Neurodevelopmental effects of postnatal lead exposure at very low levels

Abstract

This study is among the first to examine specific neurobehavioral deficits in children exposed at very low lead levels. A systematic analysis for the presence of a threshold of lead exposure was conducted. The sample consisted of 246 African American, inner-city children from whom blood lead concentrations were assessed at 7.5 years of age. The results consistently show neurobehavioral deficits in relation to low levels of lead in the areas of intelligence, reaction time, visual–motor integration, fine motor skills, attention, including executive function, off-task behaviors, and teacher-reported withdrawn behaviors. Effects were identified in the specific domains of attention, executive function, visual–motor integration, social behavior, and motor skills, which have been previously suggested as part of lead's “behavioral signature”. Visual inspection of nonparametric regression plots suggested a gradual linear dose–response relation for most endpoints. No threshold discontinuity was evident. Regression analyses in which lead exposure was dichotomized at 10 μg/dl were no more likely to be significant than analyses dichotomizing exposure at 5 μg/dl. Given that associations were found between lead levels as low as 3 μg/dl for multiple outcomes, these data provide additional evidence that there is no apparent lower bound threshold for postnatal lead exposure.

Introduction

In 1995, the World Health Organization [72] reviewed the extensive literature on postnatal lead exposure and children's neurobehavioral development and determined that the effects of lead are credible and persistent, and there appears to be no safe threshold for lead exposure. Bellinger, Dietrich, and others have proposed that the principal endpoints affected by lead exposure (“behavioral signature”) are attention, executive function, visual–motor reasoning skills, vestibular–proprioceptive control and social behavior, and Lanphear et al. have concluded that there is no safe threshold [46]. Thus, Lanphear et al. have reported significant associations where poorer cognitive performance was evident at exposure levels lower than 5 μg/dl. Despite these reports, most government agencies have continued to use 10 μg/dl as a criterion in public health advisories. For example, the U.S. Environmental Protection Agency [25] and the Center for Disease Control [14] recommend that lead levels above 10 μg/dl be avoided. As a result of extensive efforts to reduce lead levels in the environment, mean lead level in children has decreased from 15 μg/dl in the 1970s to 4 μg/dl in the 1990s [7]. Nevertheless, many children continue to be exposed above the recommended “avoid” level. For example, an assessment of Mexican-American children showed that approximately 5% of children of all ages still have lead levels above 10 μg/dl [57].

Because postnatal lead exposure is often associated with socioenvironmentally more disadvantaged homes, researchers have assessed a broad range of control variables in an attempt to distinguish the effects of lead exposure from other socioenvironmental influences. Confounding variables have been assessed so comprehensively that it has been argued that some studies have “overcontrolled” for the social environment, misattributing lead effects to the environment [3]. Even so, after controlling statistically for these confounding effects, most studies continue to find associations of lead exposure with IQ [9], [21], [30], [46], [49], [66] and visual–motor integration [19], [30], [31]. Significant associations have also been reported in the areas of achievement, including reading [27], [29], [55], [75], math, and spelling [9], [11], [28], [75]. Postnatal lead exposure has also been found to be associated with deficits in motor skills (specifically, speed and dexterity), memory abilities, advanced spatial functions and attention [20], [26], and slower reaction times in laboratory assessments [51], [55], [73], [74].

Attention is one domain that has been examined in particular detail [10], [11], [31], [46], [63], [76]. Bellinger et al. [10] attempted to identify the specific aspects of attention that are affected by exposure to lead. Using a middle-class sample whose mean blood lead levels were reported as low (79.2% of the sample was below 5 μg/dl), Bellinger et al. administered Mirsky's attention battery, which assesses sustained attention, focused attention, shift (executive function), and encoding (working memory). Significant associations were found solely in the domains of focus and executive function.

In the past few years, increased attention has been directed to examining behavior problems in lead exposed children. Byers and Lord [12] had noted that heavily exposed children are distractible, impulsive, and aggressive. More recently, antisocial, delinquent, and aggressive behaviors have also been reported in children exposed to low levels of lead [11], [23], [56], even in very young children. In a study of 1- to 3-year-olds, whose mean lead level was 10.4 μg/dl, Mendelsohn et al. [50] found that children exposed to lead had more difficulties with emotional regulation and behavior orientation. These children were more hyperactive, impulsive, and easily frustrated, more withdrawn and lacking in interest.

Although lead exposure has been linked to a broad range of neurobehavioral endpoints, the domains affected are not always consistent across studies. Faust and Brown [26] did not find consistent results on the Achenbach's Child Behavior Checklist (CBCL [1]), the measure used to identify aggressive and delinquent children in the studies cited previously. Silva et al. [63] found no association with intelligence. Needleman et al. [56] found no relation with reaction time and found associations only in the vigilance domain, but not the focused and executive function factors on Mirsky's attention battery reported by Bellinger et al. [10].

Our study examined postnatal lead exposure in a sample of children with very low levels of exposure (mean=5.4 μg/dl). The levels are markedly lower than in earlier low-level lead studies [e.g., Bellinger et al. [10] (mean=10 μg/dl); Dietrich et al. [20] (mean=11.8 μg/dl, for similar-aged children); McMichael et al. [48] (mean=19 μg/dl); Winneke et al. [74] (mean=8.2 μg/dl)] and slightly lower than some of the more recent studies of very low postnatal lead exposure [Bellinger et al. [9] (mean=6.5 μg/dl); Fergusson et al. [28] (mean=6.2 μg/dl)]. To date, only one study has reported levels lower than those presented in this research [Lanphear et al. [46] (geometric mean=1.9 μg/dl)]. In our study, we examined the specific domains where deficits have been previously identified to determine similarities and differences compared with the findings reported in previous studies. In addition, we attempted to identify a lower bound threshold for lead exposure, using nonparametric regression, in several cognitive and behavioral domains where the effects of lead have been shown to be harmful.