Temporal pattern in the effect of postnatal blood lead level on intellectual development of young children

Neurotoxicology and Teratology 22 (2000) 805 ± 810

Temporal pattern in the effect of postnatal blood lead level on intellectual development of young children Lourdes Schnaasa, Stephen J. Rothenbergb,c,*, Estela Perronia, Sandra MartõÂneza, Carmen HernaÂndeza, Reyna M. HernaÂndeza a Department of Developmental Neurobiology, National Institute of Perinatology, Mexico City, Mexico Center for Research in Population Health, National Institute of Public Health, Cuernavaca, Morelos, Mexico c Department of Anesthesiology and Center for Environmental Research, Charles R. Drew University of Medicine and Science, Los Angeles, CA, USA b

Received 23 November 1999; accepted 2 July 2000

Abstract To determine the temporal pattern of the effect of postnatal blood lead level on the General Cognitive Index (GCI) of the McCarthy Scales of Children's Abilities, we used data from 112 children of the Mexico City Prospective Lead Study with complete evaluations from 36 to 60 months of age at 6-month intervals. We measured blood lead level every 6 months from 6 to 54 months. We controlled for 5-min Apgar, birth weight, birth order, sex, socioeconomic level, maternal IQ, and maximum maternal educational level in a repeated measures ANCOVA using child blood lead level grouped by 6 ± 18 month (geometric mean 10.1 mg/dl, range 3.5 ± 37.0 mg/dl), 24 ± 36 month (geometric mean 9.7 mg/dl, range 3.0 ± 42.7 mg/dl), and 42 ± 54 month (geometric mean 8.4 mg/dl, range 2.5 ± 44.8 mg/dl) averages. There were significant interactions between the 6 ± 18 month blood lead level and age with GCI as the endpoint and between 24 ± 36 month blood lead level and age. The regression coefficient of blood lead at 6 ± 18 months became more negative with age until 48 months, when the rate of decline moderated (linear polynomial contrast p = 0.047). The regression coefficient of blood lead at 24 ± 36 months with CGI became more negative as well from 36 to 48 months but then started decreasing toward zero from 48 to 60 months (quadratic polynomial contrast p = 0.019). Significant between-subjects lead effects on GCI were found for 24 ± 36 month blood lead level at 48 months ( p = 0.021) and at 54 months ( p = 0.073). The greatest effect (at 48 months) was a 5.8-point GCI decrease with each natural log unit increase in blood lead. Significant betweensubjects lead effects on GCI were found for 42 ± 54 month blood lead level at 54 months ( p = 0.040) and at 60 months ( p = 0.060). The effect of postnatal blood lead level on GCI reaches its maximum approximately 1 ± 3 years later, and then becomes less evident. Four to five years of age appears to be a critical period for the manifestation of the earlier postnatal blood lead level effects. D 2000 Elsevier Science Inc. All rights reserved. Keywords: Blood lead level; Pb; Intelligence; Development; Child

Many studies, both cross-sectional and longitudinal, have been published over the past three decades showing damaging effects of lead at exposure levels previously considered innocuous [2,4,19,28]. These studies have provided the basis for sweeping changes in regulations defining limits for acceptable blood lead levels in children in many countries. Longitudinal studies have given investigators the opportunity to determine if there are developmental periods of * Corresponding author. Department of Anesthesiology (MP#10), Charles R. Drew University of Medical Science, 1621 East 120th Street, LosAngeles, CA 90059,USA.Tel.: +1-323-563-4839; fax:+1-310-632-9857. E-mail addresses: [email protected], [email protected] (S.J. Rothenberg).

special vulnerability to the effects of lead exposure. Such studies have used both repeated blood lead measures and repeated measures of intellectual development in a cohort studied over time. To this date there have been seven prospective studies of lead effect on intellectual development of children from a wide range of environments and socioeconomic levels [1,3,6,9 ±13,17,22,25± 27]. With some exceptions [9,13], these studies agree that early lead exposure has negative effects on early neurobehavioral development. Although the effects reported have at times been small and/or nonsignificant, no study has reported beneficial effects at higher blood lead levels. The marginal statistical significance of many of the results suggests that vulnerability to lead exposure varies among

0892-0362/00/$ ± see front matter D 2000 Elsevier Science Inc. All rights reserved. PII: S 0 8 9 2 - 0 3 6 2 ( 0 0 ) 0 0 1 0 1 - X

806

L. Schnaas et al. / Neurotoxicology and Teratology 22 (2000) 805±810

children, or that there are critical periods during which the child can be more sensitive to the toxic effects of lead. These aspects of individual variability and vulnerability to lead exposure could be linked to various factors, such as physical and social environment, duration and magnitude of lead exposure, and general health of the studied child. There is also wide variability in the pattern of results shown by different studies. Apart from study differences in reporting the age at which lead exposure is associated with decreased intellectual performance or the age at which the manifestation of decreased intellectual performance is expressed, research groups variously report long-lasting effects [8,23,26], effects at only circumscribed ages [6,12], effects concurrent with measured lead levels [14,25], and effects measured only with some latency from key lead levels [6,23]. Sample differences in moderating influences, such as socioeconomic status, quality of the home environment, maternal drug use during pregnancy, temporal pattern of lead exposure throughout the life of the child, and absolute levels of lead to which the child is exposed may all play a role in producing this mix of results. The present work examines the magnitude of the effect of postnatal blood lead level on the General Cognitive Index (GCI) of the McCarthy Scales of Children's Abilities, and describes how the effect varies with the time between the blood lead measurements and the intellectual assessments. 1. Method 1.1. Subjects We recruited 502 pregnant women attending the prenatal clinics at 12 weeks of pregnancy at the National Institute of Perinatology in Mexico City. The Ethics Committee of the institution approved the study and signed informed consents are on file. The inclusion and exclusion criteria resulted in 436 healthy women with normal pregnancies, without complications during delivery, and delivering apparently healthy babies. Details of the selection criteria and characteristics of the sample have been previously published [21]. We followed the children through structured interviews of the parents, and with psychometric tests, anthropometric measurements, and blood lead determinations every 6 months from delivery. In addition, we captured information from hospital records, determined socioeconomic level, and measured blood lead level of the mother during pregnancy (every 8 weeks from week 12 to week 36) and at delivery from the mother and umbilical cord. We report data from 112 children (47.3% males) who provided a complete record of five McCarthy evaluations at 6-month intervals between 36 and 60 months of age.

1.2. Collection of developmental data As there are no norms for the McCarthy scales in the Mexican population, we used United States norms [15] to calculate the GCI, with a Spanish translation of the test [16]. Four trained psychologists who did not know the blood lead levels of their subjects applied the tests. We evaluated interexaminer reliability by calculating the correlation in GCI scores assigned by two of the psychologists with the scores of a third psychologist whom they observed applying the test in all possible combinations with 10 subjects for each combination. As scoring criteria are well defined for the McCarthy scales, the mean observer ± examiner correlation was 0.99. We did not examine inter-examiner reliability that might vary as a result of examiner variability in application of the test, as this would have required multiple applications of the test at short intervals in the same child. 1.3. Blood lead sampling and analysis Venous blood was drawn into purple-top Vacutainers that contained ethylenediaminetetraacetic acid (EDTA). Personnel at the Environmental Associates Laboratories (ESA Labs) in Chelmsford, MA, analyzed blood samples for lead levels. Quality control information is available elsewhere [21]. The ESA Labs is a reference laboratory for the Centers for Disease Control Blood Lead Proficiency Testing Program, and participates in the New York State Department of Control blood-lead-testing program. 1.4. Data analysis We used the general linear model for repeated measures of SPSS (SPSS, Version 9, Chicago, IL) to analyze the pattern of lead effect on GCI across age. The repeated measure was GCI measured every 6 months from 36 to 60 months of age. Prenatal lead measures tested were average of 12- and 20-week maternal blood lead (first half of pregnancy) and average of 28- and 36-week maternal blood lead (second half of pregnancy), average of all maternal prenatal blood lead (12 to 36 weeks), maternal blood lead at delivery alone, cord blood lead alone, and the average of maternal blood lead at delivery and cord blood lead. Postnatal lead measures were mean blood lead concentrations calculated in three age ranges: 6± 18 months, 24 ± 36 months, and 42± 54 months. We used the same control variables in all models, selected a priori according to their association with infant development [20]. We did not use the HOME score in our model, as did other authors. We found such a high degree of collinearity between HOME scores and maternal IQ (Wechsler Intelligence Scale for Adults, Spanish Version) with GCI that we omitted the HOME score in favor of the maternal IQ variable. In addition, we used sex of child, Apgar score at 5 min, birth weight, serial order of child in the family, educational level of mother, and family socio-

L. Schnaas et al. / Neurotoxicology and Teratology 22 (2000) 805±810

807

Table 1 Differences in general cognitive index and mean blood lead level between subjects with complete and incomplete data Group with complete data

Group with incomplete data

Measure

(Meansa)

N

(Means)

N

Probability

GCI at 36 months GCI at 42 months GCI at 48 months GCI at 54 months GCI at 60 months Mean blood lead 6 ± 18 months Mean blood lead 24 ± 36 months Mean blood lead 42 ± 54 months

101.3 106.2 104.6 105.0 105.8 10.1mg/dl 9.7 mg/dl 8.4 mg/dl

112 112 112 112 112 112 112 112

99.0 102.2 97.1 100.4 102.2 9.8 mg/dl 11.0 mg/dl 9.8 mg/dl

77 65 76 49 44 121 53 91

>0.10 0.02 0.001 0.02 >0.10 >0.10 0.04 0.02

a

Blood lead levels are geometric means.

economic level as control variables. All variables, including the mean blood lead level (transformed into natural logarithmic values), were entered as a block in the analysis, regardless of their statistical significance in explaining child GCI. We constructed two series of models, one with an interaction term between sex and blood lead, the other without the interaction term. As none of the sex by lead interactions was significant, either within or between subjects, we do not report those models here. We examined the diagnostic tests for general linear models and, where indicated, adjusted degrees of freedom for within subject tests. We used planned polynomial contrasts (linear, quadratic, cubic and order 4) to examine within-subject effects (variation of lead effect with GCI at various ages). Bonferroni correction was used for post hoc comparisons of the age factor within each model (i.e., for each age range of blood lead average), but no correction for probability inflation was used for between model comparisons. We used the coefficient of the lead variable in the resulting regression models to calculate probabilities of lead effect on GCI for between-subjects effects. 2. Results There were no significant differences in 5-min Apgar, birth weight, serial order of child in family, socioeconomic status, maternal educational level, and maternal IQ between the group with complete McCarthy scales data and the group of subjects excluded from the analysis with incomplete McCarthy scales data ( p >0.10). How-

ever, we did find significant differences between the two groups in some blood lead measures and in some GCI measures (Table 1). The geometric mean of blood lead at 24 ±36 months and at 42 ± 54 months was significantly higher in the excluded group. The GCI at 42, 48, and 54 months was also significantly lower in the excluded group. We found no significant effects using any of the prenatal and perinatal blood lead measures, either within subjects (change of lead effects on CGI with age), or between subjects (effect of lead on CGI). The only significant lead effects found were those using postnatal blood lead. Table 2 shows the test for interaction between blood lead level and age at which the GCI was determined (within subject effect). Overall F tests for lead effect on GCI with age were significant for lead at 6± 18 months ( p = 0.076) and 24 ± 36 months ( p = 0.044). The linear polynomial contrast was significant for lead at 6 ± 18 months ( p = 0.047), and the quadratic polynomial contrast was significant for lead at 24 ± 36 months ( p = 0.019). The effect of lead at 6 ± 18 months on GCI tends to become increasingly negative with increasing age. The effect of lead at 24 ±36 months on GCI tends to become increasingly negative with age to 48 months and then tends to return toward baseline. We did not find a significant effect of lead at 6 ± 18 months on age-collapsed GCI in between-subjects analysis (F1,103 = 0.03, p = 0.859). The effect of lead at 24 ± 36 months on age-collapsed GCI was marginally significant (F1,103 = 2.97, p = 0.088) and the effect of lead 42 ± 54 was significant (F1,103 = 4.56, p = 0.035). Increasing blood lead between 24 and 54 months is associated with decreasing GCI over all ages.

Table 2 Tests of within-subjects effects blood lead on GCI Source

Type III sum of squares

df a

Mean square

F

p

ln(Pb 6 ± 18) ln(Pb 24 ± 36) ln(Pb 42 ± 54)

348.68 403.51 6.23

3.97 3.97 1

87.81 101.62 6.23

2.13 2.48 0.23

0.076b 0.044c 0.632

a b c

Degrees of freedom, adjusted by Huynh ± Feldt epsilon. Orthogonal polynomial contrasts: linear effect, p = 0.047. Orthogonal polynomial contrasts: quadratic effect, p = 0.019.

808

L. Schnaas et al. / Neurotoxicology and Teratology 22 (2000) 805±810

Fig. 1. Summary of the effects of blood lead level on General Cognitive Index. (Ð) Effects of mean 6 ± 18-month blood lead level on GCI. (- - -) Effects of mean 24 ± 36-month blood lead level on GCI. (. . . . .) Effects of mean 42 ± 54-month blood lead level on GCI. Vertical lines through each point represent standard errors of the regression coefficients adjusted for control variables in multiple regressions. Points are offset on the time axis for clarity only. Six- to eighteen-month mean lead level linear polynomial within-subjects contrast was significant ( p = 0.047). Twenty-four- to thirtysix-month mean lead level quadratic polynomial within-subjects contrast was significant ( p = 0.019). (+) p < 0.10 for between-subjects effect of lead. (++) p < 0.05 for between-subjects effect of lead. Lead effect on GCI generally increased in time up to 48 months, then decreased. Note latency for effect of lead level measured at earlier ages to appear at later ages.

Fig. 1 shows the pattern of lead effect at the three age ranges studied on GCI. It also shows the ages at which significant lead effect on GCI was detected for each blood lead measure. There were no ages at which a significant effect on GCI was noted for blood lead at 6± 18 months ( p > 0.10). Increasing blood lead at 24 ± 36 months was associated with decreased GCI at 48 ( p = 0.021) and 54 ( p = 0.073) months. Increasing blood lead at 42± 54 months was associated with decreased GCI at 54 ( p = 0.040) and 60 ( p = 0.060) months. 3. Discussion A temporal pattern in the manifestation of lead effect on cognitive development in children can be seen in several prospective lead studies [2,14], although only one published study has tested the temporal effect statistically [5]. This latter study examined temporal trend of cord lead effects on Bayley Mental Development Index from 6 to 24 months of age, and found no significant interaction of lead effect with age. This same study found that a statistical association between prenatal blood lead level and intellectual development did not appear until 1 year later (at 12 months) and then could not be found at lapses that exceed several years from the blood lead measurement [7]. In the same study at

later ages, blood lead at 24 months, but not earlier or later, still predicted decreased intellectual performance out to 10 years [6,8]. Another study found the strongest associations between blood lead and intellectual development measured concurrently [14]. Different profiles of blood lead with age as well as different blood lead concentrations across studies, among other differences, could partially account for the variable pattern of results, especially if lead effects at certain blood lead concentrations and at certain ages is reversible. As mentioned above, all but one of the previous results were produced with repeated but separate analyses of the cohort at various ages, but without using a formal repeated measures design. An unavoidable design defect in all studies of child development that start at less than 2 years of age and continue for years after is that no single psychometric test can cover the entire age range probed. As children's capacities increase with time, tests at different ages must probe a changing constellation of abilities, as already mentioned by others (e.g., [7,24]). Thus, an apparent latency to expression of lead effect and the apparent disappearance of effect with time may be a result of lead affecting a particular group of skills that are tested for only at certain ages. We partially avoided this problem by focusing our analyses on the age range (36 ±60 months) covered by a single test, the McCarthy Scales of Children's Abilities. However, even the McCarthy scales' test items change in character, as well as difficulty, according to the intellectual age of the child. Although our dependent variable changed in time, in contrast to other published studies, we used a repeated measures analysis that allowed us to statistically test temporal changes in lead effect with age. We found that the effect of lead at 6 ±18 months on GCI significantly increased in an inverse relationship with age at intellectual testing until at least 48 months, despite not finding any significant between-subjects lead effects at any age of testing. With lead at 24 ±36 months we found statistical evidence that the inverse relationship between lead level and GCI increases with age until 48 months, and then decreases. We also found significant betweensubjects effects for these lead levels on GCI at 48 and 54 months. Although GCI was statistically tested at only two ages (54 and 60 months) with lead levels at 42± 54 months (perhaps an insufficient number to detect within subjects trend), we also found statistically significant between-subjects lead effects at 54 and 60 months. If we ignore for the moment the changing nature of the McCarthy scales with intellectual age, our results statistically support the idea that postnatal lead exposure, at levels measured in this study, is best related to intellectual deficits measured months to years later, and that thereafter the effect of lead becomes less evident. The present data do not allow us to choose between hypotheses in which the decrease in lead effect after reaching its maximum at a certain age is due to the changing constellation of specific abilities probed with age in assessing CGI, or is due to an hypothesized

L. Schnaas et al. / Neurotoxicology and Teratology 22 (2000) 805±810

temporary lead effect on CGI with this range of exposure at these ages. In this set of results, however, peak effects of prior lead exposure occur at 48± 54 months of age, regardless of age at blood lead measurement. 3.1. Design issues We selected subjects for this analysis based on their having complete McCarthy scale evaluations from 36 ±60 months. The statistical comparison of this group with the group with incomplete data indicated that the selected sample differed in important aspects from the unselected sample. Though these factors have undoubtedly altered the results, the bias introduced works in favor of our interpretation. It is interesting to note that precisely those ages at which the lead effect on GCI was greatest, both within and between subjects, were the ages at which significant differences in lead and GCI between selected and unselected groups were found. The unselected group had significantly higher blood lead levels and lower GCI than the selected group. We suggest that the significant differences in GCI between both groups were due in part to the higher blood lead level of the unselected group. If the same factors that produced the results in the selected group operated in the unselected group, we would expect to find lower GCI at 48 and 54 months due to the higher blood lead level. We constructed several separate multivariate models, one for each age range of blood lead level. Thus, we expect some inflation of probabilities noted for lead effects due to multiple testing. Although we used a repeated measures model, the repeated measure was the GCI alone, not blood lead. Trial testing with all three postnatal blood lead measures in the model at one time indicated very high levels of collinearity, requiring the removal of all but one of the blood lead measures for each test. The high correlation among the three blood lead measures may be responsible for the similarity of pattern among the three models, not withstanding the fact that each blood lead measure produced a unique pattern of results, at least with respect to statistical significance. Furthermore, we used exactly the same subjects with each of the three models, making it impossible that our results were due to different subjects contributing to different models. The items of the McCarthy scales change with increasing intellectual age, as noted before. For example, part of the verbal scale requires first that the child identify objects by pointing to a named object in a group of objects. If the child passes all items in the previous part, he must name objects pointed to by the examiner, and similarly after that must define a word named by the examiner. If the child fails to complete any of the three sections of the test, the child is not examined on the following set. Part of the perceptual scale (right ±left orientation) is not given to any child with a chronological age of less than 5 years, no matter how well the child performs on other scales. Thus, the qualitative

809

nature of the McCarthy scales change with both increasing mental age and increasing chronological age. Part of any age-related difference in response to lead level, or other independent variable, may be affected by the changing nature of the scale itself as a function of age. We tested the same children with the same test every 6 months, as there are no alternate versions of the McCarthy scales available. There was undoubtedly some learning across time that could have affected our results. This can be seen in the highly significant change (not shown) between the GCI at 36 months and each of the subsequent ages ( p < 0.001, Bonferroni post hoc comparisons). However, these data show that there were no significant differences among the GCI scores between 42 and 60 months. Most of the test-specific learning, as defined by significant increase in test scores, took place between the first and second application of the test (between 36 and 42 months). Significant between-subject associations between GCI and lead were found for GCI between 48 and 60 months, depending upon the lead measure tested, outside the age range at which objective evidence of test learning was noted. We also note that the children in this study entered kindergarten at 48 months. The additional stimulation provided children at this time could explain some of the decrease in lead effect from 48 months described here. Similar studies in other countries with different educational plans might be examined to determine the viability of this explanation. 3.2. Public health implications Our results are in accord with some other prospective studies; effects of lead on children's intellectual development are not seen for months or years after measuring lead level, and then seem to disappear at later ages. However, we should not necessarily interpret the apparent lack of effect of lead on psychometric indices of intelligence several years after exposure to mean that children exposed to lead do not suffer lasting deficits in intellectual development. The problems of changing test requirements with advancing age discussed above, as well as the previously extensively discussed problems of measuring all other influences on intellectual development in the intervals between measurements, even with well-designed prospective studies, can obscure real relationships between early lead exposure and intellectual development years later. However, even the most conservative interpretation of the above results suggests that early childhood lead exposure is related to greatest intellectual impairment at an age when most children are just entering school. If decreased mental development at this age leads to poor school performance in the early grades, educational attainment may continue poorly, even in later years. The outcome of early poor educational achievement is often increased absenteeism, increased risk of dropping out of school, and increased risk for delinquency, conse-

810

L. Schnaas et al. / Neurotoxicology and Teratology 22 (2000) 805±810

quences already empirically associated with early lead exposure [18]. Acknowledgments This work was supported in part by the Secretariat of Health, Mexico (SecretarõÂa de Salud), and the U.S. Environmental Protection Agency. Special acknowledgements are due to the parents and children participating in the study. The opinions expressed herein are those of the authors and should not be interpreted as those of the participating institutions or funding agencies. References [1] P.A. Baghurst, A.J. McMichael, N.R. Wigg, G.V. Vimpani, E.F. Roberts, R.J. Roberts, S.L. Tong, Environmental exposure to lead and children's intelligence at the age of seven years. The Port Pirie Cohort study [see comments], N. Engl. J. Med. 327 (18) (1992) 1279 ± 1284. [2] E.C. Banks, L.E. Ferretti, D.W. Shucard, Effects of low level lead exposure on cognitive function in children: A review of behavioral, neuropsychological and biological evidence, Neurotoxicology 18 (1) (1997) 237 ± 281. [3] D. Bellinger, A. Leviton, H.L. Needleman, C. Waternaux, M. Rabinowitz, Low-level lead exposure and infant development in the first year, Neurobehav. Toxicol. Teratol. 8 (2) (1986) 151 ± 161. [4] D. Bellinger, A. Leviton, M. Rabinowitz, H. Needleman, C. Waternaux, Correlates of low-level lead exposure in urban children at 2 years of age, Pediatrics 77 (6) (1986) 826 ± 833. [5] D. Bellinger, A. Leviton, C. Waternaux, H. Needleman, M. Rabinowitz, Longitudinal analyses of prenatal and postnatal lead exposure and early cognitive development, N. Engl. J. Med. 316 (17) (1987) 1037 ± 1043. [6] D. Bellinger, J. Sloman, A. Leviton, M. Rabinowitz, H.L. Needleman, C. Waternaux, Low-level lead exposure and children's cognitive function in the preschool years, Pediatrics 87 (2) (1991) 219 ± 227. [7] D.C. Bellinger, K.M. Stiles, Epidemiologic approaches to assessing the developmental toxicity of lead, Neurotoxicology 14 (2 ± 3) (1993) 151 ± 160. [8] D.C. Bellinger, K.M. Stiles, H.L. Needleman, Low-level lead exposure, intelligence and academic achievement: A long-term follow-up study, Pediatrics 90 (6) (1992) 855 ± 861. [9] G.H. Cooney, A. Bell, W. McBride, C. Carter, Neurobehavioural consequences of prenatal low level exposures to lead, Neurotoxicol. Teratol. 11 (2) (1989) 95 ± 104. [10] K.N. Dietrich, K.M. Krafft, R.L. Bornschein, P.B. Hammond, O. Berger, P.A. Succop, M. Bier, Low-level fetal lead exposure effect on neurobehavioral development in early infancy, Pediatrics 80 (5) (1987) 721 ± 730. [11] K.N. Dietrich, P.A. Succop, O.G. Berger, P.B. Hammond, R.L. Bornschein, Lead exposure and the cognitive development of urban preschool children: The Cincinnati Lead study cohort at age 4 years, Neurotoxicol. Teratol. 13 (2) (1991) 203 ± 211. [12] C.B. Ernhart, M. Morrow-Tlucak, M.R. Marler, A.W. Wolf, Low level

[13]

[14] [15] [16] [17]

[18] [19] [20] [21] [22]

[23] [24] [25]

[26]

[27]

[28]

lead exposure in the prenatal and early preschool periods: Early preschool development, Neurotoxicol. Teratol. 9 (3) (1987) 259 ± 270. C.B. Ernhart, M. Morrow-Tlucak, A.W. Wolf, D. Super, D. Drotar, Low level lead exposure in the prenatal and early preschool periods: Intelligence prior to school entry, Neurotoxicol. Teratol. 11 (2) (1989) 161 ± 170. P. Factor-Litvak, G. Wasserman, J. Kline, J. Granziano, The Yugoslavia prospective study of environmental lead exposure, Environ. Health Perspect. 107 (1) (1999) 9 ± 15. N. McCarthy, The McCarthy Scales of Children's Abilities, The Psychological Corp., New York, 1972. N. McCarthy, Escalas McCarthy de Aptitudes y Psicomotricidad Para NinÄos, Publicaciones de PsicologõÂa Aplicada, Madrid, 1988. A.J. McMichael, P.A. Baghurst, N.R. Wigg, G.V. Vimpani, E.F. Roberts, R.J. Roberts, Port Pirie Cohort study: Environmental exposure to lead and children's abilities at the age of four years, N. Engl. J. Med. 319 (8) (1988) 468 ± 475. H.L. Needleman, J.A. Riess, M.J. Tobin, G.E. Biesecker, J.B. Greenhouse, Bone lead levels and delinquent behavior, Jama 275 (5) (1996) 363 ± 369. H.L. Needleman, A. Schell, D. Bellinger, A. Leviton, E.N. Allred, The long-term effects of exposure to low doses of lead in childhood. An 11-year follow-up report, N. Engl. J. Med. 322 (2) (1990) 83 ± 88. D. Papalia, S. Wendkos Olds, A Child's World, Infancy Through Adolescence, McGraw-Hill, New York, 1990. S.J. Rothenberg, S. Karchmer, L. Schnaas, E. Perroni, F. Zea, J. FernaÂndez Alba, Changes in serial blood lead levels during pregnancy, Environ. Health Perspect. 102 (10) (1994) 151 ± 160. L. Schnaas, S. Rothenberg, E. Perroni, R. HernaÂndez, C. HernaÂndez, S. MartõÂnez, RelacioÂn entre la exposicioÂn perinatal y postnatal al plomo y el desarrollo intelectual del ninÄo a los 42 meses de edad (Relation between perinatal and postnatal exposure to lead and intellectual development of the child at 42 months of age), Perinatol. Reprod. Humana 13 (3) (1999) 214 ± 220. K.M. Stiles, D.C. Bellinger, Neuropsychological correlates of lowlevel lead exposure in school-age children: A prospective study, Neurotoxicol. Teratol. 15 (1) (1993) 27 ± 35. S. Tong, Lead exposure and cognitive development: Persistence and a dynamic pattern, J. Paediatr. Child Health 34 (2) (1998) 114 ± 118. G. Wasserman, J.H. Graziano, P. Factor Litvak, D. Popovac, N. Morina, A. Musabegovic, N. Vrenezi, S. Capuni Paracka, V. Lekic, E. Preteni Redjepi, et. al, Independent effects of lead exposure and iron deficiency anemia on developmental outcome at age 2 years, J. Pediatr. 121 (5 Pt 1) (1992) 695 ± 703. G.A. Wasserman, J.H. Graziano, P. Factor-Litvak, D. Popovac, N. Morina, A. Musabegovic, N. Vrenezi, S. Capuni-Paracka, V. Lekic, E. Preteni-Redjepi, et. al, Consequences of lead exposure and iron supplementation on childhood development at age 4 years, Neurotoxicol. Teratol. 16 (3) (1994) 233 ± 240. N.R. Wigg, G.V. Vimpani, A.J. McMichael, P.A. Baghurst, E.F. Roberts, R.J. Roberts, Port Pirie Cohort study: Childhood blood lead and neuropsychological development at age two years, J. Epidemiol. Community Health 42 (3) (1988) 213 ± 219. G. Winneke, A. Brockhaus, U. Ewers, U. Kramer, M. Neuf, Z. Annau, Results from the European multicenter study on lead neurotoxicity in children: Implications for risk assessment. Behavioral toxicology and risk assessment, Neurotoxicol. Teratol. 12 (5) (1990) 553 ± 559.