, 48.60 and 0.06 pounds/day; Detroit, 92.40 and 1.82 pounds/day; NSP-Red Wing (Minnesota), 27.00 and 0.34 pounds/day; and Columbus, Ohio, 56.00 and 7.60 pounds/day. Analysis of the seven facilities with stack data, located in the MSA cities, shows a similar spread of 251.50 pounds/day estimated emission estimate, and 30.89 pounds/day stack test emissions, for an aggregate annual (stack test) emission of 11,275 pounds for the seven sources. Several sources, including the Chicago facility with estimated emissions of 35.20 pounds/day, appear to be significant sources and to warrant modeling. No municipal waste coinbuster has been modeled, however, due to the uncertainty of the estimated values, and as well as the significant differences between estimated emissions and stack test emission results. Consequently, the lead emissions from this categorical source was not incorporated into the study modeling and subsequent estimates of children exposed to lead. Large Prqj.ct LEAP— Phase 1 91 ------- sources, nonetheless, may prove to be a concern, when actual stack test data is derived. The Chicago, South Montgomery County (Ohio), and North Montgomery County (Ohio) facilities are planning or have recently obtained stack test data. Stack test data for those facilities may indicate the need for additional consideration. P,oJ.ct 1W- Pb 1 92 ------- TABLE XIV Municipal Waste Combuster Inventory Metropolitan Statistical Area Cities in Region 5 1 on. WUSt D n :•, :•. •.. - 1 atcd Air-Icad .. m s . .I . . . . m n c y . . . Da 1 ) .. Chi go, IL 1600 3520 N/A East (licago, IN 450 9.90 N/A Indianapolis, IN 2200 4840 0.06 Jackson, MI 200 4.40 N/A Detroit, MI 3300 92.40 1 Ł2 Grand Rapic , MI 625 13.75 N/A Duluth, MN 110 3.10 0.04 Rochester, MN 200 4.40 0.37 Rcs. Minneapolis, MN 1,000 22.00 0.06 Co. Dayton, OH 900 19.80 N/A Co. Dayton, OH 900 19.80 N/A Akmn, OH 900 25.20 20.94 Columbus, OH 2000 56.00 7.60 Shcboygan, WI 96 2.10 N/A Madison, WI 75 2.10 N/A ... - 14556 ........... .............. .. \ \ 3 $ -. Pr j.et LEAP— Phase 1 93 ------- 5.2.4. Drinking Water The Federal Data Reporting System(FDRS) data base for 1988 indicated that there were no violations of the (50 pg/ I) Maximum Contamination Level (MCL) for Pb for any of the study area cities. Only exceedances of the MCL, however, are reported to the system. Consequently, the FRDS does not contain information on actual measured concentrations less than the MCL State agency records indicate actual values from sampling results. These are summarized in Appendix I, with test results, number of PrnJ.ct LEAP— Phi.. 1 94 Municipal Waste Combus ters rand Rapids FIGURE 9 Municipal Waste Combustera in U.S. EPA Region 5 ------- samples in each test, and the drinking water concentration value for modeling, for each city. Of the 83 cities, only 10 had test results above the level of detection, while 27 cities showed non-detect levels. The drinking water suppliers for 46 cities did not report sample results in 1988. For the latter, 4.0 p g/l was assumed. The largest values reported were for Wausau, Wisconsin at 1500 p.g/l (reported value is suspect and was not used in the study) and 7 i.gfl ; Milwaukee, Wisconsin at 25.0 g/l; Youngstown, Ohio at 12.0 gfl; and Madison, Wisconsin at 10.2 g/l. Thus of all the cities sampled, only two would exceed a standard of 15.0 tgfl. 5.2.5. RCRA and Operating Landfills RCRA facilities as a category do not appear to present a significant risk of lead exposure. That assessment is qualified, however, due to the difficulty in obtaining information about a particular parameter, lead, at a given facility. Generally, the facilities may treat or otherwise process a limited number to a wide variety of pollutants, depending upon the facility’s operating permit and type of operation. A total of 27 RCRA facilities were assessed to determine whether lead was processed at the facility, and a determination was made on potential exposure. Appendix J lists the 27 RCRA facilities with comments on each and a T/F (true or false) notation as to potential for exposure. No information was obtained for seven facilities. For the 20 facilities assessed, only four appeared to have a potential for off-site exposure that could result in human exposure, generally via the air pathway from lead-contaminated piles and wind blown dust. These are Saint Louis Lead Recyclers, McLean Steel, Kemeto, and Olin, all located in Granite City, Illinois. Response action is ongoing in Granite City, Illinois, at the NLdTaracorp Site, that will result in capping a 240,000 ton lead- bearing waste pile situated adjacent to the former lead smelter, along with residential soil cleanup in a 55 square block area. Soils with lead concentration exceeding 500 ppm will be excavated and replaced with clean soil. EPA (Superfund program) is currently preparing the remedial design for the project. Granite City is also one of three study areas that is part of a tn-state lead study being conducted PrqJ.ct LEAP— Phase 1 95 ------- jointly by EPA and the Agency for Toxic Substances Disease Registry. In addition to determination of soil and dust contamination, the Illinois Department of Public Health has conducted extensive blood-lead testing for the project. This ongoing area-wide study in the city, should elucidate the potential for human exposure to lead from these facilities. Operating facilities that dispose of lead on-site were obtained from the TRI data base. Sixteen facilities in the MSA cities reported on-site disposal, ranging from a diminutive seven pounds annually to 566,000 pounds annually, for a total of 2,138,048 pounds/year disposed of on-site in 1988. Figure 10 provides a spatial representation of the largest facilities, with annual amounts indicated for the facilities shown. PmJ.ct LEAP— Ph.. 1 96 ------- Appendix K contains information on facilities having on-site lead disposal, along with a categorical judgement on the potential for off-site contamination. Of the 16 facilities, five appear to have the potential for off-site lead-contamination, as described in Table XV. As is the case for RCRA facilities, minimal data is available to characterize the concentration and spatial extent of lead-contamination that may result from landfilhing/on-site disposal at the facilities. In land Major On-Site Disposal Facilities Co.: 320 240 pounda/year FIGURE 10 Major On-Site Lead Disposal Facilities Pivjuct LEAP— Phi.. 1 97 ------- TABLE XV Toxic Release Inventory Reported On-Site Disposal in 1988 for MSA Cities Fac 1iIy L ca&ioi On4hc Remarks N fl ..:.•• . ::.:• :.: .: .:...:.. Dlsp a1 ..: t djy: : :: ::: .:. . ..: :. •:•. Keystone Steel & Wire Co. Peoria, IL 41,000 Ground water around the facility is contaminated with lead. Facility is seeking closure. Arc furnace dust pile addressed previously. Near residential area and Peoria State HospitaL Potential dust source. 1-11gb level exposure to population. Granite City Steel Granite City, IL 45,000 Facility sends 15-20 different types of waste streams to landfill, including blast furnace flue dust, settling pond sludge, etc. Some potential for off-site contamination of residents proximate to main street side of facility. Inland Steel East Qrn go, IN 560,000 No information on-site disposal operations. USX Gary Gary, IN 7,400 Good potential for electric arc furnace dust to get off site. Residential area. Cooperweld Steel Co. Warren, OH 2,001 Main waste is arc furnace dust. State is handling the closure wastc piles. Prq .ct LEAP— Ph.. 1 98 ------- 5.2.6. Abandoned Hazardous Waste Sites (Superfund ) The National Priority List (NPL) listing contained 95 sites in the six states that listed lead as a major contaminant. Of these, 17 facilities are located in the MSA cities. These are indicated on the map, Figure 11. Appendix L lists all sites in the six state area, with a designation of final or proposed pertaining to designation status as an NPL site. Appendix M provides definitive information on the sites Located in MSA cities. The data base consists of sites that were both proposed for listing and final, at that time, so that the extent of information about the sites vary greatly. In particular, the proposed sites tended to have much less information on the extent of lead contamination. Of course, lead is just one of the pollutants that could be on any given site, consequently, there is no requirement or particular reason for a file to contain more extensive data on lead. The more extensive site investigation step, development of a remedial investigation/feasibility study (to better characterize the extent of contamination and to develop alternatives for abatement) had not been initiated at many of the sites. Extensive sampling results, therefore, were not available. It is important to note, further, that the investigations are not undertaken solely to determine lead concentrations. Lead is only one of a host of contaminants of concern, and is most often not the prime pollutant being investigated. Table XVI lists the 17 sites located in the MSA cities. Most of the sites are abandoned landfills, many municipally owned and operated, and have been assessed primarily for potential groundwater contamination both on and off site. Soil contamination investigations, at this stage of investigation of site conditions, has almost exclusively focused upon on-site concentrations. Soil-lead contamination has been documented on-site for most of the facilities, although the primary route of exposure appears to be through contaminated groundwater. Information is, at best, sparse, particularly for those sites that have been PmJ.ct LEAP— Pbs .. 1 99 ------- proposed for the National Priority List, as contrasted to final NPL sites. For the former, only preliminary information, and few actual physical measurements of concentrations of groundwater, soil, or on site materials (e.g., in barrels or sludge lagoons) had been taken. There are two sites that are notable exceptions. The Barrels, Inc., site, in Lansing, Michigan, appears to have potential for off-site contamination. The NL Industries/Taracorp Lead Smelter site in Granite City, Illinois, has well documented and significant contamination in the residential area surrounding the Site and is, consequently, currently being addressed by the EPA Superfund pmgram. Janes S.E. Rocklord Grna Super fund NPL Sites ertima Ref & arreie Inc FIGURE 11 Superfund National Priority Ust Sites with Lead Contamination Pn J.ct LEAP— Ph ... 1 100 ------- TABLE XVI National Priority List Facilities in Metropolitan Statistical Area Cities with Lead as of November 1989 NL Industriestraraunp Lead Smelter Granite City, II Y Southeast Rockford Groundwater Contamination Rockford, II N MIDCOI Gary , In N MIDCOII Gary,In N - Tippecanoe Sanitary Landfill, Inc. Lafayette, In N WhitfordSalcs&Service SouthBend,In N Michigan Disposal (Cork Street Landfill) Kalamazoo, Mi N Motor Wheel, Inc. Lansing, Mi N 11. Brown Co., Inc Grand Rapids, Mi N Folkcrlsrna Refuse Grand Rapids, Mi N Barrels, Inc. Lansing. Mi Y Kaydon Corp. Muskegon. Mi N Van Dale Junkyard Marietta, Oh N ianesville Ash Beds Janesville, Wi N Janesvllle Old Landfill Janesvill; Wi N National Presto Industries, Inc. Eau Claire, Wi N Fort Howard Paper Co. Lagoons Green Bay, Wi N 12 Potential for off4lte lead amatnaficm. Y Skates flat mtendal exists. N indicates flat no potential exists for off.site lead amtaminntion. PunJ.et LEAP— Pbs 1 101 ------- 5.2.7. Environmental Data Qualitative Summary Table XVII is presented as a qualitative summary of potential routes of exposure to lead from environmental sources, for the 83 MSA cities. A positive indication for a categorical source for a particular city does not imply violations of an environmental standard or that there is necessarily an urgent public health concern caused by sources via the indicated medium. It does mean that, based upon current information, the potential for a problem exists. More definitive conclusions, in most instances, can only be drawn subsequent to on-site measurements. The qualitative summary table is presented as a quick view method of understanding where environmental sources of lead may exists in the 83 cities. The existence of a point source, RCRA facility, landfill facility, or Superfund site does not necessarily indicate that there is an environmental problem. Similarly, the cities shown with ambient air concentrations exceeding 0.2 p.g/m 3 and drinking water exceeding 4 tg/l, may be well below the standards for air and drinking water, respectively. The significance of the values in the table is merely to reflect the UBK default values. The table is presented to account for known sources of lead that may be above the norm for the 83 cities. A check mark indicates the existence of a facility, or measured air or water concentration above the concentrations shown in the table. Pr LEAP- Pb 1 102 ------- TABLE XVII Qualitative Summary of Environmental Exposures’ 3 to Lead for MSA Cities in 1988 . ;: •:: ; •x:: •: : • :: ::: . ......,. . ‘. . ‘••• • • Rock Island Moline Chicago 1 / Kankakee Peoria 1 / Bloomington Normal Champaign Urbana Rantoul Springfield E. St. Louis 1 / GraniteOty 7 1 /1 / p / / Rockford IUI! ________ Hammond 1 s . E.Cbicago 3 / S e i South Bend . Mishawaka Elkhart 13 Air point sourc.t (in or proximate to city), RCRA facility, Landfill Facility, and Superfund facility numbers Indicate facilities with potential to cause exposure to humans. Ambient air and drinking water exceedances pertain to the UBK model default values. PTOJ.ct ISA?— Pbs. 1 103 ------- C u y Total No utAwa H : Ot 1 &flS. No of Air Point SoUttts Woof flA ‘ ffif No of Landfifl : Na of Supcrftmd :: I: : S itS Ambient Mr > •: o wm ’ Dnnking Water> 4 J Gosben , Ft. Wayne LaFayette Kokomo .- An de r son - Muncie I ndianap o l is 3 1/ 1 Terre Haute Bloom ington Evansville New Athany tota iSs , : x : • : xx : • : ; : : c • : • : : : T : ’ : • : : : a 1 2 : . x : ’ :. : .x . :.z N :: : : . : . : .: . :. : 3 : : : : : : : : . : : c . x 1 ; : : : : : . : ; : : : : !t 10d Aa Saginaw ) BayCity . Midland Muskegon Grand Rapide Lansing 1 ‘ East Lansing Flint Detroit 1 1 Ann Athor . Battle Creek Jackson KMamnoo Benton Hathor * * S J : *f I ! ! PniJ.ct LEAP— Pbs 1 104 ------- City total No No of No of No of No of Ambient Mc Drmking Water> ofAiea Mr Point R RA UindtU Su md Concerns M oOrhead Sources PSciIIUeS PadlitS &a 02 tgf& 4 flfJ Duluth St. Cloud Minneapo lis 1 / SLPauI 1 P Rochester Ibtai state 2 ! ! MbOC$OW Toledo ) Cleveland Akron 1 / LotS Canton Steubenville Wheeling Marietta Youngstown 1 / Wairen 1 / Mansfield 2 / 1’ Lima Dayton Springfield Columbus 1 , , Hamilton Middletown Cincinnati Ban Claire i as *i im : fW ! SM Prqjet LEAP— Phee 1 105 ------- Pro [ ZAP— Phase 1 106 ------- 5.4. Chosen Cities 5.4.1. Minneapolis/St. Paul Environmental Sources of Lead Air quality monitoring data for NAMS stations located in Minneapolis and St. Paul indicate very low values of lead. Annual average air-lead concentrations, from the quarterly monitoring data, were 0.06 tg/m 3 and 0.05 .tg/m 3 for Minneapolis and St. Paul, respectively. Six sources in the Metropolitan Statistical Area reported air emissions to the Toxic Release Inventory. These were Gopher Smelting and Refining Co., Eagan, with 13,812 pounds/year total air emissions of lead and lead compounds; North Star Steel, St. Paul, at 12,480 pounds/year; American National Can Co., St. Paul, at 551 pounds/year, Honeywell New Hope Facility, Minneapolis, at 500 pounds/year; Bureau of Engraving, Inc., Minneapolis, at 500 pounds/year, and Whir-Air-Flow, Minneapolis, at 250 pounds/year. Both Gopher Smelting and Refining Co., and North Star Steel, were modeled to estimate air concentrations resulting from emissions. The maximum downwind concentrations were 0.18 g/m 3 for Gopher Smelting, and 0.30 .i.g/m 3 for North Star Steel. Both values were derived in close proximity, 200 meters, to the emission source. The Eagan facility, consequently, would not contribute to increased lead-air concentrations in either Minneapolis or St. Paul, due to the distance. Generally, noting the relatively de mini nus maximum concentration value, the exposures are rather limited. The Hennepin Energy Res. Municipal Waste Conibuster, located in Minneapolis, is the only other point source of lead emissions reported in the MSA. The amount of emission, 0.06 pounds per day, based upon stack test emissions, is quite small and, therefore, the source was not modeled to derive air concentrations. Drinking water test results were not required and therefore were not conducted for the drinking water supplies for 1988 because supplies were sampled for lead every other year. There is no indication of a problem with the source drinking water. Consequently, the model default value of four p.g l was PrqJ.etLEAP—Phu.1 107 ------- assumed. On-site disposal was reported in TRI only for the Gopher Smelting facility. As noted above, the facility is far enough from the two central cities such that wind-blown lead contaminated soil and dust, if there were any, would not impact/contribute to soil-lead and dust-lead concentrations in Minneapolis or St. Paul. Although the NPL Superfund sites with lead included in the data base, discussed earlier, did not list any sites in either city, a further review of NPL site summary documents and files found three sites: Union Scrap Iron and Metal Qmpany located in Minneapolis, Twin Cities Air Force Reserve Base (Small Arms Range Landfill), also located in Minneapolis, and Pigs Eye Landfill, located in St. Paul. None of the sites appear to pose a threat to residents via wind-blown off-site lead contaminated dust. The Pigs Eye Landfill is a 307 acre site that served as the City’s municipal waste landfill and also accepted industrial waste. The soil on-site is contaminated with lead and other constituents. The area immediately surrounding the site is industrial. A residential area is located one-half mile east. Lead was detected in high concentrations in one well, and in low concentrations in soil, indeed, soil samples taken near the facility indicate soil-lead concentrations of less than 150 ppm. The potential route of exposure is through contamination of 210 residential wells in the vicinity. The Minneapolis sites are both small. Union Scrap Iron is an one-acre site used to crush lead battery fragments. Reportedly, 30,000 Ions of lead-contaminated plaster and rubber fragments remain on- site, partially covered by tarp. A soil contamination study was to be conducted. The three-acre Twin Cities Air Force Base, Small Arms Range Landfill site, is located within and adjacent to the Minneapolis- St. Paul International Airport. Periodic flooding of the site has resulted in the release of lead and other contaminants into the Minnesota River. The primary potential for exposure is through contamination of drinking water wells. A hydrogeological investigation has been initiated. Three additional sites in the two cities are National Priority Last sites that do not cite lead as a Pr J.ct LEAP- PhM. 1 108 ------- contaminant. The Whittaker Corp. property is a 10-acre site located in Minnesota. The General Mills/Henkel Corp. site, also in the City of Minneapolis, poses a threat to the groundwater aquifer from solvents disposed in a dry well. The 45-acre Koppers Coke site is located in St. Paul. The removal of lead-contaminated coal tar waste and contaminated soil from the site has begun. The Minnesota Pollution Control Agency provided the raw data base for soil sampling conducted in Minneapolis and St. Paul, that provided a partial basis for the report to the Minnesota State Legislature (MPCA and MDH, 1987). Geometric mean soil concentrations are shown, by census tract, in Appendix N. Soil lead geometric mean values range from a low of 33 ppm to a high of 736 ppm. The geometric mean values are deceptive, however, in that the values do not truly represent soil concentrations in a tract. Indeed, a review of the individual samples taken for each census tract shows a wide range of values, with the highest concentrations generally from soil samples taken near house foundations. Foundation sample concentration values of 3,000 to 7,000 ppm are not uncommon, with the highest sample results in a single census tract in Minneapolis showing a value > 20,000 ppm. The two highest values of 38,850 ppm and 166,780 ppm were determined near an industrial facility in St. Paul. (It should be noted that the blood- lead counterpart data base contains no child blood-lead level measurements for that census tract). Recognizing the wide range of sample values within each census tract, for purposes of deriving modeled blood-lead values for each census tract, the foundation sample values were selected, where available, to represent soil concentration values. To assess the contribution to elevated Pb-B levels from mobile sources, the distance from the centmid of each census tract was calculated using geographic information systems applications. 5.4.2. Blood-lead Data/Demographics The Minneapolis and St. Paul blood-lead survey data contains the records of over a thousand children under the age of six for whom blood-lead levels were measured in 1986 and 1987. Table XVIII shows the number of children with elevated blood-lead levels by ethnicity. The dual heritage ethnic Pr LEAP- Phue 1 109 ------- category refers to those children listed as white/African-American, Hispanic/American Indian, or other races. Table XIX provides descriptive statistics for each ethnic group. TABLE XVIII Children Under Six Years of Age with Blood-Lead Levels Exceeding 10 p.g/dL based upon 1986 - 1987 Blood-lead Survey for Minneapolis and St. Paul 2. :r :.:. !:. :. Z ;.5 ? ! ..!........x:;.;.x:x.xx.;. • :: : •f ! - . . . - For white, African-American, and American Indian children (those ethnic groups greater than 100 children), arithmetic blood-lead levels are 7.7 zg/dL, 9.2 sg.dl, and 13.2 pg/dL, respectively. Corresponding geometric mean blood-lead levels (Appendix P) are 5.8 pg/dL, 7.2 pg/dL, and 9.8 p.g/dL, respectively. For the data set in total, blood-lead levels ranged from a minimum of 1.0 p.tg/dL to a maximum of 65 Lg/dL, with a geometric mean of 6.6 g/dL and a geometric standard deviation (GSD) of 2.2. Appendix 0 provides comparable data by census tract. Geometric mean blood-lead levels ranged from 2.0 p.g/dl (14 observations) to 65 p.tg/dL (one observation), for the census tract ‘ 4 Other ethnic groups include cthlldxen of dual heritage ethnicity and children for whom ethnicity wn not rearded. •Pr J.ct LEAP— Pb .. 1 110 ji i t i t • : _ Number> 10&g/dL 2% 151 4 64 9 8 28 Total Number 1022 667 114 13 — 121 31 1 1 65 Percent Exceeding 29.1 22.6 29.8 30.7 52.8 29.0 72.7 43.1 ------- TABLE XIX Blood-lead Values (pgIdL) by Ethnicity for Minneapolis and St. Paul, Minnesota Blood-lead Survey in 1986 - 1987 Btbrdcky Na of Miobnun Mnnnum OcometS StaaSd Not SpecifIed 14 : 3.0000 19.0000 7.2142 4.9017 White 667 1.0000 65.0000 7.6971 6.1447 African-AmerIcan 114 1.0000 37.0000 9.1666 6.6929 His p anic 13 3.0000 28.0000 8.8461 6.6061 American Indian 121 1.0000 39.0000 13.23% 92032 Asian 31 3.0000 36.0000 10.0967 7.9177 Other” 11 1.0000 44.0000 15.4545 12.1767 White/African- American 30 2.0000 34.0000 12.6666 9.2263 White/Hispanic 8 3.0000 22.0000 10.2500 6.0886 White/ American Indian 8 1.0000 21.0000 11.3750 7.4630 White/Asian 3 1.0000 7.0000 4.0000 3.0000 White/Other 3 3.0000 5.0000 4.0000 1.0000 African-American/ Hispanic 8 3.0000 24.0000 15.1250 7.6613 African-American/ American Indian 1 8.0000 8.0000 8.0000 Hispanic/ American Indian 4 3.0000 20.0000 83000 8.0208 Age of housing for each census tract in the Twin Ofies was provided from census data via geographic information systems. As counted by the 1980 census, housing in the two cities is “(blidren for whom ethnicity wn not rearded. PniJ .ctLEAP—Phaee l 111 ------- overwhelmingly built before 1949, with 78.9 percent of then-existing housing stock built prior to that year, and 92.9 percent built prior to 1960. For the blood-lead data base, the housing stock reflects an even older pattern (as a result primarily of the selection criteria for the blood-lead survey); consequently, there is very little differentiation in housing age by census tract. 5.4.3. Minneapolis/St. Paul Correlation Analysis A correlation analysis, using the Minneapolis/St. Paul blood-lead data and derived blood-lead levels from the UBK qiodel, for each record, was performed to ascertain the validity of the methodology for finding geographic areas where environmental exposures to lead would result in increased Pb-B levels. Sources have been described earlier. Selected zero-order correlations are shown in TABLE XX. The correlation of actual blood-lead levels to the corresponding modeled blood-lead levels is small, at 0.05, with a p = 0.14; consequently, the results indicate a failure to reject the null hypothesis of no correlation between the modeled and measured blood-lead values. The conclusion from this analysis would be that the modeled blood-lead levels do not predict the actual blood-lead levels at a statistically significant level. (Recognizing problems with the approach, however, a second analysis as employed. This provided better results, as discussed below.) The analysis also determined very small correlation coefficients for actual Pb-B levels with housing age category, distance from an interstate highway, and soil-lead concentration. The combined routes of exposure of air (due to ongoing emissions from mobile sources) and soils and dust contamination, due to past deposition from mobile sources, do not appear to contribute appreciably to Pb- B levels for the study population. Similarly, the expected finding of a strong correlation for soil with housing age category, was not determined. The analysis did, however, find a weak correlation of soil concentration with distance from a major highway (r= 0.13, p= 0.0001). PrqJctLEAP—Pha..1 112 ------- TABLE XX Correlation Analysis of Minneapolis/St. Paul . I • • . i . • •i • . . B -BA 1 L 0 .8198 ‘ I . .., . 1 . 03476 L L .L I 1 .8129 LPB-BM 901 0.9664 0.3013 870.7621 0.4941 3.86 46 HAC 1033 0.9990 0.1841 1032 0 3.0000 DIST 1033 1135 937.76 1 172375 0 5015 SOIL 1033 1863 2199 19244 15 0 11162 I I , . , ’ I . H ; . ; . . 1 . . I . . . .- . . : 4 Vy ? ?. V • .: . : : • . • • : : : : ; • 1: : • . . . : : : v : •: : • • . • • : • • • • : . P*atson thbtblt2an CoefflaenW PStlity > 01 , R1x n LV No of OSetvaUoas 1 I • : ; i : : :. • : .. ‘ : IP&BA • • • • ! ? . E 1.0 0 0 0 0.0487 0 01436 1033 901 I d ::. : • ; ; . . . : • d . : • s i 0.0095 -0 .0849 07593 00063 1033 1033 r • : • .:. : 0.0795 00106 1033 :. WBBM 0 0487 1 0000 01436 0 901 901 1_ 1 . i j -0 03855 .0 0049 0.2477 08825 901 901 0 7248 00001 901 flAt, ° 00095 -00385 * 07593 02477 : 1033 901 L.U t 11 10000 00093 0 07637 1033 1033 00240 04408 1033 DIS1 .00849 -00049 ‘ ••• 00063 0.8825 ? 1033 901 00093 10000 07637 0 1033 1033 01318 00001 1033 r ( tYL ri ( A r sot 00795 07248 1 I ’ V 00106 00001 •:• < 1033 901 00240 01318 04408 00001 1033 1033 10000 0 1033 1 whcrc LPBBA a ln uu blood-lead level (pg ,tH); LPBBM a modeled Wood-lead level (pgM l) HAC a housing age ategoty (range of years for oath category), as &fiIICd in she rcgreaion model; DIST a distace from an interstate highway (men); and SOIL a SO iWead conanntion (ppm). Pm LEAP— PIa 1 113 ------- A significant problem with the correlation analysis was that different physical scales were being compared. The actual blood-lead values were associated with soil-lead values at or near the home of the individual child, and thus it was on a relatively small geographic scale. The modeled blood-lead values, in contrast, depended upon a soil-lead concentration for the census tract where the child resided. This much larger geographic area (scale) could result in a soil-lead concentration much different than the actual exposure concentration. Recognizing this, a second correlation analysis was employed to compare the geometric mean of the measured blood-lead levels to the geometric mean modeled blood-lead levels calculated for each census tract. Only census tracts having nine or more observations were selected, as displayed in Table XX I. The modeled values are higher particularly in that high soil lead values were used for census tract modeling, than is thought to be the actual exposure of children living in a census tract. PrQJ.ct LEAP— Ph 1 114 ------- TABLE XXI Selected Census Tract Data from Minneapolis/St. Paul ... . .. $o+ Ř CCSI ‘fl m4atP (Ř j ) t S R j44 * !....!.!.. No OW -x : -x :yy’ •: : • ’: ’: • Cs*te Trs* Vb.R A * (GM) • - : • :xc • : • • ’ • : • : • Pb B Mod (Ok) a 26 15 9.79 (GM) 18.33 a 73 6.63 5.35 14 16 7.76 7.12 32 79 8.12 46.15 28 18 338 8.55 3) 83 10.19 11.73 9 21 8.60 13.59 14 86 5.36 11.93 14 22 8.10 8.00 30 301 536 6.49 27 25 4.78 8.77 20 325 10.84 6.49 14 28 538 6.31 16 326 931 8.06 14 29 6.62 14.61 14 335 7.08 8.06 12 33 731 7.77 11 355 8.91 3.48 16 36 4.95 7.22 46 357 431 6.27 3) 50 6.85 630 19 368 536 14.38 18 61 8.79 20.24 36 370 4.35 4.90 15 66 7.98 3.30 27 371 6.80 17.27 46 — 72 . 832 18.25 Table XXII presents the results of a correlation analysis for modeled blood-lead and actual blood- lead variables listed in the table. The correlation improved from 0.13 to 0.3. The results were not statistically significant, with p 0.10. Given the constraints inherent in the methodology, however, the approach works reasonably well. It is interesting to note that for the modeled Pb-B mean values, some census tracts were quite close to the measured Pb-B mean values. The larger estimated values from the model results in a small Pearson correlation coefficient. “Geometric mean value of rne ured blood-lead level. Geometric mean value of modeled blood-lead level. Pmj.s LEAP— Pbs . 1 115 ------- TABLE XX I I Correlation Analysis- M inneapolis/St . Paul Census Tract Level L 1 snnP !&atLe& : : VMi ilt Numbtt j Mean Stimd*d r I “ “ ‘ . . ; . ‘ . : ; : : : : : : c . . . : : : : :z ::: .. . . . : . :.. ;. . : . . . : . : : . . ;I • •! 23324 • 97 : F o io;o Thin !!•• — 28 io ns 8 5733 299120 0 46150 .11T 1 LL...LLJ ’ 1I ’I 1 1t .. LL.LJLL I U. .J. I • J] . . t . x.ooo 03167 o.o 0.1005 I 03167 1 1 100 * 01005 00 . In particular, one outlier results in a skewing of the data which causes a smaller pearson correlation coefficient and a smaller p-value. Figure 12 is a scatter plot of the 27 data points in Table XXI, comparing the geometric mean modeled blood-lead levels with the actual blood-lead levels for children in each of the 27 census tracts. As shown in the scatter plot 1 there appears to be a definite linear association between the two variables, with modeled values generally increasing with increasing measured blood-lead levels. “where LPBBA s meuumd blood4ead level, and PBBM a modeled blood-lead level. Prqj.ctLFM —Pks l 116 ------- MinneapoHs/St.Paut Census Tract Data G.řm.tr Ic M.ari -B Val u.s (u /dL) a 40 - 30 - 20 D D D D 0 D a U a a D I I I I I I I I I I I I 0 2 4 6 9 10 12 14 tust BIocd-I*ad C -B) Values FIGURE 12 Scatter Plot of Modeled Blood-lead Values Vs. Actual Blood-lead Values 5.4.4. Minneapolis/St. Paul Regression Analysis A second use of the blood-lead data was to partition the actual blood-lead levels found among environmental sources. For the data set, distance from a major thoroughfare was added to each record in order to ascertain whether lead from vehicle exhaust (thought to be primarily via lead-contaminated dust PrqJSCtLEAP—PhM.1 117 ------- and soil from historic deposition from vehicles) would explain a portion of the variation. In addition, housing age was added to each record (by category) to recognize the contribution of lead-based paint. This would also account for historic deposition from area and point sources. Results of the regression analysis yielded the following for the full model and the final model. Regression analysis for the full model shows a very small R 2 value of only 0.08, although the result is statistically significant (p= 0.0001). This is an expected result, given the minimal Peaison correlation coefficients derived for selected variables. The results are tabulated as Appendix 0. The final model was derived by using a stepwise regression procedure, with a criterion level of significance of p= 0.05 for inclusion of an independent variable in the formula. Non-significant regression coefficients were not carried forward from the full to the final model. The results are tabulated in full as Appendix R, and in summary form as TABLE XXJII. The final model is Log PbB =1.007 - 0.017 AGE - 0.092 El + 0.099 E3 + 0.001 FAT - 0.004 INC Where PbB , = Measured blood-lead level from survey ( g/dL) AGE = Age of child (years) El = 1 if ethnicity is white, 0 otherwise E3 = 1 if ethnicity is American Indian, 0 otherwise FAT = number of years of father’s education INC = family income for census tract (thousands of dollars) PrqJ.ct LEAP— Phi.. 1 118 ------- TABLE XXIII Summary of Stepwise Procedure for Dependent Variable Actual Blood-lead Concentration for Minneapolis/St. Paul I .... ..... . ........... ... .. . .. .. ..... .. .. Regression 5 &49494 1 .69698 14.88 0.0001 Error II 10 2 .20900 0 .11420 Total It 900 11 0.70394 : : : : : : . INTERCEP 1.00708 0 .0 5457 38.88745 34032 0.0001 AGE -0.01748 0.00657 0.80810 7.08 0.0800 El -0.091 68 0.02840 1. 18992 10.42 0.00 13 E3 0.09692 0.03969 0 .70 930 6.21 0.0129 FAT 0.00096 0.00048 0.45675 4.00 0.0458 INC .0.00454 0.00 191 0.64049 . 5.6 1 0.0181 aiwnsna ll*anniambet4.47Ia4 3OS I5U\ + AU nthbó In S ixaS an nflcam a the IWOGIcS Not vwats met the*OflisŘflc*n 1* cl *t sqyintothomScL J’L I ( 1 11 wiLi ahr L i d J t L J * 4J.J . aJ 4 J. Swumaty otS*p win rnted ntt uepnnSx YMIS LPb4IA l litF Lt i N1 MJ %v L 1 Ifl}t # ’ i Y ’ t ’ ’ ( ( i 1 , r F SIm 41* t4c + 1 1 t*ds1R* 4 >M dta t c® F : 4 b$;bt lI L f c; $.‘4t tZ ‘t; 1 El 1 J 0 .048 5 0.485 30.7085 45.8441 0.0001 2 INC 2 0.0096 0.058 1 23.3882 9. 1 134 0.026 3 AGE 3 0.0080 0 .0 661 17.5747 7.6970 0.0056 4 E3 4 0 .006 5 0.0726 13.2208 62961 0.0123 S FAT 5 0.0041 0.0767 11. 1980 3.9996 0.0458 a wte AGE s age ofthild: El a dummy variable fat white cthnidty E3 a dummy vati t hle for Aimri n In an eththdty FAT a number of yars of father’s cduation and INC a family inuxnc for a nct of nsidcn 21 V R I I IMe Enteted, Removed Pt J.ct LEA.?—. fla 1 119 ------- Independent variables that did not improve the model were the model estimate of the blood-lead level (PbBm ), housing age category (HAC), distance from a major highway (DIS1), African-American ethnicity (E2 = 1), gender (GEN), number of years of mother’s education (MOT), and the measured soil- lead concentration for the census tract (SOIL). The comparison population for ethnicity (El =E2=E3=0) consists of all ethnic categories except white, African-American, and American Indian. Although the final model is statistically significant (p 0.001), the R 2 value is small, indicating that the selected independent variables do not explain much of the variation in the measured blood-lead levels. With the exception of the number of years of father’s education, FAT, the signs of all regression coefficients in the final model reflect intuitive expectations. The blood-level decreases with age. This is as expected for the overall childhood population range, although within the age strata, blood-lead levels are expected to peak at two and then decrease. The negative association with white ethnicity and increasing family income is also as expected. A positive regression coefficient associated with American Indian ethnicity merely reflects the higher mean blood-lead levels for this ethnic group as a whole. Only the number of yeais of fathers education, indicating an increasing blood-lead level with increasing education, is counter to expectations. Further, it would be more logical to have a significant regression coefficient for the mother’s education attainment, not the father’s, as indicated in the final regression model. Consequently, the inclusion of FAT in the model and the positive sign of the regression coefficient is thought to be a spurious effect. A change in the ethnicity dummy variable scheme was made to ascertain whether African- American and Hispanic blood-lead values would be statistically significantly higher than values for white children, while controlling for other variables. Variable El was changed from El= white to E1 Hispanic. The change was made to make the comparison population (E1=E2=E3=0) to be white and a small number of ethnic minority children. This comparison population thus excludes Hispanic, African-American, and American Indian children. The final revised model results are provided as Appendix S. The revised final Prqfrct LEAP— Pb ... 1 120 ------- model is: Log PbB = 0.998 -0.0172 AGE + 0.151 E3 + 0.001 FAT - 0.006 INC Where PbB = Measured blood-lead level from survey ( .tg/dL) AGE = Age of child (years) E3 = 1 if ethnicity is American Indian, 0 otherwise FAT = number of years of father’s education INC = family income for census tract (thousands of dollars) 5.5. UBK City Results Based upon the use of ambient air quality data for each city, measured drinking water concentration for the city, and weight averaged soil and dust concentrations for each census tract group, the UBK model was used and city exceedance number developed. Aggregate results are displayed in Table XXIV. Appendix T, provides detailed information by census tract group area. Prcij.ctLEAP—Pha.e1 121 ------- TABLE XXIV Numbers of Children Under 7 Yeazs of Age in the Midwest ExpectS to Exceed 10 pxg/dL Blood-Lead Level in 1988 City Cbi ldhood Population Total No xceedth2 African Amenan 1Th Exceedzog Rock Island 4,910 461 103 17 Moline 4,379 434 5 37 Ch i t go 321,585 40,370 18,712 7,888 Knnkqkec 3:e l 289 87 — 3 Peoria 13,368 1,306 354 24 !omi ton 4,362 330 21 5 Normal 2,430 26 2 0 Champaign 3,979 168 34 2 Urbana 2,359 154 10 3 Rantoul - N/As Springfle]d 9,716 554 76 4 E. St. Louis 8,127 798 768 8 (3raniteCity 3,726 273 4 5 Zaate ‘ - rx 1%$S I , 1 44, $ S2J @ ) ! 198 Y 40 3% ThY ; A Gary 2o,855 831 652 69 Hammond — 10,522 - 1,059 92 100 E.Qilcago 5,073 660 189 275 SouthBend 11,441 1,084 207 26 Mishawaka 4,149 225 2 2 N/A not available. Data was not available S these thin. Percentage of total population. Numbers are based upon ambient air, drinking water and derived soil and dust lead ccnrrations used in the Uptake Biokmnctic Model. ProJ.ct LEAP— Pbs. 1 122 ------- Ci ty O ii ldhood PopU lat ion Tot al No 4 &C eed1P g African - E A m er i can H i s p a n i c Exceeding Elkhart • : • f: 4,616 : 464 Exo dfl 97 7 Goshen Ft. Wayne 18,910 1,780 414 55 LaFayette 4,146 243 5 3 Kokomo 5,437 401 27 6 Anderson 6,707 502 110 3 Muncie 6,822 522 56 4 Indianapolis 73,868 5,223 1,740 52 Terre Haute 5,250 797 71 6 Bloomington 2,775 - 141 7 — 2 Evansville 12,444 1,248 135 7 New Albany tttal$*te ót ln4Utw 3,598 •• : .. - t$ř sa Y 258 ‘ A 43 L 13 •flTfll . _ an ? ‘ . 2 614 — Saginaw 9,943 935 348 92 Bay City 4,358 564 10 — 27 Midland 3,834 45 1 1 Muskegon 4,741 603 135 18 Grand Rapids 20,064 1,942 486 99 Lansing 15,251 955 128 75 East Lansing 2,531 115 7 — 2 flint 19,923 1,446 581 38 Detroit 134,680 19,142 12,409 555 Ann Arbor 7,819 381 38 9 Battle_Creek 4,150 569 129 12 Jackson 4,588 748 119 15 Kalamazoo • 7,323 • 181 v i Benton Harbor NIA Pnd.ct LEAP s Pbs. 1 123 ------- Moorhead 2,401 61 0 1 Duluth 8,299 1,284 9 5 St.Cloud 3,577 206 1 1 Minneapolis 29,884 4,611 379 59 St. Paul 25,357 3,333 194 97 Rochester 5,774 237 1 2 rota! State 75,2W 9 ,732 584 165 !LMmAesota (13 %) Toledo 38,143 4,515 1,157 182 Cleveland 61,289 9,396 4,022 360 Akron 23,644 3,161 694 20 Lorain 8,962 465 53 70 Canton 9,739 1,342 264 18 — Steubenville 2,002 160 25 1 Wheeling N /A - Marietta N/A Youngstown 11,968 1,884 673 64 — Warren 5,742 437 69 3 Mansfield 7,180 688 7 Lima 5,972 550 116 6 Dayton 22,426 2,206 688 17 Springfield 7,745 914 175 7 Columbus 2,968 432 110 5 Hamilton 7,217 728 77 6 M lddletown 4,467 281 31 1 Cimati 38,829 5,415 1,939 41 I I I I Prqj.ct LEAP— Ph ... 1 124 ------- 1 J blOWn I a i m BauClaire 4,250 247 - :i • 1 Wausau 3,017 224 0 1 Green Bay 9 ,058 483 1 4 Oshkash 3,992 388 2 2 Neenah N /A Milwaukee 67,871 13,878 4,225 781 Racinc 9,626 819 130 56 Kenosha 7,927 494 19 22 Mad ison 12,294 759 21 11 Janesvifle 5,655 263 0 1 Beloit 3,982 421 63 5 LaQ •ossc 3 ,341 404 1 2 Sheboygan 4,810 443 1 8 II . .AP !1!t ?n . 6 ,2 5 1 286 0 — -ii The highest percentages of children exceeding 10 pg/dL Pb-B in Ill inois were derived for Chicago, where the majority of census areas were in double digit percentages, with many in the 15 to 19 percent range. The maximum value was 19 percent. A majority of the total number of children exceeding 10 pWdL in many of the conimunitia were Mńcan-Arnerican and Hispanic, reflecting the racial makeup of the neighborhoods. This factor also gives rise to the large number of children under seven years of age at risk of exposure to lead. The high exposure potential reflects primarily soil and dust concentrations Indict LEAP— Pbs 1 125 ------- (based upon housing stock age). Drinking water and air concentration values in the city were low. Two areas in East St. Louis also had high numbeis of African-American children with expected exceedances, although the percentages were not high at 11 and 12 percent (597 and 100 African-American children, respectively). Granite City, an area with widespread soil and dust contamination resulting from industrial operations in the city, had an average of 7 percent exceedances using the methodology. The relative small numbers for African-American and Hispanic children reflect the low population concentrations of these two ethnic groups in Granite City. It is clear that the study approach underestimates the risk in Granite City, however, by not using the higher actual soil and dust concentrations that are currently being determined. One area in Peoria is notable, with a 15 percent exceedance estimate corresponding to 224 African-American and 13 Hispanic children. No other Illinois city was notable for large numbers of African-American or Hispanic children under seven years of age, expected to exceed 10 .tg/dL Pb-B. Compared to other states in the Midwest, the community areas of most cities in Indiana have low percentile values for expected exceedances, and low numbers of potentially exposed African-American and Hispanic childhood populations. This is due to not only generally low derived-exceedance-percentages, but also to smaller city populations and relatively low population density for both minority ethnic groups. Five cities, East Chicago, Evansville, Ft. Wayne, Indianapolis, and Terre Haute, had community areas in the 15 to 19 percent range. The largest numbers of African-American and Hispanic children with expected exceedances of 10 tg/dL Pb-B were in Indianapolis, with 1,740 and 52 children, respectively, followed in quantitative rank by Gary, with 652 and 69 children, respectively. This ranking is generally indicative of the relatively large population size of these two cities. It is noted, in particular, that the community area percentages for the City of Gary were all less than 10 percent. Detroit closely resembles Chicago in having a number of areas with expected percentages ranging from 15*020 percent, with corresponding high numbers of African-American and Hispanic children PraJ.ct LEAP- Ph 1 126 ------- reflecting the ethnic makeup of the communities. A total of 19,142 children, including 12,409 African- American and 556 Hispanic children, are the expected exceedance numbers. For other State of Michigan cities, aside from Detroit, community areas in Ann Arbor, Battle Creek, Flint, Jackson, and Kalamazoo had percentile values in the 15 to 19 percent range. None of these areas, however, had very high numbers of African-American or Hispanic children with expected exceedances of 10 .tg/dL Pb-B. As expected, for the State of Minnesota, both the highest percentages and the greatest number of African-American and Hispanic children with exceedances were derived for Minneapolis and St. Paul. The Twin Cities expected numbers of African-American and Hispanic children with exceedances were, respectively, 379 and 59 for Minneapolis, and 194 and 97 for St. Paul. For the State of Minnesota, only Duluth, aside from the Twin Cities, had community areas with percentile ranges of 15 to 18 percent. The community area in Ohio with the largest expected exceedance percentile was located in the City of Toledo, with a value of 20 percent, corresponding to 431 African-American and 45 Hispanic children expected to exceed 10 .tg/dL Pb-B. For the city as a whole, 1,157 African-American and 182 Hispanic children are expected to exceed 10 .tg/dL Pb-B. Two cities in Ohio have higher numbers of children with exceedances. Qeveland’s numbers are 4,022 African-American children and 360 Hispanic children, and Cincinnati’s numbers are 1934 and 41, respectively. Although several other cities had community areas with percentage values in the 15 to 19 percent xunge, the only other city with more than 1,000 children potentially exceeding the criterion value was Columbus, with 1,094.African-American and 33 Hispanic children. For the State of Ohio, the highest percentile of 23 percent was derived for a low population density community area in Youngstown. Wisconsin is set apart somewhat from the other states and community areas by having several communities with levels of lead in drinking water at measured levels, above the level of detection. This factor, combined with soil and dust concentrations associated with older housing stock, resulted in the higher estimates of exceedance for four Wisconsin cities. Prqjsct IZAP- Ph. 1 127 ------- On a community total basis, Milwaukee is high both in percentile (20 percent) and in numbers (13,878 total, including 4,225 African-American and 781 Hispanic children). Several areas were in the 28 to 30 percent range, making the city the highest overall of all cities assessed, and resulting in large estimated numbers of children with exceedances. Milwaukee’s drinking water concentration also measured comparatively high, at 25 ppb for 1988. Aside from estimated percentages of 17 percent for areas in both La Crosse and Racine, neither the percentiles nor the numbers of African-American and Hispanic children exceedances were exceptional for all other communities in Wisconsin. Seven cities are in the top 10 by virtue of both overall percent exceeding and number of children exceeding 10 Itg/dL Those cities are Milwaukee, Wisconsin; Detroit, Michigan; Minneapolis and St. Paul, Minnesota; and Cincinnati, Akron, and Cleveland, Ohio. The top 10 cities, by percentile and total number of children, are shown in TABLES XXV and XXVI, respectively. TABLE XXV Top Ranked Cities by Percentile of Children Exceeding 10 jsg/dL Pb-B 1 Milwauke WI 20.4 13,878 4,225 781 2 Jackson,MI 163 748 118 15 3 Duluth, MN 15.5 1,284 9 5 4 Minneapolis, MN 153 4,611 379 5 9 5 Cleveland, OH 153 96 4,022 360 6 Terre Haute, IN 15.2 797 71 6 7 Detroit,MI 14.2 19,142 12,409 556 8 Cincinnati, OH 13.9 5,415 1,934• 41 9 BattleCreek,MI 13.7 569 129 11 1 Akron, OH 13.4 0 3,161 694 20 Pn J.d LEAP- Ph .. 1 128 ------- TABLE XXVJ Top Ranked Cities by Number of Children Exceeding 10 pg/dL Pb-B . ... . ...i..r - . b BimectedibudNo E x pected f l o -o f ExxctedNoo l . :.. . • . :. :r . r ::.ihIsS lSIW*W Ve in :. baldi c aa Lts,JJaIJa. xenc7Years I 1 i • — Chicago, IL 13 C i — • — a i 40,370 Old a 18,712 O I L) a S 7,888 2 Detroit, Ml 14 19,142 12,409 555 3 Milwaukee, WI 20 13,878 4,225 781 4 Cleveland, OH 15 9,396 4,022 - 360 S Cincinnati, OH 13 5,415 1,939 - 41 6 Indianapolis, IN 7 5,223 1,740 52 I. Minneapolis, MN 15 4,611 379 !. 8 Toledo OH 12 4,515 1,157 182 9 1 St. Paul, MN Akron,OH 13 85 3,333 3,161 194 694 97 20 0 a — The six Midwest states ranged from 8 percent exceedance estimates in Indiana to 13 percent in Minnesota, although it is noted that these percentages are not particularly meaningful at the stale level. The States of Illinois and Michigan had the largest numbers of African-American and Hispanic children under seven years of age expected to exceed 10 sg/dL Pb-B, including 28,000 and 16,000 minority children, in the respective states. Every state has community areas where elevated blood-lead levels are of concern. For the six states, all cities combined, the total childhood population (children under seven yeafl of age) was 1,359,000 in 1988. The analysis indicates that 154,000 children, or 11 percent of the total, would have blood-lead levels exceeding 10 pgldL This includes 55,000 African-American and 12,000 Project LEAP— Pha. 1 iv ------- Hispanic children. These numbers are presented for illustrative purposes, and are not a prediction of numbers of children. It is noted, however, that these numbers are conservative compared to other estimates (refer to Section 2.5). PrqJ.ct LEAP— Pha.. 1. 130 ------- 6. DISCUSSION The population comparative risk methodology developed for this study contributes to the understanding of the extent to which low-level environmental sources of lead add to elevated blood-lead levels in childhood populations. The methodology provides an assessment of the relative numbers of children at risk to the adverse health effects due to environmental lead exposure. Previous studies have been at the national level. The methodology fills a gap in research efforts on the extent of childhood lead poisoning. It provides city specific estimates to highlight possible areas of l igh numbers of children with elevated blood-lead levels. Indeed, the methodology provides comparative numbers within cities. A key value of the methodology is as a ranking tool to guide public health officials to cities and areas within cities having the highest potential for childhood exposure to lead. More definitive data would need to be obtained to confirm the initial characterization of areas as high risk. Rather than investing resources in areas found in retrospect to be low risk areas, however, high risk areas could be targeted and addressed on a priority basis. Further, the environmental pathways of exposure developed in this study, provide a clear indication of whether to gather further information on air quality, drinking water quality, or soil and dust concentrations in a given city. Such measured environmental data, together with any blood-lead data available for a community, is a fundamental step towards primary intervention actions. Removing lead from the environment will avoid the need for clinical interveetion for the individual child. That is, of course, the desired outcome. 6.1. Demograřhics Although the demographic and associated data (housing age, income) was obtained for each census tract, there were inherent imprecision in the data. The results of the 1990 census was not yet available; 1988 data (estimated from the 1980 census) were utilized. The numbers derived from these estimated data therefore have, inherently, the same level of inaccuracy. Prqj.et LEAP— Ph ... 1 131 ------- Beyond data estimation, a larger problem concerns how the data were categorized, as reported in the census data base. Because the age categories did not match the study design (e.g., childhood age strata were zero to five, six to 13 years, etc., while the Study design focused upon children less than seven years of age), an approximation was derived to reflect the number of children in the study design strata. A more problematic concern is that children in each age band were not totally disaggregated by ethnicity. Derived numbers were thus underestimates of the actual numbers of minority children in an area and, consequently, determined to exceed the criterion blood-lead value. This resulted from a procedure that calculated the number of children in a census tract by a proration of the relevant ethnic group’s portion of the total population. Such an approach is accurate only for mono- thnic populations. As ethnic diversity increases, this method of estimation becomes more and more imprecise. In particular, it results in an underestimate when the number of children in minority families (i.e., family size) exceed the number of children for the community as a whole. Similarly, the demographic data obtained at the census tract level did not include ethnicity-specific birth, rates. Because African-American and Hispanic birth rates are often higher than the general population, application of a city’s overall birth rate to the ethnicity-specific population, to estimate numbers of fetuses at risk, results in an underestimate of fetuses at risk. Problems of matching housing age categories were similar. The census data provided strata beginning with housing stock built prior to 1949. For purposes of the study, age strata for housing stock built before 1920 and before 1940, associated with lead pipes and higher concentrations of lead-based paint, respectively, would be more pertinent. Some precision in estimating soil and dust values, in particular, was sacrificed by assuming houses built before 1949 were also expected to have lead-water supply pipes and higher lead-based-paint concentrations. A key finding, well into the study, was that the geographic information systems software platform was not the most expedient in manipulation of demographic housing stock age, and other census bureau obtained information. Indeed, extraction of relevant data (census tract information only for selected cities PrQJ.ct IZAP— Phi.. 1 132 ------- within each of the six states) proceeded rapidly when processed using standard data base management software on a personal computer platform. 6.2. Environmental Data Several aspects of the usefulness, or lack thereof, of environmental data became readily apparent as the study progressed. The air route of exposure was determined to be of minor consequence as a contributor to the estimated blood-lead levels. The measured air-lead concentrations were found to be very low in an overwhelming number of cases. Even the modeled major sources; for which air concentrations were derived, proved to have little impact, beyond relative close proximity to the source. Consequently, at the large scale for which the algorithm was applied (aggregations of census tracts), those concentrations could not be included in the UBK model results, nor, consequently, accounted for in the estimated numbers of childhood exceedances. It is also noted that many of the sources are located distant from the central city populations of concern. Based upon these findings, it is apparent that the relatively small numbers of children affected by such point sources, would not change the substantive results of the comparative population (by city) risk analysis. Surprisingly, the Toxic Release Inventory data base proved to be valuable in assessing the relative importance of point sources of emissions, while the Aerometric Information and Retrieval System Facility Subsystem, which also provided emission information, was not useful for the study. The only other category of air emissions, municipal waste combusters, appeared to be of minor import as a source of exposure. The results for abandoned hazardous waste sites were also unexpected, in that the great majority (of lead contaminated sites) are located outside the central cities, and thus away from populations of concern. As was the case for air stationary sources, contamination of soils and dusts at a site would be a local phenomenon affecting only closely proximate populations. None were accounted for in the UBK modeling. An assessment of sites, however, indicated the need for little concern for the category as a whole, except as noted in the results section concerning Granite City, Illinois. PmJ.ct LF. P— Phase 1 133 ------- Operating hazardous waste facilities proved to be the most difficult to assess. This was primarily because there is no central data base for which to list Resource Conservation and Recovery Act facilities that dispose of lead and lead compounds. Generally, for the facilities assessed, the categorical source was •not deemed to be a factor in the study area cities. Drinking water contamination was found generally to not be a problem, and, except for the cities noted in the results section, the categorical source is not a major contributor to estimated elevated blood- lead levels. It is noted, however, that brass plumbing fixtures and lead contamination associated with new home construction is not addressed. The procedure, utilization of the UBK model, did not include lead-based paint concentrations. That was beyond the scope of the study, which addressed environmental sources of lead. There was also a deanh of information upon which to estimate lead-based paint contribution, for purposes of this methodology. The study focused upon environmental sources of lead to estimate chronic effects. Lead- based paint, historically, has been associated more with acute effects. (This is because at high blood-lead levels, signs and symptom are more readily discernable.) Consequently, soil and dust concentration values, derived from age of housing stock, generally predominates as the source of estimated elevated blood-lead levels. The percent exceedances and corresponding numbers of children exceeding the criterion value are driven by housing age (dust and soil concentrations) with adjustments for drinking water and air concentrations pertinent to each community assessed. 6.3 Correlation Analysis The original correlation analysis for the Minneapolis and St. Paul areas, upon application, was determined not to be adequate for testing the validity of the algorithm. A fundamental problem is in the use of the UBIC model to derive modeled blood-lead values for comparison to actual values. The model calculates a geometric mean blood-lead value for a population. The study data was of individual measurements. An individual child, even having the same exposure concentrations used in the UBK Project LEAP- Phase 1 134 ------- model run, could, of course, have a Pb-B level on either side of the mean value, and could very well be two or more standard deviations from the mean. Consequently, the derived (mean) value cannot be expected to correlate well with the actual (individual) value. Further, a different scale for the two values was used. For the actual Pb-B value, data for the child was measured specific to the child and the home. For the UBK model, information (soils data) was available only at the (much larger and consequently very much more varied) census tract level. The great deal of variability in soil concentrations at residential yards, as further varied by choice of sampling location, is also noted. Moreover, nothing in the model could account for what was undoubtedly also occurring, i.e., potential for contaminated paint and dust exposure, home habits (such as frequency of dusting), occupational exposure, cigarette smoking in the home, and other factors. These factors are all known to affect Pb-B levels. Consequently, the procedure was not deemed to be robust, in that statistical power was lost due to each of these factors. A second derived approach, using the geometric mean values for all actual blood-lead values, yielded a better result, although most of the same problems are inherent in that approach as well. 6.4 Renression Analysis Distance from a major highway was not found to be associated with blood-lead levels. The finding of statistical significance is not particularly relevant given the small R 2 value calculated for the final model. The analysis failed to find an association between blood-lead levels and distance from either an interstate highway, or with soil. The latter fmding is consistent with that of the State of Minnesota study (MPCA and MDH, 1987) (i.e., the relationship between blood-lead levels and soil concentrations is weak). Given that weakness, the lack of a relationship with distance from a highway is also an expected result. With the exception of the number of yeazs of father’s education, FAT, the signs of all regression coefficients in the final model reflect intuitive expectations. The blood-level decreases with age. This is as expected for the overall childhood population range, although within the age strata, blood-lead levels Projsct LEAP— Ph ... 1 135 ------- are expected to peak at two and then decrease. The negative association with white ethnicity and increasing family income are also as expected. A positive regression coefficient associated with American Indian ethnicity merely reflects the higher mean blood-lead levels for this ethnic group as a whole. Thus compared to minority children (other than African-American), white children have statistically significantly lower blood-lead level. American Indian children, however, have statistically significantly higher blood- lead levels. Only the number of years of fathers education, indicating an increasing blood-lead level with increasing education, is counter to expectations. Further, it would be more logical to have a significant regression coefficient for the mother’s education attainment, not the father’s, as indicated in the final regression model. Consequently, the inclusion of FAT in the model and the positive sign of the regression coefficient is thought to be a spurious effect. The revised final regression inodel, substituting Hispanic for white as dummy variable El, yielded unexpected results. After controlling for other variables, neither variable El (Hispanic) nor E2 (African- American) were associated with the actual blood-lead levels at statistically significant levels. Thus the higher mean blood-lead levels for African-American and Hispanic children compared ton white children, controlling for other variables, is not statistically significant. Compared to white children, American Indian children have statistically significant higher mean blood-lead levels. Compared to (predominately) white childhood population in the revised model, or other ethnic minorities as in the original regression model, American Indian children have blood-lead Levels that are statistically significantly higher. 6.5 City Estimates of Exceedance The derived values are thought to be minimal estimates for several reasons. As discussed under demographics, the numbers of children estimated in the ethnicity categories of interest arc low. The procedure, further, focuses upon chronic exposure from environmental sources, and does not account for additional numbers of children due to exposures to contaminated paint, including any resultant acute PrqJsd LEAP— Ph. 1 136 ------- exposures. The latter, in an area of deteriorating older housing stock, can greatly increase the numbers of affected children. Moreover, a significant concern in the procedure is that the UBK model does not account for ethnicity or socioeconomic status (nor was it designed to do so). It is well documented that such factors increase the relative risk for many in the study population. By point of comparison, for large numbers of African-American children, ATSDR (1988) postulates that fully two-thirds in the lowest socio-economic stratum would exceed 15 p.tg/dL blood-lead. This is well above the percentages derived from the algorithm. Further, a 10 g/dL criterion level would result in greater than two-thirds of the pertinent population exceeding the value. 6.6 Uncertainties As with most screening methodologies, there are a number of areas in the methodology that introduce uncertainty into the results. Due to the wide range of data that the methodology uses, mixing actual data with postulated data and then using a model, it is impossible to calculate an uncertainty in the traditional sense (derivation of a confidence interval with an associated level of statistical significance for the numbers of children cited for each city). It is not the intent of the study, however, to predict numbers of children exceeding 10 tg/dL blood-lead. Rather, it is to compare cities in order to make reasoned judgements on which geographic areas appear to have children at highest risk of exposure to environmental sources of lead. Actual measurements would then be necessary to ascertain childhood exposure. Nonetheless, it is useful to discuss uncertainties in the methodology, discussed throughout this document, in one section. That is the purpose of this discussion, to summarize uncertainties inherent in this population comparative risk screening methodology. The quality of ambient air quality data is judged to be excellent. The data is from an ongoing ambient air quality network administered by each state agency under a rigorous quality assurance program. The program is prescribed by U.S. EPA regulations. Monitors arc generally sited to ascertain peak spatial Project LEAP— Phi.. 1 137 ------- concentrations, however, not to determine representativeness of air quality in a city, per se. It is, nevertheless, often used for that purpose. Further, the limited numbers of monitors does raise concern about how representative the data are. Drinldng water data were taken from data generated by drinking water suppliers, as provided to state agencies, under a quality assurance program prescribed by U.S. EPA regulation. A limited number of samples, taken at the supply, is used to characterize exposure for the entire community serviced. Variations in samples taken over a course of time during the year indicates that actual exposure, in some instances, may be difficult to determine. There is no readily discernable pattern of variation, where lead was found at detectable levels. It is also noted that most of the supplies consistently measured non- detectable levels of lead. The soil and dust values used in the UBK model were those estimated from the ages of housing stock for each area. No measured values were used. Consequently, there is substantial uncertainty in the derived concentrations. Further, the data base from which the estimates were derived, the National Housing Survey Data discussed in Section 4.6.3., had a range of values for each housing age category. For houses built between 1961) and 1979, dust-lead concentration values ranged from Oppm to 1520 ppm, with a mean value of 20 ppm and a standard deviation of 145. For older homes, those built prior to 1940, the range was even greater, from a minimum of 0 ppm dust-lead to a maximum of 33,130 ppm, with a mean of 565 ppm and a standard deviation of 3,780. Comparable soil-lead concentrations for pre-1940 housing stock were a minimum soil-lead concentration of 1 ppm, a maximum of 6,260 ppm, a mean of 565 ppm, and a standard deviation of 1,060. Thus there is uncertainty in the values chosen to represent soil and dust values in the UBK model, based upon age of the dwelling. There are a number of routes for introduction of uncertainty via use of the UBK model. The model uses assumptions regarding behavioral and physiological parameters that affect the results (discussed in Section 4.7). Behavioral patterns assumed for each age group, for example, could miss the mark. Prqfrct LEAP— Ph 1 138 ------- Section 4.7.1. discusses the high dependence of the model on the selection of the geometric standard deviation (GSD) assumed to be applicable a modeled population. For a given set of concentrations, changing the GSD from 1.42 to 1.8 results in an estimated percentage of childhood exceedance of 10 tg/dL of 0.05 percent for the lower GSD, to 2.47 percent for a GSD of 1.8. This, consequently, would introduce a great amount of uncertainty, if the UBK model were being used as a predictive tool. It is not being used for that purpose here. Consequently, because any uncertainty introduced by selection of a GSD value is in the same direction for all cities, the uncertainty introduced via this mechanism is of less concern (when comparing populations). Additional uncertainty is introduced via this new use of the model. It has not been validated for use at the census tract level. Rather, it was developed for use at specific sites, for which environmental concentrations have been more readily obtainable. Finally, the correlation analysis, comparing the mean values of blood-lead values for groups of children in a census tract, to the UBK modeled values for the tract, resulted in a correlation coefficient of 0.3 at p > 0.10. While this result indicates a relatively weak correlation, it appears to be quite reasonable given the myriad of uncertainties associated with the methodology. In particular, it is reasonable given the use of the methodology as a population comparative risk screening tool, as opposed to as a predictive methodology. PrqJ.ct LEAP— Pb. 1 139 ------- 7. CONCLUSIONS Central city residents, particularly African-American and Hispanic children, are subject to low- level exposure of environmental sources of lead. Differential exposure exists amongst the cities. The population screening methodology provides a viable method for estimating where the greatest numbers of children at highest risk reside. Clearly soil and dust concentrations predominate as sources of lead contamination. Drinking water quality contributes in a few cities. The screening methodology is based upon using existing environmental and demographic information. Consequently, not all desired information was attainable. Several assumptions were made in order to proceed with the study. To test the impact of the assumptions (for example, the use of model default concentrations when measured environmental data were unavailable), a sensitivity analysis was conducted. l’hat analysis indicated that soil and dust concentrations, at higher concentrations, predominated as contributing to higher blood-lead levels. The dust concentration value, however, was unreliable. Nevertheless, the analysis indicated that use of the calculated dust concentration had minimum effect upon the numbers of children calculated to exceed 10 p.g/dL Pb-B. An inability to account for ethnicity and socioeconomic status resulted in an underestimate of the at-risk population in lower socioeconomic minority communities. Nevertheless, the approach is considered to be valid, even though there was only a weak cormlation between Pb-B modeled and Pb-B measured, due to the factors discussed. A fundamental factor of the analysis is that the UBK model used to derive modeled blood-lead levels is not, nor was it intended to be, applicable and appropriate for use to discern a blood-lead level for an individual child. The model is only appropriate for estimating the affects on populations of children. That is the use for which the methology uses the UBK model. Consequently, its use in this population comparative risk analysis is thought to be appropriate. Accordingly, the methodology should prove to be useful in identifying “hot spot” areas where there may be sizable numbers of children at higher relative risk to environmental lead exposure. The study approach estimates that Pmj.ct LEAP- Phi.. 1 140 ------- significant numbers of children under the age of seven years are exposed to environmental sources of lead at levels exceeding 10 g/dL The population comparative risk number of children is 163,000 in 83 cities in the six states assessed, including 56,000 African-American and 12,000 Hispanic children. The actual numbers exceeding 10 ig/dL cannot be ascertained by the population screening methodology. Additional blood-lead elevations due to lead-based paint exposure is not accounted for in the methodology. Consequently, the calculated numbers are believed to be conservative. Although soil and dust are the most important determinants of modeled Pb-B levels, there is a paucity of information about the extent of lead-contamination caused by operating and abandoned hazardous waste facilities that could cause such contamination. Generally, off-site soil, dust, and air sampling for lead has not been conducted. Nonetheless, there may be a relatively small number of residents potentially exposed. Except for Granite City, Iffinois and Lansing, Michigan, this category of sources does not appear to warrant significant concern for most areas. Extensive sampling, however, around each site, would be required to make a definitive fmding. Unless there is strong indication of contamination, however, such sampling is not generally deemed to be prudent or cost effective. Major air sources are of concern only for residences near emission sources. Municipal waste combusters, as a whole, do not appear to constitute a serious concern. More information is required, however, to make that judgement. Modeling of the air sources did not add value to the methodology, aside from confirming the lack of wide-spread impact. The ambient air, drinking water supply, and toxic release inventory data were useful in development of the methodology and, for the former two data bases, for calculation of mean blood-lead values. One would expect to find a stronger correlation for distance from a major highway and the actual blood-lead measurements only if there were a strong correlation for soil concentration and distance from a major highway. That association, although statistically significant, was weak. The conclusion is that, Pr .etLEAP—Pbu.s1 141 ------- for the population (recall that high soil-lead values were the criterion for sclection of tested children for the blood-lead survey), the distance from a major highway does not correlate with actual blood-lead levels (or soil concentrations). Consequently, other factors (e.g., lead-based paint) appear to contribute more to the elevated Pb-B found, in the survey. It is noted, however, that the average distance from the center of the census tracts to an interstate is greater than one km, and that the maximum distance exceeds five km. Accordingly, most of the children appear to be too far distant to be exposed via the mobile source route. These results are applicable only for the Minneapolis,St. Paul area studied, and can not be generalized to the study area as a whole. The majority of data analysis should be conducted on a personal computer, in lieu of manipulation using geographic information systems. Data manipulation on the personal computer proceeds with relative ease. The latter computer platform should be utilized to obtain the census bureau data, as well as to map the results of the analysis. The demographic and housing information was adequate for purposes of developing the methodology and for estimating the spatial and numerical dimensions of minority children at risk for low- level exposure from envimnnientaj sources of lead. Precise population and housing estimates for optimal stratification were not available. Nevertheless, that factor was not deemed crucial to the study results, in that estimated and representative environmental exposures were used. Further, the UBK model itself is an approximation. Consequently, for population risk screening purposes, the data were satisfactory. PreJ. * LEAP— Pb 1 142 ------- 8. RECOMMENDATIONS Repeat the derived methodology for other EPA regions and states, or for smaller geographic areas where targeting is desired to rank, prioritize, and better characterize the numbers and extent of at risk minority populations exposed to lead. The derived methodology recognizes the efforts that did not contribute to the screening approach. For example, account for abandoned and operating waste sites, municipal waste combusters, and stationary sources of air emissions spatially and qualitatively, with follow-up if a facility is located within a high percentage exceedance area, as identified via the methodology. It is not worthwhile, however, to include the modeled air-lead concentrations as input to the UBK model. Include the contribution of lead-based paint to elevated blood-lead levels by using procedures as derived for soil- and dust-lead concentrations, based upon age of housing stock. This would better estimate expected blood-lead values. Calculated values would be based upon better knowledge of the association of lead-based paint contributions to daily intake, with housing age. Select areas within the top 10 cities with the highest numbers of children at risk, for on-site sampling and investigation, in order to determine the actual extent of residential lead contamination. Develop and implement a public outreach and awareness strategy, pertinent to African-American and Hispanic communities, in particular, but inclusive of any population at high risk of exposure, in selected cities. Work with public health departments to coordinate outreach and education efforts to targeted communities. Determine if census tract level data are available from the Bureau of the Census, stratified by ethnicity for children under seven years of age, for ethnicity specific birth rates, and housing age categories more relevant to lead usage in residential areas. Further investigate hazardous wastes sites in Granite Qty, East St. Louis, Lansing Michigan, and any city PrqJ.ct LEAP-- 1 143 ------- where a site falls within an area with large numbers of children expected to exceed 10 .tg/dL blood-lead. Complete the remedial design work for the NLTFaracorp Corp. site in Granite City, Illinois, pursuant to on-site abatement and replacement of contaminated soil in the 55 square block residential area. Review major sources with high (relatively) high modeled air values to ensure nearby residents are not exposed to excessive air-lead concentrations. Obtain results of stack test information, when available, for the Chicago, Illinois, North Montgomery County, Ohio, and South Montgomery County, Ohio, municipal waste incinerators, to ensure that lead emissions do not pose an unacceptable risk to local residents. Ascertain the current drinking water lead concentrations for the Cities of Wausau, Milwaukee, and Madison, Wisconsin, and Youngstown, Ohio, and consider whether additional education or other action is warranted. PrqJ.ct LEAP— Ph s 1 144 ------- CITED LITERATURE American Academy of Pediatrics. 1987. Statement on childhood lead poisoning. Committee on Environmental Hazards/Committee on Accident and Poison Prevention. Pediatrics 79(3):457-462. Angle CR, Marcus A, Cheng I-H, Mclntire MS. 1984. Omaha childhood blood-lead and environmental lead: A linear total exposure model. Environ Res 35:160-170. Angle CR, Mclntjre MS, Swanson MS, Stohs SJ. 1978. Low level lead and inhibition of erythrocyte pyriniidine nucleotidase. Environ Res 17:296-302. Angle CR, Mclntire MS, Swanson MS, Stohs, SJ. 1982. Erythrocyte nucleotides in children—increased blood-lead and cytidine triphosphase. Pediatr Res 16:331-334. ATSDR (Agency for Toxic Substances and Disease Registry). 1988. The Nature and Extent of Lead Poisoning in Children in the United States: A Report to Congress . AThDR, Public Health Service, Department of Health and Human Services, Atlanta, Ga. ATSDR (Agency for Toxic Substances and Disease Registry). 1990. Toxicological Profile for Lead . ATSDR, Public Health Service, Department of Health and Human Services, Atlanta, Ga. Baker, EL, Folland DS, Taylor TA, Frank M, Peterson W, Lovejoy G, Cox D, Housworth J, Landringan PJ. 1977. Lead Poisoning in Children of Lead Workers: House Contamination with Industrial Dust. Engi J Med 296(5):260-261. Baghurst PA, Robertson EF, McMichail AJ, Vinipani GV, Wigg NR, Roberts RR. 1987. The Port Pine Cohort Study: Lead Effects on Pregnancy Outcome and Early Childhood Development. Neurotoxicologv 8(3):395-402. Bellinger D, Leviton A, Waternaux C, AlIred E. 1985a. Methodological Issues in Modeling the Relationship Between Low-level Lead Exposure and Infant Development: Examples from the Boston Lead Study. Environ Research 38:119-129. Bellinger D, Leviton A, Waternaux C, Needleman H, Rabinowitz M. 1985b. A Longitudinal Study of the Developmental Toxicity of Low-level Lead Exposure in the Prenatal and Early Postnatal Periods. In: Lekkas TD, ed. International Conference: Heavy Metals in the Environment. Vol. 1 . September, Athens, Greece. Edinburgh, U.K.: CEP Consultants, L*d.:32-34. Bellinger D, Leviton A, Needleman HI ., Watcrnawc C, Rabinowiiz M. 1986a Low-level Lead Exposure and Infant Development in the First Year. Neurobchav Toxicol Teratol 8(2):151-161. Bellinger C, Leviton A, Rabinowitz M, Needleman H, Waternauz C. 1986b. Correlates of Low-level Lead Exposure in Urban Children at Two Yeats of Age. Pediatrics 77(6):826-833. Beuinger D, Leviton A, Waternaux C, Needleman H, Rabinowitz M. 1987. Logitudinal Analyses of Prenatal and Postnatal Lead Exposure and Early Cognitive Development. N Enal J Med 316(17):1037- 1043. PrqJ.ct LEAP— Phass 1 145 ------- Bellinger D, Leviton A, Sloman J. 1990. “Antecedents and Correlates of Improved Cognitive Performance in Children Exposed in Utero to Low Levels of Lead. In: Lucier GW, Hook E, eds. Advances In Lead Research, Environmental Health Perspectives . Research Triangle Park, North Carolina. U.S. Department of Health and Human Services, Public Health Service— National Institutes of Health, National Institute of Environmental Health Sciences. November 1990. 89:5-11. Brunekreef B, Noy D, Biersteker K, Boleij J. 1983. Blood-lead Levels of Dutch City Children and Their Relationship to Lead in the Environment. J Air Pollut Control Assoc 33(9):872-876. Bruekreff B. 1984. The Relationship Between Air-lead and Blood-lead in Children: A Critical Review. Sci Total Environ 38:79-123. CDC (Centers for Disease Control). 1985. Preventing Lead Poisoning in Young Children, A Statement By The Centers For Disease Control . Atlanta, GA.:CDC, U.S. Department of Health and Human Services, Public Health Service, Center for Environmental Health, Chronic Diseases Division. Pubi No 99-2230:7- 19. CDC (Centers for Disease Control). 1986a. East Helena. Montana. Child Lead Study . Summer, 1983:Final Report. July, 1986. Chisoim JJ Jr, Harrison HE. 1956. The Exposure of Children to Lead. Pediatrics 18(6):943-958. Chisoim JJ Jr. 1962. Aniinoaciduria as a Manifestation of Renal Tubular Injury in Lead Intoxication and a Comparison With Patterns of Aminoaciduria Seen in Other Diseases. J Pediatr 60:1-17. Chisohn JJ, Thomas DJ, Hamill TG. 1985a. Eiythmcyte Porphobilinogen Synthase Activity as an Indicator of Lead Exposure to Children. Clin Cheni 31(4):601-605. Chisoim JJ, Mellits ED, Quaskey SA. 1985b. The Relationship Between the Level of Lead Absorption in Children and the Age, Type and Condition of Housing. Environ Res 38:31-45. Chisolm JJ. 1986. Removal of Lead Paint from Old Housing: The Need for a New Approach. Am J Pubi Health 76(3):236-237. Clark CS, Bornschein RL, Succop P, Que Hee SS, Hammond PB, Peace B. 1985. Condition and Type of Housing as an Indicator of Potential Environmental Lead Exposure and Pediatric Blood-lead Levels. Environ Res 38:46-53. Clark CS, Bornschein RL, Succop P, Hammond PB, Peace B, Krafft K, Dietrich K. 1987. Pathways to Elevated Blood-lead and Their Importance in Control Strategy Development In: Lindberg SE, Hutchinson TC, eds. International Conference: Heavy Metals in the Environment. Vol. 1 . September, New Orleans, La. Edinburgh, U.K.:CEP Consultants Ltd. 159-161. Davis MJ, Svendsgaard DJ. 1987. Lead and Child Development Nature 329:297-300. DHHS (U.S. Department of Health and Human Services). 1991a. Healthy People 2000, National Health Promotion and Disease Prevention Objectives . Public Health Service, U.S. Department of Health and PmJsct LEAP- Pha.e 1 146 ------- Human Services. 1991. DHHS (U.S. Department of Health and Human Services). 1991b. Strategic Plan for the Elimination of Childhood Lead Poisoning . Centers for Disease Control, Public Health Service, U.S. Department of Health and Human Services. February 1991. DHHS (U.S. Department of Health and Human Services). 1991c. Preventing Lead Poisoning in Young Children . Centers for Disease Control, Public Health Service, U.S. Department of Health and Human Services. Atlanta, GA. October 1991. Danford DE, Smith JC, Huber AM. 1982a. Pica and Mineral Status in the Mentally Retarded. American Journal of Clinical Nutrition . 35:958-967. Danford DE. 1982b. Pica and Nutrition. Annual Review of Nutrition . 2:303-322. DHUD (Department of Housing and Urban Development). 1990. Comprehensive and Workable Plan for the Abatement of Lead-Based Paint in Privately Owned Housing Report to Congress . Office of Policy Development and Research, U.S. Department of Housing and Urban Development. Washington, D.C. December 7, 1990. Dietrich KN, Kiaffi KM, Bier M, Succop PA, Berger 0, Bornschein RL 1986. Early Effects of Fetal Lead Exposure: Neurobehavioral Findings at 6 Months. mt J Biosoc Res 8:151-168. Dietrich KN, Krafft KM, Bornschein RL Hammond PB, Berger 0, Succop P, Mariana B. 1987a. Effects of Low-level Fetal Lead Exposure on Neurobehavioral Development in Early Infancy. Pediatrics 80(5):721-730. Dietiich KN, Krafft KM. Shukla R, Bornschein R1 Succop PA. 198Th. The Neurobehavioral Effects of Early Lead Exposure. Monogr Am Assoc Ment Defic 8:71-95. Dietrich K, Succop P, Bornschein R, Krafft K, Berger 0, Hammond P, Buncher C. 1990. “Lead Exposure and Neurobehavioral Development in Later Infancy”. In: Lucier GW, Hook E, eds. Advances In Lead Research. Environmental Health Perspectives . Research Triangle Park, North Carolina. U.S. Department of Health and Human Services, Public Health Service— National Institutes of Health, National Institute of Environmental Health Sciences. November 1990. 89:13-19. Elias RW. 1985. Lead Exposures in the Human Environment. In: Mahaffey K, ed. Dietary and Environmental Lead: Human Health Effects. Topics in Environmental Health . Cincinnati, Oh. National Institutes for Occupational Safety and Health. 7:79-107. EPA (Environmental Protection Agency). 1986a. Air Quality Criteria for Lead. June 1986 and Addendum. September 1986 . Research Triangle Park, N.C.:Offlce of REsearch and Development Office of Health and Environmental Assessment Environmental Cziteria and Assessment Office, EPA. EPA 6OO -83-O18F. EPA (Environmental Protection Agency). 1986b. Reducing Lead in Drinkina Water: A Benefit Analysis . Washington, DC: Office of Policy Planning and Evaluation, EPA Report No. EPA-230-09-86-019. PmJ.ct LEAP— Pb 1 147 ------- EPA (Environmental Protection Agency). 1989a. Risk Assessment Guidance for Superfund Volume 1 Human_Health_Evaluation Manual (Part A) Interim Final. Office of Emergency and Remedial Response, U.S. Environmental Protection Agency. EPAJ54O/l-89/002. Washington, D.C. December 1989. EPA (Environmental Protection Agency). 1989b. Review of the National Ambient Air Quality Standard for Lead: Exposure Analysis, Methodology and Validation . OAQPS Staff Report, EPA, Office of Air Quality Planning and Standards. Research Triangle Park, N.C. EPA-450/2-89-O11. June 1989. EPA (Environmental Protection Agency). 1989c. Exposure Factors Handbook Final Report . United States Environmental Protection Agency, Office of Health and Environmental Assessment, Exposure Assessment Group. EPA/600/8-89/043. Washington, DC. March 1989. EPA (Environmental Protection Agency). 1988b. Corrosivitv Monitoring Data From U.S. Public Water Systems . Task No.5, Contract No. 68-01-7088. Office of Drinking Water, U.S. Environmental Protection Agency. Washington, D.C. December 1988. EPA (Environmental Protection Agency). 1990a. Toxics in the Community. National and Local Perspectives. The 1988 Toxics Release Inventory National Report . Economics and Technology Division, Office of Toxic Substances, U.S. Environmental Protection Agency. Washington, D.C. September 1990. EPA (Environmental Protection Agency). 1990b. Superfund: Focusina on the Nation at Large. A Decade of Progress at National Priorities List Sites . EPA/540/8-90/009. Office of Program Management, Office of Emergency and Remedial Response, United States Environmental Protection Agency. Washington, D.C. September 1990. EPA (Environmental Protection Agency). 1991a. Technical Support Document on Lead . First Draft ECAO-cin-757. Environmental Criteria and Assessment Office, Office of Health and Environmental Assessment, U.S. Environmental Protection Agency. Cincinnati, Oh. January 1991. EPA (Environmental Protection Agency). 1991b. Risk Assessment on Lead Leachina From Plumbina Draft. Office of Toxic Substances, U.S. Environmental Protection Agency. Washington, D.C. September 16, 1991. EPA (Environmental Protection Agency). 1991c. Integrated Risk Information System (IRIS) Carcinogenicity Assessment for Lifetime exposure for Lead and Compounds (inorganic) . On-line. (Verification date May 1, 1991. Office of Health and Environmental Assessment, Environmental Criteria and Assessment Office, U.S. Environmental Protection Agency. Cincinnati, Oh. EPA (Environmental Protection Agency). 1991d. Users Guide For Lead: A PC Software Am)lication of the Uptake/Biokinetic Model Version 0.50 First Draft . Environmental Criteria and Assessment Office, Office of Health and Environmental Assessment, Office of Research and Development, U.S. Environmental Protection Agency. Cincinnati, Oh. ECAO-CIN-. January 1991. Erenberg G, Rinsler SS, Fish BG. 1974. Lead Neuropathy and Sickle Cell Disease. Pediatrics 54(4):438- 441. Ernhart CB, Wolf AW, Kennard MJ, Filipovich, Sokol RI, Erhard P. 1985. Intrauterine Lead Exposure and the Status of the Neonate. In: Lekkas TD, ed. International Conference: Heavy Metals in the •P LEAP- Phase 1 148 ------- Environment, September, Athens, Greece, Vol. 1 . Edinburgh, U.K.: CEP Consultants, Ltd. 35-37. Fernandez AM, McElvaine MD, Orbach HG, and Pulido AM. 1990. An Assessment of Childhood Lead Poisoning: A Demographic Profile of Ten Community Areas in Chicago. Paper presented at the National Minority Health Conference: Focus on Environmental Contamination. Sponsored by the Agency for Toxic Substances Disease Registry. Atlanta, Georgia. December 6, 1990. Fulton M, Raab G, Thomson 0, Laxen D, Hunter R, Hepburn W. 1987. Influence of Blood-lead on the Ability and Attainment of Children in Edinburgh. Lancet 1 (8544):1221-1226. Grant LD, Davis JM. 1989. Effect of Low-level Lead Exposure on Pediatric Neurobehavioral and Physical Development: Current Findings and Future Directions. In: Smith M, Grant LD, Sors A, eds. Lead Exposure and Child Development: An International Assessment . Dordrecht, Netherlands. Klumer Academic Publishers. 49-115. Griffin TB, Coulsion F, Golgerg L, Wills H, Russell JC Knelson JH. 1975. Clinical Studies on Men Continuously Exposed to Airborne Particulate Lead. In: Griffin TB, Knelson JG, eds. Lead . Stuttgart, West Germany: Georg Theme Publisher, New York, Academic Press. 221-240. Gunderson EL. 1988. FDA Total Diet Study, April 1982-April 1984, Dietary Intakes of Pesticides, Selected Elements, and Other Chemicals. Journal Association of Official Analytical Chemists . November/December 1988. 71(6): 1200-1209. Hawk BA, Schroeder SR, Robinson G, Otto D, Mushak P, Kleinbauni D, Dawson G. 1986. Relation of Lead and Social Factors to 10 of Low-SES Children: A Partial Replication. American Journal of Ment Defic . 91(2):178-183. Hernberg S, Nikkanen J. 1970. Enzyme Inhibition by Lead Under Normal Urban Conditions. Lancet 1 (7637):63-64. IARC (International Agency for Research on Cancer). 1987. IARC Monovaphs on the Evaluation of the Carcinogenic Risk of Chemicals to Humans. Overall Evaluations of Carcinogenicity: An Updating of the IARC Monographs . Vols. 1 to 42. Lyon, France. IARC, World Health Organization. (Suppl 7):230-232. Lin-Fu J. 1980. Lead and Children: A Historical Review. In: Needlemin Hi, ed. Low Level Lead Exposure: The Clinical Implications of Current Research . New York: Raven Press, 1980:3-16. Lin-Fu iS. 1982. The Evolution of Childhood Lead Poisoning as a Public Health Program. In: Chisoim JJ, O’Hara DM eds. Lead Absorption in Children: Manaaemeni Clinical and Environmental Asnecis . Baltimore, MD: Urban and Schwartzcnberg. 1982: 1-10. Lin-Fu iS. 1992. Modern History of Lead Poisoning: A Century of Discovery and Rediscovery. In: Needleman HI., ed. Human Lead Exposure . Boca Raton , florida. Q C Press. 1992:23-35. Marcus AN. 1991. Use of Site Specific Data Models for Lead Risk Assessment and Risk ManagemenL In: Beck BD, ed. Symposium Overview, An Update on Exposure and Effects of L -ad. Fundamental and ADolicd Toxicology . 18:1-16. Pi dsct LEAP— Ph... 1 149 ------- Marino P, Landrigan P, Graef J, Nussbaum A, Bayan G, Bock K, Boch S. A Case Report of Lead Poisoning During Renovation of a Victorian Farmhouse. American Journal of Public Health . October 1990. 80(10):1183-1185. Mahaffey K. 1990. “Environmental Lead Toxicity: Nutrition As a Component of Intervention. In: Lucier GW, Hook E, eds. Advances In Lead Research, Environmental Health Perspectives . Research Triangle Park, North Carolina. U.S. Department of Health and Human Services, Public Health Service— National Institutes of Health, National Institute of Environmental Health Sciences. November 1990. 89:75-78. McMichael AJ, Vimpani GV, Robertson EF, Baghurst PA, aark PD. 1986. The Port Pine Cohort Study: Maternal Blood-lead and Pregnancy Outcome. J Epideni Commun Health 40:18-25. Mielke HW, Blake B, Burroughs S, Hassinger N. 1984. Urban Lead Levels in Minneapolis: The Case of the Hniong Children. Environ Res 34:64-76. Moore MR, Goldberg A, Pocock SJ, Meredith A, Steward IM MacAnespic H, Lees R, Low A. 1982. Some Studies of Maternal and Infant Lead Exposure in Glasgow. Scott Med J 27:113-122. MPCA (Minnesota Pollution Control Agency). 1987. Soil Lead Report to the Minnesota “State Legislature. A Statement by the Minnesota Pollution Control Agency and the Minnesota Department of Health . Minnesota Pollution Control Agency. St. Paul, Minnesota. Minnesota Department of Health. Minneapolis, Minnesota. June 1987. NAS (National Academy of Science). 1972. Lead: Airborne Lead in Perspective. Biologic Effects of Atmospheric Pollutants . Washington, D.C.: NAS, pp. 71-177, 281-313. Needjeman HL, Rabinowjjz M, Leviton A, Linn S, Schoenbaum S. 1984. The Relationship Between Prenatal Exposure to Lead and Congenital Anomalies. J Am Med Assoc 251(22):2956-2959. Needleman HL and Gatsonis CA. 1990a. Low Level Lead Exposure and the 10 of Children, A Mets- analysis of Modern Studies. JAMA . February 2, 1990. 263(5):673-678. Needleman HL 1990b. The Future Challenge of Lead Toxicity”. In: Lucier GW, Hook E, eds. Advances In Lead Research. Environmental Health Perspectives . Research Triangle Park, North Carolina. U.S. Department of Health and Human Services, Public Health Service— National Institutes of Health, National Institute of Environmentaj Health Sciences. November 1990. 89:85-89. Nye LJJ. 1929. An Investigation of the Extraordinary Incidence of Chronic Nephritis in Young People in Queensland. Med I Ansi 2:145-159. Piomelli S, Corash L, Corash MB, Seaman C, Mushak P, Glover B, Padgeti R.. 1980. Blood-lead Concentrations in a Remote Himalayan Population. Science . 210:1135-1 37. Pope A. 1986. Exposure of Children to Lead-based Paints . Research Triangle Park, NC: U.S. Environmental Protection Agency, Strategies and Air Standards Division, EPA Contract No. 68-02-4329. Rabinow z MB, Wetherill OW, Kopple ID. 1976. Kinetic Analysis of Lead Meiabolism in Healthy PrqJ.ct LEAP— Phi.. 1 150 ------- Humanas. J Cli i i Invest 58(2):260-270. Rabmowitz MB, Wetherill GW, Kopple ID. 1977. Magnitude of Lead Intake from Respiration by Normal Man. J Lab Clin Med 90(2):238-248. Robinson G, Baumann S. Kleinbaum D, et aL 1985. Effects of Low to Moderate Lead Exposure on Braunstem Auditory Evoked Potentials in Children . Copenhagen, Denmark: World Health Organization Regional Office for Europe 177-182. (Environmental Health Document 3). Roels H, Buchet I-P Lauwerys R, et a!. 1976. Impact of Air Pollution by L ad on the Heme Biosynthetic Pathway in School-age Children. Arch Environ Health 31:310-316. Rosen JF, Chesney RW, Hamsira A, DeLuca HF, Mahaffey KR. 1980. Reduction in 1,25- dihydioxyvitamin D in Children With Increased Lead Absorption. N En21 I Med 302(20):1128-1 131. Schwartz I, Angle C, Pitcher H. 1986. Relationship Between Childhood Blood-lead Levels and Stature. Pediatrics 77(3):281-288. Schwartz I, Otto D. 1987. Blood-lead, Hearing Threshold, and Neurobehavioral Development in Children and Youth. Arch Environ Health 42(2):153-160. Secchi GC, Erba L, Cambiaghi G. 1974. Delta-amiolevulinic Acid Dehydratase Activity of Erythrocy*es and Liver Tissue in Man: Relationship to Lead Exposure. Arch Environ Health 28:130-132. Succop A, O’Flaherty EJ, Bornschein RL, et al. 1987. A Kinetic Model for Estimating Changes in the Concentration of Lead in the Blood of Young Children. In: Lindberg SE, Hutchinson TC eds. International Conference: Heavy Metals in the Environment . Vol. 2, September, New Orleans, La. Edinburgh, U.K.:CEP Consultants, Ltd.:289-291. Sullivan LW. 1991. Remarks by Louis W. Sullivan, M.D., Secretary of Health and Human Services. In Preventing Childhood Lead Poisonina The First Comprehensive National Conference October 6.7 and 8, 1991 Final Report . Alliance To End Childhood Lead Poisoning. Washington, D.C. 1991.A-1-A-10. Thornton I, Davies D, Watt J, Quinn M. 1990. Lead Exposure in Young Children front Dust and Soil in the United Kingdom. In: Lucier OW, Hook E, eds. Advances In Lead Research. Environmental Health Perspectiv . Research Triangle Park, North Carolina. U.S. Department of Health and Human Services, Public Health Service— National Institutes of Health, National Institute of Environmental Health Sciences. November 1990. 89:55-60. Vimpani GV, Baghurat PA, Wigg NR, Robertson EF, McMichael AJ, Roberts RR. 1989. The Port Pine Cohort Study—Cumulative Lead Exposure and Neurodevelopmenial Status at Age 2 Years: Do HOME Scores and Maternal 10 Reduce Apparent Effects of Lead on Bayley Mental Scores. In: Smith MA, Grant LD, Sors A, eds. Lead Exposure and Child Development: An International Assessment Dordrecht Netherlands. Klumer Academic Publishers. 332-344. Wads 0, Yano Y, Ono T, Toyokawa K. 1973. The Diagnosis of Different Degrees of Lead Absorption; In Special References to Choice and Evaluation of Various Parameters Indicative of an Increased Lead Absorption. md Health 11.55-67. Prq sct LEAP— Pta., 1 151 ------- Watson WS, Hume R, Moore MR. 1980. Oral Absorption of Lead and Iron. Lancet 2 (8188):236-237. Winneke G, Beginn U, Ewert T, Havestadt C, Kraemer U, Krause C, Thron HL Wagner HM. 1985a. Comparing the Effects of Perinatal and Later Childhood Lead Exposure on Neurobehavioral Outcome. Environ Res 38:155-167. Winneke G, Grockhaus A, Collet W, et al. 1985b. Predictive Value of Different Markers of Lead- exposure for Neuropsychological Performance. In: Lekkas Td, ed. International Conference: Heavy Metals in the Environment, September, Athens, Greece, Vol. 1. Edingburgh, U.K.:CEP Consultants, Ltd., pp. 44- 47. Wolf AW, Ernhart CB, White CS. 1985. Intrauterine Lead Exposure and Early Development. In: Lekkas TD, ed. International Conference: Heavy Metals in the Environment . September, Athens, Greece, Vol. 2. Edinburgh, U.K.:CEP Consultants, Ltd., pp. 153-155. Yankel AJ, von Lindern IH, Walter SD. 1977. The Silver Valley Lead Stu4y: The Relationship Between Childhood Blood Lead Levels and Environmental Exposure. J Air Pollut Control Assoc 27(8):763-767. Ziegler EE, Edwards BB, Jensen RL, Mahaffey KR, Foinon SJ. 1978. Absorption and Retention of Lead by Infants. Pediatric Res . 12(1):29-34. Zündahl RL Hasse JJ. 1977. In Lead in the Environment A Report and Analysis of Research at Colorado State University. University of Illinois at Urbana-Champaign and University of Missouri at Rolla, Prepared for the National Science Foundation . NSFIRA-770214. Boggess WR and Wixson BO, ed. National Science Foundation. U.S. Government Printing Office, Washington, D.C. P Jstt lEAP— Pi . 1 152 ------- BIBUOGRAPHY Alliance to End Childhood Lead Poisoning. 1991. Childhood Lead Poisoning Prevention A Resource Directory (2nd Edition’) . Washington, D.C. :National Center for Education in Maternal and Child Health. Alliance to End Childhood Lead Poisoning. 1991. Resource Guide for Financing Lead-Based Paint Cleanup . Washington, D.C. Alliance to End Childhood Lead Poisoning. 1991. Guide to State Lead Sčreening Laws . Washington, D.C. City of Chicago. 1991. U.S. Census of Chicago. Race and Latino Statistics for Census Tracts. mmunity Areas and City Wards:1980-1990 . Department of Planning, City of Chicago. Chicago, Ii. February 1991. DHUD (Department of Housing and Urban Development). 1990. National Survey of Lead-based Paint in Housing Documentation of Analytical Data Files . U.S. Department of Housing and Urban Development. Washington, D.C. DHUD (Department of Housing and Urban Development). 1990. Lead-Based Paint; Interim Guideline for Hazard Abatement in Public and Indian Housinn: Notice . Office of Assistant Secretary for Public and Indian Housing, U.S. Department of Housing and Urban Development. April 18, 1990. Federal Register 55(75): 14555-14789. EPA (Environmental Protection Agency). 1985a. Health and Environmental Effects Profile for Lead Alkyls . ECAO-CIN-P133. Cincinnati, Ohio:Environniental Criteria and Assessment Office, EPA. EPA (Environmental Protection Agency). 1985b. National Primary Drinking Water Regulations; Synthetic Organic Chemicals, Inornanic Chemicals and Microorganisms; Proposed Rule. 40 CFR Part 141 . Fed Regist 50(219):46935-47025. EPA (Environmental Protection Agency). 1985c. Notification Requirements: Reportable Quantity Adjustments; Final Rule and Proposed Rule. 40 CFR Parts 117 and 302. Fed Regist 50(65):13456-13475 , 13489-13490. EPA (Environmental Protection Agency). 1985d. Rc lation of Fuels and Fuel Additions: Gasoline Lead Content. Final Rule. 40 CFR Part 80 . Fed Regist 50:9386-9399. EPA (Environmental Protection Agency). 1987. Industrial Source Complex (ISC Dispersion Model Users Guide- Second Edition ( Revised’) Volume 1 . EPA-450/4-88-002a. Office of Air Quality Planning and Standards, Office of Air and Radiation, U.S. Environmental Protection Agency. Research Triangle Park N.C. December 1987. EPA (Environmental Protection Agency). 1989. Exuosure Factors Handbook Final Report . EPA/600//8- 89R)43. Office of Health and Environmental Assessment Office of Research and Development, U.S. Environmental Protection Agency. Washington, D.C. March 1989. Projact LEAP— Pb 1 153 ------- EPA (Environmental Protection Agency). 1991. Three City Urban Soil-Lead Demonstration Project Midterm Project Update . Office of Solid Waste and Emergency Response, U.S. Environmental Protection Agency. May 1991. EPA (Environmental Protection Agency). 1991. 4OCFR Parts 141 and 142, Maximum Contaminant Level Goals and National Primary Drinking Water Regulations for Lead and Copper; Final Rule . Federal Register. June 7, 1991. 56(11O):26459-26564. Kleinbaum, DG, Kupper LL. 1978. Applied Regression Analysis and Othes Multivariable Methods . The University of North Carolina at Chapel Hill. Duxbury Press. Boston, Mass. 1978. Randerson D, ed. 1984. Atmospheric Science and Power Production . Technical Information Center, Office of Technology Information, Unites States Department of Energy. DE84005177 (DOE/TIC-27601). SAS Institute Inc. 1985. SAS Introductory Guide for Personal Computers. Version 6 Edition . Cary, NC:SAS Institute Inc. 1985. SAS Institute Inc. 1987. SAS/STATTM Guide for Personal Computers. Version 6 Edition . Cary, NC: SAS Institute Inc. 1987. SAS Institute Inc. 1988. SAS Language Guide for Personal Computers, Release 6.03 Edition . Cary, NC: SAS Institute Inc. 1988. U.S. Department of Commerce. 1988. County and City Data Book 1988 . Bureau of the Census, U.S. Department of Commerce. U.S. Government Printing Office. Washington, D.C.: 1988. P j.1t LEAP— Phi.. 1 154 ------- |