Document 6bgrV61EDxOaGxJ5K4pKN3Bqd

INTERNATIONAL LEAP ZIFfO RESEARCH ORGANIZATION. INC. POST OFFICE BOX 12036 RESEARCH TRIANGLE PARK. N.C. 27709-2036 TELEPHONE (919) 361-4647 TELEX: 261533 FACSIMILE: (919)361-1957 TO: ILZRO Lead Environmental Health Technical Committee Members of the Sydney Cohort Study Steering Committee FROM: Craig J. Boreiko, Ph.D. Manager, Environmental Health PATE: November 4,1992 SUBJECT: Lead and Child Development / Enclosed you will find a variety of items that may be of interest to you. The Port Pirie Child Development study has just published their seven-year follow-up in the New England Journal of Medicine. A copy of the publication is enclosed, along with an editorial by Katherine Mahaffey which appeared in the same issue of die journal. The Port Pirie Child Development study reports that there is a small, but statistically significant, effect of lead upon the intellectual development of children. The deficits reported are related to the lifetime average blood lead concentration in the study cohort. In essence, this mode of data expression suggests that prolonged elevated exposure to lead, as opposed to transient increases, are those which should be attributed greater significance from a public health perspective. The Port Pirie findings are difficult to interpret in isolation - results from the other prospective studies will greatly facilitate our interpretation of the newly emerging data base. The Boston and Cincinnati Prospective studies will be publishing their findings in the journal Pediatrics over the next several months. The Boston study will be reporting similar results from their ten-year follow-up with a dose-response that may be similar to that of Port Pirie. The Cincinnati Prospective study will also be reporting effects of lead upon intellectual development. In the Cincinnati study, we anticipate that the doseresponse curves will suggest that the effects of lead are minimal below 20 pg/dl Cumulative exposure. This 20 pg/dl "threshold" is interesting in light of the Port Pirie findings that there may be a discontinuity in the dose-response curve at approximately 17.5 jig/dl (figure 1). Although effects on IQ shall be reported by all studies, careful attention should be paid to the level of consistency observed (e.g. verbal vs. performance, male vs. female, etc.) in the outcome measures reported to he impacted by lead exposure. I have also enclosed a copy of a review article produced by researchers at the Center for Disease Control. This article by Thacker et al. reviews the previously published data of the five prospective studies, discusses the limitations of meta-analysis in the analysis of the prospective study data, and concludes that a weight of evidence approach is optimal for interpretation of the data. In essence, this review concludes that definitive statements regarding the effects of low-lead lead exposure on IQ cannot be derived from the existing longitudinal data. However, using a weight of evidence approach which factors in the ILZRO Pb Bnv. Health Comm, Sydney Cohort Steering Comm. November 4,1992 Page - 2 * results of other studies, the authors suggest that there is an adverse effect of lead upon intellectual development in children. I would like to draw your attention to Table 3 of the Thacker et al. paper. The Sydney cohort study has consistently reported the absense of lead effects. A variety of individuals have criticized the study for "methodological weaknesses" and cite this as a reason for the failure of the study to detect a lead effect. The CDC researchers conducted an interesting exercise during the course of their review. The methodologies employed during the course of the various prospective studies were stripped of all identifiers and sent to an independent panel of scientific experts for review. The Sydney study was judged to be the study with the strongest design characteristics by all reviewers. The Port Pirie and Cleveland studies "tied" for second. Boston was next, followed by Cincinnati. I hope you find the enclosures to be of interest. Please do not hesitate to contact me should you have any questions or comments. Best regards. Craig J. Boreiko, Ph,D. Manager, Environmental Health CJB/mf Vol.327 No. 18 EXPOSURE TO LEAD AND INTELLIGENCE AT AGE 7 -- BAGHURST ET AL. 1279 EI'TYIRONMENTAL EXPOSURE TO LEAD AND CHILDREN'S INTELLIGENCE AT THE AGE OF SEVEN YEARS The Port Pirie Cohort Study Pe t e r A. Ba g h u r s t , Ph .D., An t h o n y J. Mc Mic h a e l , Ph .D., Ne il R.Wig g , M.B., B.S., Gr a h a m 'V. Vimp a n i, Ph .D., Ev e l y n F. Ro ber t s o n , M.B., Ch ,B., Ru s s e l l j. Ro b e r t s , M.Cu n .Ps y c h ., a n d Sh i-Lu To n g , M.P.H. Abstract Background. Exposure to lead in early child hood is thought to result in delayed neuropsychologi cal development. As yet there is little longitudinal evi dence to establish whether these effects, persist into later childhood. Methods. We measured IQ scores in 494 seven-yearod children frGm the: lead-smelting community of Port Pirie, Australia, in whom developmental deficits asso ciated with elevated blood lead concentrations had al ready been reported at the ages of two and four years. Exposure to lead was estimated from the lead concentra tions in maternal blood samples drawn antenatslly and at delivery and from blood samples drawn from the children at birth (umbilical-cord Wood), at the ages of 6 and 15 months and 2 years, and annually thereafter Data relating to knowi covariates of child development were collected systematically for each chiid throughout the first sever years of life.; Results. We found inverse relations between IQ at the age of seven years and both antenatal and postnatal blood lead concentrations. After adjustment by multiple regres- sion for sex, parents' level of education, maternal age at delivery, parents' smoking status, socioeconomic status, qualify of the home environment, maternal IQ, birth weight, birth order, feeding method (breast, bottle, or both), duration of breast-feeding, and whether the child's natural parents were living together, the relation with lead exposure was still evident for postnatal blood samples, particularly within the age range of 15 months to 4 years. For an increase in blood lead concentration from 10m9 per deciliter (0.48 /imol per liter) to 30 MS per deciliter (1.45 jumol per liter), expressed as the average of the concen trations at 15 months and 2,3, and 4 years, the estimated reduction in the IQ of the children was in the range of 4.4 points (95 percent confidence interval, 2.2 to 6.6) to 5.3 points (95 percent confidence interval, 2.8 to 7.8). This reduction represents an approximate deficit in IQ of 4 to 5 percent. Conclusions. Low-level exposure to lead during early childhood is inversely associated with neuropsychological development through the first seven years Of life. (N Engl J Med 1932;327:1279-84;) EXPOSURE to low levels of lead in childhood may found to be inversely associated with postnatal blood result in impaired neuropsychological develop lead concentrations at the ages of two and four years, ment and classroom performance.1 The extent of thisafter adjustment for confounding factors.6,7 It was esti relation, however, after adjustment for the confound mated that a child with a blood lead concentration of ing effects of socioeconomic and environmental covar 30 Mg per deciliter (1.45 Mmol per liter) had a deficit of iates, has been debated,2'17 Taken together, the epide 3.3 points (approximately 3.2 percent) on the Bayley miologic studies indicate a moderate inverse relation Mental Development Index at the age of two years, between the body burden of lead (measured as blood and of 7.2 points (approximately 6.7 percent) on the or tooth lead1 concentrations) and the neuropsycholog McCarthy General Cognitive Index at the age of four ical or cognitive performance of children.18'25 Whether years, as compared with a child with a blood lead the effects disappear once lead exposure ceases or concentration of 10 Mg per deciliter (0.48 Mmol per whether eariy exposure to lead has effects on neuro liter). The cohort has now been followed into the pri psychological development that persist into later life is mary-school age range. This paper examines intelli uncertain. Longitudinal studies are clearly the best gence at the age of seven years in relation to lifetime way to address this question. exposure to lead. The Port Pirie Cohort Study began in 1979. Early results showed considerable interindividual variation Me t h o d s in blood lead concentrations during early child The Cohort hood,With approximately one third of the children in this lead-smelting community having concentra The original study population comprised 723 singleton infants bom in and around Port Pirie, South Australia, during the three- tions above 25 jig per deciliter (1.21. /umol per liter) on one or more occasions. Mental development was year period from 1979 to 1982. These infants represented an esti mated 90 percent of all singleton live births in the community dur ing this period. Of the 516 children who remained in the study through the age of seven years, developmental status (including Rom the Division of Human Nutation, CommonwealthSdciruffc Industrial ..Research Organis-ttion (P.A.B.), the Department of Community Medicine, tJnjk intelligence) was assessed in 494 within the specified age range of seven to eight years. versity of Adelaide (A.J.M.), site Child. Adolescent and Family Health Service (N.R.W.), the Department of Chemical pathology, Adelaide Const.for Wom Data Collection en's and Childrens Health (E.F.R.), and the School of Nursing, Flinders Univer sity (R.J.R.), all in Adelaide, South Australia; the Faculty of Medicine, Universi ty of Newcastle, New South W;Ues, Australia (G.V.V.); and the Department of Special Programmes. National Health Education institute, Beijing, China (S.-L.T.). Address reprint requests to Dr. Baghurst at C5IRO Division of Hu man Nutrition. F.O. Bo r 10041, Gouger Si.,Adelaide. SA 5000, Australia. Four trained nurse-interviewers collected up to three venousblood samples from each mother before delivery, a sample from the umbilical cord at birth, and capillary-blood samples from each child at, the ages of 6 and 15 months and 2 years, and annually there after.23-24 A pilot study demonstrated that thelead concentrations in DUP040011265 N 29305.01 1280 THE NEW ENGLAND JOURNAL OF MEDICINE Oct. 29, 1992 capillary-blood samples, collected according to a strict protocol, wire highly correlated (r = 0.97) with the lead concentrations de termined in venous-blood samples taken from 47 children who were two to four years of age. At the time of each blood sampling, the nurse-interviewer also conducted a structured interview to obtain information on a range of demographic, psychosocial, medical, and environmental factors. Measureme nt of Blood Lead Concentrations Blood lead concentrations were measured by electrothermal atomization atomic-absorption spectrometry.56 The analyses were subject to internal and external quality-control procedures, with consistently satisfactory results.11 The results were standardized to a packed-cell volume of 35 percent for ail samples except cord Mood, for which a value of 50 percent was used. developmental Assessment The IQ of each child was measured under uniform conditions by means of the revised version of die Wechsler Intelligence Scale for Children (WISC-R).77 The WTSG-R comprises 10 sequentially ad ministered siufasrales. Although they do not in themselves measure discrete neuropsychological functions, the first five subscales (infor mation, similarities, arithmetic, vocabulary, and comprehension) ai e used to estimate a verbal IQ, and the remaining subscales (pic ture completion, picture arrangement, block design, object assem bly, and coding) are used to estimate a performance IQ..All childi tn were evaluated by the same research psychologist, who was unaware of each child's lead-exposure statins. Although the psychol ogist had also assessed the child's abilities at the ages of two and fo ar years, he was not formally aware of the earlier results. The median age of the children on the day of testing was 186 days after tJif'.r seventh birthday (the 25th and 75th percentiles were i32 and 246 days, respectively). Covarlate Measures Other;factors likely to confound the relation between lead expo sure and IQ;, and for which ancillary information was collected, included1 socioeconomic status (established with use of Daniel's Scale of Prestige of Occupations in Australia78), the care-giving environment (assessed by the Home Observation for the Measuremen t of the Environment [HOME] inventory79), maternal intelli gence (measured with the Wechsler Adult Intelligence Scale71'), pa rental smoking habits and years of secondary education, whether the. parents w c t c living together, the birth weight and birth order of the child, and the duration of breast-feeding during infancy. Statistical Analysis Statistical analyses were performed on the natural logarithm of the blood lead concentration, and all reported mean values are geometric. For each child, a curve plotting the blood lead concentra tion against age was constructed. The lifetime average blood lead concentration up to a particular age was estimated by dividing the appropriate area under the curve by the specified age. The effects of potential confounding factors were investigated by multiple regression analysis with (log) blood lead concentration as a continuous explanatory variable. The covariates used in the final models included sex, birth weight, birth order, feeding method (breast, bottle, or both), duration of breast-feeding, parents' level of education; ma trrnal age at delivery, parents' smoking status, sodoecoaomic status, quality of the home environment, maternal IQ, and whether the child's natural parents were living together. Re s u l t s Loss io Follow-up The 207 children born into the cohort but subse quently lost to follow-up were similar to the 516 who remained in the study with respect to 12 of the 15 variables studied. The socioeconomic status of the children lost to follow-up was slightly lower, more of their mothers smoked (35 percent vs. 27 percent), and fewer were breast-fed during infancy (32 percent vs. 37 percent). The mean umbilical-cord blood lead concen trations for the two groups were almost identical (9.3 vs. 8.9 pig per deciliter [0.45 vs. 0.43 pimol per liter]). Blood Load Concentrations The geometric mean lead concentrations in mater nal blood collected both antenatally and at delivery, in cord blood, and in capillary samples collected throughout childhood are shown (according to quartile) in Table 1. The blood lead concentrations were highest at the age of two years; by the age of], seven years the mean values had fallen by over 40 percent. Lifetime averages at each age are also shown. Age-Specific Blood Lead Concentration and Children's IO The mean scores for verbal IQ, performance IQ, and full-scale IQ were 103.1 (95 percent confidence interval, 101.8 to 104.4), 105.9 (95,percent confidence interval, 104.6 to 107.1), and 104,7 (95 percent confi dence interval, 103.5 to 106.0), respectively. There was a consistent inverse relation between the blood lead concentrations and scores on all IQ scales (Table 2):, The mean IQ scores differed by 2.7 to 12 percent between the children with values in the highest and lowest quartiles foT blood lead concentration. The unadjusted relation between IQ and lifetime average blood lead concentration at the age of seven years is shown in Figure 1. For each IQ scale, there was an inverse gradient across most of the range of blood lead concentration. There is a suggestion that this gradient was less steep at higher exposures (>20.0 p,g per deciliter [>0.97 pmol per liter]), but because of the paucity of children with such exposure levels, there is greater statistical variability associated with these higher exposures. The gradient is steeper for verbal IQ than for performance IQ. The propor tion of the variance of full-scale IQ that could be accounted for by the blood lead concentration at dif ferent ages (without consideration of the covariates) varied from 1.4 to 6.1 percent. Other Covariatos and Children's IQ The unadjusted mean IQ scores for subgroups di vided according to-covariates that, may confound the relation between lead exposure and IQ are shown in Table 3. Many sociodemographic factors and neona tal or infant characteristics were strongly related to the child's IQ, in the anticipated direction. Socioeconom ic status, the quality of the home environment, and matcrpalintelligence were the variables most strongly associated with IQ and in simple regression analysis accounted for 8.4,12.0, and 13.3 percent, respectively, of the variance of the full-scale IQ. Sex and birth order of the child were not significant correlates of IQ. In simple regression analyses all measures of bipod lead concentration (antenatal, delivery, and postnatal averages) were significantly inversely associated with DUP040011266 VoJ. 327 No. 18 EXPOSURE TO LEAD AND INTELLIGENCE AT AGE 7 -- BAGHURST ET AL. 128) verbal, performance, and full-scale IQ. The inverse relation of the Table 1. Mean Blood Lead Concentration (within Quartiles) in Maternal Samples Tak en Antenataisy and in Children's Samples at Various Ages. blood lead concentration to verbal IQ was consistently stronger than its relation to performance IQ. In Qu a k t il e .!. AVERAGE ANTSKATAL _______ UtttXUCU. cor d Bl o o d Lead Co n c en t r at io n *______ ag e multiple regression c ntlvses (Table 6 ms 15 ino 2yr 3yr Ayr Syr 6yr Jyr 4), the efleu of adjusting for the covariates in IVblr* 3 was i.c attenuate markedly fe apparert association of the efci 11*t IQ v.'th the -arious measures of slmd lead concentra tion. In curricular, the regression coeJficisnt! associated with the lead concenfrrt c-r.s in ..ntcratal, deliv ery, and tord-hlood samples be came in-gudcanr. TV covariates Mean concentration I (low) 6.2 n 8,7 mi 10.6 JV (high) 14.3 4.3 7.4 9.9 15.0 Mean lifetime average concentration I (low) -- "a -- .-r- -- IV (high) -- microgramsper deciliter 8.3 11.8 12.6 lg.6 168 24.4 24,2 34.4 13.0 18.6 24,2 33.S 11.6 17.4 22.4 30.2 9.5 14.7 19.0 26.5 8.3 12.6 17.2 23.6 7.2 11.2 14.7 20.5 6.6 10.1 13.7 20.0 7.4 9.9 10.6 14.3 1315 18.0 18,4 23.8 11.6 16.6 203 27.1 12,2 17.4 21.7 28.2 12.2 17.6 213 27.7 11.8 17.0 21.1 26,9 11.2 16,4 20.3 25,9 10.8 15.7 19.7 24.8 cc>mr:bt-ttra rm=t to rhir direnu- To convert whies for lead ra mkromojw per liter, divide by 20.7. aune efrec were f,iose identified as, ceint' most closely 1 ciatcd to IQ: socioeconomic tions. However, the apparent sensitivity of the compo s"ur, TOt E score, and maternal IQ. However, nents of the WISC-R id the blood lead concentration for lifetime vraiv hood Jeaa concentrations rang differed. The associations between lifetime average ing from birTh through 15 months to birth through blood lead concentration and the information and 4 ; ''Srr tht jf were sTan :ncalh sign ficant inverse as- block-design subscales were stronger than those for sociatjcn: Mth erbal IO and full-scale IQ. On the any other subscales (Table 5). basis of rtcressjon a at 1/si-:, an increase in blood lead concern a nor. from 0 to 30 /xg per deciliter (0.48 to Dis c u s s io n 1.45 jxr.i'l per liter i was associated with a deficit in These results indicate that the inverse associations veilxl iQ that ean.tl xicorGmg to age from 5.5 to 6.4 between blood lead concentration and indexes of de p<- :n 3 ir ( 2 'r'rcfnt'. and in full-scale IQ the velopment reported earlier for this cohort6,7 persist e-ti or.red :ef cu v a; - - to >.j poirts (4.2 to 5.1 per- into the primary-school years. The strong correlation cfir'i Tic ierr's: ' naag',s m response to adjustment (r *= 0.65, P<0,001) of the full-scale IQ at the age of for .-n\ariaus v k re -n ine reiatiormof blood lead con seven years with the McCarthy General Cognitive In centration o oerfornarco IQ. dex at the age of four years6 indicates that many of the The cs-iiratcc linear Inst.'se relation between log children who scored poorly initially have not had great avciage I>tf"id lead .'onreniraiion and full-scale IQ, improvements in their Overall ranking by the age of determined Mt i us; of the covariate-adjusted coeffi seven years. cients at -hr a-;- ci 3 ' -ars in Table 4, is shown Two prospective studies have found that the lead in -Voif 2 (t-'we ore -imilar relations at other concentration in the umbilical-cord blood is predictive age: but me r go of ' ;s centrally located in the of developmental progress in early childhood.2,3 How range of apDarr otlv maximal sen sitivity -- 15 mei ths to 4 years of a gel. The g -h -rif i ore sen sitive to *h -fleets cf 1-ad than Table 2. Mean Verbal, Performance, and Full-Scale IQ Scores of the Seven-YearOld Subjects, According to the Quartile for Blood Lead Concentration at Various Ages.* were the bo/s Fin as in-rease in blood '3r nnren lation from If1 xe pe- d"rihit' (0 48 u-mol pet liten m 50 uz per deciliter (1 -*5 itriivi per liirrj. ,he cxpcctco "ov'n.i e-aruust'-d o-cr-meat in iuli-suae IQ \.i> 7.3 points for the girls and 2.6 points for the boys. Simple and Multivariable Analyses of Subscale Scores Simple and multivariable anal yses of WISC-R subscale scores showed that the mean subscale QU*AflL Av ek ag e An t en at al . Ums zu c a l Co r d Verbal IQ score I (tow) n IK IV (high) 107.3 103.9 102.8 98.6 Performance IQ score I (tow) 108.2 II 107.0 ni 107.0 W (high) 101.4 Full-scale IQ score I (low) 108.5 B 105.7 111 105.1 IV (high) 99.8 106.9 104.6 99.4 101.7 108.7 106.1 103.4 105,8 108.4 105.6 101.3 103.9 6 mo 107.9 102.8 101.7 98.0 109.5 106.0 104.1 102.4 109,4 104.7 102.9 100.0 15 TDO Ag e 3* 107.9 105.4 101.0 99.8 109.2 IQ6.7 104.7 102.9 109.3 1063 102.9 101.3 109.0 -104.4 100.7 98.5 109.6 107.8 103.8 101.8 1103 106.5 102.2 100.0 5yr 108.4 104.5 102.2 96.8 108.7 106.8 105.6 101.6 109.3 106.1 104.1 98.8 7.jr 108,2 106.4 1003 97.1 109.4 107.9 105.0 100.9 109,6 107.7 102.7 98:7 scores varied inversely with the life time average blood lead concentra -Different*, between qeattilcs were nauvticeily significant (P<Q.QI) inever, instance except for.ttae variationrtfperform, aace IQ scam with cord-blood lead concentrations (P 0.05). DUP040011267 1282 THE NEW ENGLAND JOURNAL OF MEDICINE Oct. 29,1992 ever, in a subsequent report from one of these study groups, the scores of the children on Use McCarthy Scales of Children's Abilities at the age of 57 months were inversely related to the blood lead concentration at 24 months of age, but not to the umbilical-cord blood lead concentration.12 This result is compatible with our overall finding of no significant relation be tween IQ and antenatal or perinatal blood lead con centration after other covariates are taken into ac count. The conclusions reached in this study are likely to be ecn.cr-.Dtjvr fer several reasons. First, the chii- dic.1. 'I'nairiug it 'he cohort had slightly more advan taged cackprouid' than those lost to follow-up. Since < hildren from disadvantaged families may be b c i e "u'nerable to tnc effects of lead than those frerc i,tor* fa-.erafck backgrounds,3'31 the inverse re lation oi lead to IQ migru have been stronger among in :h.Idre' lest v follow-up ,han among those re- .'nilrhiv in tt r cohort Second, the use of certain co- variaiu in a j t ratine analy.es may result in an o''"T.cjv.i'rirent :n studies of the association between le- i niiJ IQ 0 ` J'arental .smoking status, the quality of tiii home environment, ard maternal IQ may themrch ' s cm rihun. ro a child's lead burden,2-3,s,7JU in vlnci cos*' some of the *n.e 3fleet of lead would h- c-c'i rr.ino'-' d ny conn'..hug for these factors. T n rardor'iy riirnbuted Imprecision or misdassi- fv ,',n ii the me me emetic of lead exposure (or its ;-t ') jnders, rtili Ota.: rhe estimate of the effect toward me null value 3 .-' lice- c r estimates presented here are likely to L.' .on i-t aii p fer rhe reasons described above, the pcismlit, rhat v t did not completely control for posi- I" - 'p. -iC.ir -> enables and subsequently overesti- niar d i*> .rt e cannot be completely eliminated. Foi example, the developmental effects of anemia cjtud bias the study.3*35 After the exclusion of the 10 Table 3. Effect of Potential Confounding Variables on Mean IQ Scores at the Age of Seven Years.* COVAJUATE BU0 LEAfit Vs*i. IQ PMUPOBMANCE IQ Fuu-ScitU.lQ J*w Sex Si female MnrW. /pAucxtirm levclt !3yr 18.2 >3yr 15.5 104.4*0.9 102,1*0.9 100 0*0-8 iW0S <3 yr >3yt n level* 17.2 16.1 1014*1.0 1W49.9 Motto's age at child's birth fyir) '<423: 19.2 101.04:15 23-28 16.2 102.94:1.1 >28 16.4 105.64:1.1 No. of parents smoking None 15.9 One 17.4 Both 19.2 105.54:0.9 101,3 *14 99.7*1.8 105.9*1.0 106.0*0-9 105.4*0,9 104.3*0.9 103.4*0.9 108.8*0.9 101.7*0.8 108.4*0.9 104.8*1.0 103.0*0.9 108.6*0.9. . 108.3*0.9 104.5*1.2 106.1*1.0 107.1*1.1 102.8*1.2 104,8*1.0 106.8*14 107.6*0.8 1044*15 104.3*1.6 107.1*0.8 102.7*1-1 102.0*1.7 Lower Mitfrff* Higher HOME scores <40 . 40--45 >45 11 15.1 19.7 17.4 14.1 Mother's Q <85 85-95 >95 20.7 17.0 15,5 Binh weight (g) <2500 2500-3500 >3500 19.2 17.6 165 Binh order 1st 17.4 16,6 16.4 Feeding style Breast Mixed Bottle 15.7 14.7 18.4 T* of breast-feed^ 1-6 17.4 >6 15.5 98.0*1.1 iQ2:*.o 106.6*0.9 102.4*1.2 104.2*15 109,1*1,0 96.54:1.1 105.7*1.0 107.5*15 100.3*14 108,5*0.9 108.9*1.3 97.1*1.4 101.0*15 110.6*1,0 99.6*1,6 106.1*1.3 111.5*1.0 995*3.4 102.4*0,9 104.6*0,9 1024*3.7 104.7*0.9 107.9*1,0 1035*0.9 102.7*15 103.4*1.4 105.8*1.0 106.8*1.1 105.0*1,4 106.6*1.0 105.4*2.3 100,3*0.9 108.4*1.0 105.0*2.7 104.2*0.9 98.3*1.3 102.7*0.9 106.6*1,1 103.0*1.4 105.2*0.9 108.8*14 99.9*14 103.8*1,3 10&.5 2:0.9 98.0*1.1 107.6=0.9 108.9*1.2 97.8*1.3 103.6*1.2 112.1*1,0 11X1.5*3.6 103.7*0.9 106.7*0,9 104.8=0.9 105.0*1.1 104.4*1,4 1084*1.0 105,6*2.6 1025=0.8 100.4*1.3 104.2*0.9 108,3*1.1 Yes 16,6 1035*0.7 106.4*0.7 105.3*0.7 No 20.3 100.8*1.8 102.4*1.9 101.5=1.8 fVfUues ire the tifetirije svenge bJood feed.cancftmiioot t (be ige of.sevta yem. To convert values fpr ld to micromoles per iir, divide by 20.7. tYean refer; cumber ofytm of high school completed. Figure 1, Lifetime Average Blood Lead Concentration and IQ at the Age of Seven Years. To convert blood lead concentrations to micromoles per liter, divide by 20.7. children who had a packed-cell volume of less than 34 percent at the age of seven years, however, the esti mated regression coefficients for the blood lead con centrations changed little. Because a child with a high blood lead concen tration at one age is likely to have high concentra tions at other ages (a phenomenon referred to as "tracking''), it is difficult to determine critical or sensitive periods for the effects of lead. However, the maximal effect of lead on IQ was found for life time average blood lead concentrations from birth to any age between 15 months and 4 years. This sag- DUP040011268 Vol.327 No. 18 EXPOSURE TO LEAD AND INTELLIGENCE AT AGE 7 -- BAGHURST T AL. 1283 gests that exposure to lead in the preschool age range has a maximal effect on IQ. Among the WJSC-R subscales, the information and the block-design subscales were the most sensitive to the effects of lead. Since the information subscale may be culturally dependent, its effects on the relation of the more objective biock-c csign score; to the blood lead conc*nt -'tton were ti sted by including the infor mation score ? s an additional covariate in the regres sion mode.. "i"b<- aisccear.oa between fog biood lead corcentranoii trd the block-design scores was only partially attenuated iregression coefficient, -- l.00; 95 percer.i oonfdence interva , --0.42 to --1.58), which dis jds the notion that tar results are an artifact of the cul rural envirc nrnent The block-design subscale tests a person's pei<'ct>ttcl orgar ization and synthesis, spa tial 'Asuairzaucn,- nonv si La* concept, formation, and Table 4. Adjusted Coefficients of Log Lifetime Average Blood jaad Concentration from Multiple Regression Analyse? of iQ Scores at the Age of Seven Years.* Time. Bl o o d 5a mr e Ob t a in e d Before birtht After birih (cord biood) 0--15 Mo of age 0-2 Yr of age 0-3 Yrofage 0-4 Yrqf age 0-7 Yr of age VjEMAj. IQ --1.5--2.0 (0.46) -0.141.4 (0.04) -5.042.1 (0 03) --5.8425 <<0.0I) --5.74:2.4 (0.03) --5.04:2.5 (0.04) -4.34:2.6 (0.10) jpEXTOftMANCX IQ -1.04:2.1 (0.62) 1.04:1.5 (050) -2.34:2.2 (0.30) -2.44:2.2 (0.28) -2.842.4 (0.24) -3.34:2.5 (0.18) -2.34:2.6 (0.37) Fu l l -Sc al e IQ -1.4* 2.0 (0.48) 0.64:1.4 (0.68) --4.04:2.0 (0 04) -4.64:2.1 (0.03) -4.842.3t (0.04) -4.64:2.4 (0.05) -3.74:2.5 (0.14) The eoefficidiu were adjusted-fora!} covinwes shown In Table 3. Plus-minus values are means $E. P value at*.shown in parentheses. tBfood sample were obtained imeoiuliy from the mother. mis means that die expected decrement in fuHsca!e IQ associated with an increase in avenge blood lead concentration from 10 to 30 ivg per deciliter (0.48 to 1.45 /Amo! per liicr) dunagihe firn .three yean of lifer is S.3 points: 4.8 x (Jog[30j - logilO]) * 5-3. Figure 2. Estimated Relation between Full-Scale IQ at the Age of Seven Years and the Lifetime Average Blood Lead Concentration up to the Age of Three Years. The line of best fit is shown, as estimated by multiple regression analysis. To convert blood iead concentrations to micromoles per liter, divide by 20.7. visual motor coordination. It also reflects the degree to which right and left cerebral functioning is integrat ed.36 Impaired spatial discrimination has also been reported in adult monkeys exposed to low levels of lead for long periods.37 The results indicate that the deleterious effects of environmental: lead are not large, and that only a small fraction of the overall variation in IQ can be attributed to lead exposure. Nevertheless, the so cial consequences of such an effect are not negligi ble. If a child with an IQ of less than 80 requires educational assistance, then the number of children requiring assistance in a community whose children are expected to have a mean IQ of 105 will be dou bled if the mean is reduced by approximately 5 per cent as a result of lead exposure. Health authorities in the United States have recently estimated that for each reduction of 1 /i.g per deciliter (0.05 panol per liter) in the Hood lead concentration due to lead-expo sure-abatement programs, there would be a net sav ing to society of approximately $2,000 per child, be cause the need for clinical attention and remedial education and the expense of lost productivity could be avoided.38 The fact that an inverse association between the blood lead concentrations and abilities has been ob served longitudinally at the ages of two, four, and seven years within this cohort suggests that even a low level of exposure to lead has an independent and en during effect on neuropsychological development in childhood. From a public health perspective, the early detection and abatement of lead in the environment are highly desirable. Table 5. Mean Age-Adjusted Subscale Scores and Estimated Regression Coefficients for Lifetime Average Blood Lead Con centration up to the Age of Three Years.* SUSSCALE BLOOD LEA& QUASmSf i n in IV GOEHHOEHT ESTIMATE Pv a l v e information Similarities Arithmetic Vocabulary Comtrrchensior; Picntre completion. Bienne arrangement Block design Object assembly Codidg 11.7 12.2 11.2 10.5 10.0 10.8 10.2 9,2 9.6 9.7 95 8.5 12.5 12.7 12.0 12.3 10.0 9.6 9.6 9.2 11.2 11.4 11.0 10.6 10.2 10.7 105 9-9 11.6 12.0 10.S 105 11.4 11.7 11.4 10.9 .10.8 10.1 10.9 9,9 -- 1.4540,57 -0.904:0.62 -0.5140.63 -0.4440.51 -0.8840.54 -0.1540.45 -0.3440.60 -I.6!40.62t -0.0840.S3 -0.2240.53 0.01 0.14 0.42 0.40 0.10 0.74 0.56 0.01 0.86 0.66 *The coefficients wen adjusted fortchpo! year, age attesting, and die covsriates listed in Table 3. Plus-minus values are means sSE. tTbjs lowest .values as the first guarttie, and die highest values an in the fourth quartile. , tThis meansdm theexpected decrement indie block-design score for an increase in average blopdkad concentration ftom 10 to 30/ig per tiedliter(0,48 to 1.45 junoi per.littr) during the fast three years of life is 1.8 points: 1* 0ogl3b] - JogtlOJ) 1.8, 1284 THE NEW ENGLAND JOURNAL OF MEDICINE . Oct. 29, 1992 We are indebted to the families who participated in this study, their doctors, the local hospital staff, and the nurse-interviewers (Ms. liiarbara Hobson, Ms. Chris Mavromatis, Ms. Mary-Anne Lange, Ms. Bronwen Morgan, and Ms. Louise Thomson); to the staff of the Department of Chemical Pathology, Adelaide Chil dren':; Hospital (Ms. B. King, Mr. R- Oldfield, and Dr. A.C. Pot lard); to Ms. M. Padhye and Ms. A. Bartlett for data management; and to Mr. C. Greeneklee, Commonwealth Health Department Laboratory in Port Pine. Re f e r e n c e s 1. lieeiHeaiaa HL,, Cannes C, Leviton A, etai. Deficits in psychologic and classroom performance ofchildren with elevated dentine lead levels. N Engl J Men !979;3C0689-95. 2. Bellinger D, i,:viton A. Wacernaux C, Nccdlznvin H, RtbinowisM. Lon- gituiituii analyses of prenatal and postnatal lead exposure and early cogni tive development. N Engl J Med 1987;316:1037-43. 3. Dieizich KN. JCraffi KM, Bornschein RL. et al. Low-level fetal lead expo sure. effect on neurobehavioral development in eariy infancy. Pediatrics 1987;80:721-30. ' :"T . ' 4. pulton M. Raab G. Thomson G, Laxcr. D. HunterR, Hepburn W_ Influence of blood lead on the abilityand attainment of children in Edinburgh, Lancet .. 1987;1:1121-6...................... ..' ! * > 5. Haraakis A, Kokkevi A. Maraveliaa; G, et al. Psychometric intelligence deficits m lead-exposed children. In: Smith MA, Giant LD. Sors Al, eds. Lead exposure and child de velopmem. an intemanonai assessment. Lancas ter. England; Kluwer Academic) 1989:211-23. 6. Vic Michael AJ, Baghnrst PA. Wigg NR, Vimpani GV, Robertson EF, Robens RJ. Pott Piric Cohort Study: environmental exposure to lead and children's abilities at die age offour years. N Engl J Med 1988,319:468- 73. ............' " ' I .: 7. 3V eg NR. Vimpani GV. McMidiasl Al, Grghurst PA.. Robertson EF, Robens RJ. Port Fine Cohort study: childhood blood lead and neuropsycho logical development at age two yens. 1 Epidemiol Community Health 1988;42:213-9. 8. Winneke G. Kramer U, Btockhaus U, et al. Neuropsychological studies in children with elevated tooth-lead concentrations. )1. Extended study. lot Ach Occup Eivunn Hea th 1987,51 231-52. 9. Smith M, Dt Ives T. Lansr own P-, Chyton B, Graham P The effects of lead exposure on urban children, the Institute of Child Heaith/Southampian Sludy Dev Med Child tfeurol Supf I 1963,47 1-54 10. Harvey PG, Hamlin'MW, KimnarR, Delves HT. Blood lead, hehavioorand intelligence test performance in preschool children. Sci Total Environ l`4:40:45-60. 11. Needleman HL, Schell A, Bellinger D, Leviton A. Allred EN. The long term effects of exposure to low doses of lead in childhood: an 11-year follow-up report- N Engl J Med 1990:322*83-8 12. EellingerB, Slomanf. Leviton A, PabinowitzM, Needleman HL. Water- naux C. Low-level, lead exposure and children's cognitive function in the preschool years Pediatrics i99i;87:219-27. 13. Laitsdcwn R, vule W, Lnbanowicz M-A, Blunter J. The relationship be tween blood-lead concentrations, m eiligence. attainment and behaviour in a school population: the second London stuay. lat Arch Occup Environ Health 1985.57:225-35. 14. Erabart CB. Morrow-T lucak M, Wolf AW. Low level lead exposure and intelligence in the preschool years. Sci Total Environ 1988:71:453- 9,.; 15. .Pccock SJ, Ashby- D. Smith MA. Lead exposure and children's inteUectual performance. Int 1 Epidemiol 1987;16:57-67. 16, EmhanCB, Morrow-Tluok M, Wolf fcW, Super D.Dtotar D. Low level feed exposure in prenatal and early preschool periods: intelligence prior to School entry, Ncurotoxicol Teratol 1989:11:161-70. 17: Cooney GH. Bell A, McBride W, Carter C. Neurobehavioural conse- queaces of prenatal low level exposures to tend. Neurotoxicdl Teratol 1989;11:95-104. 18. Environmental Protection Agency. Air quality criteria for lead. Research Triangle Park, N.C.: Environmental Criteria and Assessment Office, 1986. (EPA-600/S-83/028aF.) 19. Childhood lend poisoning -- United States: report to the Congress by the Agency for Toxic Substances and Disease Registry. JAMA 1988 260:1523- .33. 20. Lee WR, Moore MR. Low level exposure to lead: the evidence for harm accumulates. BMJ 1990;301:504-6. 21. Needleman HL, Gitsonis CA. Low-level lead exposure and the IQ of chil dren: a meta-analysis of modem studies. JAMA 1990;263:673-8. 22. Needleman HL, Bellinger D. The health effects of low level exposure to lead. Anna Rev Public Health 1991:12:111-40. 23. Baghuzst PA, Oldfield R, Wigg NR, MeMichael AJ Robertson EF, Vim- prai GV. Some characteristics and correlates of blood lead in early child hood: preliminary results from the Port Piric Study. Environ Rex !985:38: 24-30, 24. MeMichael AJ. Baghurst PA, Robertson EF. Vimpani GV, Wigg NR. The Port Pine Cohort Study: blood lead concentrations in early childhood. MedJ Aiol 1985:143:499-503, 25. Baghurst PA. Tong SL, McMich tel At, Pob-rtson EF. Wigg NR, Vimpani GVe. Determinants of bloc-- lead cnncattreutTS to age 5 years in a birth cohortistudy of chi.dre.-i Irnrg in me ie.d -[r-cllin: city of Port. Pine and suirounding areas Arch Environ Hechii 1992,47 203*10: 26. Whole blood--leterminstion ot lean -- rlectrothctmal atomization atomic absorption speetre-netr.; memud V'h'i cney Standards Association of Australia, 1985 'Australian Stanturo V ,"-1955 j 27 WechderD. W1SC-R m-nu.l 1 'echsler lr.te'licence scale for children -- revised. San Antor.io. Tex Psycho rgicit Corporation. 1974. 28. Daniel A. The measuremen- m" v a ii' riaas Community Health Stud 1984.8 218-22. 29 Bradley RH. Caldwell BM. Hon- ab-eroiu. (Lr measurement of the envi- ranmem; a revis: m of me preschool -ta- \.r.J MenrSefic.1979:84:235- ,,:p.44,..: . i 'W 3C '1 ecnsler D WJIS-R mni.il w-cinle'-oLi' intelligence scale--revised. New Yonc PsychjlogicrJ Lortorairn 19-1 hi 31 3*!1 nger D. Le-i.oi. a , 'V. e-nrux C s. IQ and social class. Int J Ep'de-ru'l i<9 16 180-5 32 Needle,eui HL Bel: user DC Typ- 1! taJ'acies in ihe study of childhood cxf-i ure *n lead ' iuw dose -1 rtiejl rj,d quantitative review. In: Smith MA. Grant LD Sore Al. eds Lead exposure and child development: an iniemauonal .isressmeiit Lancaster, Evsl ud. Kluwer Academic. 1989:293- 304 : 33 Ro'hmjp <J Medero m Jc t u log; Easton: Little. Brown, .1986. 34 V Jtn T. Le '.uth-icn i, CL due p. P'aiexeG,iron deficiency anemia: advert'- effects on ii-faui o vehemoroe development; Bsdiatrics 1989:84:7'j:'fIlf!,!lWW l:1 3i Luzon R Jimenez E. Wclf A`V Leg term developmental outcome of .f.tana *idi .ran -sfic,r-.cy. N Lpl J Me>1 1991:325.687-94. 36 Kaufman AS Inrriligent testing with the WJ5C-R New York: John Wiley, 1979. 1 37. Gilbert SG. Price DC. Low-level lifetime lead exposure produces behavioral toxicity (spatial discrimination reversal) in adult monkeys. Toxicol Appl Fhatroacoi 1987;91:484-90. 38. Binder 5, Falk M. Strategic plan for the eliminauon of childhood lead poisoning. A tlanta; Department of Health and Human Services. 1991. DUP040011270 1308 THE NEW ENGLAND JOURNAL OF MEDICINE Oct. 29, 1992 unknown. Amyloid Is formed in the patient's brain because of changes in the proteolytic processing of APR, and not because of alterations of genetic or mo lecular properties. There are two known pathways for processing APP -- constitutive secretion and internal processing within the endosomal--lysosomal system.50 In constitutive secretion a membrane-associated endoprotcase cleaves the 0/A* peptide into nontoxic fragments, whereas amyloidogenic fragments are like ly to accumulate as a result of APP metabolism in the lysosomal pathway. APP processing appears to be reg ulated by ncurotransmitter-rcceptor-couplcd activa tion. In the resting state the lysosomal pathway is dominant, whereas cholinergic stimulation of protein kinase C--linked muscarinic receptors preferentially increases the release of nontoxic fragments.11 With better understanding of factors that influence APP processing, it may be possible to develop, pharmaco logic: treatments designed to enhance constitutive se cretion or block the lysosomal pathway. Either ap proach would decrease /3/A4 deposition and, if started early enc.igh, might prevent neuronal degeneration and I he evolution of clinical dementia. I*- ;i*i' there are. no safe :ncl effect]/e treatments the mpr" e fogiut'on in Alzielmer s ciscase. There ar- -eironc br.-wr, to be cptirn su- ab-'ur die fu ture 1 inipo nd" embodying restonthe, protective, and pieverurn strategies are under development. The pre" cf reteorch into the ca_?ei 2nd mechanisms of neuronal degitiieratiorl in Alzh eimer's. disease is quick ening, and die investigations will yield additional drugs. Finally, as the Tacrine Collaborative Study showed, patients with Alzheimer's disease, their fam ilies, and their physicians stand ready to test promis ing drugs in rigorous clinical trials. Missadnuetts Geaentl Hospital Boffloa, MA021K __ _ JOHN H. Gr OWDOH. M.D. Re f e r e n c es 1. Gfowdon JH. Neturopluffmacobgy ofdegeperarivc diseases issnciared with aging. Med Res Rev 19832:237*57i i 2. Barms RT, Dean RL m, BeerB, Lippa AS. The cholinergic hypothesis of geriatric memory dysfunction. Science 1982217:408-14. 3. Davis KL, Thai LJ, Gamzu Er et al. A double-blind, placebo-controlled muldcenter study of tacrine for Alzheimer's disease. N Bag! J Med 1992227;I2S3*$, I r !- 4] Hefti F, Hamkka J, Kousel B Function of neurotrophic factors in the adult and aging brain anil their possible useintbe treatment of neurodegeneradve diseases. Neurobid Aging 1989;10:515-33. 5. Tbszynsld MH, Sang H, Yoshida K, Gage FH. Recombinant human nerve growth factor infusions prevent cholinergic neuronal degeneration in die adult primate brain. Ann Neurol 1991;30:625-36. 6. Koliatsos VE, CUrttertmck RE, Nauta HJW, ct al. Human nerve growth factor prevents degeneration of basal forebraia cholinergic neurons in pri mates. Ann Neurol 199120:831-40. 17. Schapira AHV, Cooper JM, Dexter D, Clark JB, Jenner P, Mazsden CD. Mitochondrial complex I deficiency in Parkinson's disease. J Neurochero 199024:823-7. ;"V: ! !,, . "i ; - 8. Beal MF. Does impairment of energy metabolism result in excitotoxic neu ronal death in neurodegeneradve illnesses? Ann Neurol 1992;31:119-30. 9. The Parkinson Study Group. Effect of deprenyl on the progression of dis ability in early Parkinson's disease. :N Engl X Med 1989221:13.64-71. 10. Golde TE. Estus S, Younkin LH, Selkoe DJ. Younkin SG. Processing of the amyloid protein precursor to potentially1 amybidogenic derivatives. Science 1992255:723-30. 11. Nitsch RM. Slack BE, Wurtman RJ, Growdon JH. Release of Alzheimer amyloid precursor derivatives stimulated by activation cf muscarinic acetyl choline receptors. Science (in press). EXPOSURE TO LEAD IN CHILDHOOD The Importance of Prevention Ba g h u r s t et al.1 are to be congratulated for con tinuing their long-term studies of intellectual deficits in children who were exposed to lead in early child hood. These investigators have identified deficits in intelligence, ii.ot just delays in neurobehavioral or mo tor development, among school-age children that are associated with exposure to lead in early childhood. That children's intellectual performance can be di minished by environmental exposure to a chemical as common as lead appears to startle many. This associ ation has been attacked in both the popular press and the scientific literature. The principal counterasser tion is that such deficits are due to other factors com mon among children with the highest prevalence of exposure to lead, such as poverty, urban environt mentsv lower social class, having a mother with too few resources to cope with multiple problems, and deterio rating bousing. Even when these social factors have been considered, whether through the type of popula tion studied or through the use of statistical tech niques, the assertion has persisted that not all the covariables have been considered or that the cohort studied has another as yet unidentified problem.2 The .assertion that the effects of lead are limited to socially and economically disadvantaged children repudiated by the findings of Baghurst et al. re ported in this issue of the journal,1 as well as by the results of another longitudinal cohort study of an affluent population in Boston.3 Child-development experts have predicted that the adverse effects of exposure to lead on intellectual development will be greatest among disadvantaged children.* The re sults of a longitudinal study of a cohort of severely disadvantaged children in Cincinnati verify this pre diction.4-7 Taken. together, the results of three longitu dinal studies1,3*4-7 demonstrate that the lead-associated decrements; in intelligence are persistent across cul tures, racial and ethnic groups, and social and eco nomic classes. Although the results of cross-sectional studies of lead-associated effects on IQ are inconsistent, longitu dinal studies have permitted a more thorough consid eration of the covariables that might interact with lead exposure. The higher social and economic status of the children studied by Baghurst et al.1 is in marked contrast to the traditional image of lead-exposed chil dren as those from poor and, medically underserved communities. The intellectual deficits associated with exposure to lead that were identified by Baghurst et DUP040011271 Vol 327 No. IS ed it o r ial s 1309 al.1 seem especially impelling because the cohort was drawn from families of skilled workers as well as from middle-class families, the testing instrument -- the revised version of die Wedisler Intelligence Scales for Children (W1SC-R) -- has been widely used ;md well standardized, and the children's integrated blood lead concentrations were comparable to those often en countered in the United States. In addressing the im portance of social and familial covariab] es, it is critics! to recogn'zr that the reduction in intellectual scores at jibuted to exposure to lead in early childhood was present after aatistical adjustment for the covariables known to Irfhienc: intellectual development. The Ioogi' id nal evalurtion of severely disadvan- ta'teri ch'jo ;i n C'^cinnati identifieo central <.udi- to-j-proees:,ii.g deheii_-. and an eight-piint decrement Tio.n 91 ~o 'll) n the D^riomance IQ (based on tire WlbC-R set'es a'ter rd,uu:nent for covanablcs) as th; hfetirnr a\er.ige olood lead concentration in creased frerr 10 m 35) fig per deciliter (from 0.'-8 to l.'O (.'.mo! p<; iit.'rj These children also hac de- cica'-ed f.ne "ii<or skills' a'ter adjustin' nt for eevaria- bles tedeoting sue factors a= motherm-f skills and so- cioeccnor!-'- < tarns A cohort of children from affluent families m Eoston- had a six-point decline in the vVfSC-P. full -calc TQ when restea at th- age cf l0 yea's tnat ,> n. a^ratiaitri 'virh an increase in the blood lead corc_nfT?ro- ->T >0 fig y r deciliter iC-'g yttmol per liter! wrhm the tangf of 0 to 25 ftg per dec liter (0 to 12' p.r'0 m at 'hr of 29 month!,. Both ceboi*! ol cejlclrcr> iheretoif, nan adverse .c < lea ilit .:o _ntd *ith epesurs Ulead m ear1" ih Lh'or -c.vever, in the most aisad\ai- U.ted chiiri-r i the '''ac-a-'ociatcd ir.ie'itctual deficits re 3.csenior rn.* c'h a _- ate frs'ninal .os; of IQ prints, b- * sis' c. !;1 ,r t.. l" el; s.-oua'cd with cln.ir;lly e vr, t ~>yv .notion le.c . IQ =co-e s ir the low 80s;. Blood lead coo rent ".no is ^ed^ctinp the degr'-e of lead e.\posur i art hicnej among disadvantaged chil dren in the I nited States." The approximate magni tude cf the dvtrtir.cnt in IQ associated with exposure to I'-iv-to-rtc'li.ta'-- Ie .1: af leac. donay carl', coild- hi od appear, in rr r.g' fror t 5 to 10 pei ttrt of the IQ sc ores deterrnneo e i' > i" elementary -chooi with the t/lSC-P."1 dc.','n ...i'1 suit ur c .o 10 trams > > comparable .} that produced by other environment:; 1 variables uzdnionall' - roTiizto as afiectinc chil- d" Ti': aevU'p I'm T-if iognj`ude ol the decrement as >o~i uec * ih e? it: aosure m.i" bt t mater because a umber of rcwcnalle* .on1 idered ir:3v share /an- anre v'iri-Jc'f1 n n h'-1 my the fruit.or 3 the ."rim e to the co ari'hi: rather thin to -end is likely lo cause one to u''d'/";ri:-c m trt' meg-bude of thf cffh-'t of Jes d 1 Thr L S. to 'er-jTien* has launched a major cam paign ii "o'i 'c. -'l-.hocc o'sire in lead 10 rirer. our *982 re-f'ir* on -he prevalence o' I'lesamd blood lecd' concent-s'" r : h uoi.rd Sr;. s. ihc concrn- ti: uons base detn .-seel. This dtcrease ieffects *he <dr- tual elimination of lead solder from the seams of food and beverage cans produced in the United States and marked decreases in the use of leaded gasoline. Al though these changes have helped the general popula tion, children continue to be exposed to lead. Leadbased paint and dust containing high levels of lead remain a health menace in an estimated 3.8 million U.S. homes housing children seven years of age or younger.11 The high cost of removing environmental sources of lead, however, has often been cited as a reason for delaying the removal of lead-based paint. The problem of excessive exposure to lead is cer tainly not unique to the United States, although this country has been in the forefront of attempts to reduce lerd exposure. International programs for chemical safety and the control of exposure to chemicals have lagged behind those of the United States. Clinical lead poisoning remains an internationally recognized problem. In coming decades, maximizing the intellectual and educational capacity of our population will be critical to cur success in dealing with many social and eco nomic challenges. The importance of safeguarding the in*elircrual and educational development of our chib drtn against preventable disease is difficult to overes timate Chile hood exposure to lead is preventable. The oara of Baghurst et al.1 emphasize the intellectual ccst 'ti no: preventing it. National Institute of Environmental Health Sciences Research Triangle Perk, n c 27709 Ka t h r y n R. Ma h a f f e y , Ph .D. Re f e r e n c e s 1. Baghurst PA, McMichael AJ, Wigg NR, et al. Environmental exposure to lead and children's intelligence at the age of seven years -- the Port Pine Cohort Study. N Engl J Med 1992;327:1279-84. 2. Ernhart CB. A critical review of low level-prenatal lead exposure in rite human. 2. Effects on the developing child. Reproductive Toxicology 1992; 6:21-40. !I 3. Bellinger DC, Stiles KM* Needleman HL. Low-level lead exposure, intelli gence and academic achievement: a long-term foUow-up study. Pediatrics (in press). rr .4. Yule W, Rutter M. Effect of lead on children^ behavior and cognitive performance: a critical review. ln:i Mahaffey KR, ed..Dietary and environ mental lead: human health effects. Amsterdam: Elsevier Press, 1985:211- 59, 'hi!.: : 5. Dietrich KN, Berger OG Sucwip PA, Hammond PB. The developmental consequences of low to moderate prenatal and postnatal lead exposure: intellectual anainnoent in the Cincinnati Lead Study Coboit following school entry, Neuxotoxicol Teratol (in piress). 6. Dietrich KN.i Lead exposure and central auditory processing in children. Teratology 1992;4S:522. abstract. 7. Dietrich KN, Berger OG, Succop PA. Lead exposure and the motor devel opmental status of urban 6 year-old children in the Qnemmwi prospective Study. Pediatrics (in press). 8. Mahaffey KR, Annest JL, Roberts it Murphy RS. National estnnaies of blood lead levels: United States, 1976--1980: association iwith seletned dem ographic and socioeconomic huctors. N Engl J Med 1982;307:573-9. 9. Hunt JV. Environmental risks in fetal and neonatal life as biological deter minants of infant intelligence, fa: Lewis M, ed. Origins of intelligence: infancy and early childhood. 2nji ed. New York: Plenum Press, 1983:255- 304. ' 1:1 i;: ' i0. Centers for Disease Control. Preventing lead poisoning in young children: a statement by the Centers'for Disease Control. Atlanta: Centers for Disease Control, 1991. iv. h it i' ll. Department of Housing and Urban Development. Comprehensive and work able plan for the abatement of lead-based paint in privately owned housing: repon t o Congress. Washington, D.C.: Government Printing Office, 1990: 3-18. 'i; I-- DUP040011272 Effect of Low-level Body Burdens of Lead on the Mental Development of Children: Limitations of Meta-analysis in a Review of Longitudinal Data STEPHEN B. THACKER Centers for Disease Control Atlanta, Georgia DANIEL A. HOFFMAN JAY SMITH KAREN STEINBERG Center for Environmental Health and Injury Control Atlanta, Georgia MATTHEW ZACK Centers for Disease Control Atlanta, Georgia ABSTRACT. The effect of low-level body burdens of lead on the intelligence of children, as measured by intelligence quotient (IQ), was assessed. We reviewed 35 reports from five longitudinal studies conducted in the United States and Australia. In each of these studies, infants were followed for 58 mo or less. The study populations consisted of low- and middle-socioeconomic-class infants who had low-level exposure to environmental lead. Blood-lead levels were measured in a standard fashion at various times, beginning in the prenatal period, and intelligence was first measured at 6 mo of age and was followed by subsequent assessments. Studies were assessed for quality by a review panel blinded to the identity of the investigators and their affiliations. Efforts were made to pool the data with meta-analytic techniques, but efforts were unsuccessful because the methods used to analyze and report data were inconsistent. Inconsistencies were as follows: (a) there were few instances in which IQ and hlood-iead levels were measured at comparable times in dif ferent studies; (b) incompatibilities existed among the studies, including differences in in dependent variables, data transformations, and statistical parameters reported; (c) results conflicted when measurement intervals were comparable (i.e., heterogeneity); (d) patterns of regression and correlation coefficients were inconsistent; and (e) data were insufficient to interconvert the parameters reported. Consequently, definitive conclusions regarding the effect of low-level body burdens of lead on IQ could not be determined from the longitudinal data. Examination of the weight of the evidence from this and other studies, however, suggests an adverse relationship of lead on the intelligence of children. LEAD is absorbed readily by humans and affects the central nervous system, peripheral nerves, red blood cells, liver, and bone.1 The absorption of high doses of 336 lead has adverse health effects on children and adults.1'3 Since the clinical description in 1943 of the effects of very high lead levels in children without encephalopa- Archives .of Environmental Health DUP040011273 N 29305.02 thy, the existence of more subtle effects of lower levels of lead has been hypothesized and studied/ In 1965* Patterson's observations of the distribution of high levels of lead within the population of the United States caused attention to be directed at determining a "safe" threshold for lead in humans.5 In 1985, the Centers for Disease Control fCDQ established a threshold of in tervention for children at 25 pg/dl (x 0.04826 mmol/I).6 Whether a threshold of effect for lead exists in children is a matter of current debate. In particular, the potential impact of low blood-lead levels (i.e., < 25 jUg/d!) on early neurodevelopment, is of increasing con cern. In a recent meta-analysis (I.e., systematic method used to compare results quantitatively across studies7), a statistically significant inverse correlation was found between exposure to low levels of lead in the teeth of children (PbT) and their intelligence quotient (IQ) levels.8 Controversy centers on defining the exposure levels that produce these effects, the duration of these effects, and the role of concomitant factors such as parental IQ, home environment, and other types of ex posures. Although iiterature is extensive regarding the rela tionship between asymptomatic exposure and neurodevelopmental effects, most published reports are either retrospective or are limited to research in which blood lead (PbB) and neurodevelopmental outcomes are measured at one time (i.e., cross-sectional studies). Longitudinal studies of children, beginning before birth (i.e., in utero), offer several advantages. The population at risk is defined at the outset, which enables in vestigators to estimate the magnitude of potential selec tion bias associated with children lost to follow-up. Pro spective, data collection can minimize recall and other types of information bias, and it allows investigators to measure temporal changes in outcome relative to prior levels of exposure. Finally, repeated measures of PbB enable investigators to analyze the effects of lead in a child at different times and to estimate cumulative ex posure. Considerations of cost and effort have limited the size and numbers of such studies, the effect of which has limited both the power of studies to detect smail differences and the generalizability of results to other populations. Materials and methods We reviewed 40 reports from five longitudinal studies in which the relationship between PbB and mental development in children was examined."5 These reports were identified in a search of the MED LINE database for the period 1966 through 1988, in a review of major literature summaries, and in bibliog raphies from identified articles. We included all studies for which data were published prior to September 1, 1989. We created a database that contained all rele vant quantitative statistical information from the five studies, which enabled us to evaluate the potential for a meta-analysis. An epidemiologist, a statistician, and a laboratorian-- all of whom were unfamiliar with the five longitudinal studies--made up a panel that assigned quality scores to the five studies. They read only the methods sections of the relevant publications for which ail study iden tifiers had been removed. The panel scored the studies independently on a scale from 0 to 100 in accordance with the criteria provided in Table 1. The panel met and discussed the strengths and weaknesses of each study and adjusted their individual scores. The final score for each study was the arithmetic mean of the score assigned by each reviewer. Studies were scored higher if (a) the source population was a defined, representative population; (b) exclusion criteria were defined; (c) completeness of follow-up was described; (d) examiners were blinded and irrterexaminer varia tion was addressed; (e) selection and detection bias were addressed; (f) appropriate covariates were includ ed; (g) laboratory quality control was adequate; and (h) statistical analyses addressed issues such as multiple comparisons, distributional assumptions, colinearity, bivariate screening, dose-response approach to data analysis, and outliers. After we reviewed these scores, one of us--who had read the publications in their en tirety but had not served on the panel--suggested that pertinent methodologic information given in sections other than the results sections of publications should be considered. The panel, still blinded to identifiers and outcomes, reviewed the added information and adjusted scores as appropriate. In this review, We addressed study design, data-coiiection methods, laboratory methods* and statistical analyses, and we placed particular emphasis on the elimination of bias and on confounding. The results of .these studies, a discussion of methodological concerns, and an attempt to reconcile apparent discrepancies in the data follow. Also presented is an examination of the strength of the evidence for a causal association be tween lead and decreased mental development. Final ly, the implications of our findings for current public policy are discussed. Results Analyses were limited to five longitudinal studies in Which pregnant women were identified, and serial PbB levels ir) their infants were measured for 2-5 y after birth (Table 2). These studies varied in size, i.e., from 249 births11 to 745 births in the most recently reported study/1 Three of the studies were conducted in pre dominantly middle-class and working-class white pop ulations! (Boston,9-16 Port Pirie,37-44 and Sydney/"9 re spectively); two were conducted in low-income popu lations in Cincinnati17-26 and Cleveland29-36; and one study included many blacks in Cincinnati.24 Typically, a maternal and/or cord-blood sample was obtained, and during many years, PbB samples were obtained at various intervals after birth. PbB levels in all these studies fended to be low (mean PbB, < 25 jig/dl). The investigators of the five studies examined the re lationship between PbB and mental development through the first 2 y of life with Bayiey's mental devel opment index (MDI). This index was designed to assess "sensory-perceptual acuities, discriminations, and the ability to respond to these: the early acquisition of 'ob- September/October 1992 [Vo!. 47 (No. 5)3 337 DUP040011274 Table 1.--Assessment of Data from Five Longitudinal Studies That Evaluated the Association between AsymptomaticChildhaod Lead Exposure and intelligence Boston Longitudinal studies Cincinnati Cleveland Port Pirie Sydney Prior hypotheses Power calculation Eligibility criteria Exclusion criteria Drop-out assessment Laboratory quality control 11 Y Y N N *N(Y) N YY Y Y VY Y hi YY Y Y YV Y Y Blinding Interviewers Statisticians Laboratorians YY UU uU Y U U Y u u Assessment of interexaminer variability iQ measure Standardized Quality control Y Y Y Y Y U N Y U N Y U Appropriate statistics Main effects Subgroups Multiple comparisons YY YY NN Y Y Y Y Y u Multivariate adjustment for confounding Age Set Parental IQ Socioeconomic status Caregiving Maternal smoking Y Y Y Y Y Y Y Y N Y Y N Y) Y YY YY YY YY YY Effect modification assessed YY Y Y Notes: 1 - implicit, Y - yes, N - no, U - unknown, and NA - not applicable, No power calculations were reported in the initial studies, but a later study included power calculations. i N uV Y Y Y U U Y Y Y Y Y N Y Y Y Y Y Y N ject constancy' and memory, learning, and problem solving ability; vocalizations and the beginnings of ver bal communication; and early evidence of the ability to form generalizations and classifications, which is the basis of abstract thinking."50*' The test evaluates devel opmental 'status and consists Of mental, motor, and be havioral ratings scales. The MD1 measures mental abil ity, whereas the psychomotor development index (PD!) measures gross- and fine-motor abilities. Investigators from the Boston, Port Pirie, and Sydney studies used the McCarthy scales to assess mental development in older children. The Cleveland study used the StanfordBinet intelligence scale for children 3 y of age and used the Wechsier preschool and primary scale of in telligence (WPPSi) for children 58 mo old. The Cincin nati investigators used the Kaufman assessment battery for children 4 y of age. Each study included data on various combinations of potential confounding varia bles, including socioeconomic status', caregiying at home, parental IQ (except the Cincinnati study), and material smoking, as weil as Other potentially Important variables such as age, sex, race, and obstetrical factors. Ddta frqm the entire population were used in three of the studies; the Boston investigators focused their anal ysis on the highest, lowest, and central PbB groups of the population; and the Sydney investigators examined the children at the highest and lowest PbB deciles of the population. All five studies appeared to have been carefully de signed and executed. In each study, the investigators described eligibility and exclusion criteria; analyzed to varying degrees data on persons who dropped out or who were removed from the study; reported careful quality-assessment and quality-control procedures in the laboratory; "blinded" both interviewers and per sons who administered the Intelligence tests; and used appropriate statistical methods for main effects, sub group analysis, and assessment of interactions (Table 2). Investigators from the Boston, Sydney, and Cincin nati studies reported analysis of Interobserver variabil ity in psychological tests. In Port Pirie, almost all the tests were conducted by the same person. Only the Sydney investigators, however, reported quality-cont rol procedures for the psychological tests. None of the investigators reported power calculations in the initial publication of their study methods, and none "blinded" the statisticians, labqratorians, or persons who evaluated caregiving at home. Only the Cleveland investigators expressed concern regarding statistical problems with multiple comparisons Of data, and only in that study's latest publications were power calcula tions reported.32,33 338 Archives of Environmental Health DUP040011275 Table2.--Results of Five Longitudinal Studies That Evaluated the Impact of Low-dose Human Exposure to Environmental Lead (United States and Australia, 1979-1988) Boston (1979-1983) Cincinnati (1979-1983) Oeveland (1971-1981) Port Pirie (1979-1988) Sydney . (1982-present) No. of subjects: In original sample* Birth 6 mo 12 mo 24 mo 36 mo 48 mo 58 mo Race (% White/Black) Socioeconomic status Child lead: Type of measure Exposure level (pg/dl) Pb, maternal (mean) Pb, cord (mean) Pb, postnatal (mean-range) ii'b, postnatal (peak-range) MDI change (points)* 6 mo 12 mo 24 mo IQ change (points) 36 mo 48 me SB mo 11 837 249 216 204 188 NA NA 170 85/15 Middle/high Capillary NA 6.6 6.2-7J 24,9-48.9 45.8 47.31 44.8 NA NA 41.8## 1.973 305 305 NA NA NA 258 NA 15/85. Mainly Venous 8.1 6.3 4.8-21.1 26.0-85.0 422.7# NA NA NA 4 NA 543 359 127 145 142 138 NA 234 65/35 Low venous 6.5 6.0 6.5-16.7 11.0-41.8 ill 4// ill ill NA ill 831 745 652 619 601 NA 438 NA 100/0 Middle/low 318 298 274 259 234 215 207 NA'100/0 Middle Capillary 9.5 8.3 9.3-24.3 14.7-67.0 Venous, capillary 9.1 8.1 10.1-16.4 5.0-79.0 47.1" 46.5"** 410.0** 410.1" 44.5** NA 4+t 4t+t 4m m 4+T NA Notes: MD1 - Mental Development Index of the Bayley developmental tests (measures obtained after adjustment for confounding); WPPSSI * Wechsier Preschool and Primary Scale of Intelligence; PbB * blood lead level; NA - not available; and SC study completed. 'Differences between the number of subjects in the original -sample and the number included for follow-up at birth varied in each cohort. In Boston, the birth cohort was a sample of eligible children available after exclusion. In Cincin nati, the difference between the initial sample and those followed at birth resulted solely from exclusion whereas that difference in the Cleveland study included both exclusions and some pregnant women lost to follow-up. In Port Pirie, repored numbers in the original cohort of pregnant women varied, but the difference between the Original Sample and the birth cohort presumably included both exclusions and those lost to follow-up. The Sydney cohort was apparently selected at birth, and the number excluded was not available in papers published. tAi! 6-mo blood samples were capillary; up to 24 mo, approximately halfthe samples were capillary; almost all subse quent samples were venous, *Related to maternal or cord PbB, adjusted for confounding. ^Adjusted difference in means of upper and lower umbilical cord PbB exposure deciles. ^Decrement across the range of prenatal PbB for males. Each log unit of prenatal PbB is associated With a covariateadjusted decrease of 5,7 MDI points. IIWhereas the coefficients for the association between PbB and MDI (Stanford-Binet at 3 y and verbal scaale of the Wechsier Preschool and Primary Scales of Intelligence at 58 mo) were negative, MDI changes were not quantitated, These decreases were statistically significant at 6 mo only, and none remained significant after adjustment for con founding. Difference in means of upper and lower exposure quartiles of average or antenatal PbB based on McCarthy Scales of Children's Abilities. tfReports of correlations varied in different reportsrM'<4 adjusted analyses at 48 mo were not published for maternal or cord PbB. **15 mo, rather than 12 mo. At 48 mo, there was no statistically significant association between prenatal or neonatal PbB and mental develop ment (Kaufmann), although postnatal PbB at 2 y was significantly associated with a decrease in mental development #*AT 57 mo, the inverse association between cord PbB and IQ (McCarthy) was not statistically significant. the review panel agreed in their assignments of high est and lowest quality scores: in all cases, the Sydney study scored the highest and the Cincinnati .study scored the lowest (Table 3). Even though the ranks of the three remaining studies varied among the review- Sejitember/October 1992 [Vdl. 47 (No. 5)] ers, quality scores did not vary more than 15 points among the three studies for any one of the three re viewers. Information provided in additional to the methods sections of the papers did not affect the panel's relative rankings of the studies. 339 DUP040011276 Table 3.--Quality Scorn, Assessed by Panel of three "Blinded" Reviewers, of Five Longitudinal Studies Th.st Evaluated tlie Input of Low-dose, Asymptomatic Human Exposure to Environmental Lead (United States and Australia, 1979-1988) Study Boston Cincinnati Oevsiand Port Pirie Sydney Reviewer 1r;' X 1 "3' 61tf 41, 5;1' 618 8.3 SO." 35* 65 6.5 70 55 45 60 54 65 Mean ,55.0 41.0 59j3 59.7 72.7 Median 55 43 60 60 70 Table 4.---Reported Parameters in Blood-Lead Studies ' Port Boston Gncinnati Cleveland Pine Sydney Regression coefficient Change in standardized scores Correlation ves* Yes Yes# Vest Yes* Yes Yes// 'Coefficient!! given, where (1) independent variable represents three lead concentration groups and dependent variable is MDL adjusted for covariates; (2) independent variable is change in Pb concentration, dependent variable Is change in standardized scores for MDl and McCarthy tests. Ages at men tal tests were 6, 1.2,18, 24, and 57 mo. tCoeffirients are adjusted for confoundere; and dependent variable is MDL Ages at mental test were 3, 6, and .24 mo. The Kaufman Assessment Batten/ for Children was administered at 48 mo. Coefficients are adjusted for confoundere, dependent variables are MDl, and McCarthy measures of mental develop ment. Ages at which mental tests were conducted were 24 and 48 mo. Change in standardized scores for MDl (24 mb) and McCar thy tests (57 mo), Correlation between MDl (6-24 mo) cord PbB in three categories. //Correlation between MDl (6-36 mo) and maternal and cord PbB. More than 250 reported statistical parameter values were obtained from the five studies. These parameters were examined for feasibility of meta-analysis so that the magnitude of the effect of increased blood lead levels on intelligence could he estimated from all studies. The statistical measures reported most fre quently in each of the five studies included regression coefficients, correlations, and changes in standardized scores (Tables 4-6). Although the Boston, Cincinnati, arid Port Pirie investigators reported regression coeffi cients, PbB was collapsed into three categories in the Boston study, which made estimates of regression coef ficients unsuitable for potential pooling with coeffi cients from continuous PbB values. Also, the Boston study used changes in Z-scores as a dependent varia- ble--a measure incompatible With that used in other studies. The Boston, Cincinnati, and Port Pirie in vestigators reported the regression on mental develop ment of PbB at the same age only once, i.e., 24 mo. The regression coefficient was positive in the Cincinnati study only. When mental ability was measured at 24 mo and was related to PbB measured at 6 mo, the Bos ton and Port Pirie studies obtained regression coeffi cients of opposite sign. Although the Boston, Cleve land, and Sydney studies reported correlations, the Boston study again computed its value, based on col lapsed PbB categories, which prevented pooling ofjte correlation coefficient with those of the other two studies. On six occasions, the Cleveland and Sydney in vestigators correlated mental development or IQ with PbB at the same time. In three of the six occasions, quite different correlations were obtained. A quantitative estimate of the common effect of PbB on mental development from these five studies was not made because there were different statistical param eter;, PbB was treated differently in the analyse;, and conditions under which resuits were obtained iage at mental development test ana blood lead measure ments) diffe-ed. In the few instances of similar resorting cond'tions, the results were not comparable; however, in three instance;, similar negative coefficients were obtained for the Cleveland and Sydnev studies. An inwc-se relationship between prenatal and post natal e^pofure to lead and the MD! in unadjusted anal yses was shown ir, all five studies. Even after adjustment was mane for confounding, no statistically significant relationshio was seen between PbB and the MD in the Cle* eland and S-dnev studies (Table 2i. Statsticaily significant differences remained only :n the studies f'om Boston, Cincinnati (white infants;, and Port Pirie. The Cinciffat1 ipvesrigotors found no evidence of an adverse erect oi prcnaiai or neonate! lead exposure by the time childrer :eac.ned 2 > o: age, except for chil dren from the poorest families. Blood-lead levels at 1, 2, 3 antMvctapa vere associated inveisely with men- ra1 de ciofment t 4 bu' th association was not sta tistically significant when adjustment was made for conroundmg In the Eoston and Port P'rie studies; how ever, coTPction for confounding increased the magni tude '! i*v n.f'-r 't'jricr shm n :rrl,uis : Port Pirie children based on the cumulative measure cf postnatal exjjocuir. measure of children at ^ v age. showed a 9.6-ooin. decrement in mental developmen. between the 'owes1 arid highest uuarhi'w. The Port P rie in- v *ir ifr mr'-'Lit'-i th, .* -_n atcr controlling for comoundmg, children w5th an average postnatal PbB cr; 1.3 rg/dl du-irg the first 4 v of rtfs -iq s mental de- v_rp->" c-i i _ ml loir rnildren who had an average PbB concentration or 70 4 pg/di during the sam*1 pence. The Boston fobow-uD at 57 mo deter mined th?" decreased lead exposure, as measured by Pc? iro soc'odemograomc. charactenstics previously ass^oaf*c with increased mental development, ap- p*>a ed 1 cr t r'-n-p^c f~r <--3, mfi e defki*. attributed to crenata) lead exposure.'7 This same study riemonsrraic a a stall it " ally s'strvcani asso'iat'or b-'tween 340 Archives of Environmental Heafih DUP040011277 Table 5.--Reported Regression Coefficients* for Age at Mental Development Test (MDI) versus Age at Blood Lead Concentration Measurement in five Longitudinal Studies That Evaluated the Impact of Low-dose, Asymptomatic Human Exposure to Environmental Lead (United States and Australia, 1979-1388) Age lead Age (mo) MDI or IQ measured measured (mo) Study 3 "6........ ...... is' " 'W"' ` '48' '' ..............57" ' ,0 Bostonf -2.89 (0.92) Cincinnati -0.60 -0.66 -.63 Port Pirie 0.10 3 Cincinnati -0.23 -0,48 0.24 6 Boston Port Pine 0.23(1.29) -0.16 (p - .09) 12 Boston Cincinnati -1.43 (1.25) .03 15 Port Pirie -0.03 18 Boston -1.62 (1,39) 24 Boston Cincinnati Port Pirie -2.95 (1.42) ai3 -0.05 -.02 36 Cincinnati -.08 48 Cincinnati -,io .57 Boston -Z28 (1,88) Regression coefficients were not available in published reports from studies in Cleveland .and Sydney. These data were not available beyond 24 mo in any of the studies. Standard errors or p values are given within parentheses. ICategorized lead values were used to determine coefficients for the Boston study; continuous values were used in other studies. # Table 6.--Reported Correlation Coefficients* for Age at Mental Development Test (MDI or IQ) versus Age at Blood Lead Concentration Measurement (United States and Australia, 1979-1988) Age lead measured (mo) Study Age (mo) at which MDI or IQ measured 6 12 24. 36 .58 0 Cleveland -0.14 -0.24 -0.15 -0.21 Sydney -0.16 0,15 0.05 Bostonf -p.il 6 Cleveland -0.09 0.08 -0.03 -0.04 -0.06 Port Pirie -0.12 12 Sydney 0.11 0,09 18 Sydney -0.05 24 Cleveland Sydney Port Pirie -0.24 -0.02 -0.18 -0.31 -0.38 -038 36 Cleveland Sydney -0.29 -0.32 -0.32 Correlation coefficients were not available'in published reports from the Cincinnati study. tCategorized lead values were used to determine coefficients tor the Boston study; continuous values were used in other studies. PbB at 24 mo and decreased mentai development (Mc Carthy scale) at 57 mo. The negative association found in the Boston study between cord PbB and IQ at 57 mo was small (1,8 points) and was not statistically signifi cant. In none of the studies, however, was an effect on the PDi demonstrated. September/October 1992 [Vpi. 47 (No. 5)] Data reported from the Sydney study differed from those in the other longitudinal studies in that, apart from unadjusted data for 6-mo-pids, PbB was not con sistently related inversely to either MDI or IQ. Results at 12 mo and 24 mo . differed in direction in various reports from this study, but PbB was generally directly 341 DUP040011278" related to MDI or IQ through 48 mo in both adjusted and unadjusted analyses. Methodological issues Several methodological concerns affect the inter pretation of studies regarding the healt h'effects of lowdose exposure to environmental lead..' Each of the in vestigations reported here must be understood in light of these concerns. Measures of tody-lead burden. The transient nature of PbB level-; lsce^sitates repeated PbB testing to deter mine measures of PbB. Alternative methods for the de velopment ot summary measures from serial PbB meas urements have rot been validated as measures of expo sure to lead. :-nd PbT has been regarded as a more ac curate measure 01 cumulative exposure/body burden^ Moreover the longitudinal study design requires that laboratory methods be stable over time. 'Serial PbB mersuremeni'.. however, allow evaluation of the ef fects 01 varyii i exposure to environmental lead. Blood lead measurements that exceed 10 pg/dl have been ou-mtiiied pimsel/, although at levels below this the precision dec eases substantially. The relative accuracy of PhB, brsed cn -;apillary {"finger-stick") samples that are pcssib'y (ontaminated with external surface lead, ris; rued: tc x accounted for in the analysis and inter pretation of a=t-'. The Port Pirie investigators examined `.hi. potential source of bias and found no effect. The poisihte aifr-r+ of surface contamination in Boston, Cie/eland and Sydney cannot be assessed, although, in mar- ncun'*es the Sydney investigators obtained duplicate venous samples. Measures of mental development. The Bayley, VV.SC, Slarfcrd-binet, and McCarthy scales are reliable, -eil standardized scales of mental develops mtrt or intell-genr e. Their use facilitates comparison of resulTS fro- the several studies. Whereas the best av='laMe meas-it of infant development is the MD! on thr Bayley =::!c, t does not correlate well with the Starford-Einei Fo<-*n L-M.S1 Similarly, although concur rent validity :f the McCarthy scale is satisfactory, large difierrnces have teen reported between it and either Stf-mord-binet Form L-M or the WISC-R IQ measures. The Kaufman assessment battery for children is standar dized adeocatriy and has high reliability. The correla te with ::!?iif:>rc Binet Form .L-M and the WISC-Riare mc-aerate, bu: ro verbal comprehension component is co "stained in he mental-processing section of the test. O'er??, this test is not a good measure of intellectual ab-litics. Sample size. None of the studies reported to date are sufficiently laige to evaluate the impact of environmen tal lead on these outcomes. This lack of power is also a function of; the low levels of exposure measured and the minimal changes in expected neuropsychological outcomes. Bias. Selection bias could easily account for apparent differences among children with different PbB levels or could 'mask irue differences between groups. In pro spective studies, e.g., those that examine PbB over time, children excluded or lost to follow-up must be Studied carefully to ascertain the impact of their ab sence on analysis and interpretation (e.g., the exclusion of children with low birth weight who tend to have ele vated levels of PbB). If the possibility of transgenerational effects is ignored, true relationships between ex posure and adverse outcomes may be obscured. Moth ers (or fathers), whose IQs were lowered by exposure to lead, may have parented children with postnatal lead exposure. Control for parental IQ in this setting may overcontrol for lead exposure and possibly ob scure a true relationship between PbB and lowered mental development. All five longitudinal studies are at risk of selection bias because of loss of subjects froth the original samples (Table 23. The effects of such bias are difficult to assess,, however, because the two Aus tralian studies with the fewest children lost, to follow-up showed opposite effects of lead exposure on mental development. Information bfas occurs when measurement of ex posure 'i.p. PbB) or outcome (i.e., mental develop ment and behavior) is incorrect, increased accuracy and precision >n measurement, standardization of study instruments, and "blinding" of interviewers to study hypotheses and of laboratorians and statisticians lo the identities of cases and controls will help to reduce this bias All five studies included laboratory qua'ity control activities, and investigators also made etorts to blind interviewers to the study hypotheses in si! the studies (Table 1). None of the studies, however, LTnded the laboratorians or the statisticians to the suid- hypotheses, which potentially intrbduced biases 1 hat could obscure an association between exposure cnc outcome. Confounding. In these studies, one cannot adjust analytically for" confounding of the effect exposure to low |p` ei< of lead has on mental development without eor Gartering risk of overadjustment or underadjustment. it ir necessary, therefore, to review the efforts of control for confounding and to assess the overall effect c-f confounding on the, postulated relationship in the context of all the evidence for causation, as discussed below. Although evidence is consistent concerning a crude inverse relationship between lead levels in the body and decreased mental development, skeptics argue that this relationship results primarily from social deprivation, poverty, and parental, intelligence, espe cially if PbB levels are < 25 pg/dl. All of the investiga tors who, authored the five studies reviewed in this report controlled for confounding in their analyses, but the cofactors and methods used varied (Table 1). In three studies, control of these variables diminished the association between PbB and IQ, and, in Cleveland and 'Sydney, control df the variables accounted for most of the association. In Boston and Port Pirie, how ever, correction for confounding strengthened the in verse association between PbB levels and intelligence. Although analysts must .account for such confound ing. they( must also be aware of potential interactions among variables and be careful not to overadjust and obscure an exposure-outcome relationship by control ling for a variable either intermediate in the causal 342: Archivesof Environmental >-*o=>nh DUP040011279 pathway of proxy for an exposure.52 In this situation, a child's social class might convey some information about lead contamination in the environment and thus about a child's PbB level. The association between so cial class and mental development includes two com ponents, one of which is dependent on lead and the other not. Simulation demonstrated how the associa tion between lead and mental development can be ob scured (or exaggerated) by the way in which social class is handled in the analysis.53 Multiple comparisons of data will, or the other hand, tend to produce statis tical associations that have little or no biological plausibility. Comparability. Differences exist in populations (e.g., nat'onpliiy, socioeconomic status,, and race), and measures of exposure (e.g., PbT versus PbB) and ana lytic approaches also differ; therefore, studies may not be compsraole. Longitudinal studies, on the other hand, aie remarkably similar in design and implementaton. and, ax sept for the Sydney study, the findings arc consistent. Discussion The understanding of the effects on humans of ex posure to environmental lead is an important public health issue. Whereas childhood lead poisoning and occupational iead intoxication in adults are weli- recognized clinical entities that result from high body burdens of lead, the effects that lower levels of lead have on humans continue to be controversial. In this review, we, have focused on the effects that low levels of lead (< 25 pg/dl) may have on mental development. At the outset of the overview, we intended to com bine data from the longitudinal studies in a statistical meta-analysis; we were unable to pool the data for sev eral reasons. First, the statistics used by the study authors (i'.d., correlations, regression coefficients, and changes in standardized scores) were different and could not be interconverted because of a lack of sup porting data. For example, correlations could not be converted to equivalent regression coefficients because standard deviations of the independent arid dependent variables Vjere not provided. Grouping values of an in dependent'variable into categories and regressing the independent variables for these calegories1'3 will yield different Egression coefficients from those based on continuous] measurements of the variable of interest. Second, values were treated differently in the vari ous studies. Bbod-leadi values were categorized prior to statistical regression analysis in the Boston study, and either lo^-iransformed or original values weie used in other studies. Also, the ages at which mental tests were giVen and jages at which lead levels were measured were different for the studies. In the few instances where comparable measurement intervals were report ed, the sigrjs of the coefficients differed. In addition, the stated goii^ of the 'studies were not identical. For exam ple, the 'tlleveiand study was designed originally as a calse-comparison study, and it investigated the impact ofimaternb!;' lifestyle factors (e.g., alcoholism) on the oc currence of birth defects.29 Finally, except for the September/October 1992 [Vot. 47 (No. 5)1 Sydney study, any reported absence of PbB effect on mental development was apparent only when analyses were adjusted for confounding. For these reasons, we did not pooi or stratify even the unadjusted data. The authors of the Boston, Cincinnati, and Port Pirie studies concluded that increased exposure to iead at levels below the current threshold for intervention (25 pg/dl) slowed mental development in children and lowered IQ. But the Cleveland investigators concluded that their data did not show such an effect. Careful ex amination of their data, however, showed a consistent adverse effect of lead exposure on MD1 and IQ, al though the differences measured were not significant statistically. The Cleveland findings, therefore, are con sistent with studies that have shown a statistically signif icant toxic effect of iead on mental development/PerhaDS there was insufficient statistical power to detect; such an effect. The results of the Sydney study, however, are more difficul* to reconcile with the findings of the other iongi:udir.al studies. With the exception of the unad- justeo results on the effect of maternal PbB on the MDt at 6 mo and results reported inconsistently at 12 mo and 2-1 mo in separate publications, the findings of the Sydney investigators are consistent either with no effect or v'lth a direct (but not statistically significant) effect of F'-B on MO' at all rimes of exposure up until 48 mo. v.o'eo -er, this finding was apparent in both adjusted anc! unadjusted analyses. Some methodologic con cerns set this stud' aoart from the others, e.g., incon sistent!', reported results noted above, inconsistency of lepc-rteo PbB measurements within subgroups of the population, potential selection bias that resulted from the nojpital-based nature of the populations, and the limited scope o1' reported analyses. In additior, the re porter! effect of environmental factors on mental devel opment measures suggests that social1 fatten; in this m'odie-dass copulation may have played a role in compensating for adverse effects of exposure to iead or the fetus and child. At the same time, it is not dear vvn, mis study, mdged by blinded reviewers to be of lh highest duality methodologically, showed results di'Vrenf from 'he other studies and other sources of (".fareere. Final judgment must await further publica tions from these investigators; Whc**as investigators from the five longitudinal studies exchanged ideas and results during a period of several vears and agreed on certain aspects of study de sign arm* implementation, apparently no such consen sus on ne analyses and reporting of results occurred. Our Tpbil'tv to petorm a jmeta-anaiysis should sen sitize investigators to the need for publication of com mon!'' reponed statistical results or, at a minimum, suf ficient information tr allow ''interconversion of param- (450,. 1 ne alternative that the rreta-anaiyst seek the prima-v data from ft^ir|rjHftfiej5qw^ n&. some times impractical, and may introduce serious bias from inror'isrecf efforts in the r^triewf sr*t'. trie ir.fdequacv of old files.54es||ifc>lishmpnt of a-- ip iern.-t'di'.-l registry of stijjdles,- analoeruf to registries proposed rat randomized con- vulleu Enaisr4'^ wppld held to address concerns of 343 .DUP040011280. comparability and would facilitate meta-analysis. Re cent studies in Yugoslavia, Mexico, and Scotland should be included in such a registry/'*47 The among-person variability should be reduced if the statistical modeling for longitudinal studies takes advantage of the longitudinal nature of the results for each person.58 This variability could account for much of the observed variation. The slope of the mental de velopment index versus elapsed time should be obtain ed for each person. The null hypothesis dictates that this slope should average zero if testing is consistent. These individual slopes should then be considered in a model versus their respective mean PbB values and potential confounding variables. The second slope would indicate the low-level effects of leac. If this ap proach were ine.s. the power of the analysis to detect even low-leyei effects might be increased in longitudinal studies. Assessment of the evidence for causation Currently, efforts are being made to eliminate envi ronmental lead as a risk factor for disease and disability; therefore, it ,s important to place these five longitudinal studies in the context of available scientific knowledge. There are issues described in this report that limit generalizations from these studies; therefore, the evi dence for a causa! association between low-dose expo sure to lead and impaired mental development will be discussed. The evidence will be reviewed, using the fol lowing features proposed first by Hill, to evaluate ob servational data for the nature of a relationship be tween exposure and outcome49 strength of associa tion, consistency, specificity, temporal relationship, bi ological gradient, biological plausibility, coherence of! the evidence, experimental evidence, and reasoning by analogy. Strength of association. Strength of association would be best measured by the change in mental de velopment that results from a change in PbB. Unfortu nately this1 .information is available from the Boston study only. Moreover, the relative risk of having a low IQ (e.g., < SO on the W15C verbal) has not been calcu lated in any of thb studies, and the published reports do not provide sufficient date on which to base such a cal culation. Consistency, investigators who authored four of the five longitudinal studies from two countries demon strated an inverse relationship between PbB arid MDI, although .adjustment for confounding reduced or re moved the effect in two of the studies. The data from th= PbT studies are consistent with an inverse associa tion Between PbT and mental development.8 Temporal relationship. Longitudinal studies provide the best evidence for a temporal relationship because e?wure can be most dearly demonstrated to orecede oitcome Biological gradient. A dose-related effect of lead on mantel development was seen in the Cleveland and G:inrnzt: si relies, in which PbB was used as a continu01 s variable. and in the German and English PbT studies,8 in the Boston and Port Pirie longitudinal studies, PbB was found to be correlated inversely with IQ. Similar results were noted for the Philadelphia and Boston PbT studies.6041 The well-recognized clinical syndrome of encephalopathy that is associated with elevated PbB is also consistent with a biological gradient.1 Biological plausibility. The evidence for biological plausibility has been documented independently of this review. The known neurologic effects of lead in hu mans on nerve conduction velocity and EEC patterns the uptake of lead in human tissue---primarily in bone and teeth, at very low levels, and the experimental evidence in animals, are all consistent with a rieurpbiological effect of low PbB levels on intelligence and behavior// Experimental evidence. Although intentionally ex posing children to environmental lead is unethical, studies that examine the effects of removing or reduc ing iead levels can be regarded, for the purpose of this discussion, as an experiment in lead removal, in which the child is his/her own control. The few rather small reoorts or: PbB reduct'on do not clearly show improved neuropsychological performance.4 This observation does not contradict a causal role for lead in inhibiting rreite1 development, because it could indicate a longlading or perhaps irreversible effect of lead. As noted above animal experiments support a causal role for lead in the inhibition of mental development. Lend exp isure ir. rats leads to a reduction of axoriai pa way development in the hippocampus, which affects neur onal development and entu'lh' eriets a reduc !d in tellectual ootenria1 u lead a'sc compt-tes with and dis places calcium in the neural membrane, winch disrupts choimerpc wnrtioning.63 Tins efieri was most ap parent in the brair, especially in the fiipnocampLs and de/e'e-pirn cerebellum. These bade mechar'.-ns of lead action at fhe neuronal level >v\e reduced long term memory in experimental arimJ' A reciuct'on in lengterm memory, resulting from iead exposure, could c-vcUm dferences in mental development and ed'icalicral atteirment tests lhai reflect learning and '^rentier over long pe-iods. Coherence of the evidence. Adverse effects of d-lchoocl lead exposure on mental development and neha'-loi, seen with different types of research in a vrirfe arietv of srtingc, are cons^c-nt with a causal roie for leac. The studies with exoerimental animals nleri shore ere also consistent F.eafom'r.s; bv analogy. Lead is the Drototypic heavymete.' trxirarr in environmental health, and few substances can tv. compared with it. Although other mete's ruch as mercury have known effects on the human nervous system, sfuriie* of low-dose chronic expdlure:aire ndt availa&l^'foriHtem. Conclusion Although data from the longitudinal studies remain inconclusive, current evidence from ail sources is con sistent with a small (2-10 points), but biologically signifi cant, .effect on IQ of children exposed to low levels of iead {< 25 pg/dl). This significance stems from the dramatic effect that a shift in the mean in a population .344 Archives of EnvironmentaDUP040011281 distribution has on the tails of that distribution. For ex ample, a decrease oi: 4 points in mean population in telligence increases the number of children with IQs of < 80 by threefold and decrease the percentage of children with IQ scores of > 125 from 5 to 0. This seemingiy small effect on individuals can, therefore, have very dramatic effects on a population.64 We should continue to examine critically future data collected from the longitudinal studies reviewed in this article, as y u pII as other studies that are planned or are t-nd-:-1" V'V- 'Jp.-eflheiess, the evidence for the ex istence of adverse haalth effects directly related to lead exposure >s more consistent than some skeptics have arquea. art: this evioerce has led 'erne dinicam to r^co-nrrcrd treatment V asvmDtoma ic children v-ho rave iw lf<ek of PbB.rr hem a public health perspei*t-e, the sirurnent o" betr. prima-v anr secondary: frev^ntior of had c<r.'`cure and its sequelae is compelline bec<i:se lead has not been shown to benefit humans and "Torts ic reduce PbB levels in children by r-mrviny, If'id frm urhnri dwellings can be s< ccessfui. Lead poi^-ninr is ai entne'y prevariable cmirormenlA disease, ihe c'imina'.on of which should he one of fip naricr. oublic heati eoals The nature and '/tent rf public r>( ill f-otrsn .Kat aim to reduce lead in *he epvirormen ` noun refect tne continuing review or t 'e nesr a* i !?b e sci^n >iic intoTna ion, the crfs to _ o p ;* i* rliem.fvf irt irv"ntions jnti `he social ara e.nnomifius" & fa lure Ir .nten/^n'- apprnpria'cl*' ***** * * Submitted for publication August 16, 1991; revised; accepted for publication 'December6,: 1991. Requests for reprints should be sent to: Stephen B. Thacker, M.D., Epidemiology Program Office, Centers for Disease Control, Mailstop COS, Atlanta, GA 30333. ***** References 1. U.S. Department of Health and Human Services. The nature and extent of lead poisoning in children in the United States: a report to Congress. Agency for Toxic Substances and Disease Registry. U.5- Public Health Service; July 1988. 2. Yule W, Rutter M. Effect of lead on Children's behavior and cognitive performance: a critical review. In: K R Mehaffey, Ed, Dietary and environmental lead: human health effects. Amster dam: Elsevier, 1985; pp. .211-59. 3. Lansdown R. Lead, intelligence, attainment and behavior. In: Lansdown R, Yule W. Lead toxicity: history and environmental impact. Baltimore, MD: Johns : Hopkins Press, 1986; pp. 235-70. 4. Starr RH, Jr. Current research of the development of ecology of lead exposure during childhood. Environ Res 1985; 38:197-210. 5. Patterson CC. Contaminated and natural lead environments of man. Arch Environ Health 1965; 11:34460. 6. Centers for Disease Control. Preventing lead poisoning in young children: a statement by the Centers for Disease Control, January 1985. Atlanta, GA: U.5. Department of Health and Human Ser vices, No. '99-2230. 7. Thacker S. Meta-analysis. A quantitative approach to research in tegration. JAMA 1988; 259:1685-89. 8. Needieman HL, Gatsonis CA. Low-level lead exposure and the IQ of children. JAMA 1990; 263:673-78, 9. Bellinger DC, Needieman Hi, Leviton A, Watemaux C, Rabinowitz MB, Nichols ML. Early sensory-motor development and prenatal exposure to lead. Neurobehav Toxicol Tetroi 1984; 6-387-402. 10. Bellinger D, Leviton A, Needieman HL, Watemaux C, Rabinowitz M. Low-ievel lead exposure and infant development in the Brat year. Neurobehav Toxicol Terato! 1986; 8:151-61. 11. Beiiinger D, Leviton A, Rabinowitz M, Needieman H, Watemaux C. Correlates of low-level lead exposure in urban children at 2 years of age. Pediatrics 1986; 77:826-33. 12. Bellinger D, Leviton A, Watemaux C, Needieman H, Rabinowitz M. Longitudinal analyses of prenatal and postnatal lead exposure and early cognitive development. N Engl J Med 1987; 316: 1037-43. 13. Bellinger D, Sloman J, Leviton A, Watemaux C, Needieman H, Rabinowitz M. Low-level lead exposure and child developmentassessment at age 5 of a cohort followed from birth, in: Undberg SE, Hutchinson TC (Eds). Heavy metals in the environment, Vo! 1. Edinburgh; CEP Consultants, 1987; pp 49-53. 14. Beiiinger D, Leviton A, Watemaux, Needieman H, Rabinowitz M. Low-level lead exposure, social class, and infant develop ment. Neurotoxicol Terato! 1988; 10:497-503. 15. Bellinger D, Leviton A, Sloman j. Antecedents and correlates of improved cognitive performance in children exposed in utero to low levels of lead. Presented at the Conference Advance:; in Lead Research;: Implications for Environmental Research, National in stitute of Environmental Health Sciences. Research Triangle Park, NC, January 1SS9, pp. 1-26. 16. Bellinger D, Sloman J, Leviton A, Rabinowitz M, Needieman HL, Watemaux C. Low-level lead exposure and children's cognitive function in the preschool years. Pediatrics 1991; 87:219-27. 17. Bomschein RL, Hammond PB, Dietrich KN, et ai. The Cincinnati prospective study of low-level lead exposure and Its effects on child development: protocol and status report. Environ Res 1985; 38:4-18. 1$. Dietrich KN, Krafft KM, Pearson DT, Eornschein RL, Hammond PB. Succop PA. Postnatal lead exposure and early'sensorimotor development. Environ Res 1985; 38:130-36. T9. Padich RA, Dietrich KN, Pearson DT. Attention, activity level, and leadlexposure at 18 months. Environ Res 1985; 38:137-43. 20. Dietrich KN, Krafft KM, BierM, Succop PA, Berger O, Bomschein R. Early effects of fetal lead exposure: neurobehavioral findings at 6 months. Im J Biosocial Res 1986; 8:151-68. 21. Krafft KM, Eornschein RL, Dietrich KN, Peace B Naraire 5. The ule of social and behavioraitl ftacicvttowrias iirni*i exposure to environmen- saiiiiiis metals dinburgh: CEP Consultants, 1987; pp. 162-64. 22. Die*rich KN Krafft KM, Bier M, et al. Early effects of fetal lead ex- ivvure. de'-elopmertal 'InMtindberg E, 'i:tct..nsnn rC. Heavy metals Edih- Jurg'i CtP Consultants, 1987 pp. 63-65. 23. Dietrich KN, Krant KM, Shuxla R, Padich RA, Succop PA. The r eurobchavioral effects of prenatal and early postnatal lead ex- nesurr in rber.. -.E, Eberts CG (Eds). Trends i<| ergonomics/hu- ni.-i, ac'or,. '1. Amsterdam- El^ifir iSpepce 'PiiBlish^fS B.V., 985 00 477-83 ; 24. Di'-trirh KN. Krafft KM, Boms-c1h--e:i~n RL, et ai. Low-ievel fetal lead expowm offer* on nteeuurcbeha. viorai development in early infen- o*. PH atrics 1987: 80.721-30. 25. Dietrich KN, Kreff- KM Shuxla R, Bomschein RC, Succop PA. f *tr- ,"hay o*?' effort, of early:|i|e|id.|i^fo|d|j; (n: ^Chreieder SR. 'ii-ic substances and mental retardation. American Association ni Mert-1 Deficier-y Wjjt inston, DC, 1987, pp. 71-95. 26. Dietnrn KtJ, Su cop PA, Rcrnscheir. RL Hammond PB, Krafft KM. De-'elopnentai irffecKj; gsiS|i:j^astih^a! 'iJtead x- eOSir cn nfdiff rr e'epment Environ Health Prcsr-e-l (in prejs^, 27. Br.rnsrfffin Rl , Cfark CS, Dietri&j|3ff; et ai,;fect6rs influencing 'iigtiidina! ue-rio^rnenf-,' j^< ;}|ftrfilei in children. Pn-enteri at nets in lead Research: Implications for Health Noti-rji Institute oi rnvimnm|^j|! Health: .SCierices, Research "range Para. NC, lanuarv 9-11| 1^89,, 28. Di-'iicn KH, Succop p a , Berger i, jjjamnjbhd PS, Bqrnschein PL Lead n>oo;,.jrr ind tne :.^llppriiert of urban rj-esrhoc -hildmr the C nnnnjmljij^; Istcppy' ledhort at age 4 "-ir Ncurntoxicr.l Terafoi T^'l, J3.203-11 29. ir ten Cs, Vo AW Lmn BL|i left i|al^ 4i)ddbol-feiated birth detects: 'syndromai anomalies, intrauterine powth retardatiorti September/October 1992 [Vol. 47 (No. 5)1 345 ,DUP04001 1282k and neonatal behavioral assessment. Alcohol Clin Exp Res 1985; 9:447-53. 30. Emhart CB, Wolf AW, Sokol R|, Brittenkam CM, Erhard P. Fetal lead exposure: antenatal factors. Environ Res 1985; 38:54-66. 31. Emhart CB, Woif AW, Kennard Ml, et ai Intrauterine exposure to tow levels of lead: the status of the neonate. Arch Environ Health 1986; 41:287-91. 32. Emhart CB, Morrow-TIueak M. Low level lead exposure in the prenatal and early preschool years as related to intelligence, just prior to school entry. In: Lindberg 5E, Hutchinson TC {Eds}. Heavy metals in the environment. Vol 1. Edinburgh: CEP Con sultants, 1937; pp. 150-53. 33. Emhart 03, Mcrrow-Lucak M, Matter MR, Wolf AW. Low level lead exposure in the prenatal and early preschool periods: early preschool development Neurotoxicoi Teratol 1987; 9:259-70. 24. Ernhar* f3. Morrov'-TIurak M, Wolf AW. Low level lead ex posure and intelligence in the preschool years. Sci Total Environ 1588; 7". :e5 3-59. T 35. Emhart C 3 Mn-'ow-Tiucak M, Wolf AW, 5uper D, Drotar D. Iw lo <" lead eviyrsi rt: in the prenatal and early preschool centals: -ree licence pi nr to school entry. Neurotoxicoi Teratol 1989. 1':' 61 -70 36. Emhart 'IP, Lrnda B. Wolt AW. Subdinical tear! level and cevelopr.-.a-ilal ceric':: reanah'ses of data. J Learning Oisabil i98:, !fi-i:;.7I 57. i:r,p:.n (I Mc.Aichae1 /-, Robertson E, Wigg N. The Port Pirie Stunv: a preset c*ive ruoy of pregnancy outcome arid early cm-c` ;rni'th and de-.elopmert in a lead-exposed corrrvjr ivwi*- rcol and status report: Environ:j ReSi 1985; 38:19-23.'' 38. Vimpar,1 (IV, w.gg NR, Robertson AJ, etial. The Port Pirie Cohort -r'lf. bi'jcd tead concentration land ichiidhoods deveiopment ass-.s i-.rt 'r-.'.ent"J et Lead;Environmental Health: The Sur- r-'ii Issue- Du1 e University. Durham, NC; May 1985. 39. naghur* f C lonela R. Wigg N, McMichaei A, :Robeitson E, Vim- pan- G. ioine enarae'ensnes arducorrelates of blood .leadt in early ch'.dho.:: ore:i nlnsry "suits from the (Port Piriesitudy. Environ re i ' j O. 40. rAr` Vc'rui / J. fir-gnurs* PA. Pobertsori.EF, Vimparii GV, Wigg IS. F-<* Am 1 rtcr* St.dy blood lead concentiations :in r r| i -j J Acsi 1985; 143:499-503. I 41. fcayjr 1 hi cert,or. Er, McMichaeii'A, Vintpanii GV, Wigg tip., ~rit* rs h* The Port Pine1 Cohort a Study: lead: effects on f ic-na-vir p `f'jmj and early childhood development. Neuiotu. co 'i93'; 6:395-402. 42. ."IF 7u--ari C'-. McMichaei AJ, et ai. Port Pirie Cohort JiuJ.-: di.lc'i'-H tI,od tend and neuropsychclogical develop ment at ajf. .wo years. I Epidemiol Comm Health 1988; 42:213-9. 43. M'-mi-Iv .1 Baghurv P*-, Wigg NE, Vimpam CV, Robertson rf, !?-,r,"r.: 'J. Jrn "'.riv Cohort Study, environmental exposure to teat! ar J cmldren : aLiiliries at the agei of four years.i N Engl ) tic,-! I980; 21 j '65 75 44. 7nn,rri Q ?; 'hurst P, McMichaei Alt: et al, Port Pirie Cohort '"ru J-. :l i- r :-f-; of cumulative lead exposure: on pregnancynout- coire an: chidhood development during the first four iyears. ! iv.than ret n Lead' Research: Implications fon En- ircrmnit:. -te:l!h f onterence. National Institute of fin- "imrn.- it; i -laalth Sciences, Research Tnangle Park, NC; ,anu:-ry I. 1989. 45. vVG, i|C.'3.-ter C.J, Cooney G, Bratel JR. A study of blood lead1 lev's Is: inlelfertual and motor development of ar group of Sydney urban children. Report to Foundation 41, Surry Hills, New South Wales, Australia, July 28, 1986. 46. McBride WG> Cooney GC, Bell A. Blood lead levels in Sydney ur ban children. In: Lindberg SE, Hutchinson TC. Heavy metals in the environment Vol JL Edinburgh: CEP Consultants, 1987; pp. 153-55, i 47. McBride WG, Carter CJ, Bratel Jr, Cooney G, Bel! A. The Sydney study of health effects of lead in urban children. In Lindberg SE, Hutchinson TC. Heavy metals in the environment. Voi. 1. Edin burgh: CEP Consultants, 1987; pp. T-5. 48. Cooney GH, Bell A, McBride W, Carter C Neurobehavioral con sequences of prenatal low level exposures to lead. Neurotoxicoi Teratol 1989; 11 -.95-104. 49. Cooney GH, Bell A, McBride W, Carter C. Low level exposures to lead and neurobehavioral development the Sydney load study. Trace Subst Environ Health 1989; 22:357-67. 50. Bayley N. Bayley Scales of Infant Development New York, NY: Psychological Corpora'ion. 1369. 51. Sat! ter !M. Assessment of intelligence and intent development: with 'pecialized measures in Assessment of children In San Diego; Jerome Saittei, Publisher, 1988 (Third edition), Chapter 12- pp. 294-326. 52. Pott-mrn K Modern epidemiology. 'Boston, MA: Little Brown and Co, Inc, 1q26, pc. 8G-9J. 53. Bellinger D, Le> non *, Woternaux C. Lead, IQ and Social Class. Inr J Epicem-ol 1989 18 '3085. 54. Chalmers TC, Levin H, Sacks HS, et al. Meta-analysis of clinical fial* a- a soen iric discipime !. Control of bias and comparison with large cvopoi. tivs. trick Stct Med 1987; 6:315-25, 55. Chalmers I, H"'henngton J, Nedwick M, et al. The Oxford anabases oi p'-rinata' trial-: developing a register of published repirts a' c-.ntroiUd trait. Osniro-led Clin Trials 1986; 7:306-24. 56. Pothenbnrt'l, Casino-Ortiz S, et al. Neurobehavioral deficits arter Ir.v le/el lead exposure in neonates. The Mexico Ci- t/ pi or stiidv Ncu i'Mmc o I Terarol 1989: 11:85-93. 57. Moore MP, 3i ,1-nul l/v'R Goiobergi A. Ai prospective study of Me -es i'ts or t.rwf n om ronmentaiitlead exposure in children m Glasgow, in- v nn M. Grant LD. Sore A (Eds). Lead exposure ,vio cm.A r evr-l(,ur'"ir an .nemationai assessment. Boston, mA Kkv.e- Acnrienic Pjoushcrs, 1989. 58. Can Im SP, Srrrth 5: Cholesterol-oased personal risk assessment m rur'rvlpri i "tat Med TS89; 3:295-309. 59. hili At. * 'her' `r^iboov of medical statistics. London: Hodder ana Stoughton 1CJ77, po 188-94. . 60. Needlemen ri1. Davidson I, Sewell EM, Shapiro IM. Subclinical =d'i-nsir ir rhifadelphia school children. N Engl J Med 1974; 250:245-i8. i 61. NiiifT-,n `-iL Gunnoe C, Leviton A, et al. Deficits in iBi'cho'ogic and classroom performance of children with tne >;ad levels. N Engl | Med 1979; 300:689-95. 62. ,\ifarw> DP LeBouiliter |C, Petit TL. Hippocampal mossy fiber oath-v.-w develooment in'normal and postnatally lead-exposed rat-: -aer Neuro' 1982; 75:308-19. i ; 63. Pound: JG. Effect sof ilead intoxication on calcium homeostasis and c'te.um-meri ated cell funcion: a review. Neurotoxicoi .'C-ix j . s ri 64. Netdlerrap He, Leviton A, Bellinger D. Lead-associated intellec tual deficit fLetter). JAMA 1982; 306:367. 65. Shannon M, Grace A, Graef JW. Use of penicillamine in children m-it sen j ! tead burdens (Letter). N Engl I Med 1989; 321:979-80. 346 Archives of EnvjrohmeiSH^PdS.?.11283