Blood concentrations of 11 volatile organic compounds (VOCs) were measured up to four times over 2 years in a probability sample of more than 150 children from two poor, minority neighborhoods in Minneapolis, Minnesota. Blood levels of benzene, carbon tetrachloride, trichloroethene, and m-/p-xylene were comparable with those measured in selected adults from the Third National Health and Nutrition Examination Survey (NHANES III), whereas concentrations of ethylbenzene, tetrachloroethylene, toluene, 1,1,1-trichloroethane, and o-xylene were two or more times lower in the children. Blood levels of styrene were more than twice as high, and for about 10% of the children 1,4-dichlorobenzene levels were [greater than or equal to] 10 times higher compared with NHANES III subjects. We observed strong statistical associations between numerous pairwise combinations of individual VOCs in blood (e.g., benzene and m-/p-xylene, m-/p-xylene and o-xylene, 1,1,1-trichloroethane and m-/p-xylene, and 1,1,1-trichloroethane and trichloroethene). Between-child variability was higher than within-child variability for 1,4-dichlorobenzene and tetrachloroethylene. Between- and within-child variability were approximately the same for ethylbenzene and 1,1,1-trichloroethane, and between-child was lower than within-child variability for the other seven compounds. Two-day, integrated personal air measurements explained almost 79% of the variance in blood levels for 1,4-dichlorobenzene and approximately 20% for tetrachloroethylene, toluene, m-/p-xylene, and o-xylene. Personal air measurements explained much less of the variance (between 0.5 and 8%) for trichloroethene, styrene, benzene, and ethylbenzene. We observed no significant statistical associations between total urinary cotinine (a biomarker for exposure to environmental tobacco smoke) and blood VOC concentrations. For siblings living in the same household, we found strong statistical associations between measured blood VOC concentrations. Key words: biomarkers, blood concentrations, children's health, cotinine, environmental justice, environmental tobacco smoke, exposure assessment, interchild variability, intrachild variability, personal exposure, volatile organic compounds. doi:10.1289/ehp.7412 available via http://dx.doi.org/[Online 22 November 2004]
Volatile organic compounds (VOCs), many of which exhibit acute and chronic toxicity in people, are common constituents of cleaning and degreasing agents, deodorizers, dry-cleaning processes, paints, pesticides, personal care products, and solvents. Numerous VOCs are also components of automotive exhaust, industrial emissions, and environmental tobacco smoke (ETS), and they can be released into the air during showering or bathing in chlorinated water. Airborne VOCs are therefore ubiquitous in urban and nonurban environments, in indoor and outdoor settings, and in occupational and nonoccupational situations (Adgate et al. 2004a, 2004b; Edwards et al. 2001b; Kim et al. 2002; Sexton et al. 2004a, 2004b, 2004c; Wallace et al. 1985, 1987, 1988).
Although data on nonoccupational exposures to VOCs are scarce, it is apparent that concentrations of many VOCs tend to be higher indoors than outdoors and that personal (breathing zone) exposures are likely to be higher than matched in-home concentrations (Adgate et al. 2004a, 2004b; Edwards et al. 2001b; Kim et al. 2002; Sexton et al. 2004b, 2004c; Wallace et al. 1985, 1987, 1988). Research also demonstrates that nonoccupational exposures can produce corresponding blood VOC concentrations in the parts-per-trillion to parts-per-billion range (Ashley et al. 1992, 1994, 1996, 1997; Brugnone et al. 1989, 1992, 1995; Churchill et al. 2001). Children are a potentially at-risk population because they may be both more exposed to VOCs and more susceptible to adverse effects than adults. It is well established, for example, that children can be affected by different sources, pathways, and routes of exposure than adults; that children often have greater intake of air, food, beverages, soil, and dust per unit body weight and surface area; and that children differ from adults in terms of important pharmacokinetic and pharmacodymanic parameters (Aprea et al. 2000; Bearer 1995; Guzelian et al. 1992; Needham and Sexton 2000). Yet despite these concerns, it is difficult to estimate VOC-related health effects accurately because there is a paucity of information on childhood VOC exposures (Adgate et al. 2004a, 2004b; Morello-Frosch et al. 2000; Sexton et al. 2004a; Wallace 2001; Woodruff et al. 1998). In this study, we examined longitudinal measurements of blood VOC concentrations for a probability sample of elementary school-age children from two economically disadvantaged neighborhoods in Minneapolis and explored correlations with matched measurements of personal exposure to airborne VOCs and total urinary cotinine levels.
Materials and Methods
The School Health Initiative: Environment, Learning, Disease (SHIELD) study examined children's exposure over time to complex mixtures of environmental agents, including VOCs, ETS, metals, pesticides, and allergens.
Subjects. The children and families participating in the SHIELD study were from two of the most disadvantaged and ethnically diverse neighborhoods in Minneapolis: Lyndale and Whittier. For the 150 children/families in the study, total annual household income was < $9,999 for 27% of the households, between $10,000 and $19,999 for 30%, and between $20,000 and $29,999 for 21%. Just 3% of the households earned > $50,000 annually. Forty-four percent of the participating households had no occupant with a high school degree or equivalent, 32% had at least one occupant with a high school degree or equivalent, and 23% had at least one occupant who was a college graduate or technical certificate holder. In fall 1999, of the 558 children enrolled in either the Lyndale or Whittier elementary schools, 43% were African American, 20% were recent immigrants from Somalia, 20% were Hispanic (primarily Mexican American), 7% were white, 6% were Asian, and 3% were Native American. Just over half of the children (54% at Lyndale and 52% at Whittier) lived in a household where English was the primary language. As a further indicator of poverty, > 75% of the children attending each school received either free or reduced-cost meals through the National School Lunch/Breakfast Program.
Data collection. This study was approved by the University of Minnesota Research Subjects' Protection Program Institutional Review Board: Human Subjects Committee. Only a brief synopsis is provided here because details of the study design (Sexton et al. 2000) and recruitment, retention, and compliance results (Sexton et al. 2003) have been published previously. A stratified random sampling strategy was used to select SHIELD participants from students in grades 2-5 (age range, 6-10 years) at either the Lyndale or Whittier elementary schools in south Minneapolis, and age-eligible siblings were also allowed to participate. In fall 1999, children and their families selected for SHIELD were contacted based on enrollment information provided by the Student Accounting Department, Minneapolis Public Schools. After successful contact, recruiters met with children and caregivers in their homes to explain the study and answer any questions. Recruiters obtained verbal and written consent/assent and administered the baseline questionnaire (which asked questions about demographic, socioeconomic, and housing attributes) to the 152 children/families who agreed to be in the study, plus 51 siblings. At enrollment the primary caregiver was asked a series of questions about smoking status and behavior, as well as questions about socioeconomic status, residential characteristics, and the child's health.
During winter (January-February) and spring (April-May) of both 2000 and 2001, children were asked to give blood samples, which were collected at school by a trained phlebotomist. The phlebotomist attempted to obtain a 33-mL venipuncture blood sample from each child during each of the four monitoring sessions. Urine samples were also collected at the same time.
For the 2 days preceding collection of a blood sample, children, with the help of caregivers, interviews/translators, and field technicians, were asked to maintain a time-activity log, which recorded the location and approximate time they spent in seven different microenvironments. They also were asked to answer questions about the location and approximate time they spent in the presence of an active smoker. During winter and spring 2000, children also were asked to wear or carry a small passive sampler throughout the same 2-day period to measure airborne VOC concentrations. At times when it was impractical to wear or carry the monitor, such as while sleeping, children/families were instructed to place the monitor as near as possible to the child's head (e.g., on a nightstand next to the bed). For year 1 of SHIELD, the enrollment rate was 57%, the retention rate was 85%, and > 80% of children provided requested blood …