Association between lower-level of environmental lead exposure and reactive and proactive aggression in youth: Sex differences

Abstract

Lead exposure during childhood has been associated with a variety of negative outcomes, including antisocial/aggressive behavior. However, different subtypes of antisocial behavior have been found to have different neurobiological correlates, and it is unclear whether lead exposure is related to specific subtypes of aggressive behavior. The objective of the study was to examine relationships between childhood blood lead levels (BLL) and proactive and reactive aggression. Further, given prior findings of sex differences in the effects of lead exposure, we examine whether there are sex differences in these relationships. In a sample of 818 youth (47.2% girls) ages 10–13 in China, we assessed BLL and administered the Reactive Proactive Aggression Questionnaire. Results show that BLL were associated with reactive but not proactive aggression. There was a significant interaction between BLL and sex in predicting aggression; boys with higher BLL scored higher in both proactive and reactive aggression than boys with lower BLL, but these differences were not present for girls. These findings suggest that lead exposure may have broad effects on antisocial behavior, but that boys may be more susceptible than girls. These findings may provide insights to identifying protective factors that could be potential targets for intervention.

Introduction

Lead is an environmental health hazard that has been associated with many negative health outcomes¹. In children, lead exposure has been associated with disrupted cognitive and behavioral development.^2–4 Lead exposure is known to have effects on the developing brain, and even low levels of lead exposure have been found to impair fronto-executive functions.⁵ Previously, the threshold of 10μg/dL was considered toxic⁶. However, in 2021 the Centers for Disease Control (CDC) updated the blood lead reference value to 3.5μg/dL⁷ and stating that no safe BLL in children exist and that even low levels cause harm.

A growing number of studies have documented associations between childhood lead levels and externalizing behaviors (including aggression, delinquency, and hyperactivity) as measured by the Child Behavior Checklist (CBCL) in youth during late childhood.^9–12 Additional studies have found associations between lead levels and hostile distrust, oppositional defiant behaviors¹³ and antisocial behavior¹⁴ in youth. Bone lead levels have been found to be higher in arrested delinquents compared to controls,¹⁵ and childhood BLL have been associated with higher arrest rates in adulthood, including for violent crimes.¹⁶ In an analysis of the impact of lead exposure in 800,000 children differentially exposed to the phaseout of leaded gasoline in Sweden, Grönqvist et al¹⁷ found that early lead exposure above a particular threshold was associated with both property and violent crime in adulthood. It has been hypothesized that lead exposure induces alterations in the brain vasculature, specifically in nitric oxide synthases, which then affect the serotonergic system, leading to heightened aggressive behavior.¹⁸

However, aggressive behavior is not a uniform construct. An existing gap in the literature is that few studies have examined relationships between lead levels and subtypes of aggressive behavior. Different types of aggressive behavior have been found to have distinct biological correlates.^19,20 Therefore, understanding which type of aggression is most affected may be important in furthering our understanding of the mechanisms by which lead affects behavior. Here we differentiate between proactive and reactive aggression. Reactive aggression is considered an aggressive response to a perceived provocation, and is more associated with impulsivity.²¹ Proactive aggression is planned and pre-meditated, and highly associated with callous-unemotional traits.²² Proactive and reactive aggression have been found to differ in heritability, and it is likely that different genetic mechanisms are involved. A recent review genetic and epigenetic studies has proposed that an overstimulation of serotonergic and dopaminergic pathways may be associated with increases in both reactive and proactive aggression, but that proactive aggression may be specifically associated with dopamine related increases in reward anticipation, while endocrine system may modulate reactive aggression.²⁰ Reduced amygdala responses have also been found to mediate the relationship between callous-unemotional traits and proactive aggression, despite being positively correlated with externalizing behavior generally.²³ Fanti et al¹⁹ found that children with conduct problems who are high in callous-unemotional (CU) traits (associated with the presence of proactive aggression) show blunted physiological responding to fear, compared to those high in conduct problems who are low in CU traits (associated with primarily reactive aggression) who show high physiological responsiveness to fear. Jiang et al²⁴ have also identified structural brain differences in boys with conduct disorder with either impulsive or premeditated aggression.

Thus, in order to fully understand the relationships between lead exposure and aggressive behavior, it may be valuable to determine whether lead exposure is associated with proactive aggression, reactive aggression, or both. To the author’s knowledge, only one study has examined subtypes of aggression. In adolescents, Nkomo et al²⁵ found that BLL were associated with direct aggression, which included attacking others, meanness, threatening to hurt others, and getting into fights. However, BLL were not associated with indirect aggression, which involved stubbornness, having a hot temper, argumentativeness, and teasing others. This study suggests there may be differences in the types of aggression that lead levels are associated with. However, direct and indirect aggression were determined through principal component analysis, and therefore may not replicate in other studies. The present study focuses on examining relationships between lead and reactive and proactive aggression, which are well established constructs with validated measurement tools, and in which biological correlates have been examined.

An additional gap in the literature is that few studies have examined sex as a potential moderator of the relationships between lead levels and aggression. Burns et al⁹ found that lifetime lead exposure was associated with externalizing problems in both boys and girls. However, Grönqvist et al¹⁷ found that early lead exposure had worse effects on long-term outcomes in boys, suggesting that boys are more strongly affected by lead exposure. Emerging research suggests that the developmental effects of lead exposure on the brain may be sex-dependent.²⁶ Thus, the present study also examined potential interactions with sex.

The goal of the present study was to examined relationships between BLL and reactive and proactive aggression in a large sample of youth, and to examine the potential moderating effect of sex.

Methods

Study Design

Data from a single wave of a longitudinal study, the China Jintan Child Cohort Study, were used to examine relationships between BLL and reactive and proactive aggression, controlling for covariates. The China Jintan Child Cohort Study is an ongoing longitudinal project that aimed to study the effect of environmental exposures, such as lead, on children’s neurobehavioral outcomes.

Study Site

The study took place in Jintan, China.

Study Participants

The current study utilizes data collected during Wave II (2011–2013) of the China Jintan Child Cohort Study when children were 10–13 years old. Complete data were available for 818 youth (47.2% female; mean age 12.0 ± 0.4 years old ). Participants were initially recruited six years prior (2005–2007) when in preschool. Four preschools were chosen to be representative of the geographic, social, and economic profile of the city of Jintan. All children attending the four preschools were invited to participate in the study.^27,28

Study Variables

BLL was the independent variable in the study, and proactive and reactive aggression were the dependent variables.

Blood Lead Level

BLL were measured using a graphite furnace atomic absorption spectrophotometer (AA100 - Perkin-Elmer Company). This laboratory has participated successfully in the Blood Lead Proficiency Testing Program, a CDC-administered quality control program for the measurement of lead in whole blood. Blood collection was conducted by trained pediatric nurses using a strict research protocol to avoid lead contamination. Samples were frozen and shipped to the Research Center for Environmental Medicine of Children at Shanghai Jiaotong University for lead analyses. Each specimen was analyzed twice for blood lead, and the mean of the repeated measurements was taken as the final measure. Kaulson Laboratories provided blood lead reference materials for quality control (QC). QC samples were inserted blindly among the study samples (one QC sample in every 10 study samples). The limit of detection was 1.8 μg/dL; half of the limit of detection was imputed for three samples (0.2%) under the limit of detection. More details on this process are provided in the study by Liu et al²⁹ and Halabicky et al.³⁰

Reactive and Proactive Aggression Questionnaire (RPQ)

The Chinese version of the Reactive and Proactive Aggression Questionnaire³¹ was used to measure reactive and proactive aggression. This is a Mandarin version of the RPQ, a well-validated 23-item self-report questionnaire designed to measure reactive and proactive aggression in children and adolescents beginning at age eight.³² Each item has a three-point response format (0=never, 1=sometimes, 2=often). Scores were summated to provide three scores: proactive aggression (11 items), reactive aggression (12 items), and total scores (the summation of reactive and proactive aggression).The Mandarin version of the RPQ has been validated in a sample of 1352 Chinese participants with a mean age of 11.65; it was found to have good construct validity, internal consistency, and test-retest reliability.³¹. The questionnaire was administered by paper and took approximately 10 minutes to complete. Cronbach’s alphas were as follows: reactive aggression: 0.810; proactive aggression: 0.846; total score: 0.893.

Covariates

Seven sociodemographic variables were chosen as covariates. This selection was Based on our previous work as well as a review of the literature of variables associated with aggression. These covariates include: age, parental education levels, and parental occupations (1 = unemployed, 2 = physical work, 3 = professional work). Region of family residence was also included, because in Wave 1 of the study (preschool age), children living in rural areas had higher mean BLLs than children living in urban and suburban areas because of lead emission from a cement factory. Children provided information about their sex and age. Parents provided information about their education levels, family region, and occupation.

Data Collection

All children in Wave II (ages 10–13) were invited to receive a physical examination which included a blood draw that assessed heavy metal concentrations and to participate in psychophysiology testing. The research team obtained written informed consent from parents, and verbal informed assent from participants prior to the second wave of data collection. Blood samples were collected at school health clinics. Children completed the RPQ and provided general information about their sex and age in classrooms during school hours; this was completed within one month of the BLL assessment. Parents provided demographic information. Institutional Review Board approval was obtained from the University of Pennsylvania and the ethical committee for research at Jintan Hospital.

Statistical Analyses

Independent samples t-tests were used to examine sex differences in reactive aggression, proactive aggression and BLL. Because both proactive and reactive aggression were not normally distributed, a Mann-Whitney U test was also used to test the differences. Because of high expected correlations between reactive and proactive aggression based on prior studies,^32,33 proactive aggression was controlled for in models predicting reactive aggression, and vice versa, in order to examine independent effects. Hierarchical linear regression analyses were used to test whether BLL positively predicted proactive aggression and reactive aggression, and whether sex and BLL interacted in predicting reactive and proactive aggression. As both predictors and criteria were not normally distributed, a robust estimator in Mplus version 8.0 MLM³⁴, was used. In Step 1, reactive/proactive aggression was regressed on age, residence father educational level, mother educational level, father occupation, mother occupation and proactive/reactive aggression (as covariates), and sex, and BLL. In Step 2, the interaction term (sex×BLL) was added to the model to assess the moderation effect by sex. To alleviate multicollinearity due to addition of the moderator and interaction term, mean centering was performed on BLL. The regression models were evaluated by R² and predictors were evaluated by standardized coefficients, β.

To examine the differences in BLL among subtypes of aggressors and non-aggressors (i.e., reactive aggressors, proactive aggressors, reactive–proactive aggressors, and non-aggressors), participants were classified based on their scores on the RPQ. Reactive–proactive aggressors scored z > 1 for both reactive aggression and proactive aggression; reactive aggressors scored z > 1 for reactive aggression only; proactive aggressors scored z > 1 for proactive aggression only; and non-aggressors scored z < 1 for both reactive and proactive aggression. This classification criterion was adopted from previous studies.^35,36

Results

Descriptive Statistics

Characteristics of the children and their families are described in Table 1. The mean BLL in the sample was 3.1 (SD = 1.2) μg/dL. The mean of reactive aggression in the sample was 5.34 (SD = 4.03), and the mean of proactive aggression was 1.41 (SD = 2.98). These means are comparable to levels observed in a large sample of 11–15 year-olds in Hong Kong (reactive aggression mean = 4.71; proactive mean = 1.33)³⁷. Proactive and reactive aggression were positively correlated (r = 0.598, p < .01). As shown in Table 2, boys had significantly higher BLL than girls. In addition, boys showed more proactive aggression than girls, but there were no sex differences in reactive aggression.

Table 1.

Sociodemographic Characteristics of Study Participants

		n	%
Sex	Boys	432	52.8
	Girls	386	47.2
Residence	Rural	137	16.7
	Suburban	341	41.7
	City	340	41.6
Grade	4th	265	32.4
	5th	287	35.1
	6th	266	32.5
Marital status of parents	Married	699	96.5
	Divorced/separate	25	3.5
Father’s occupation	Unemployed	8	1.2
	Physical work	507	77.6
	Professional work	138	21.1
Mother’s occupation	Unemployed	66	11.3
	Physical work	336	57.6
	Professional work	181	31.0
Father’s educational level	Middle school or less	232	28.5
	High school	232	28.5
	College or more	349	42.9
Mother’s educational level	Middle school or less	321	39.5
	High school	194	23.9
	College or more	298	36.7

Table 2.

Means, Standard Deviations and Sex Differences in BLL and Aggression Variables

	Boys	Girls
	(n=432)	(n=386)	t	p	95% CI	U	p
	M (SD)	M (SD)
BLL (μg/dL)	3.21(1.19)	2.97(1.12)	2.909	0.004	(0.077, 0.395)	71631.00	<0.001
Proactive aggression	1.93(3.64)	0.83(1.82)	5.516	<0.001	(0.705, 1.483)	70413.50	<0.001
Reactive aggression	5.44(4.15)	5.23(3.89)	0.736	0.462	(−0.345, 0.758)	81230.00	0.523

Note. BLL = blood lead levels; M = mean; SD = standard deviation; CI = Confidence Interval.U reflects comparisons made with a Mann-Whitney U test because variables were not normally distributed. The 95% CI reflects the confidence interval for the difference between means.

Hierarchical Linear Regression

As shown in Table 3, there was a main effect of sex for reactive and total aggression, but not for proactive aggression. There was a main effect of BLL in predicting reactive aggression, and a trend toward BLL predicting total aggression. Analyses indicated an interaction between sex and BLL in predicting proactive, reactive, and total aggression. Models with sex and BLL predicted approximately 34% of the variance in proactive aggression, and 33% of the variance in reactive aggression. To probe the interaction, a median split was performed on BLL and values were plotted at levels above and below the median (Figure 1). Compared to boys scoring below the median on BLL, boys scoring above the median BLL reported significantly higher levels of proactive aggression (t =3.546, p <0.001) and reactive aggression (t =2.454, p = 0.015). There was no significant difference between girls scoring above and below the median on BLL for either proactive aggression (t = 0.672, p = 0.502) or reactive aggression (t =00.193, p = 0.847).

Table 3.

Hierarchical Linear Regression Models Examining the Interaction between Sex and BLL on Proactive, Reactive, and Total Aggression

	Proactive (R2= 0.343)				Reactive (R2=0.326)				Total aggression
	β	SE	p	95%CI	β	SE	p	95%CI	β	SE	p	95%CI
Sex	−0.594	0.686	0.387	(−1.939, 0.752)	−2.782	0.994	0.005	(−4.731, −0.833)	−3.376	1.489	0.023	(−6.294, −0.457)
BLL	−0.085	0.153	0.581	(−0.385, 0.216)	−0.489	0.222	0.028	(−0.924, −0.054)	−0.573	0.332	0.084	(−1.225, 0.078)
Sex*BLL	0.461	0.209	0.027	(0.052, 0.870)	0.869	0.302	0.004	(0.276,1.461)	1.329	0.453	0.003	(0.442, 2.217)

Note. CI = Confidence Interval;Age, residence, parents’ occupation, and parents’ educational level were included as covariates in these models.

Figure 1.

Interaction Between Sex and BLL on Proactive and Reactive Aggression

Chi-square test and sex differences

Based on the classification criteria for the subtypes of aggressors/non-aggressors, 665 participants were categorized as non-aggressors, 80 as reactive aggressors, 27 as proactive aggressors, and 46 as reactive–proactive aggressors (Table 4). The Chi-square analysis was significant for boys and for the total sample. Boys with higher BLL (BLL ≥2.90μg/dL) were more likely to be categorized in the reactive-proactive aggressor group. This was also true for the total sample.

Table 4.

Chi-Square Analysis of High/Low BLL by Aggressor/Non-Aggressor Category

	BLL	Non-aggressor (n (%))	Proactive aggressor (n (%))	Reactive aggressor (n (%))	Reactive-proactive aggressor (n (%))	χ2	p
Boys	<2.90	154 (46.4)	9 (39.1)	19 (45.2)	5 (14.3)	13.517	0.004
	>2.90	178 (53.6)	14 (60.9)	23 (54.8)	30 (85.7)
Girls	<2.90	181 (54.4)	3 (75.0)	21 (55.3)	6 (54.5)	0.686	0.887
	>2.90	152 (45.6)	1 (25.0)	17 (44.7)	5 (45.5)
Total	<2.90	335 (50.4)	12 (44.4)	40 (50.0)	11 (23.9)	12.31	0.006
	>2.90	330 (49.6)	15 (55.6)	40 (50.0)	35 (76.1)

Note. BLL = blood lead level

Discussion

Lead exposure during childhood is a major public health concern that is associated with many negative outcomes. In the present study, we examined whether BLL was associated with either reactive or proactive types of aggression, which have been found to have different biological correlates.³⁸ We found that BLL was associated with both reactive and proactive aggression. However, these relationships were only present in boys; for girls, the relationships were not significant. This finding is consistent with a growing body of literature suggesting that the effects of lead exposure may depend on sex.²⁶ For example, in the study examining the impact of lead exposure in Swedish children, Grönqvist et al¹⁷ found that early lead exposure had worse effects on neurocognitive skills in boys, suggesting that boys are more strongly affected by lead exposure. Similarly, Malavika et al³⁹ found that correlations between BLL and neurobehavioral indicators (a combined measure of depression, anxiety, hyperactivity, aggression) were much stronger in boys than girls.

These findings are consistent with a study by Cecil et al⁴⁰ which found that early childhood lead exposure was correlated with a lower brain volume in adulthood in males but not females. This effect was observed primarily in the prefrontal cortex, which is highly involved in behavioral regulation and social and emotional decision-making processes that are relevant to aggression.⁴¹ Other studies have found that in boys, but not girls, BLL are associated with poorer attention and visuoconstruction,⁴² and poorer planning and rule learning and reversal.⁴³ Singh et al²⁶ review both human and animal literature supporting sex differences in outcomes in response to lead exposure. They suggest that sex differences in outcomes may arise from a combination of complex factors, including sex-specific molecular and genetic mechanisms, and external risk factors, including sex-specific responses to environmental stressors. Further studies are needed to better understand the biological mechanisms by which males may be more susceptible to the effects of lead exposure than females.

The exact mechanism underlying of the relationship between aggression and BLL is not clear. However, the fact that both reactive and proactive aggression were associated with BLL may reflect the fact that lead exposure has effects in the prefrontal cortex,⁴⁰ a region that is associated with both proactive and reactive aggression.^44,45 It is also in line with studies showing that the effects of lead exposure on mental health symptoms are widespread, as it is associated not only with externalizing problems, but a variety of internalizing problems as well, including depression, anxiety, and social problems.^9,10,46,47

The mean BLL in the current study was 3.1μg/dL (standard deviation 1.2), which is close to the newly updated blood lead reference value of 3.5 μg/dL by the CDC in the United States.⁷ The CDC states that there is no known safe blood lead concentration, and indicates that even blood lead concentrations as low as 3.5μg/dL may be associated with decreased intelligence, behavioral difficulties, and learning problems in children.⁷ Our findings of associations between BLL and proactive and reactive aggression in a sample with BLLs near this threshold provide further support that relatively lower BLL are still associated with negative outcomes.

In the present study we found that BLL were higher in boys than in girls. This is consistent with a number of previous studies throughout various countries,^25,48,49 though other studies have found higher BLL in female children.³⁹ Higher BLL in boys may be due to factors associated with both behavioral characteristics (e.g., hand-mouth activity^29,50) and biological processes including increased levels of absorption, distribution, and excretion of pharmacological agents being greater in men than women.⁵¹

It is important to discuss several potential limitations of the present study. First, the study assessed BLL and aggression cross-sectionally, and therefore does not provide information about how lead levels may predict aggressive behavior in the future. Although several studies have found that lead exposure or levels during childhood are predictive of adult behavior,^16,17 that is not always the case. For example, Desrochers-Couture et al¹¹ found that child BLL was associated with concurrent childhood externalizing behavior, but not with later adolescent outcomes. There is a need for additional longitudinal studies exploring the relationship between lead levels and behavior over time. A second limitation is that reactive and proactive aggression were measured via a self-report questionnaire. More confidence in the findings could be gained by using a multi-informant approach.

Finally, while we do not have information about the precise source or timing of lead exposure in these participants, possible sources of lead exposure in this sample include residual of leaded gasoline, which was phased out and then banned in China in 2000, industrial emissions, lead-contaminated paints (e.g., in houses, toys, and stationeries), family members’ take-home lead from work, coal combustion, and medications.^6,29,52,53 In this region a cement factory is also a source of lead emission, and lead emitted from the factory may have contaminated the locally grown food.

Despite these limitations, this study provides new information about the nature of relationships between lead exposure and aggression. The finding that lead exposure is associated with increased proactive and reactive aggression highlights the importance of mitigating environmental lead exposure. The finding that these relationships existed in boys but not girls warrants further exploration. This sex difference in susceptibility to lead exposure may provide clues to factors which may act in a protective manner, which could be explored as potential targets for intervention.