Early lead exposure (<3 years old) prospectively predicts fourth grade school suspension in Milwaukee, Wisconsin (USA)

Environmental Research 126 (2013) 60–65

Contents lists available at ScienceDirect

Environmental Research journal homepage: www.elsevier.com/locate/envres

Early lead exposure (o 3 years old) prospectively predicts fourth grade school suspension in Milwaukee, Wisconsin (USA)$, $$ Michael S. Amato a,b,n, Sheryl Magzamen c,d, Pamela Imm c,e, Jeffrey A. Havlena f, Henry A. Anderson e, Marty S. Kanarek b,c, Colleen F. Moore a,b,g a

Psychology Department, University of Wisconsin–Madison, 1202 West Johnson Street, Madison, WI 53706, USA Nelson Institute for Environmental Studies, University of Wisconsin–Madison, 550 North Park Street, Madison, WI, USA c Department of Population Health Sciences, University of Wisconsin–Madison, 610 North Walnut Street, Madison, WI, USA d Department of Biostatistics and Epidemiology, University of Oklahoma Health Sciences Center, 801 N.E. 13th Street, Oklahoma City, OK, USA e Division of Public Health, Wisconsin Department of Health Services, 1 West Wilson Street, Madison, WI, USA f UW-Madison Medical School, Department of Surgery, 600 Highland Avenue, Madison, WI, USA g Montana State University, P.O. Box 173440, Bozeman, MT, USA b

art ic l e i nf o

a b s t r a c t

Article history: Received 21 February 2013 Received in revised form 16 July 2013 Accepted 23 July 2013 Available online 12 August 2013

School suspensions are associated with negative student outcomes. Environmental lead exposure increases hyperactivity and sensory defensiveness, two traits likely to increase classroom misbehavior and subsequent discipline. Childhood Blood Lead Level (BLL) test results categorized urban fourth graders as exposed (2687; lifetime max BLL 10–20 mg/dL) or unexposed (1076; no lifetime BLL Z 5 mg/dL). Exposed children were over twice as likely as unexposed children to be suspended (OR ¼2.66, 95% CI ¼ [2.12, 3.32]), controlling for covariates. African American children were more likely to be suspended than white children, but lead exposure explained 23% of the racial discipline gap. These results suggest that different rates of environmental lead exposure may contribute to the racial discipline gap. & 2013 Elsevier Inc. All rights reserved.

Keywords: Lead exposure School suspension Discipline gap Disparities Blood lead School discipline

1. Introduction One of the most frequently used disciplinary actions in United States schools is suspension. Nationally, approximately 1 out of 14 public school students (7%) are suspended each year (Planty et al., 2009), although variability between individual schools is high. In a comprehensive review of school disciplinary data in the state of Colorado for the years 2008–2010, Pfleger and Wiley (2012) found that out-of-school suspensions were the most commonly used form of discipline (53% of all actions taken), and in-school suspensions were the second most common (32%).

☆ Funding: This research was made possible by funding from the Wisconsin Partnership Program—Medical Education and Research Committee (MERC), School of Medicine and Public Health, University of Wisconsin–Madison. ☆☆ Human Subjects Institutional Review Board: The protocol for this study was approved by the University of Wisconsin–Madison Education Research Institutional Review Board. n Corresponding author at: Psychology Department, University of Wisconsin– Madison, 1202 West Johnson Street, Madison, WI 53706, USA. Fax: +1 608 262 4029. E-mail address: [email protected] (M.S. Amato).

0013-9351/$ - see front matter & 2013 Elsevier Inc. All rights reserved. http://dx.doi.org/10.1016/j.envres.2013.07.008

Several quantitative studies have raised questions about the long term negative effects of suspension on students. Even after controlling for a variety of individual, family, and community level risk factors, students who receive one or more suspensions are at greater risk for dropping out of high school, compared to students who have never been suspended (Suh et a., 2007; Lee et al., 2011; Arcia, 2006). The number of times a student is suspended has been shown to be negatively related to subsequent reading achievement (Arcia, 2006). Negative consequences associated with suspension are not limited to educational outcomes, but extend to health and public safety as well. Using a longitudinal design and controlling for covariates, Hemphill et al. (2012) found that students suspended from school during the previous year were roughly twice as likely to use tobacco 12 months later as were students who had not been suspended. School suspension has also been found to be a strong predictor of violent behavior one year later (Hemphill et al., 2009). African American students in the United States are suspended at higher rates than white students. The 2009–2010 Civil Rights Data Collection, a nationally representative survey of approximately 7000 school districts and more than 72,000 schools conducted annually by the U.S. Department of Education, found

M.S. Amato et al. / Environmental Research 126 (2013) 60–65

that both African American boys and girls were three times more likely to receive an out of school suspension than their white peers (Office for Civil Rights, 2012). Similarly to national rates, in Colorado Pfleger and Wiley (2012) found that 17% of African American students received at least one disciplinary action, compared to 11% of Latino students and 7% of white students. Although the cause of the racial disparity in student suspension rates is unclear, three broad, non-exclusive hypotheses exist. The differential behavior hypothesis suggests that students from disproportionately suspended groups participate in misbehavior at higher rates than white students. The differential selection hypothesis suggests they are more likely to be subject to discipline than are white students committing the same offense. The differential processing hypothesis suggests the disparity is the result of more severe punishment for African American, Latino, and Native American students committing the same offense as white students. Gregory et al. (2010) reviewed the literature and found evidence suggesting that all three hypotheses may be correct. Partial support for each was also found in McCarthy and Hoge (1987), which drew on findings from the domain of juvenile justice to suggest that consideration of factors other than the offending behavior itself may bias school officials in the process of discipline. Using longitudinal data from six public schools in a mid-Atlantic city, McCarthy and Hoge (1987) found that a student′ s likelihood of being suspended was primarily influenced by the seriousness of their misconduct, followed by the number of disciplinary incidents they had been involved in the previous year, followed by their grades, followed by their teach-rated demeanor. When those four constructs were included in a regression, measures of race, gender, and socio-economic status were no longer significant predictors of suspension. However, although in their model a student′s race was not directly predictive of suspensions, race was still correlated with student-level variables that were directly predictive of suspensions. African American students in their sample had poorer grades and were perceived by teachers as less well behaved, resulting in a greater number of prior disciplinary incidents, which predicted a higher likelihood of future suspension. McCarthy and Hoge′s (1987) results link the racial disparity in suspension rates to the racial achievement gap, and push the question of causality from “why are African American students suspended at higher rates” to “why do African American students have (1) poorer grades and (2) worse behavior as perceived by teachers?” One causal variable that has recently received attention as a partial contributor to both academic achievement and perception of student behavior, which is more prevalent in nonHispanic black children than children from other groups in the United States, is environmental lead exposure (Jones et al., 2009; Advisory Committee on Childhood Lead Poisoning Prevention (ACCLPP), 2012; Wheeler and Brown, 2013). Several studies have demonstrated a negative relationship between lead exposure during the first years of life and subsequent school exam scores. With a sample of students enrolled in Milwaukee, Wisconsin, public schools, Amato et al. (2012) found that moderately elevated blood lead levels (BLL) during development were associated with end-of-grade exam deficits at 4th grade in reading, mathematics, language arts, science, and social studies. Similar negative relationships between lead exposure and standardized test scores have been found in New Orleans (Zahran et al., 2009), in North Carolina (Miranda et al., 2007), in New York State (Strayhorn and Strayhorn, 2012), and in Massachusetts (Reyes, 2011). There is also reason to believe that children exposed to elevated BLLs may be perceived by teachers as less well behaved than unexposed children. Pioneering work by Needleman and colleagues found a negative association between dentine lead and

61

teacher-rated classroom behavior in a sample of 3329 first and second grade students in Massachusetts (Needleman et al., 1979). More recently, lead exposure has been linked to increased risk of an Attention Deficit Hyperactivity Disorder diagnosis (Braun et al., 2006), and increased risk of a Learning or Behavioral Exception Designation (Miranda et al., 2010) in children. Those epidemiological results are supported by animal research, which suggest that disrupted sensory gating processes may be the neurological mechanism by which lead exposure reduces attention and focus (Moore et al., 2008). Students who have difficulty paying attention and controlling their impulses are likely to be disruptive in class, increasing the likelihood that their teacher will perceive them as poorly behaved. A survey of middle and high school students who had been suspended found that a majority attributed their suspension to a lack of self-control (Costenbader and Markson, 1998). The current study examines student-level disciplinary data for 4th graders in an urban school district to investigate the role of childhood lead exposure in the racial discipline gap. We operationalize the discipline gap using an unduplicated suspension count, which is the most common operationalization among the studies cited above. An unduplicated count provides a binary classification of students into those never suspended, and those suspended one or more times. We use a series of logistic regressions to model the relationship between lead exposure and school suspension, and the extent to which apparent racial differences in suspension rates can be attributed to differences in exposure.

2. Method 2.1. Participants We used data from the Wisconsin Childhood Lead Poisoning Prevention Project (WCLPPP) to identify children who had a maximum blood lead test result during the first 3 years of life indicating either moderate exposure (exposed group: BLL Z 10 mg/dl and o 20 mg/dl), or a maximum test result indicating low or unquantifiable exposure (unexposed group: BLL o 5 mg/dl). We chose to examine data from children who resided in Milwaukee at the time of their blood test because Milwaukee (1) is the largest city in Wisconsin, providing an adequate sample size, and (2) has the highest prevalence of childhood lead poisoning (6.1%) of any municipality in the state (Wisconsin Department of Health and Family Services [DHFS], 2008). The protocol for this study was approved by the University of Wisconsin–Madison Education Research Institutional Review Board. The upper bound of the exposed group was operationalized based on the level required for state mandated intervention in Wisconsin as well as several neighboring states (20 mg/dl; although many counties and municipalities intervene at lower levels). The lower bound was chosen based on the Centers for Disease Control and Prevention (CDC) Level of Concern at the time of data collection (10 mg/dl). In 2012 the CDC accepted the recommendation of the Advisory Committee on Childhood Lead Poisoning Prevention (ACCLPP, 2012), and replaced the previous level of concern (10 mg/dL) with a new reference value calculated every 4 years as the 97.5th percentile of BLLs in the two most recent National Health and Nutrition Examination Survey datasets. This change was made in response to an accumulation of evidence that there is no safe level of lead exposure (Centers for Disease Control and Prevention (CDC), 2012), with the effect of increasing the number of children that the CDC considers at risk for negative consequences of lead exposure. Although the criteria used to create the sample of exposed children in the current study do not reflect the new CDC reference value, they nonetheless demarcate a range of BLLs that are above the reference value but below the level required for state mandated intervention, and represent a group of children that may be considered “moderately exposed”. 2.2. Data Blood lead data were available because health care providers and laboratories are required to report all BLL results to WCLPPP. There is substantial variation in the limits of quantification (LOQ) established by individual laboratories. The Wisconsin State Laboratory of Hygiene (WSLH), which analyzes more than one-quarter of all samples, established an LOQ of 5 mg/dl for all blood lead analysis prior to 2000, and we used this to anchor the low end of the quantifiable lead exposure spectrum. Although the WSLH has lowered their LOQ since 2000, there remains a high degree of uncertainty in individual results at BLL concentrations below 5 mg/dl. The CDC

62

M.S. Amato et al. / Environmental Research 126 (2013) 60–65

supports this interpretation: “The uncertainty associated with laboratory testing is too great to ensure that a single blood lead test reliably classifies individual children at levelso10 mg/dl” (CDC, 2005). Our definition is more stringent and based on the WSLH LOQ at the time of testing to refer to a population of children we define here as “unexposed.” School discipline data were obtained by matching eligible children from the WCLPPP database with records from Milwaukee Public Schools (MPS). Study staff provided a list of students that met the exposed or unexposed definitions from the WCLPP database to an MPS data manager, along with each student′s date of birth and BLL information in a coded string that were indecipherable to non-study personnel. From the 5779 children identified, MPS was able to match 3796 with records of 4th grade enrollment. Those children for whom MPS was unable to find a match had most likely moved out of the district before the 4th grade. Children for whom a match was unable to be found had a lower prevalence of exposure (47%) than children with records of 4th grade enrollment in MPS (71%). Information on educational differences between the groups was unavailable. Fig. 1 shows a flowchart of sample construction. MPS returned a de-identified dataset to study researchers. The anonymized dataset of student records included number of days enrolled, number of suspension incidents, and information on student characteristics including race/ethnicity, gender, and year of birth. All students were listed with a single race/ethnicity status. Eligibility for free or reduced lunch during 4th grade was used as a marker for income. The numbers of students identified as Native American (23) or Asian (10) were too few for reliable analysis; therefore data from those students were not included in analysis, resulting in a final sample of 3763 students.

such that African American students, Hispanic students, and boys would have higher suspension rates than white students or girls. A third model, the full model, combined all variables from the basic and demographic models. Our primary measure of interest was change in OR for each independent variable across the different models. We predicted that the OR for lead exposure would decrease in the full model relative to the basic model, but still remain greater than 1 because lead exposure poses additional risk beyond that explained by the other demographic variables. We also predicted that ORs for the African American and Hispanic variables would be reduced in the full model relative to the demographic model, because the racial discipline gap is hypothesized to be partially attributable to differing rates of lead exposure. In contrast, we predicted that the OR associated with gender would remain constant across the demographic and full models, because the gender asymmetry in suspension rates is not hypothesized to be related to lead exposure. In addition to changes in odds ratios, models were also compared based on the degree to which they improved overall fit to the data using the Vuong test (Vuong, 1989). The Vuong test allows comparisons of both nested and non-nested models, in contrast to the chi squared test which is only appropriate for nested models. The null hypothesis tested is that neither model is a significantly better fit to the data. We predicted that the full model would be a better fit than either of the other models. All analyses were conducted using R version 2.12.2. The Vuong test was implemented using the pscl package.

3. Results 2.3. Analysis The bivariate relationships between lead exposure and the covariates were evaluated using chi square tests of independence, as were the bivariate relationship of lead exposure with unduplicated suspension, and the bivariate relationships of race/ethnicity with unduplicated suspension. Effect sizes were calculated as ɸ¼ √(χ2/N). We then modeled unduplicated suspension as the dependent variable in three logistic regressions. A basic model included lead exposure (exposed, unexposed) as the only independent variable. The logistic coefficient associated with exposure was exponentiated in order to obtain an odds ratio (OR). The OR represented the odds that an exposed student would be suspended at least once in the fourth grade, relative to the odds that an unexposed student would be suspended at least once in the fourth grade. An OR greater than 1 suggests a positive association between lead exposure and suspension. Standard errors were used to construct 95% confidence intervals around each OR. We predicted a robust relationship between lead exposure and likelihood of suspension. A second model, the demographic model, included other student characteristics known to be associated with suspension rates (gender, income, race/ethnicity). Interactions were not included because the very small sample sizes in some cells would likely cause overfitting and lead to spurious conclusions. Independent variables were binary coded so that a value of 1 indicated the characteristic was true for a student. Gender was entered into the model as “male”; income was entered as “eligible for free or reduced lunch”. Race/ethnicity was entered as two variables, one coding African American membership and the other coding Hispanic/ Latino membership. The data obtained from MPS listed each student with a single group membership. We predicted that our sample would replicate national trends,

Our data replicated the overall national pattern (Wheeler and Brown, 2013) that African American and Hispanic children were more likely to have elevated BLLs than non-Hispanic white children (African American 80%, Hispanic/Latino 64%, white 38%), χ2(2, N¼ 3763)¼364.82, po0.001. Among all covariates, lead exposure was most strongly associated with race/ethnicity (ϕ¼0.31) and with assisted lunch eligibility, χ2(1, N¼3763)¼301.77, po0.001, ϕ¼0.28. Children who had been exposed to lead were nearly three times as likely to be suspended in the 4th grade (31%) than were children who had not been exposed (11%), χ2(1, N¼3763)¼171.15, po0.001, ϕ¼0.21. Crosstabs are presented in Table 1. Our data were also consistent with national trends (Office for Civil Rights, 2012) in that African American students were approximately three times more likely to be suspended at least once (33.92%) than white students (9.72%). However, our data diverge from national trends in that Hispanic/Latino students were the least likely to be suspended (7.39%). Separate 2  2 chi square tests revealed that the difference between African American and white students was significant, χ2(1, N ¼ 2978) ¼107.56, p o0.001, ɸ ¼0.19; however the difference between Hispanic/Latino students and white students was not χ2(1, N ¼1248)¼1.79, p ¼0.18. Table 1 Bivariate associations with lead exposure.

5,799 children with: a) Milwaukee residence at time of BLL test b) BLL results in eligible range c) Recordof 4th grade attendance with Wisconsin Department of Public Instruction

3,796children matched to Milwaukee Public Schools (MPS) records

3,763 children in sufficiently populated racial/ethnic groups for analysis

1,076 children classified as unexposed(29%)

2,687 children classified as exposed (71%)

Student characteristic

Total Gendera Male Female Race/Ethnicityb Black/African-American Hispanic/Latino White Incomeb No Assisted Lunch Assisted Lunch Program Suspension incidentsb Suspended in 4th Grade Not Suspended in 4th Grade

Exposed

Effect Size

n

(%)

n

(%)

ɸ

1076

(29)

2687

(71)

536 540

(27) (30) -

1444 1243

(73) (70)

508 279 289

(20) (36) (62) -

2007 506 174

(80) (64) (38)

366 710

(57) (23) -

278 2409

(43) (77)

115 961

(11) ↓ (89) ↓

841 1846

(31) (69)

0.03

0.31

0.28

0.21

Arrows indicate whether percentages sum horizontally or vertically. a

Fig. 1. Sample creation flowchart.

Unexposed

b

Denotes p o0.05 for chi-square test of independence. Denotes p o 0.0001 for chi-square test of independence.

M.S. Amato et al. / Environmental Research 126 (2013) 60–65

63

Table 2 Odds ratios and confidence intervals for the 3 models. Characteristic

Odds ratio Basic

Intercept Lead exposed Male African Amer. Hispanic/Latino Free/reduce lunch Null deviance Resid. deviance % Improvement

0.12 3.81

95% Confidence interval Demo. 0.03 2.44 3.87 0.58 2.59

4265 4071 4.5

4265 3755 12.0

Models, odds ratios, confidence intervals, and deviance values are presented in Table 2. All effects were significant at the p o0.001 level by the Wald test, except where otherwise noted. As predicted, the OR for lead exposure in the basic model suggested that children in the exposed group were 3.81 times more likely to be suspended at least once in the fourth grade than were children in the unexposed group. For the demographic model, results were similar to the bivariate tests above despite the simultaneous inclusion of race/ethnicity, gender, and income in a single model. African American students were more likely than white students to be suspended (OR ¼3.87, 95% CI ¼ [2.78, 5.38]), while Hispanic students were less likely than white students (OR ¼0.58, 95% CI ¼[0.38, 0.88], p¼ 0.01). Boys were more likely than girls to be suspended (OR ¼2.44, 95% CI ¼ [2.08, 2.87]), and students eligible for free or reduced lunch were more likely to be suspended than ineligible children (OR ¼2.59, 95% CI ¼ [1.97, 3.41]). Odds ratios in the full model differed from the basic and demographic models largely as predicted. The effect of lead exposure was reduced but remained robust (OR ¼2.66, 95% CI ¼ [2.12, 3.32]). The OR associated with African American membership was reduced to 2.98 (95% CI ¼[2.12, 4.17]) when lead exposure was included in the model, a 23% reduction. The effect of free/reduced lunch eligibility was also reduced in the full model (OR ¼2.11, 95% CI ¼[1.59, 2.79]), while the effect of gender remained effectively unchanged (OR ¼2.45, 95% CI ¼[2.08, 2.89]). However, contrary to predictions, the magnitude of the effect of Hispanic membership on suspension rates increased when lead exposure was added to the model (i.e., moved further from 1), with an OR of 0.48 (95% CI ¼[0.32, 0.74]) in the full model suggesting that Hispanic students were roughly half as likely as white students to be suspended. ORs for the full model are presented in Fig. 2. Pairwise model comparison with the Vuong test revealed that the full model was more likely to reflect the true structure of the data than either of the other two models (test statistic: basic ¼11.14, full ¼4.65; p o0.001 for each).

4. Discussion The data reveal that lead exposure in the first three years of development is positively associated with increased risk of school suspension during the fourth grade. When covariates were considered, children in the exposed group were over twice as likely as children in the unexposed group to be suspended. The current study suggests that lead exposure may be a contributing factor to discipline problems within schools. The study also suggests that exposure may contribute to increased risk of the negative

Full 0.02 2.66 2.45 2.98 0.48 2.11

Basic

Demo.

Full

0.10, 0.15 3.09, 4.70

0.02, 0.05

0.02, 0.04 2.12, 3.32 2.08, 2.89 2.12, 4.17 0.32, 0.74 1.59, 2.79

2.08, 2.87 2.78, 5.38 0.38, 0.88 1.97, 3.41

4265 3674 13.9

Fig. 2. Odds ratios (OR) of suspension for each student characteristic in the full model. Each characteristic was True/False coded, so that ORs indicate the likelihood of suspension for students for whom the characteristic is true, compared to students for whom it is false. An OR 41 indicates a positive relationship. Error bars show 95% confidence intervals.

academic, health, and social outcomes for individual children that are associated with suspension (Suh et al., 2007; Lee et al., 2011; Arcia, 2006; Hemphill et al., 2012, 2009), and is consistent with other research that has found an association between early lead exposure and violent crime rates at the city (Mielke and Zahran, 2012), state (Reyes, 2007), and national (Nevin, 2007) scales. The detrimental effects of lead exposure on attention, causally established through animal experiments (Moore et al., 2008), are consistent with disruptive classroom behaviors and the loss of control that suspended students frequently cite as a reason for their behavior (Costenbader and Markson, 1998). However, the current cohort design is unable to differentiate whether lead exposure itself is causally related to likelihood of suspension, or whether it is an indicator of other mechanisms. Establishing causality is particularly difficult because lead exposure is often comorbid with exposure to other environmental pollutants including metals, particulate matter, and noise (e.g., Zahran et al., 2012; Moore, 2009; Evans and Kantrowitz, 2002). Further research that includes measures of other common pollutants is necessary to more conclusively establish the independent relationship between lead exposure and school suspension. Similarly, future research should include more refined measures of income, because one limitation of the current study was the single dichotomous measure of income that was available. Eligibility for free or reduced lunch may not have captured important differences among eligible children. If so, one possibility is that lead exposure in our models may have served as an indicator variable for other factors affecting children at the more extreme end of household income, or as an auxiliary measure of income for children who would have been eligible but whose parents did not apply for the free/reduced lunch program. It is also possible that

64

M.S. Amato et al. / Environmental Research 126 (2013) 60–65

the observed relationship between exposure and suspension in our data was due to a combination of both a true causal effect of lead as well as other unmeasured factors. Although lead exposure and household income can both be considered student level characteristics, there is a critical theoretical difference in the types of variables that they are. Lead is a neurotoxicant that experiments have causally demonstrated to increase hyperactivity in rats (Moreira et al, 2001), increase reactivity to startling auditory stimulation in rats (Commissaris et al., 2000), and increase defensiveness to tactile stimulation in rhesus monkeys (Moore et al., 2008). Taken together, these animal results propose a direct mechanism for linking lead exposure in children′s physical environment to disruptive classroom behaviors likely to result in suspension. Such a mechanism does not exist for student household income; rather, income is understood in the model as a marker for other unmeasured factors which are correlated with income. The same is true for the variables in the model coding race and ethnicity. Statistics on disparities in school discipline, such as those collected as part of the Civil Rights Data Collection, are descriptive of a problem but are unable to specify its cause. A primary goal of the current research was to investigate the extent to which racial and ethnic disparities in lead exposure may causally contribute to the national discipline gap in U.S. schools. For our sample, we found that differences in lead exposure were able to account for 23% of the disparity in suspensions between African American and white students. Our data only roughly approximated the national pattern of racial/ethnic disparities in suspension rates reported in the 2009– 2010 Civil Rights Data Collection (Office for Civil Rights, 2012). One possible explanation is that the population pattern in Milwaukee is different from the national pattern, for reasons our data are unable to address. Alternatively, the exclusion of many children from families with the inclination and financial means to move out of the school district may have decreased the representativeness of our sample. The finding that prevalence of exposure was higher for students whom MPS was able to match educational records supports that speculation. The true reasons for the divergence are unclear, but could be a valuable subject of future research. While across population subsamples it would be unsurprising to find a range of correlations between lead exposure and broad social categories (such as race and ethnicity), we would expect that the effects of exposure on suspension rates would be relatively constant for all children. In 2012 the CDC changed its guidelines on lead to reflect the growing body of evidence that even very low amounts of exposure are associated with substantial risk, replacing the previous “level of concern”(10 mg/dl) with the new “reference value” (currently 5 mg/dl). Primary prevention is the only effective method for reducing the harm caused by environmental lead (Needleman, 1998; Jacobs et al., 2002). Furthermore, a recent study estimated expected change in population mean BLL as the result of different hypothetical standards of environmental contamination in floor dust, soil, and water, and found that unless standards were strict only a small number of children at the highest levels of exposure would benefit (Oulhote et al., this issue). The current study adds to our understanding of the costs of lead′s continued presence in the environment, by linking it to school suspension rates. Suspension is associated with negative outcomes for the suspended student, as well as a disrupted and potentially unsafe classroom for other students. School discipline is a vastly complex and controversial issue. These results suggest that part of the challenge exists entirely outside of educational policy, and that long-term efforts to reduce disruptive behavior must also consider children′s physical environments as a necessary part of the solution.

Acknowledgments The authors are grateful to Margie Coons and the Wisconsin Childhood Lead Poisoning Prevention Program; to Milwaukee Public Schools; to the Wisconsin Department of Public Instruction; and to Noel Stanton, WI State Lab of Hygiene. This research was made possible by funding from the Wisconsin Partnership Program—Medical Education and Research Committee (MERC), School of Medicine and Public Health, University of Wisconsin–Madison.

References Advisory Committee on Childhood Lead Poisoning Prevention, 2012. Low level lead exposure harms children: a renewed call for primary prevention. Atlanta: CDC. Retrieved from: 〈http://www.cdc.gov/nceh/lead/ACCLPP/Final_Document_ 030712.pdf〉. Amato, M.S., Moore, C.F., Magzamen, S., Imm, P., Havlena, J.A., Anderson, H.A., Kanarek, M.S., 2012. Moderate lead exposure and educational proficiency on end-of-grade exams. Annals of Epidemiology 22, 738–743. Arcia, E., 2006. Achievement and enrollment status of suspended students: Outcomes in a large, multicultural school district. Education and Urban Society 38, 359–369. Braun, J.M., Kahn, R.S., Froelich, T., Auinger, P., Lanphear, B.P., 2006. Exposures to environmental toxicants and attention deficit hyperactivity disorder in U.S. children. Environmental Health Perspectives 114 (12), 1904–1909. Centers for Disease Control and Prevention, 2005. Preventing Lead Poisoning in Young Children. Atlanta: CDC. Retrieved from: 〈http://www.cdc.gov/nceh/lead/ publications/prevleadpoisoning.pdf〉. Centers for Disease Control and Prevention, 2012. CDC response to Advisory Committee on Childhood Lead Poisoning Prevention recommendations in “Low Level Lead Exposure Harms Children: A Renewed Call for Primary Prevention”. Atlanta: CDC. Retrieved from: 〈http://www.cdc.gov/nceh/lead/ ACCLPP/CDC_Response_Lead_Exposure_Recs.pdf〉. Commissaris, R.L., Tavakoli-Nezhad, M., Barron, A.J., Pitts, D.K., 2000. Effects of chronic low-level oral lead exposure on prepulse inhibition of acoustic startle in the rat. Neurotoxicology and Teratology 22 (1), 55–60. Costenbader, V., Markson, S., 1998. School suspension: A study with secondary school students. Journal of School Psychology 36, 59–82. Evans, G.W., Kantrowitz, E., 2002. Socioeconomic status and health: the potential role of environmental risk exposure. Annual Review of Public Health 23, 303–331, http://dx.doi.org/10.1146/annurev.publhealth.23.112001.112349. Gregory, A., Skiba, R.J., Noguera, P.A., 2010. The achievement gap and the discipline gap: two sides of the same coin? Educational Researcher 39 (1), 59–68. Hemphill, S.A., Smith, R., Toumbourou, J.W., Herrenkohl, T.I., Catalano, R.F., McMorris, B.J., Romaniuk, H., 2009. Modifiable determinants of youth violence in Australia and the United States: a longitudinal study. Australian & New Zealand Journal of Criminology 42 (3), 289–309. Hemphill, S.A., Heerde, J.A., Herrenkohl, T.I., Toumourou, J.W., Catalano, R.F., 2012. The impact of school suspension on student tobacco use: a longitudinal study in Victoria, Australia, and Washington State, United States. Health Education & Behavior 39 (1), 45–56. Jacobs, D.E., Clickner, R.P., Zhou, J.Y., Viet, S.M., Marker, D.A., Rogers, J.W., Zeldin, D.C., Broene, P., Friedman, W., 2002. The prevalence of lead-based paint hazards in U.S. housing. Environmental Health Perspectives 110 (10), A599–A606. Jones, R.L., Homa, D.M., Meyer, P.A., Brody, D.J., Caldwell, K.L., Pirkle, J.L., Brown, M.J., 2009. Trends in blood lead levels and blood lead testing among US children aged 1 to 5 years, 1988–2004. Pediatrics 123 (3), e376–e385, http://dx.doi.org/10.1542/ peds.2007-3608. Lee, T., Cornell, D., Gregory, A., Xitao, F., 2011. High suspension schools and dropout rates for black and white students. Education and Treatment of Children 34 (2), 167–192. McCarthy, J.D., Hoge, D.R., 1987. Racial disadvantage out of universalistic process. Social Forces 65 (4), 1101–1120, http://dx.doi.org/10.1093/sf/65.4.1101. Mielke, H.W., Zahran, S., 2012. The urban rise and fall of air lead (Pb) and the latent surge and retreat of societal violence. Environment International 43, 48–55. Miranda, M.L., Kim, D., Overstreet Galeano, M.A., Paul, C.J., Hull, A.P., Morgan, S.P., 2007. The relationship between early childhood blood lead levels and performance on end-of-grade tests. Environmental Health Perspectives 115 (8), 1242–1247. Miranda, M.L., Maxson, P., Kim, D., 2010. Early childhood lead exposure and exceptionality designations for students. International Journal of Child Health and Human Development 3 (1), 77–84. Moore, C.F., 2009. Children, Pollution, and Why Scientists Disagree. Oxford University Press, New York, New York. Moore, C.F., Gajewski, L.L., Laughlin, N.K., Luck, M.L., Larson, J.A., Schneider, M.L., 2008. Developmental lead exposure induces tactile defensiveness in Rhesus monkeys (Macaca Mulatta). Environmental Health Perspectives 116 (10), 1322–1326. Moreira, E.G., Vassilieff, I., Vassilieff, V.S., 2001. Developmental lead exposure: behavioral alterations in the short and long term. Neurotoxicology and Teratology 23 (5), 489–495. (pmid:11711252).

M.S. Amato et al. / Environmental Research 126 (2013) 60–65

Needleman, H.L., 1998. Childhood lead poisoning: the promise and abandonment of primary prevention. American Journal of Public Health 88 (12), 1871–1877. (pmcid:PMC1509047). Needleman, H.L, Gunnoe, C., Leviton, A., Reed, R., Peresie, H., Maher, C., Barrett, P., 1979. Deficits in psychologic and classroom performance of children with elevated dentine lead levels. New England Journal of Medicine 300 (13), 689–695. Nevin, R., 2007. Understanding international crime trends: the legacy of preschool lead exposure. Environmental Research 104, 315–336. Office for Civil Rights, 2012. The Transformed Civil Rights Data Collection (CRDC). United States Department of Education, Washington, DC. (Retrieved from: 〈http://www.ed.gov/ocr〉). Oulhote, Y., Le Tertre, A., Etchevers, A., Le Bot, B., Lucas, J., Mandin, C., Le Strat, Y., Lanphear, B. and Glorennec, B., Implications of different residential lead standards on children's blood lead levels in France: Predictions based on a national cross-sectional survey, International Journal of Hygiene and Environmental Health, http://http:dx.doi.org/10.1016/j.ijheh.2013.02.007, this issue. Pfleger, R. & Wiley, K., 2012. Colorado Disciplinary Practices, 2008–2010: Disciplinary Actions, Student Behaviors, Race, and Gender. Boulder, CO: National Education Policy Center. Retrieved from: 〈http://nepc.colorado.edu/publication/ colorado-disciplinary-practices〉. Planty, M., Hussar, W., Snyder, T., Kena, G., KewalRamani, A., Kemp, J., Bianco, K., Dinkes, R., 2009. The Condition of Education 2009 (NCES 2009-081). Washington, DC: National Center for Education Statistics, Institute of Education Sciences, U.S. Department of Education. Reyes, J.W., 2007. Environmental policy as social policy? The impact of childhood lead exposure on crime. B.E. Journal of Economic Analysis & Policy 7, 1.

65

Reyes, J.W., 2011. Childhood lead and academic performance in Massachusetts. New England Public Policy Center. Federal Reserve Bank of Boston, Boston, Massachusetts. Strayhorn, J.C., Strayhorn Jr., J.M., 2012. Lead exposure and the 2010 achievement test scores of children in New York counties. Child and Adolescent Psychiatry and Mental Health 6, 4. Suh, S., Suh, J., Houston, I., 2007. Predictors of categorical at-risk high school dropouts. Journal of Counseling and Development 85 (2), 196–203. Vuong, Q.H., 1989. Likelihood ratio tests for model selection and non-nested hypotheses. Econometrica 57 (2), 307–333 〈http://www.jstor.org/stable/ 1912557〉. Wheeler, W., Brown, M.J., 2013. Blood lead levels in children aged 1–5 years-United States, 1999–2010. Morbidity and Mortality Weekly Report 62 (13), 245–248. Wisconsin Department of Health and Family Services, 2008. The legacy of lead: The report on childhood lead poisoning in Wisconsin. Madison, Wisconsin. Retrieved from: 〈http://www.dhs.wisconsin.gov/lead/LegacyofLead/Legacyo fLead45109.pdf〉. Zahran, S., Mielke, H.W., Weiler, S., Berry, K.J., Gonzales, C., 2009. Children′s blood lead and standardized test performance response as indicators of neurotoxicity in metropolitan New Orleans elementary schools. NeuroToxicology 30 (6), 888–897, http://dx.doi.org/10.1016/j.neuro.2009.07.017. Zahran, S., Mielke, H.W., Weiler, S., Hempel, L., Berry, K.J., Gonzalez, C., 2012. Associations between standardized school performance tests and mixtures of Pb, Zn, Cd, Ni, Mn, Cu, Cr, Co, and V in community soils of New Orleans. Environmental Pollution 169, 128–135.