2000 Progress Report: School-Based Study of Complex Environmental Exposures and Related Health Effects in Children Part A - Exposure

EPA Grant Number: R825813
Title: School-Based Study of Complex Environmental Exposures and Related Health Effects in Children Part A - Exposure
Investigators: Sexton, Ken , Adgate, John L. , Church, Timothy , Greaves, Ian , Ramachandran, Gurumurthy , Tweedie, Richard L.
Institution: University of Minnesota , The University of Texas Health Science Center at San Antonio
EPA Project Officer: Saint, Chris
Project Period: March 1, 1998 through March 1, 2001 (Extended to January 25, 2003)
Project Period Covered by this Report: March 1, 1999 through March 1, 2000
Project Amount: $899,264
RFA: Issues in Human Health Risk Assessment (1997) RFA Text |  Recipients Lists
Research Category: Human Health , Health Effects , Human Health Risk Assessment , Health


The objectives of the study are (1) to document complex exposure patterns involving multiple exposures to single chemicals and exposures to chemical mixtures for elementary-aged school children, (2) to examine temporal variability by monitoring complex exposures in these children over time, (3) to apportion the relative contribution to measured personal exposures of outdoor community air, air inside the child's school, and air inside the child's residence, (4) to evaluate the relationship between measured exposures and internal dose using biological markers of exposure in blood and urine, and (5) to compare children's complex exposures between a new school designed and operated to improve indoor air quality and an older school with more conventional architecture, mechanical systems, and furnishings.

Progress Summary:

The SHIELD (School Health Initiative: Environment, Learning, and Disease) study embodies two overarching goals: to characterize exposures to multiple environmental agents for elementary school children from two economically disadvantaged neighborhoods in south Minneapolis; and to explore related effects on respiratory health and learning outcomes.

Study Population -The SHIELD study is designed to measure important environmental health variables for more than 550 children in grades 2 through 5 enrolled at two elementary schools, Lyndale and Whittier, in south Minneapolis. The Lyndale elementary school is located 8 blocks south of the Whittier school and similar numbers of children (269 at Lyndale, 289 at Whittier) were enrolled in grades 2 - 5 at the start of the 1999/2000 school year. Each is a neighborhood school and, although most of the children live in relatively close proximity (within a few blocks), virtually all ride buses to school. The residential neighborhoods of Lyndale and Whittier are adjacent to each other and both are home to many low-income families and people of color, including recent immigrants from Africa and Southeast Asia.

The 558 children eligible for SHIELD comprise a diverse mix of ethnic and racial backgrounds: 356 (63.8%) black (including 114 Somali); 114 (20.4%) Hispanic; 37 (6.6%) white; 35 (6.3%) Asian; and 16 (2.9%) Native American. Most of these children (more than 75% at each school) receive either free or reduced-cost meals through the National School Lunch/Breakfast Program. Participation in this program is an indicator of poverty; for example, a four-person household must demonstrate total earnings of less than $596 per week to qualify. Just over half the children, 146 at Lyndale (54.3%) and 149 at Whittier (51.6%), live in households where English is the primary language. The rest of the children's families speak a variety of other languages, primarily Somali or Spanish.

Personal, Environmental, and Biological Exposure Measures - The main objective of the exposure assessment component of SHIELD is to characterize children's exposure to multiple environmental stressors, including: volatile organic chemicals (VOCs), allergens, airborne particulate matter (PM2.5), environmental tobacco smoke, metals, pesticides, polychlorinated biphenyls (PCBs), and phthalates. Primary emphasis was placed on obtaining comprehensive information about children's inhalation exposure to selected VOCs because these compounds are ubiquitous in both indoor residential and outdoor community air, and because they may affect children's neurobehavior development and/or respiratory health. The data were collected (1) to allow for determination of the relative contributions of residential, school, and ambient environments to measured VOC exposures, (2) to estimate the additional VOC exposure resulting from contact with environmental tobacco smoke, and (3) to compare environmental and personal VOC concentrations with matched VOC levels in blood.

Respiratory Health - Respiratory tract illnesses, including infections, allergic rhinitis, asthma, and non-specific respiratory irritation, are common causes of morbidity in elementary school children. The prevalence and severity of asthma, for example, has increased over the past two decades in the United States, with the greatest increase occurring for poor minority children living in the inner city. Although the etiologies of asthma and other respiratory tract illnesses tend to be complex and not well understood, among the contributing factors are environmental exposures to allergens (e.g., cockroaches, dust mites) and hazardous chemicals (e.g., environmental tobacco smoke), along with genetic predisposition and other factors that play a role in this growing public health problem.

Results from questionnaires on allergies, asthma, infections, and non-specific respiratory illness, as well as results from lung function tests (spirometry and peak flow) were used to characterize respiratory health. The SHIELD study explores possible causal relationships between these variables and exposure-related measurements, including: (a) measured environmental concentrations (in-home and in-school levels of allergens in carpets); (b) measured personal exposures (breathing zone levels of VOCs); and (c) measured values for biological markers of exposure (e.g., urinary cotinine, IgE levels in blood).

Learning Outcomes - Neurobehavioral effects from low-level environmental exposures to lead, methylmercury, and PCBs have been documented in children. It is postulated that certain pesticides, VOCs, and polyaromatic hydrocarbons also cause such effects. Research to understand the nature and magnitude of related health risks in children is complicated by the difficulty of distinguishing chemically induced problems with coordination, perception, and cognitive ability from the effects of genetic, social, and cultural factors. The SHIELD study explores possible causal links between measures of learning outcomes (scores on standardized tests administered every spring, academic performance as measured by grades) and measured levels of neurotoxins in environmental samples (levels of VOCs inside/outside the child's residence and school), personal exposure samples (levels of VOCs in the child's breathing zone air), and blood (VOCs, lead, mercury, persistent pesticides, PCBs) and urine (organophosphate pesticides, metals) samples.

Study Design - The SHIELD study has three primary objectives: (1) to compare exposures to multiple environmental agents for children attending two elementary schools; (2) to compare respiratory health and learning outcomes for children attending two elementary schools; and (3) to model associations between and among environmental measurements (e.g., outdoor, indoor, and personal VOC levels), biological markers of exposure, susceptibility, or effect (e.g., blood VOCs), and respiratory health effects (e.g., lung function) and learning outcomes (e.g., standardized test scores). Recruitment for SHIELD occurred from November 1999 through January 2000, and the data collection portion of the study was accomplished during two monitoring sessions, winter (February-March 2000) and spring (April-May 2000).

The design takes advantage of similarities in sociodemographic characteristics between residents of the two neighborhoods and between students enrolled at Lyndale and Whittier to reduce related effects on the outcomes of interest. Although observational in nature, the study utilizes students from the two schools as non-randomized, concurrent comparison groups to assess differences in both exposures and effects. Children at each school were randomly sampled in a way that accounts for over-representation of families with multiple children and assures adequate representation by gender, age, and ethnic/racial group. Determining associations such as those between exposures to different environmental agents or between exposures and respiratory health requires statistical modeling of complex relationships to account for various confounders. The epidemiological portion of the study is essentially ecologic, but ecological bias is reduced by measuring intermediate individual variables (biological markers of exposure) to verify school-based comparisons of the link between environmental exposures and health or learning outcomes. Data were also collected on other factors that may affect the comparisons, such as socioeconomic status, home environment, diet, and exposure to tobacco smoke, in order to adjust for these potential confounders.

Results from the Pilot Study - A pilot study was conducted at the Lyndale and Whittier schools to test and refine methods for recruiting children/families (January - April 1999), and for monitoring exposures and performing lung function tests (May 1999). A random sample of 78 "index" children, stratified by school, grade, and language, was selected from the 2nd, 3rd, and 4th grades. Any siblings of these index children who were in grades 2 - 4 were also asked to volunteer for the pilot study. The total recruitment pool therefore consisted of 96 children: 78 index children and 18 siblings.

Of the 78 randomly selected children/families, 8 (10%) had transferred their children to another school and were therefore ineligible, 27 (35%) could not be contacted, 9 (12%) were contacted but recruitment visits could not be scheduled/completed, and 12 (16%) refused after talking with recruiters. A total of 22 index children (plus 7 siblings) was enrolled in the pilot study, an overall response rate of 31.4%% (22/70 eligible index children: the 8 children transferred were not eligible). The response rate was similar at Lyndale (29%) and Whittier (26%), but there were more transfers at Lyndale (16% versus 5%) and more refusals at Whittier (23% versus 8%). There was a dramatic difference in response rates between English- (15%) and non-English- (67%) speaking families, with English-speaking families more likely to transfer (14% versus 0%), and to pose contact (41% versus 17%) and follow-through problems (15% versus 0%).

We obtained informed consent/assent and completed in-home baseline interviews for the 29 children (22 index + 7 siblings). The mobility of these children/families is illustrated by the relatively short time they reported being at their present residence (median 8 months, average 15 months, range 2 to 63 months). Two children/families subsequently dropped out prior to the May 1999 monitoring period (one transferred to another school and one could not be contacted). Of the 27 children who participated in the SHIELD pilot study, all (100%) completed spirometry testing, wore a personal VOC monitor, kept a time-activity diary, and provided a urine sample. Twenty-four of 27 children (89%) agreed to provide a venipuncture blood sample (3 refused at the time of the baseline interview; 1 was absent from school during the period of sample collection). A 33 ml blood sample (4 tubes) was obtained from 18 children (67%), flow stoppage resulted in partial samples (approximately 20 - 30 mls) from 2 others (7%), and for 3 children (11%) no sample was obtained even though blood collection was attempted (in one case the first attempt was unsuccessful and the child refused a second attempt; in another there was a problem with collection tubes and the child refused a second attempt; and one child's veins were too small).

Results from the SHIELD Study - Lessons learned in the pilot study were used to refine the recruitment protocol for SHIELD. The major changes implemented for the full study were designed primarily to increase response rates among English-speaking families (although we also modified the sampling protocol), and included: (1) conducting community outreach activities and attending school-sponsored meetings to encourage participation; (2) distributing a brochure, sometimes as a hanger on residential door handles, that described the study; and (3) using school employees as recruiters for English-speaking families. Recruitment for SHIELD began in November 1999 and continued through January 2000.

A random sample of 311 index children, stratified by school, grade, and language was selected from the 2nd, 3rd, 4th, and 5th grades at both schools. If the index child had siblings in grades 2 - 5 they were also asked to participate in SHIELD (51 siblings were enrolled). Of the 311 randomly selected children, 41 (13%) had transferred to another school by the time we tried to contact them, which made them ineligible for the study (therefore these children are not included when calculating response rates). For the 270 index children eligible for SHIELD, 60 (19%) were not enrolled because we had problems contacting their guardians. We were able to contact the remaining 210 guardians but 24 (8%) were not enrolled because we had problems scheduling or completing recruitment, and 33 (11%) declined to volunteer after talking with recruiters. The overall enrollment (response) rates were comparable at the Lyndale (46%) and Whittier (52%) schools, but the rate for non-English-speaking families/children was higher at Whittier (85.9% versus 55.2% at Lyndale), while the rate for English-speaking families/children was higher at Lyndale (47.8 % versus 34.4% at Whittier).

A comparison of response (enrollment) rates between the pilot and SHIELD is presented in Table 1. Rates for non-English-speaking families went up slightly (from 66.7% - 71.0%), while rates for English-speaking families increased considerably (from 19.2% to 41.7%). The overall response (enrollment) rate increased from 31.4% in the pilot to 56.7% in SHIELD. These results indicate that the changes made to the SHIELD recruitment protocol, based on our experience in the pilot, were successful in increasing enrollment rates for English-speaking families.

Preliminary results from data collection activities are available from the first of two five-week monitoring periods, which occurred during January-February 2000. Of the 153 index children participating in SHIELD, 139 (90.9%) volunteered at the time they were enrolled to later provide blood samples while the other 14 (9.1%) declined but agreed to complete most other monitoring protocols. One hundred and twenty-three (88.5%) of the 139 showed up at the clinic on schedule, and a 33 ml sample was obtained for 106 (76.3%) while partial samples (<33 mls) were obtained for another 11 (7.9%). Blood draws were attempted on 4 (2.9%) but no sample was obtained due to technical difficulties, and 2

Table 1. Comparison of Enrollment (Response) Rate for the Pilot and SHIELD studies.

  Enrollment (Response) Ratea
Primary Language
Spoken by Family
Pilot Study (May 99)
n = 22 index children
SHIELD (Nov 99 - Jan 00)
n = 153 index children
  enrolledb enrolledc
Non-English 66.7% (12/18) 71.0% (98/138)
English 19.2% (10/52) 41.7% (55/132)
Total (English + Non-English) 31.4% (22/70) 56.7% (153/270)


a - Enrollment rate = number of index children actually enrolled divided by the total number eligible (i.e., not including transfers) times 100
b - In addition to the 22 index (randomly selected) children enrolled, 7 siblings were also included in the pilot study
c - In addition to the 153 index (randomly selected) children enrolled, 51 siblings were also included in SHIELD (16 Hispanic, 15 Somali, 9 black, 5 white, 4 Cambodian, 1 Laotian, 1 Native American)

(1.4%) did not give their assent for the phlebotomist to draw blood. No blood draws were attempted on 16 (11.5%) of the 139 children because: they moved (7); did not show up for scheduled monitoring appointments and could not be contacted (5); refused prior to coming to the clinic (3); or were unavailable on scheduled blood-draw days (1). Of the 51 siblings enrolled in SHIELD, 49 had previously agreed to provide blood samples. Complete 33 ml blood samples were obtained for 39 (79.6%) and partial samples for 2 (4.1%). No blood collection was attempted for 8 children (16.3%) because 3 had moved since enrolling in SHIELD and 5 refused prior to coming to the clinic.

The 153 index children and their 51 siblings were also asked to provide urine samples. School nurses attempted to collect urine samples from 133 index children (86.9%) and were successful in obtaining at least 10 mls in all cases. No attempt was made to collect samples from 20 index children (13.1%) because 7 had moved, 5 did not show up for scheduled appointments, 3 refused prior to coming to the clinic, and 5 were unavailable on scheduled sampling days. School nurses tried to collect urine samples from 44 siblings (86.3%) and successfully obtained at least 10 mls for 43 (84.3%). No samples were attempted for 7 siblings (13.7%) because 2 had moved and 5 refused prior to coming to the clinic.

All of the 153 index children volunteered to wear small passive VOC dosimeters, and 135 (88.2%) were available to start the first 48-hour monitoring period (18 were unavailable because they had moved, we could not schedule appointments, or they had changed their mind). Dosimeters were retrieved from 128 children (83.7%) (4 were lost, 1 was left at school, 1 was returned to the school nurse, and 1 child dropped out of the study). In addition, 3 of the returned dosimeters were invalid because they either had not been worn by the child or had been taken off prior to the end of the 48-hour monitoring period. Of the remaining 125 (81.7%), the field team raised validity questions about 16 (10.5%) because there were obvious signs of tampering or damage, foreign material was on the badge, or there was greater than 48-hour exposure. Overall, acceptable personal 48-hour VOC samples were obtained for a total of 109 (71.2%) out of a target of 153 children.

Conclusions - Knowledge gained from SHIELD provides insight into the effectiveness of a school-based design for recruiting and monitoring children from economically disadvantaged neighborhoods. Our experience suggests this approach offers a practical and affordable way to address many of the inherent obstacles that often hinder environmental health research in this population. The primary advantages of a school-based design are numerous: (1) the process of identifying households with age-eligible children is direct, simple, and relatively inexpensive; (2) contact information (i.e., names, telephone numbers, addresses) and sociodemographic information (e.g., race/ethnicity of child, language spoken at home) is readily available, provided appropriate safeguards are in place to protect privacy; (3) the involvement of school personnel (e.g., recruitment letter from the principals, use of bilingual education assistants as recruiters) lends credibility to the study and increases the likelihood that children/families will volunteer to participate; (4) information available from the schools (e.g., race/ethnicity of child, language spoken at home, academic performance, standardized test scores) makes it easier to assess differences in responders and non-responders; and (5) the in-school collection of biological samples (i.e., blood, urine) and testing of lung function (i.e., spirometry, peak flow) is a convenient and effective way to monitor children's environmental health. The key to successful implementation is developing close working relationships, in effect a partnership, with school personnel and residents of the local community, ensuring that all partners are kept informed and meaningfully involved.

Notwithstanding these advantages, response rates for English-speaking, predominantly African American families were not as high as we had hoped. Although the school-based approach allowed us to identify all eligible children/families, obtain the best address information available, and involve school personnel in recruitment, the overall response (enrollment) rate for SHIELD was still only 56.7% (71.0% non-English-speaking families and 41.7% English-speaking families). These findings highlight the continuing need to improve our understanding of factors (e.g., cultural, economic, psychological, social) that encourage or discourage participation among this population, with special emphasis on identifying cost-effective methods for study recruitment.

Once enrolled, however, the vast majority of SHIELD children/families participated fully in this relatively burdensome study, doing their best to comply with sometimes-demanding study protocols and willingly providing blood and urine samples. These results suggest that, despite certain obstacles, it is feasible to make probability-based assessments of children's environmental exposures and related health effects in economically disadvantaged neighborhoods.

Journal Articles on this Report : 2 Displayed | Download in RIS Format

Other project views: All 5 publications 5 publications in selected types All 5 journal articles
Type Citation Project Document Sources
Journal Article Sexton K. Socioeconomic and racial disparities in environmental health: is risk assessment part of the problem or part of the solution? Human and Ecological Risk Assessment 2000;6(4):561-574. R825813 (2000)
R825813 (2001)
  • Abstract: Taylor & Francis - Abstract
  • Journal Article Sexton K, Greaves IA, Church TR, Adgate JL, Ramachandran G, Tweedie RL, Fredrickson A, Geisser M, Sikorski M, Fischer G, Jones D, Ellringer P. A school-based strategy to assess children's environmental exposures and related health effects in economically disadvantaged urban neighborhoods. Journal of Exposure Analysis and Environmental Epidemiology 2000;10(6 Pt. 2):682-694. R825813 (2000)
    R825813 (2001)
    R826789 (2000)
    R826789 (2001)
    R826789 (2002)
    R826789 (Final)
  • Abstract from PubMed
  • Other: Nature PDF
  • Supplemental Keywords:

    children's health, environmental justice, environmental tobacco smoke, indoor air quality, metals, PCBs, pesticides, sample recruitment, volatile organic chemicals, RFA, Health, Scientific Discipline, Air, Toxics, air toxics, Environmental Chemistry, Chemistry, Epidemiology, pesticides, VOCs, Risk Assessments, Susceptibility/Sensitive Population/Genetic Susceptibility, Children's Health, genetic susceptability, indoor air, school based study, pesticide exposure, cumulative risk, sensitive populations, multiple acute exposures, exposure and effects, adolescents, multiple acute exposure, indoor exposure, health risks, measuring childhood exposure, exposure, chemical mixtures, Human Health Risk Assessment, air pollution, children, PAH, assessment of exposure, human exposure, minorities, cigarette smoke, environmental toxicant, environmental stressors, exposure pathways, harmful environmental agents, toxic environmental contaminants, indoor air quality, tobacco smoke, acute exposure, heavy metals, school-based study, environmental hazard exposures, metals

    Progress and Final Reports:

    Original Abstract
    2001 Progress Report