Printer Friendly

Children's exposure to environmental tobacco smoke: using diverse exposure metrics to document ethnic/racial differences.

Four metrics were used to assess exposure to environmental tobacco smoke (ETS) for a probability sample (n = 152) of elementary school-age children in two economically disadvantaged neighborhoods: a) caregiver responses to a baseline questionnaire (BQ) about smoking status and behavior; b) 48-hr time-activity (T-A) data on location and time spent by children in the presence of tobacco smoke; c) total urinary cotinine as a marker for nicotine uptake; and d) urinary NNAL [4-(methylnitrosamino)-1-(3-pyridyl)-1-butanol] + NNAL-Gluc [4-(methylnitrosamino)-1-(3-pyridyl)-1-(O-[beta]-D-glucopyranuronosyl)butane] as a marker for uptake of the tobacco-specific lung carcinogen 4-(methylnitrosamino)-1-(3-pyridyl)-1-butanone (NNK). Consistent differences in ETS exposure by ethnicity and race were observed. Although data were insufficient to determine differences for NNAL + NNAL-Gluc, BQ responses, T-A data, and cotinine levels all indicated that average ETS exposure was highest for African-American children, moderately high for those designated "other" (white, Southeast Asian, Native American), moderately low for Hispanic children, and lowest for Somali immigrant children. For example, in February 2000, mean cotinine levels were 14.1 ng/mL for African Americans, 12.2 ng/mL for other, 4.8 ng/mL for Hispanics, and 4.4 ng/mL for Somalis. The BQ and T-A data together were reasonably good predictors of total cotinine levels (adjusted [r.sup.2] = 0.69), and based on limited data, measured total cotinine values were a relatively good predictor of NNAL + NNAL-Gluc (adjusted [r.sup.2] = 0.73). The results suggest that when children are exposed to ETS primarily in their homes, questionnaires and T-A logs might be effective screening tools for identifying those likely to experience higher uptake of nicotine. Key words: children's health, cotinine, environmental tobacco smoke, ethnicity, questionnaires, race.

**********

Environmental tobacco smoke (ETS) is a carcinogenic mixture of more than 4,000 chemicals [National Toxicology Program 2000; U.S. Department of Health and Human Services (DHHS) 1986; U.S. Environmental Protection Agency (EPA) 1992] and a potentially important determinant of children's environmental health (Carroquino et al. 1998; Hecht et al. 2001; Landrigan et al. 1998). Although available data are inconclusive (Boffetta et al. 2000), there are legitimate concerns that childhood exposure to ETS might lead to cancer later in life (Sandler et al. 1985; Tang et al. 1999). In addition, ETS may directly harm children by exacerbating preexisting asthma (Gold 2000; Jaakkola et al. 2001) and contributing to neurodevelopmental problems (Eskenazi and Castorina 1999).

Because ETS is a common constituent of the air inside many buildings and vehicles (Cummings et al. 1990; Samet 1999; Subramaniam et al. 2001), most children are exposed to some level of ETS during their normal daily activities (Hecht et al. 2001; Scherer et al. 1999). Although there are many sources of ETS in the environment, parental smoking status (and related in-home ETS concentration) is a strongly associated with biologic markers of exposure for many children, and it is well established that smoking prevalence rates vary by ethnicity/race, immigrant status, and country of birth (Baluja et al. 2003). Relatively little is known, however, about how these demographic variables affect the magnitude, duration, and frequency of children's exposure to ETS.

Better estimates of childhood ETS exposure, including delineation of the effects of important sociodemographic variables, are needed to help inform decisions about protection of children's environmental health. For example, accurate assessment of ETS exposure (as a causative agent, cofactor, confounder, or some combination) is necessary for ensuring the validity of most epidemioiogic studies designed to investigate causal links between air pollution and children's health. Exposure misclassification errors (e.g., erroneously classifying a child as "highly exposed" or "not exposed") can introduce information bias that affects the validity of findings by either exaggerating or understating the true causal link between exposure and effect (Rothman 1986). Diverse methods are available to assess children's ETS exposure, ranging from parental questionnaires about smoking status, which are relatively easy to obtain, inexpensive to analyze, and less accurate, to biomarkers of exposure (cotinine in urine), which are comparatively harder to collect, expensive to analyze, and more accurate (Benowitz 1999; Daisey 1999; Jarvis et al. 1984; National Research Council 1986, 1991). Given the goals of a particular study and the resources available, investigators must make trade-offs between using exposure metrics that deliver better accuracy at higher cost (e.g., biomarkers) versus those that provide less accuracy at lower cost (e.g., questionnaires).

In the School Health Initiative: Environment, Learning, Disease (SHIELD) study, children's exposure to ETS is assessed using two relatively inexpensive metrics: questionnaires about caregiver smoking status/behavior and time-activity (T-A) logs reporting the time and place of ETS exposure; and two relatively expensive metrics: measurement of urinary total cotinine [cotinine plus pyridyl-N-[beta]-D-glycopyranuronosyl-(S)-(-)-cotininium inner salt] as an uptake marker of nicotine, and measurement of NNAL [4-(methylnitrosamino)-1-(3-pyridyl)-1-butanol] + NNAL-Gluc-[4-(methylnitrosamino)-1-(3-pyridyl)-1- (O-[beta]-D-glucopyranuronosyl)butane] as an uptake marker of the tobacco-specific lung carcinogen NNK [4-(methylnitrosamino)-1-(3-pyridyl)-1-butanone]. Associations between measures in a probability sample of elementary school-age children enrolled in SHIELD are estimated, and ethnic and racial differences in ETS exposure are examined.

Materials and Methods

Subjects and data collection. This study was approved by the University of Minnesota Research Subjects' Protection Program Institutional Review Board: Human Subjects Committee. The participating children were part of the SHIELD study, the details of which have been published previously (Sexton et al. 2000, 2003). Two hundred sixteen students in grades 2 through 5 (6-10 years of age) at either the Lyndale or Whittier elementary schools in south Minneapolis, Minnesota, were selected based on a stratified random sampling strategy, and eligible siblings were also allowed to participate. In the rail of 1999, children and their families eligible for SHIELD were identified and 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. For the 152 who agreed to be in the study, recruiters obtained verbal and written consent, and assent as appropriate, and administered the baseline questionnaire (BQ).

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. For example, they were asked "In the past month how many packs of cigarettes did you smoke inside the home in the presence of the child?" and "How many smokers who live with the child smoke inside the child's home?"

During winter (January and February 2000) and spring (April and May 2000) of the next year, children were asked to give 40-mL urine samples at school. In addition, for the 48 hr preceding collection of a urine sample, children, with the help of caregivers, interviews/translators, and field technicians, were asked to maintain a T-A log recording the location and approximate time they spent in seven different microenvironments and also to answer questions about the location and approximate time they spent in the presence of an active smoker. A summary of smoking-related data from questionnaires and logs used in SHIELD is provided in Table 1.

Laboratory analyses. Total cotinine was measured by gas chromatography--mass spectrometry, as described in previous publications (Hecht et al. 1993, 2001). Because of cost considerations, nor all samples could be analyzed for NNAL and NNAL-Gluc. Because we considered 10 ng/mL cotinine to be an indicator of potentially elevated ETS exposure, all usable samples with total cotinine [greater than equal to] 10 ng/mL as well as a selection of those with total cotinine < 10 ng/mL were analyzed for NNAL and NNAL-Gluc, using methods described previously (Hecht et al. 2001). Analysis was by gas chromatography--nitrosamine selective detection, using methods described previously (Hecht et al. 2001).

Statistical analysis and related considerations. Index children were sampled with selection probabilities designed to equally represent strata defined by school, grade, ethnicity (dichotomized as English-speaking vs. non-English-speaking homes), and sex, and analyses were weighted to account for selection and response probabilities. Race/ethnicity was broken down further for analyses addressing this factor specifically, by aggregating categories with fewer than 15 children into the "other" category. Analyses were performed on logtransformed laboratory values to normalize the distributions and to equalize variances, and transformed means were exponentiated to obtain geometric means. Values below detection limits were set to half the limit. Previous analyses showed that results are relatively insensitive to the choice of a substitute value. Confidence intervals were calculated in the transformed scale and back-transformed by taking logs.

Multiple linear regression modeling was applied to the smoking-related variables in the BQ and the T-A log to develop predictive equations for total urinary cotinine within each monitoring session, as well as for the average over both sessions in the subset of children who provided two samples. In addition, questionnaire and T-A variables, both individually and in combination with cotinine values, were used in a similar manner to predict urinary NNAL + NNAL-Gluc, but for winter only because an inadequate number of analyzed samples was available from the spring monitoring session. All possible regressions were examined, and the adjusted multiple [r.sup.2] was used as the penalized optimality criterion because it represents an estimate of the percentage of variability explained by the regression, adjusting for the number of covariates. Some variables were not included in these models because extreme collinearity made the computations unstable. To assure applicability to the entire target population, all regressions were weighted by stratumspecific selection and response probabilities.

Use of multiple predictors can lead to optimistic estimates of the model's predictive capability. An empirical validation of the prediction equation for cotinine was performed by examining how the predicted values might perform relative to the actual values. One potential use of the measured cotinine levels would be to predict future cotinine levels for a cohort of children. With this in mind, we first examined the ability of dichotomized measured cotinine concentrations in the winter to predict dichotomized cotinine values measured in the spring using a two-by-two table. Each winter or spring cotinine value was classified by whether it was above or below 5 ng/mL, and the concordance of winter categorization with that in the spring was quantified. Next, an analogous two-by-two table was constructed based on predicted winter cotinine values (from the BQ) dichotomized above or below 5 ng/mL versus the same spring categorization. Again, the concordance of predicted winter cotinine values with measured spring values was quantified and compared with the previous table. The comparison of model-based predictors versus actual values on this simple prediction problem, summarized by the cross-product ratio, provides a measure of the validity of the model.

Results

Data from BQs were collected for 150 of the randomly selected children (referred to as "index" children) enrolled in the SHIELD study. Two urine samples, one in winter and one in spring, were obtained from 86% of these children, and 66% provided both 48-hr T-A logs (Sexton et al. 2003).

A summary of smoking-related responses on the BQ (administered at the beginning of the study) and the T-A log (completed in both the winter and spring monitoring sessions) is provided in Table 1, including a breakout by ethnic/racial group. Overall, 22% reported on the BQ that the caregiver had ever smoked inside the home, and 21% said that other occupants also smoked inside the home. Fourteen percent stated that their child had at least some exposure to ETS in vehicles, and 13% acknowledged that their child had some ETS exposure in other indoor environments besides the home. More than 20% of those completing T-A logs reported ETS exposure inside the home on both the first and second sampling day for winter and spring.

As shown in Table 1, there were substantial and consistent differences for smoking-related responses among the four ethnic/racial groups. In all instances, a considerably higher percentage of African-American children lived in a home that reported smokers (46% of caregivers smoked, 33% of other occupants smoked) and were reportedly exposed to ETS on both the first and second monitoring day in winter (> 40%) and spring (> 45%). Typically, children in the "other" category (white, Native American, Southeast Asian) were nominally the next most exposed group, followed by Hispanics and Somali immigrants. The evidence shows that Somali families almost never reported ETS exposure for their children on either the BQ or the T-A log.

Mean total cotinine levels by season are summarized in Table 2. Geometric mean cotinine concentrations were < 5 ng/mL for the entire cohort, as well as for each of the ethnic/racial groups, and no major seasonal differences were apparent. Geometric mean cotinine levels were substantially higher for African-American children (3.4 ng/mL in winter, 3.6 ng/mL in spring) compared with children classified as "other" (2.2 ng/mL in winter, 1.4 ng/mL in spring). Mean cotinine values for "other" children were, in turn, substantially higher than for Hispanic (0.6 ng/mL in winter and spring) and Somali immigrant children (0.7 ng/mL in winter, 0.4 ng/mL in spring). This general pattern (African American > "other" > Hispanic and Somali) was similar to that observed for the responses to smoking-related questions on the BQ and information derived from the T-A log (Table 1).

More detailed information on the distributions of urine cotinine by ethnic/racial group is provided in Figure 1. These data show that the general pattern of African-American > "other" > Hispanic and Somali immigrant children persisted for the middle percentiles. At the upper percentiles, total cotinine values began to converge, for example, for 95th percentiles: "other" (46 ng/mL) > African American (38 ng/mL) > Hispanic (35 ng/mL) > Somali immigrant (30 ng/mL). However, estimates of extreme percentiles are less stable, so this ordering may be arbitrary.

A summary of mean cotinine concentrations by season is provided in Table 3 for the stratification variables used to select the SHIELD probability sample. Cotinine levels were relatively constant across season, except for children in the third (2.2 ng/mL in winter, 0.8 ng/mL in spring) and fifth (1.5 ng/mL in winter, 2.8 ng/mL in spring) grades and children enrolled at the Lyndale school (3.0 ng/mL in winter, 1.5 ng/mL in spring). Mean cotinine values tended to be higher a) in girls than in boys (1.9 vs. 1.4 ng/mL in winter, 1.9 vs. 1.2 ng/mL in spring), b) for students at Lyndale than for those at Whittier in the winter (3.0 vs. 1.0 ng/mL), and c) for children from English-speaking compared with those from non-English-speaking families (3.0 vs. 0.9 ng/mL in winter, 2.9 vs. 0.6 ng/mL in spring).

Regression models were used to validate responses to smoking-related questions on the BQ, information from the T-A log, or a combination of the two as predictors of urine cotinine concentrations. The estimates of proportion of explained variance ([r.sup.2]), adjusted for the number of covariates, are compared in Table 4 for the models tested. Individually, both the BQ and T-A log predicted cotinine with reasonable reliability (adjusted [r.sup.2] > 0.45). The BQ was about as good as the T-A log, even though the BQ was administered several weeks before sample collection, whereas the T-A log was completed during the 2 days preceding sample collection. In combination, the BQ and T-A log were better yet. For example, the BQ plus the T-A log (winter and spring) explained 69% (adjusted [r.sup.2]) of the variance in average (winter and spring) urine cotinine levels compared with 58% for the BQ alone and 66% for the T-A log alone.

We also examined whether concentrations of NNAL + NNAL-Gluc measured in urine could be predicted by total cotinine values or a combination of total cotinine values and either the BQ of T-A log (Table 4). Because relatively few urine samples (< 20) were analyzed for NNAL + NNAL-Glue from the spring (Hecht et al. 2001), there were insufficient degrees of freedom available to run the full regression model. Therefore, only results from the winter monitoring sessision are summarized in Table 4. Findings indicate that total cotinine in the winter was a reasonably good predictor (adjusted [r.sup.2] = 0.73) of measured NNAL + NNAL-Gluc in the winter. A combination of either total cotinine plus BQ smoking variables (adjusted [r.sup.2] = 0.65) or total cotinine plus T-A log smoking variables (adjusted [r.sup.2] = 0.33) was less effective.

Several previous studies suggest that subjects with urinary cotinine values less than about 5 ng/mL are typically unaware of any ambient ETS exposure (Cummings et al. 1990; Jarvis et al. 1984; Pirkle et al. 1996). Consequently, despite the presence of cotinine in their urine, they would be unlikely to report ETS exposure, presumably because they were oblivious to inhalation of low levels during routine activities. Using this 5-ng/mL "awareness threshold" as a cutoff, we compared measured cotinine concentrations (< 5 ng/mL vs. [greater than equal to] 5 ng/mL) in winter and spring (Table 5) and found that children who had urine cotinine values < 5 ng/mL in the winter also tended to have values < 5 ng/mL in the spring, and rice versa (misclassification error: 8/65 = 0.12). Then we compared measured cotinine concentrations (< 5 ng/mL and [greater than equal to] 5 ng/mL) in the spring versus predicted winter values from the BQ full regression model. The results (Table 6) show that the smoking variables from the BQ did almost as well at predicting spring cotinine values (misclassification error: 9/63 = 0.14).

Discussion

This article builds on and expands earlier findings among SHIELD children. We focus here on comparing results using four different exposure metrics to assess children's exposure to ETS: BQ, T-A log, measured urinary levels of total cotinine, and measured urinary levels of NNAL + NNAL-Gluc. All four metrics suggest that a substantial fraction of elementary school--age children from two economically disadvantaged and ethnically diverse neighborhoods were routinely exposed to ETS. Moreover, three of the exposure metrics (BQ, T-A log, cotinine) indicated a consistent ethnic/racial pattern for childhood ETS exposure, with African American > "other" (Southeast Asian, white, Native American) > Hispanic > Somali immigrant. The BQ and T-A log data, although not conclusive, suggest that the home environment (e.g., caregiver or others smoking indoors in the presence of the child) was the primary source of ETS exposure for children in all ethnic/racial groups.

There was comparatively little difference in the ability of the BQ and the T-A log to predict urinary cotinine levels, with both doing a reasonably good job (adjusted [r.sup.2] > 0.45). A combination of the BQ and T-A log did better than either alone (adjusted [r.sup.2] = 0.69 for average cotinine). Because of the relatively short half-life of nicotine in the body, one might logically expect the T-A log (which covers the 48 hr preceding urine collection) to be a better predictor of urinary cotinine. The observed results (BQ [congruent to] T-A log) are consistent with a scenario wherein children are exposed to ETS primarily at home, the time they spend at home in proximity to smokers is relatively constant, and the number of smokers in the home and the amount they smoke in the child's presence are relatively stable over time. This scenario is also compatible with the fact that children whose urinary cotinine concentrations were below (or above) the "awareness threshold" (5 ng/mL) in winter were also likely to be below (or above) that value in the spring.

Although fewer urine samples were analyzed for NNAL + NNAL-Gluc, it was possible to use data from the winter monitoring session to examine statistical relationships between these biomarkers of NNK and cotinine as well as smoking-related variables. For this limited data set (winter only), urinary cotinine did a reasonably good job of predicting NNAL + NNAL-Gluc (adjusted [r.sup.2] = 0.73). However, a combination of cotinine plus the smoking variables from either the BQ or T-A log was less effective in predicting NNAL + NNAL-Gluc.

When choosing an appropriate metric to assess childhood ETS exposure, it is necessary to evaluate the relative costs and benefits of various indicators within the context of study objectives and resource constraints. Although the SHIELD study was not expressly designed to compare the BQ and the T-A log, there were some obvious differences in logistics. The BQ was administered once at the beginning of the first year of SHIELD, whereas the T-A log was collected twice, once in each monitoring session. The sole respondent to the BQ was the child's caregiver, which made it comparatively easy to administer. Maintaining an accurate 48-hr T-A log of a child's activities was a challenging undertaking that often required the combined efforts of the child, parents, and field staff.

Urine cotinine is commonly considered the most direct and therefore the best indicator of ETS exposure. But collecting urine from children is always challenging, and laboratory analysis of large numbers of samples can be expensive. Moreover, there may be more interest in measuring a urinary biomarker (NNAL + NNAL-Gluc) for the uptake of a tobacco-specific carcinogen (NNK) rather than one for nicotine uptake (cotinine), for example, as in a childhood cancer study. Although more urine is needed from exposed individuals for analysis of NNAL + NNAL-Gluc (10-20 mL vs. 1-2 mL for cotinine), the costs of collecting the requisite urine samples are similar. The costs of laboratory analysis for NNAL + NNAL-Gluc are, however, 3- to 4-fold higher than for total cotinine.

In summary, two major findings from the data are presented here. First, there were apparent ethnic/racial differences in children's exposure to ETS in two economically disadvantaged neighborhoods. Based on multiple exposure indicators (BQ, T-A logs, total urinary cotinine), a clear and consistent pattern emerged: African-American children tended to have the highest exposure, children classified as "other" (white, Southeast Asian, Native American) tended to be intermediate, and Hispanic and Somali immigrant children typically had the lowest exposure. Second, both the BQ and T-A log did a reasonably good job of predicting urine total cotinine levels, and measured urine total cotinine levels were a comparatively good predictor of urinary NNAL + NNAL-Gluc, based on analysis of a relatively small number of samples. Our results demonstrate a) the importance of considering differences in smoking prevalence by ethnicity and race when conducting children's ETS exposure studies, b) the value of measuring biomarkers of uptake for accurate assessment of children's exposure to ETS, and c) the potencial worth of questionnaires and T-A logs as screening tools or adjunct exposure metrics.
Table 1. Summary of responses to smoking-related questions from the BQ
and the T-A log (includes all smoking-related questions used in the
regression models examined in this article).

 Responded yes or no > 0 (a) [Frequency
 (%)] (b)
 African Somali
Question and source All American immigrant

BO (total no. of respondents) 150 35 39
 Caregiver ever smoked 25 (22.3) 16 (46.4) 0 (0)
 inside the home
 Current smokers in the home 25 (20.7) 9 (33.1) 0 (0)
 besides caregiver
 Packs/day in past month by 22 (19.4) 14 (40.2) 0 (0)
 caregiver
 Pack years smoked before 22 (18.8) 14 (37.8) 0 (0)
 past month by caregiver
 No. of current smokers who 23 (18.2) 8 (27.4) 0 (0)
 smoke in the home besides
 caregiver
 No. of days in month smoked 38 (29.4) 20 (51.9) 0 (0)
 Don't know 5 (2.2) 1 (2.1) 4 (7.5)
 No. of smokers inside the 35 (27.2) 20 (49.91) 0 (0)
 home in past month
 No. of hours in home with 36 (30.5) 21 (58.3) 0 (0)
 smoker
 Missing 2 (0.6) 0 (0) 2 (3.3)
 No. of hours in vehicle 14 (14.2) 10 (32.4) 0 (0)
 with smoker
 Missing 3 (1.2) 0 (0) 2 (3.3)
 No. of hours indoors with 10 (12.5) 7 (29.7) 0 (0)
 smoker not counting the
 home
 Missing 4 (2.0) 1 (2.3) 2 (3.3)
T-A log (winter; total no. of 107 23 23
 respondents)
 Day 1--tobacco smoked in 25 (26.8) 14 (52.7) 0 (0)
 home
 Day 2--tobacco smoked in 21 (23.5) 11 (43.7) 0 (0)
 home
 Missing 2 (2.1) 1 (4.0) 1 (4.7)
 Day 1--no. of cigarettes 26 (27.1) 14 (52.7) 1 (4.0)
 smoked in child's presence
 Day 2--no. of cigarettes 20 (20.1) 9 (32.3) 1 (4.0)
 smoked in child's presence
 Missing 2 (2.1) 1 (4.0) 1 (4.7)
 Day 1--no. of cigars smoked 1 (0.8) 1 (2.6) 0 (0)
 in child's presence
 Day 2--no. of cigars smoked 1 (0.8) 1 (2.6) 0 (0)
 in child's presence
 Missing 2 (2.1) 1 (4.0) 1 (4.7)
 No. of hours day 1-- 20 (21.6) 13 (49.2) 0 (0)
 indoors with smoker
 Missing 1 (1.3) 0 (0) 0 (0)
 No. of hours day 2-- 17 (17.2) 10 (34.6) 0 (0)
 indoors with smoker
 Missing 3 (3.4) 1 (4.0) 1 (4.7)
 No. of hours day 1--in 5 (4.0) 2 (6.6) 2 (6.6)
 vehicle with smoker
 Missing 1 (1.3) 0 (0) 0 (0)
 No. of hours day 2--in 4 (3.6) 2 (6.6) 1 (4.0)
 vehicle with smoker
 Missing 3 (3.4) 1 (4.0) 1 (4.7)
T-A log (spring, total no. of 100 22 21
 respondents)
 Day 1--tobacco smoked in 20 (22.6) 14 (49.1) 0 (0)
 home
 Day 2--tobacco smoked in 18 (20.9) 13 (46.7) 0 (0)
 home
 Day 1--no. of cigarettes 19 (21.0) 13 (45.4) 0 (0)
 smoked in child's presence
 Day 2--no. of cigarettes 19 (21.5) 13 (46.7) 0 (0)
 smoked in child's presence
 Day 1--no. of cigars smoked 2 (2.0) 1 (4.1) 0 (0)
 in child's presence
 Day 2--no. of cigars smoked 1 (0.6) 0 (0) 0 (0)
 in child's presence
 No. of hours day 1-- 21 (22.2) 13 (45.8) 0 (0)
 indoors with smoker
 Missing 1 (0.9) 1 (3.7) 0 (0)
 No. of hours day 2-- 18 (19.5) 11 (39.4) 0 (0)
 indoors with smoker
 No. of hours day 1--in 4 (3.8) 1 (2.7) 0 (0)
 vehicle with smoker
 Missing 1 (0.9) 0 (0) 0 (0)
 No. of hours day 2--in 3 (2.9) 1 (2.7) 0 (0)
 vehicle with smoker

 Responded yes or no >
 0 (a) [Frequency
 (%)] (b)

Question and source Hispanic Other (c)

BO (total no. of respondents) 46 30
 Caregiver ever smoked 3 (8.5) 6 (18.4)
 inside the home
 Current smokers in the home 9 (19.9) 7 (20.3)
 besides caregiver
 Packs/day in past month by 3 (8.5) 5 (14.9)
 caregiver
 Pack years smoked before 2 (6.0) 6 (18.4)
 past month by caregiver
 No. of current smokers who 8 (17.7) 7 (20.3)
 smoke in the home besides
 caregiver
 No. of days in month smoked 8 (20.5) 10 (29.0)
 Don't know 0 (0) 0 (0)
 No. of smokers inside the 6 (15.9) 9 (26.9)
 home in past month
 No. of hours in home with 6 (15.9) 9 (28.6)
 smoker
 Missing 0 (0) 0 (0)
 No. of hours in vehicle 0 (0) 4 (12.7)
 with smoker
 Missing 0 (0) 1 (2.5)
 No. of hours indoors with 3 (9.5) 0 (0)
 smoker not counting the
 home
 Missing 0 (0) 1 (2.5)
T-A log (winter; total no. of 38 23
 respondents)
 Day 1--tobacco smoked in 4 (12.2) 7 (26.1)
 home
 Day 2--tobacco smoked in 3 (10.6) 7 (26.3)
 home
 Missing 0 (0) 0 (0)
 Day 1--no. of cigarettes 5 (14.1) 6 (22.7)
 smoked in child's presence
 Day 2--no. of cigarettes 4 (12.4) 6 (22.9)
 smoked in child's presence
 Missing 0 (0) 0 (0)
 Day 1--no. of cigars smoked 0 (0) 0 (0)
 in child's presence
 Day 2--no. of cigars smoked 0 (0) 0 (0)
 in child's presence
 Missing 0 (0) 0 (0)
 No. of hours day 1-- 4 (12.2) 3 (9.7)
 indoors with smoker
 Missing 0 (0) 1 (5.2)
 No. of hours day 2-- 3 (10.6) 4 (12.9)
 indoors with smoker
 Missing 0 (0) 1 (5.2)
 No. of hours day 1--in 1 (3.0) 0 (0)
 vehicle with smoker
 Missing 0 (0) 1 (5.2)
 No. of hours day 2--in 1 (3.0) 0 (0)
 vehicle with smoker
 Missing 0 (0) 1 (5.2)
T-A log (spring, total no. of 37 20
 respondents)
 Day 1--tobacco smoked in 2 (7.9) 4 (15.3)
 home
 Day 2--tobacco smoked in 2 (7.9) 3 (11.6)
 home
 Day 1--no. of cigarettes 3 (10.0) 3 (11.6)
 smoked in child's presence
 Day 2--no. of cigarettes 3 (10.0) 3 (11.6)
 smoked in child's presence
 Day 1--no. of cigars smoked 1 (2.1) 0 (0)
 in child's presence
 Day 2--no. of cigars smoked 1 (2.1) 0 (0)
 in child's presence
 No. of hours day 1-- 5 (13.7) 3 (11.6)
 indoors with smoker
 Missing 0 (0) 0 (0)
 No. of hours day 2-- 4 (12.02) 3 (11.6)
 indoors with smoker
 No. of hours day 1--in 2 (6.5) 1 (4.7)
 vehicle with smoker
 Missing 0 (0) 1 (3.7)
 No. of hours day 2--in 1 (4.6) 1 (3.1)
 vehicle with smoker

(a) Number (weighted percentage) of respondents either answering "yes"
to a particular question or reporting a value > 0. (b) Percentages are
weighted to adjust for selection and response probabilities.
(c) Includes white, Cambodian, Laotian, Native American, and "other."

Table 2. Mean urine total cotinine concentrations (ng/mL) (a) for
SHIELD children (b) by ethnicity and race.

 All SHIELD African Somali
Season children American immigrant

February
 No. of children 113 25 29
 Mean cotinine (ng/mL) (d) 9.9 14.1 4.4
 Geometric mean (ng/mL) (d) 1.6 3.4 0.7
 95% CI (d) 1.0-2.6 1.2-10.2 0.4-1.5
May
 No. of children 86 24 19
 Mean cotinine (ng/mL) (d) 9.5 15.2 0.8
 Geometric mean (ng/mL) (d) 1.5 3.6 0.4
 95% CI (d) 0.9-2.6 1.2-11.1 0.2-0.7

Season Hispanic Other (c)

February
 No. of children 33 26
 Mean cotinine (ng/mL) (d) 4.8 12.2
 Geometric mean (ng/mL) (d) 0.6 2.2
 95% CI (d) 0.3-1.2 0.9-5.8
May
 No. of children 21 22
 Mean cotinine (ng/mL) (d) 7.2 7.2
 Geometric mean (ng/mL) (d) 0.6 1.4
 95% CI (d) 0.3-1.4 0.5-3.9

(a) Samples below the detection limit were assigned a value of 0.2
ng/mL, which is half of the minimum detectable level. (b) Includes all
randomly selected (index) children with at least one urine sample in
either the winter or spring. (c) Includes children classified as
white, Cambodian, Laotian, Native American, and other. (d) Means and
percentages are weighted to account for selection and response
probabilities.

Table 3. Mean urine total cotinine concentrations (ng/mL) by season
for SHIELD stratification variables.
 Cotinine concentration (ng/mL) (a)
 February May
Sex
Male
 No. 61 43
 Mean 9.0 5.6
 Geometric mean 1.4 1.2
 95% CI 0.8-2.5 0.6-2.2
 Female
 No. 52 43
 Mean 11.1 13.4
 Geometric mean 1.9 1.9
 95% CI 0.8-4.4 0.8-4.8
Grade 2
 No. 31 23
 Mean 8.3 8.1
 Geometric mean 1.2 1.1
 95% CI 0.5-3.3 0.3-3.5
Grade 3
 No. 26 21
 Mean 13.8 6.4
 Geometric mean 2.2 0.8
 95% CI 0.7-6.9 0.3-2.3
Grade 4
 No. 27 20
 Mean 11.8 9.4
 Geometric mean 2.0 2.4
 95% CI 0.8-5.8 0.9-6.7
Grade 5
 No. 29 22
 Mean 6.74 14.4
 Geometric mean 1.5 2.8
 95% CI 0.6-3.6 0.9-8.8
Lyndale school
 No. 53 49
 Mean 14.1 7.5
 Geometric mean 3.0 1.5
 95% CI 1.6-5.8 0.8-2.7
Whittier school
 No. 60 37
 Mean 6.6 11.6
 Geometric mean 1.0 1.5
 95% CI 0.5-1.9 0.6-3.9
English-speaking
 No. 38 37
 Mean 13.7 13.1
 Geometric mean 3.0 2.9
 95% CI 1.3-7.0 1.2-7.0
Non-English-speaking
 No. 75 49
 Mean 6.1 4.7
 Geometric mean 0.9 0.6
 95% CI 0.5-1.4 0.4-1.0

(a) Samples below the detection limit were assigned a value of 0.2
ng/mL, which is half of the minimum detectable level. Results are
weighted to adjust for selection and response probabilities.

Table 4. Comparison of regression models (various self-reported ETS
exposure metrics versus urine total cotinine or NNAL+ NNAL-Gluc)
usinq proportion of explained variance ([r.sup.2]).

 Adjusted
Model No. p-Value [r.sup.2]

BQ vs. urine total cotinine
 BQ smoking variables vs. average 63 < 0.0001 0.58
 cotinine (a)
 BQ smoking variables vs. cotinine 83 < 0.0001 0.49
 (spring)
 BQ smoking variables vs. cotinine 110 < 0.0001 0.49
 (winter)

T-A log vs. urine total cotinine
 T-A smoking variables (winter + 42 < 0.0001 0.66
 spring) vs. average cotinine (a)
 T-A smoking variables (spring) vs. 64 < 0.0001 0.50
 cotinine (spring)
 T-A smoking variables (winter) vs. 83 < 0.0001 0.46
 cotinine (winter)

Combination of BQ and T-A log vs. urine
 total cotinine
 BQ and TA (winter+ spring) vs. average 41 < 0.0001 0.69
 cotinine (a)
 BQ and TA (spring) vs. cotinine (spring) 62 < 0.0001 0.66
 BQ and TA (winter) vs. cotinine (spring) 61 < 0.0001 0.54
 BQ and TA (winter) vs. cotinine (winter) 80 < 0.0001 0.48
 Urine total cotinine vs. urine NNAL+NNAL-
 Gluc
 Cotinine (winter) vs. NNAL+NNAL-Gluc 44 < 0.0001 0.73
 (winter)
BQ and cotinine vs. urine NNAL+NNAL-Gluc
 BQ smoking variables + cotinine (winter) 42 < 0.0001 0.65
 vs. NNAL + NNAL-Glut (winter)
T-A log and cotinine vs. urine NNAL+
 NNAL-Gluc
 T-A smoking variables + cotinine 30 0.023 0.33
 (winter) vs NNAL+NNAL-Gluc (winter)

(a) Average of urine cotinine from winter and spring monitoring
sessions.

Table 5. Association of measured winter cotinine
with measured spring cotinine [n (%)], when both
are dichotomized as urine cotinine < 5 ng/mL versus
[greater than or equal to] 5 ng/mL.

 Spring cotinine

Winter [greater than or
cotinine < 5 ng/mL equal to] 5 ng/mL Total

< 5 ng/mL 38 1 39
 (58.46) (1.54) (60.00)
[greater than 7 19
 or equal to] 5 ng/mL (10.77) (29.23) (40.00)
Total 45 20 65
 (69.23) (30.77) (100.00)

Misclassification error 8/65 = 0.12.

Table 6. Association of predicted winter cotinine
(from BQ regression model) with measured spring
cotinine [n (%)], when BQth are dichotomized as
urine cotinine < 5 ng/mL versus [greater than or equal to] 5 ng/mL.

 Spring cotinine

 [greater than or
 < 5 ng/mL equal to] 5 ng/mL Total

< 5 ng/mL 38 4 42
 (60.32) (6.35) (66.67)
[greater than 5 16 21
 or equal to] 5 ng/mL (7.94) (25.40) (33.33)
Total 43 20 63
 (68.25) (31.75) (100.00)

Misclassification error: 9/63=0.14.

(a) from BQ regression model.


This study was supported by two STAR grants (825813 and 826789) from the U.S. Environmental Protection Agency, National Center for Environmental Research; a grant from the Legislative Commission on Minnesota Resources; and grant CA-81301 from the National Cancer Institute.

The authors declare they have no competing financial interests.

REFERENCES

Baluja KF, Park J, Myers D. 2003. Inclusion of immigrant status in smoking prevalence statistics. Am J Public Health 93(4):642-646.

Benowitz NL. 1999. Biomarkers of environmental tobacco smoke exposure. Environ Health Perspect 107(suppl 2):349-355.

Boffetta P, Tredaniei J, Greco A. 2000. Risk of childhood cancer and adult lung cancer after childhood exposure to passive smoke: a meta-analysis. Environ Health Perspect 108:73-82.

Carroquino MJ, Galson SK, Licht J, Amler RW, Perera FP, Claxton LD, et al. 1998. The U.S. EPA conference on preventable causes of cancer in children: a research agenda.

Environ Health Perspect 106(suppl 3):867-873. Cummings KM, Markello S J, Mahoney M, Bhargava AK,

McElroy PD, Marshall JR. 1990. Measurement of current exposure to environmental smoke. Arch Environ Health 45:74-79.

Daisey JM 1999. Tracers for assessing exposure to environmental tobacco smoke: what are they tracing? Environ Health Perspect 107(suppl 2):319-327.

Eskenazi B, Castorina R. 1999. Association of prenatal maternal or postnatal child environmental tobacco smoke exposure and neurodevelopmental and behavior probLems in children. Environ Health Perspect 107:991-1000.

Gold DR. 2000. Environmental tobacco smoke, indoor allergens, and childhood asthma. Environ Health Perspect 109(suppl 41:643-651.

Hecht SS, Carmella SG, Murphy SE, Akerkar S, Brunnemann KD, Hoffmann D. 1993. A tobacco-specific lung carcinogen in the urine of men exposed to cigarette smoke. N Engl J Med 329:1543-1546.

Hecht SS, Ye M, Carmella SG, Fredrickson A, Adgate JL, Greaves IA, et al. 2001. Metabolites of a tobacco-specific lung carcinogen in the urine of elementary school-aged children. Cancer Epidemiol Biomarkers Prev 10:1109-1116.

Jaakkola JJK, Nafstad P, Magnus P. 2001. Environmental tobacco smoke, parental atopy, and childhood asthma. Environ Health Perspect 109:579-582.

Jarvis M, TunstalI-Pedoe H, Feyerabend C, Vesey C, Salloojee Y. 1984. Biochemical markers of smoke absorption and self reported exposure to passive smoking. J Epidemiol Community Health 38:335-339.

Landrigan PJ, Carlson JE, Bearer CF, Cranmer S, Bullard RD, Etzel RA, et al. 1998, Children's health and the environment: a new research agenda for prevention research. Environ Health Perspect 106(suppl 3):787-794.

National Research Council. 1986. Environmental Tobacco Smoke: Measuring Exposures and Assessing Health Effects. Washington, DC:National Academy Press.

--. 1991. Human Exposure Assessment for Airborne Pollutants: Advances and opportunities. Washington, DC:National Academy Press.

National Toxicology Program. 2000. Environmental tobacco smoke. Ninth Report on Carcinogens. Research Triangle Park, NC:National Toxicology Program, U.S. Department of Health and Human Services, III-33-111-34.

Pirkle JL, Flegal KM, Bernert JT, Brody DJ, Etzel RA, Maurer KR. 1996. Exposure of the U.S. population to environmental tobacco smoke. The Third National Health and Nutrition Examination Survey, 1988 to 1991. JAMA 275:1233-1240.

Rothman KJ. 1986. Modern Epidemiology. Boston, MA:Little Brown.

Samet JM. 1999. Workshop summary: assessing exposure to environmental tobacco smoke in the workplace, Environ Health Perspect 107(suppl 2):309-312.

Sandler DP, Everson RB, Wilcox AJ, Browder JP. 1995. Cancer risk in adulthood from early life exposure to parents' smoking. Am J Public Health 75(5):487-492.

Scherer G, Meger-Kossien I, Riedel K, Renner T, Meger M. 1999. Assessment of the exposure of children to environmental tobacco smoke (ETS) by different methods. Hum Exp Toxicol 19:297-301.

Sexton K, Adgate JL, Church TR, Greaves lA, Gurumurthy R, Fredrickson AL, et al. 2003. Recruitment, retention, and compliance results from a probability study of children's environmental health in economically disadvantaged neighborhoods. Environ Health Perspect 111:731-736.

Sexton K, Greaves lA, Church TR, Adgate JL, Ramachandran G, Tweedie RL, et al. 2000. A school-based strategy to assess children's environmental exposures and related health effects in economically disadvantaged urban neighborhoods. J Expo Anal Environ Epidemiol 10:682-694.

Subramaniam RP, Turim J, Golden SL, Kral P, Anderson EL. 2001. An exploratory study of variations in exposure to environmental tobacco smoke in the United States. Risk Anal 21(3):561-574.

Tang D, Warburton D, Tannenbaum SR, Skipper P, Santella RM, Cereijido GS, et al. 1999. Molecular and genetic damage from environmental tobacco smoke in young children, Cancer Epidemiol Biomarkers Prev 8:427-431.

U.S. DHHS. 1986. The Health Consequences of Involuntary Smoking: A Report of the Surgeon General. PHS 87-8398. Washington, DC:Department of Health and Human Services, Public Health Service, Office of the Assistant Secretary for Health, Office of Smoking and Health.

U.S. EPA. 1992. Respiratory Health Effects of Passive Smoking: Lung Cancer and Other Disorders. EPA/600/6-90/006F. Washington, DC:U.S. Environmental Protection Agency, Office of Health and Environmental Assessment, Office of Research and Development.

Ken Sexton, (1) John L. Adgate, (2) Timothy R. Church, (2) Stephen S. Hecht, (3) Gurumurthy Ramachandran, (2) Ian A. Greaves, (2) Ann L. Fredrickson, (2) Andrew D. Ryan, (2) Steven G. Carmella, (3) and Mindy S. Geisser (2)

(1) School of Public Health, University of Texas, Brownsville, Texas, USA; (2) Division of Environmental and Occupational Health, School of Public Health, University of Minnesota, Minneapolis, Minnesota, USA; (3) University of Minnesota Cancer Center, Minneapolis, Minnesota, USA

Address correspondence to K. Sexton, Environmental and Occupational Health, School of Public Health, University of Texas, Brownsville Regional Campus, 80 Fort Brown, RAHC Bldg., Brownsville, TX 78520 USA. Telephone: (956) 554-5168. Fax: (956) 554-5152. E-mail: ksexton@utb.edu

Received 21 May 2003; accepted 12 November 2003.
COPYRIGHT 2004 National Institute of Environmental Health Sciences
No portion of this article can be reproduced without the express written permission from the copyright holder.
Copyright 2004, Gale Group. All rights reserved. Gale Group is a Thomson Corporation Company.

Article Details
Printer friendly Cite/link Email Feedback
Title Annotation:Children's Health
Author:Geisser, Mindy S.
Publication:Environmental Health Perspectives
Date:Mar 1, 2004
Words:6880
Previous Article:In utero pesticide exposure, maternal paraoxonase activity, and head circumference.
Next Article:Toward the effective surveillance of hypospadias.
Topics:


Related Articles
Children's exposure to ETS: race and ethnicity matter.
Living in a fog: secondhand smoke may dull kids' wits.
Racial differences in exposure to environmental tobacco smoke among children.
Outdoor, indoor, and personal exposure to VOCs in children.
Exposure to environmental tobacco smoke and cognitive abilities among U.S. children and adolescents.
Environmental tobacco smoke linked to behavior problems.
Effects in infants from tobacco smoke, mold, and older siblings.
The economic impact of early life environmental tobacco smoke exposure: early intervention for developmental delay.
Adding up to ADHD: effects of early exposures.

Terms of use | Privacy policy | Copyright © 2019 Farlex, Inc. | Feedback | For webmasters