Vol. 324 No. 6 CORRESPONDENCE 415

hood and adolescence and the subsequent risk of lung cancer mightbe influenced by various potential confounding factors: the numberof persons in the household, /3-carotene concentration, and parentaloccupation. We cannot directly evaluate confounding due to the sizeof the household and parental occupation, because the requisiteinformation was not collected. However, we doubt that either is astrong confounder. Our findings are not altered materially when weadjust for the available sociodemographic variables that are prob-ably related to the size of the household - e.g., the odds ratio forexposure to smoke for 25 or more smoker-years during childhoodand adolescence shifts from 2.07 to 1.92 (95 percent confidenceinterval, 1.03 to 3.56) after adjustment for education, and to 2.49(95 percent confidence interval, 1.27 to 4.86) after adjustment forincome. Therefore, it seems unlikely that household size is an im-portant enough independent predictor of lung cancer to account forour findings. Parental occupation has been found to be associatedwith certain childhood cancers, but we know of no data indicating acausal effect of parental occupation on the risk of lung cancer inadult offspring.

Regarding ,8-carotene as a potential confounding factor, adjust-ment for a crude index of dietary sources of /3-carotene yields anestimate of 2.21 (95 percent confidence interval, 1.17 to 4.19) for theeffect of exposure for 25 or more smoker-years during childhood andadolescence. Kalandidi et al. have recently shown that their ob-served association between lung cancer and exposure to a spousewho smokes was likewise not attributable to case-control differ-ences in diet.'

Lee makes additional points beyond the issue of confounding.Unfortunately, his assembled data regarding the association be-tween exposure to environmental tobacco smoke in childhood andthe risk of lung cancer have little bearing on our finding of a twofoldelevation in risk for persons exposed for 25 or more smoker-years.Since less than 25 percent of the exposed controls in our study had alevel of exposure this high, an elevated risk for this subgroup wouldbe considerably obscured in any analysis (such as all those tabulat-ed by Lee) that lumped together all exposed persons.We disagree with Lee that we should have incorporated an ad-

justment for multiple comparisons. In line with current epidemio-logic practice and recent recommendations, 2-4 we simply provided95 percent confidence intervals for each of the odds ratios in thepaper in order to give the reader quantitative information on theprecision of our estimates. More definitive evaluation of whetherhigh levels of exposure to environmental tobacco smoke duringchildhood and adolescence causes lung cancer requires additionalstudies in which exposed persons are subclassified according to lev-els of exposure.

Borelli summarizes some of the negative findings from earlieranalyses of our study by the late Dr. Luis Varela. We note inaddition that Dr. Varela reported evidence of a positive associationbetween household exposure and the risk of lung cancer. We alsonote that although we know of no specific mechanism to account forheightened susceptibility to tobacco smoke in childhood, the possi-bility of such a susceptibility warrants further study, especiallysince children are far less equipped than adults to protect them-selves from any harmful effects of indoor air pollution.

DWIGHT T. JANERICH, D.D.S., M.P.H.New Haven, CT 06510 Yale University School of Medicine

W. DOUGLAS THOMPSON, PH.D.Portland, ME 04103 University of Southern Maine

PETER GREENWALD, M.D., DR.P.H.Bethesda, MD 20814 National Cancer Institute

SHERRY CHOROST, M.S.CATHY Tucci, B.S.

Albany, NY 12237 New York State Department of Health

MUHAMMAD B. ZAMAN, M.D.MYRON R. MELAMED, M.D.

New York, NY 10021 Memorial Sloan-Kettering Institute

MAUREEN KIELY, R.N.Albany, NY 12208 Albany Medical College

MARTIN F. McKNEALLY, M.D.Toronto, ON M56 2C4, Canada Toronto General Hospital

1. Kalandidi A, Katsouyanni K, Voropoulou N, et al. Passive smoking and dietin the etiology of cancer among non-smokers. Cancer Causes Control 1990;1:15-21.

2. Rothman KJ. No adjustments are needed for multiple comparisons. Epidemi-ology 1990; 1:43-6.

3. Thompson WD. Statistical criteria in the interpretation of epidemiologic data.Am J Public Health 1987; 77:191-4. [Erratum, Am J Public Health 1987;77:515.]

4. Walker AM. Reporting the results of epidemiologic studies. Am J PublicHealth 1986; 76:556-8.

THE LONG-TERM EFFECTS OF EXPOSURE TO LOWDOSES OF LEAD IN CHILDHOOD

To the Editor: The follow-up study by Needleman et al. (Jan. 1 1,1990, issue)' purports to demonstrate enduring cognitive andneurobehavioral effects of relatively low levels of exposure to lead inchildren previously studied in 1975 through 1978.2 What it maydemonstrate with equal if not greater likelihood are the enduringeffects of social, developmental, and medical risk factors associatedwith cognitive and neurobehavioral deficits. The issue of confound-ing factors in this complex subject has been highlighted previous-ly,3-5 yet Needleman et al. aver that "the persistent toxicity oflead [at relatively low levels] was seen to result in significant andserious impairment of academic success" and that their data are"nonspurious" - i.e., that the association observed is not due toconfounding.'While the authors state that "the correlation between socioeco-

nomic status and dentin lead levels in this sample was not great,"they also cite evidence that "children from families in lower socio-economic groups are more vulnerable to the effects of lead thanchildren from more favored economic backgrounds." It is thus quitepossible that other vulnerability factors may be causal, with low-level exposure to lead an associated factor that is not etiologic. Inthat regard, their argument against possible selection bias in thefollow-up sample, which tended to have fewer subjects of lowersocioeconomic status, is not entirely reassuring. Furthermore, it iscurious that they no longer exclude subjects with certain medicalrisk factors because the factors reportedly were no longer related tooutcome scores. In certain other studies, medical risk factors havebeen associated with low performance and with low lead absorp-tion.5'6 Such an association is a confounding effect that if present inthe authors' study, could add statistical weight to a spurious rela-tion between low levels of lead and performance deficits.The epidemiologic approach to the problem of possible effects of

low-level exposure to lead is indeed complex and fraught with meth-odologic and statistical snares; the authors are to be congratulatedon their persistence. It may be clarifying, however, to examine theissue by studying groups of children selected on the basis of havingan attention-deficit or hyperactivity disorder, which Needleman etal. appear to believe is "a consistent effect of lead exposure."2 Davidet al.,6 for example, studied hyperactive children in the inner cityand found that those with serious perinatal complications consid-ered "highly probable" causes of hyperactivity had mean blood leadlcvels essentially the same as those of controls. The number of caseswith a highly probable cause was not large, however. Gittelman andEskenazi,7 in a study of 103 nondisadvantaged hyperactive chil-dren, failed to find a significant elevation of lead levels in relation tothose of normal children, though their levels were higher than thoseof their own siblings. There was no association between the severityof hyperactivity and lead levels, and no definite relation betweenlead levels and cognitive-test performances. These reports illustratethe difficulty of separating socioeconomic from medical risk factorsin. studies of the effects of low levels of lead in children.

Although it is possible that a low-level lead burden in childhoodcontributes to neurobehavioral and cognitive deficits, it is not clearthat it has consistent or uniform effects or that it can act in isolationfrom other factors. While the environmental threat of lead is real,from a scientific point of view Needleman et al. might have givenfurther qualification to the cause-and-effect conclusions they drew.

MICHAEL I. GOOD, M.D.Boston, MA 02131 Massachusetts Mental Health Center

1. Needleman HL, Schell A, Bellinger D, Leviton A, Allred EN. The long-termeffects of exposure to low doses of lead in childhood: an 1 I-year follow-upreport. N Engl J Med 1990; 322:83-8.

The New England Journal of Medicine Downloaded from nejm.org on July 9, 2015. For personal use only. No other uses without permission.

Copyright © 1991 Massachusetts Medical Society. All rights reserved.

THE NEW ENGLAND JOURNAL OF MEDICINE

2. Needleman HL, Gunnoe C, Leviton A, et al. Deficits in psychologic andclassroom performance of children with elevated dentine lead levels. N Engl JMed 1979; 300:689-95.

3. Lead levels and children's psychologic perfornance. N Engl J Med 1979;301:161-3.

4. Smith M. Recent work on low level lead exposure and its impact on behavior,intelligence and learning: a review. J Am Acad Child Psychiatry 1985; 24:24-32.

5. Lyngbye T, Hansen ON, Grandjean P. Neurological deficits in children:medical risk factors and lead exposure. Neurotoxicol Teratol 1988; 10:531-7.

6. David 0, Clark J, Voeller K. Lead and hyperactivity. Lancet 1972; 2:900-3.7. Gittelman R, Eskenazi B. Lead and hyperactivity revisited: an investigation

of nondisadvantaged children. Arch Gen Psychiatry 1983; 40:827-33.

To the Editor: Needleman et al. report strikingly large effects oflow lead levels on several outcomes in late adolescence. For exam-ple, an estimated 7.4-fold increase in the odds of school failure wasattributed to childhood lead dentin levels above 20 ppm. Such mas-sive effects contrast sharply with the results of other studies relatinglow lead levels to earlier developmental outcomes."3 The authorsargue that the estimated effects represent causal relations becausetheir analysis controlled for 10 sociodemographic covariates. Thisconclusion of causality may be premature, however, because thecovariate set did not include measures of the quality of child care(i.e., parental responsiveness, involvement with the child, and pro-vision of books and suitable playthings), a primary confounder inprevious studies of developmental lead effects. Thus, the reportedeffects of lead may be partly due to a spurious association inducedby variations in the caretaking environment.

Indexes of child-care quality, such as the Home Observation forthe Measurement of the Environment (Caldwell BM, Bradley R:unpublished data) and the Childhood Level of Living Scale,4 haverepeatedly been found to be strongly related to lead levels in poorand working-class children. 3'5'6 The quality of child care is alsostrongly associated with developmental outcome,7 including schoolperformance through adolescence.8 These confounding effects areconceptually distinct from and only partly accounted for empiricallyby sociodemographic variables such as maternal IQ and parentaleducation,9 which were included as covariates by Needleman et al.The fact that none of the reported effects of lead were attenuated bythe inclusion of their covariates, as is usually the case in observa-tional studies of low lead levels, indicates that confounders such aschild care may not have been fully controlled for.

To the Editor: Lead in food, particularly in processed food, isa major source of overall lead intake in the entire population.Lead contaminates food through the deposition of airborne lead,the use of lead-bearing water and equipment in processing, the useof lead-based solders in canning, and contact with serving vesselsglazed with lead-containing compounds. In 1986, the Environmen-tal Protection Agency (EPA) attributed most of the contribution toatmospheric deposition (50 percent) and solder (42 percent).' Wehave analyzed Food and Drug Administration data from the1982-1983 and 1984-1985 Total Diet Studies (based on an analysisof market baskets) to ascertain whether solder and other leadsources - for example, gasoline additives - remain importantsources of lead in the U.S. food supply. Since the early 1980s,voluntary food-industry programs to replace lead solder wereassumed to be sufficient to protect the food supply. However, about20 percent of domestically filled food cans are still soldered withlead.2

In the United States, the impetus to ban lead additives in motorvehicle fuels has been delayed on the grounds that leaded fuelsare important for farm vehicles. Plants grown near highwaysand other sources of lead are significantly contaminated (largelyby surface deposition). Despite EPA efforts to phase out leadedgasoline, first- and second-quarter figures for 1988 show that leadedgasoline accounts for nearly 20 percent of the gas sold in theUnited States.3 The declining use of lead solder and gasoline addi-tives is reflected in reductions of lead contamination in food overthe years 1982 through 1985. However, the FDA data indicatethat these two sources of lead in food are still problems. Ofthe 234 categories in the 1982-1983 and 1984-1985 market-basketsurveys, we found seven pairs for which the only difference waswhether or not the foods were canned. Lead levels in the cannedsamples (Table 1) were on average more than 29 times higher thanthose in fresh-frozen samples of the same food in 1982-1983;by 1984-1985 this value had declined to 14. All the pairs fell with-in the "produce" classification of the U.S. Department of Agricul-ture; however, we found similar ratios of the average lead levelsin canned to those in fresh-frozen food in other USDA categories(Table 2).4On the basis of the USDA's average daily intake figures for

each of the categories shown in Table 2, a person consumingcanned products from each category whenever the option existedwould have a daily lead intake of 107 and 40 ,ug in 1982-1983and 1984-1985, respectively. For a person who avoided canned

Cleveland, OH 44109

CLAIRE B. ERNHART, PH.D.TOM GREENE, PH.D.

THOMAS A. BOYD, PsY.D.Case Western Reserve University

1. McMichael AJ, Baghurst PA, Wigg NR, Vimpani GV, Robertson EF,Roberts RJ. Port Pirie Cohort Study: environmental exposure to lead and chil-dren's abilities at the age of four years. N Engl J Med 1988; 319:468-75.

2. Fergusson DM, Fergusson JE, Horwood LU, Kinzett NG. A longitudinalstudy of dentine lead levels, intelligence, school performance and behaviour1H. Dentine lead and cognitive ability. J Child Psychol Psychiatry 1988;29:793-809.

3. Emhart CB, Morrow-Tlucak M, Wolf AW, Super D, Drotar D. Low levellead exposure in the prenatal and early preschool periods: intelligence prior toschool entry. Neurotoxicol Teratol 1989; 1 1:161-70.

4. Polansky NA, Borgman RD, De Saix C. Roots of futility. San Francisco:Jossey-Bass, 1972.

5. Dietrich KN, Krafft KM, Pearson DT, et al. Contribution of social anddevelopmental factors to lead exposure during the first year of life. Pediatrics1985; 75:1114-9.

6. Hunt TJ, Hepner R, Seaton KW. Childhood lead poisoning and inadequatechild care. Am J Dis Child 1982; 136:53842.

7. Bradley RH, Caldwell BM, Rock SL, et al. Home environment and cognitivedevelopment in the first 3 years of life: a collaborative study involving sixsites and three ethnic groups in North America. Dev Psychol 1989; 25:217-35.

8. Hess RD, Holloway SD. Family and school as educational institutions. In:Parke RD, ed. The family. Chicago: University of Chicago Press, 1984:179-222.

9. Schroeder SR, Hawk B. Psycho-social factors, lead exposure, and IQ. In:Schroeder SR, ed. Toxic substances and mental retardation: neurobehavioraltoxicology and teratology. Washington, D.C.: American Association of Men-tal Deficiency, 1987.

Table 1. Lead Levels in Canned and Fresh-Frozen Food.

FOOD MEAN LEAD LEVEL1982-1983 1984-1985

RATio OF LEAD INCANNED TO LEAD INFmSs-FtOZEN POOD

1982-1983 1984-1985

pgl8

Snap green beanCanedFresh fozen

CornCannedFresh frozen, boiled

PeachesCand in heavy syrupRaw

PlumCanned in heavy syrupRaw

Green peasCannedFrozen, boied

SpinachCmed.Fresh frz

TomatoesCannedRaw

Average

0.1075 0.03370.6175 0.0076

0.1100 0.03500.0050 0.0065

0.2225 0.14000.0275 0.0052

0.16870.0162

0.15000.0137

0.65750.0662

1.19120.0087

o.o8250.0087

0.02000.0170

0.18640.0437

0.22970.0050

6.14 4.43

22.00 5.38

8.09 26.92

10.41 9.48

10.95 1.18

9.93 4.27

136.92 45.94

29.21 13.94

416 Feb. 7, 1991

The New England Journal of Medicine Downloaded from nejm.org on July 9, 2015. For personal use only. No other uses without permission.

Copyright © 1991 Massachusetts Medical Society. All rights reserved.

Vol. 324 No. 6 CORRESPONDENCE

Table 2. Average Lead Levels in Canned and Fresh-Frozen Food.*

MEAN LEAD LEVEL

1982- 1984- percent1983 1985 change

iA,,,

USDADAILYINTAKE

glday

MEAN DAILYLEAD INTAKE

MEAN DAILY LEAD INTAKEBy FOOD TYwF

FRESH-CANNED FOOD FROZEN FOD

1982- 1984- 1982- 1984- 1982- 1984-1983 1985 1983 1985 1983 1985

,Aglday

RATIO OF INTAKE OFCANNED TO INTAKEOFFRESH-FROZEN FOOD

1982- 1984-1983 1985

8glday

Dairy productsFresh cow's milk 0.0067 0.0055Other 0.0259 0.0169

Eggs 0.0119 0.0024Meats

Beef and veal 0.0298 0.0134Pork 0.0387 0.0250Poultry 0.0115 0.0161Other 0.0278 0.0294

FishFin fishShellfish

ProducetLeafy vegetablesExposed produceProwcted produceOther

GrainBreadsCerealsOther

BeveragesTap waterWater-based drinksSoupsOther

MiscellaneousTOtal

-17.5-34.6-80.2

-54.9-35.539.95.7

0.0635 0.0626 -1.40.0287 0.0525 82.9

0.1293 0.03800.1063 0.03400.0337 0.01620.1506 0.0691

-70.6-68.0-51.9-54.2

0.0266 0.0177 -33.50.0177 0.0085 -51.90.0178 0.0400 124.7

253.555.126.9

1.71.40.3

87.6 2.628.2 1.131.3 0.425.1 0.7

1.40.90.1

1.20.70.50.7

NA4.5NA

NA3.7NANA

NA2.0NA

NA2.3NANA

1.71.30.3

2.60.70.40.7

1.40.90.1

1.20.70.50.7

5.0

5.5

14.7 0.9 0.9 2,3 2.6 0.2 0.9 10.2 26.02.6 0.1 0.1 NA NA 0.1 0.1 - -

39.2 5.186.0 9.1150.4 5.17.0 1.1

147.3 3.929.9 0.522.9 0.4

1.52.92.40.5

2.60.30.9

23.329.320.21.1

NANANA

3.88.87.10.5

NANANA

1.01.61.9NA

3.90.50.4

0.0025 0.0016 -36.0 662.5 1.7 1.1 NA NA 1.70.0025 0.0013 -50.0 457.1 1.1 0.6 NA NA l.10.0416 0.0238 -42.8 45.9 1.9 1.1 1.9 1.1 NA0.0256 0.0123 -52.0 269.9 6.9 3.3 6.2 6.7 2.40.0244 0.0150 -38.5 34.6 0.8 0.5 NA NA 0.8

- - -55.7 2477.7 46.9 24.2 106.7 44.8 26.3

1.52.92.4NA

2.60.30.9

1.10.6NA3.30.5

24.2

*USDA denotes U.S. Deportment of Agriculture, and NA not availabl.tProtecsedproduceconsistsoffoodcrps whoseedible portionsaressn_uly covered: catns, beets, tnips, parsnips, citmrsfrits, coin, legumes, melons, onions, and potatoes. Leafy vegetables

incudes all vegetables in which dhe leaf is consumed: cabbage, cauliflower, broeccoli, ceety, lettuce, and spinach. Exposed produce includes all nonleafy fruits and vegetables whoseedible poatonsare natually exposed: apples, pears, berries, cucumber, squah, grapes, peaches, apricots, plums, paunes, string beans, pea pods, and tomatoes.

foods whenever possible, daily dietary lead intake would havedecreased fivefold (from 107 to 26 ug) in 1982-1983 and to lessthan threefold (40 to 17.4 Mug) in 1984-1985. This illustratesthe importance of continuing efforts to reduce the use of lead solderin cans.4

The average lead levels in the produce categories also revealthe effect of airborne lead, which is overwhelmingly derivedfrom leaded gasoline. In 1982-1983, the overall average level,including canned and fresh-frozen samples, in leafy vegetables(0.1293 Mug per gram) and exposed produce (0.1063 Mug per gram)was, respectively, 4.3 and 3.5 times higher than the average inprotected produce (0.0301 Mg per gram). Protected produce consistsof food crops whose edible portions are naturally covered.4 The1984-1985 data show a decrease to 2.4 in the ratio of lead in leafyproduce to that in protected produce and to 2.2 in the ratio of leadin exposed produce to that in protected produce.4 Lead levels inexposed and leafy produce decreased more rapidly than levels inprotected produce, reflecting the reductions of gasoline-derived leadin the atmosphere.The levels of lead found in some foods are sufficient to result in

intake considered excessive on the basis of the latest research, whichindicates that blood lead levels should be kept below 10 Mig perdeciliter.` The slope of the model equation of Sherlock et al. relat-ing an increase in the blood lead level (PbB, in micrograms perdeciliter) to dietary intake (PbD, in micrograms per day) is 0.092 Mgper deciliter per microgram per day.2 The 1982-1983 average dailyintake of 46.9 Mg would have resulted in an elevation in blood leadof 4.3 Mug per deciliter. By 1984-1985, this had been reduced to an

elevation of 3.5 Mg per deciliter. Despite the advances in the pastfew years, to protect the public health against unacceptably high

exposure to lead, it is appropriate to recognize the continuing needto control lead levels in food processing and gasoline.

Baltimore, MD 21205MARION R. SILLS

Johns Hopkins Medical School

KATHRYN R. MAHAFFEY, PH.D.National Institute of Environmental

Research Triangle Park, NC 27709 Health Sciences

Washington, DC 20036ELLEN K. SILBERGELD, PH.D.Environmental Defense Fund

I. Environmental Protection Agency, Office of Research and Development,Office of Health and Environmental Assessment, Environmental Criteria andAssessment Office. Air quality criteria for lead, 1986:7-52.

2. Department of Health and Human Services, Public Health Service, Agencyfor Toxic Substances and Disease Registry. The nature and extent of leadpoisoning in children in the United States: a report to Congress, 1988:1-30,IX-17, Table IX-5.

3. Environmental Protection Agency, Office of Mobile Sources. Data from leadadditive reports required to be filed by refiners and importers of gasoline.

4. Environmental Protection Agency, Office of Radiation Programs. An Esti-mation of the Daily Food Intake Based on Data from the 1977-78 USDANationwide Food Consumption Survey, 1984: Tables 1 and 3.

5. Needleman HL. The persistent threat of lead: a singular opportunity. Am JPublic Health 1989; 79:643-5.

6. Davis JM, Svendsgaard DJ. Lead and child development. Nature 1987;329:297-300.

7. NIEHS conference on "Advances in Lead Research" (January 1989). EnvironHealth Perspect (in press).

8. Sherlock J, Smart G, Forbes GI, et al. Assessment of lead intakes and dose-response for a population in Ayr exposed to a plumbosolvent water supply.HumToxicol 1982; 1:115-22.

USDA CATEGORY

417

3.3

7.7

23.1 5.418.3 12.611.3 5.5

6.9 1.4

2.6 3.5

The New England Journal of Medicine Downloaded from nejm.org on July 9, 2015. For personal use only. No other uses without permission.

Copyright © 1991 Massachusetts Medical Society. All rights reserved.

418 THE NEW ENGLAND JOURNAL OF MEDICINE Feb. 7, 1991

The above letters were referred to Dr. Needleman, who offers thefollowing reply:

To the Editor: Many of the questions raised by the correspondentswould have been answered by a careful reading of our paper and theliterature they cited. Good misses the central points of both Davidet al.' and Gittelman and Eskenazi.2 David et al. compared bloodlead levels and urinary lead excretion after provocative chelation inchildren classified as "non-hyperactive," "hyperactive/probablecause" (e.g., severe prenatal or postnatal risk of central nervoussystem trauma), "hyperactive/possible cause" (e.g., prenatal ill-nesses of lesser risk), or "hyperactive/no identifiable cause." Chil-dren in the "hyperactive/probable cause" category did not differfrom the nonhyperactive children, but hyperactive children in the"possible cause" or "no cause" category had both higher lead levelsand provoked urinary lead excretion. David et al. inferred that insubjects whose hyperactivity had no other cause, lead was associat-ed with hyperactivity.'

Gittelman and Eskenazi2 found no difference between learning-disabled and hyperactive subjects, but reported that hyperactivesubjects had significantly higher lead levels than their normal sib-lings. We and the same Danish investigators that Good cites3 havereported that lead exposure is associated with learning difficulties.This disqualifies learning-disabled subjects as suitable controls.Siblings, who have a similar genetic endowment and rearing histo-ry, clearly qualify as controls. Siblings have lower lead burdens thanhyperactive subjects.Although Ernhart et al. cite our original 1979 Journal paper4 and

criticize us for not controlling for parental care of the child, theyhave repeatedly overlooked our statement in that article that weadministered a 58-item parent-attitude scale that evaluated, amongother factors, the parents' academic expectations for the child, thenumber of books in the home, and the frequency of reading, playingboard games, and attending outside activities with the child. Thesefactors did not differ at all between exposure groups. This finding,presented in a separate table (Table 6) in the paper, has againescaped notice or mention by Ernhart et al.

If early rearing did not affect early outcome, it is unlikely it wouldhave an effect on academic success 11 years later. For an effect of thesize of the one we reported for school failure (odds ratio, 7.4) tobe confounded sufficiently by social factors to make it not signifi-cant, social factors would have to have both a strong influence onoutcome and a strong correlation with lead levels.5 A strong con-vergence of findings of almost all modern studies on the neuro-toxicity of lead at a low level of exposure in childhood has beendemonstrated.6

HERBERT L. NEEDLEMAN, M.D.University of Pittsburgh

Pittsburgh, PA 15213 School of Medicine

1. David OJ, Hoffman SP, Sverd J, Clark JO. Lead and hyperactivity: leadlevels among hyperactive children. J Abnorm Child Psychol 1977; 5:405-16.

2. Gittelman R, Eskenazi B. Lead and hyperactivity revisited: an investigationof nondisadvantaged children. Arch Gen Psychiatry 1983; 40:827-33.

3. Lyngbye T, Hansen ON, Trillingsgaard A, Beese I, Grandjean P. Leamingdisabilities in children: significance of low level lead-exposure and confound-ing factors. Acta Paediatr Scand 1990; 79:352-60.

4. Needleman HL, Gunnoe C, Leviton A, et al. Deficits in psychologic andclassroom performance of children with elevated dentine lead levels. N Engl JMed 1979; 300:689-95.

5. Schlesselman JJ. Assessing effects of confounding variables. Am J Epide-miol 1978; 108:3-8.

6. Needleman HL, Gatsonis CA. Low-level lead exposure and the IQ of chil-dren: a meta-analysis of modem studies. JAMA 1990; 263:673-8.

CORTICOSTEROIDS IN CHILDREN WITH CORROSIVEINJURY OF THE ESOPHAGUS

To the Editor: In the study by Anderson et al. (Sept. 6 issue)* Inotice that the esophagoscopy performed in all the children wasdone with rigid esophagoscopes. In this day of fiberoptic esopha-

*Anderson KD, Rouse TM, Randolph JG. A controlled trial of corticosteroidsin children with corrosive injury of the esophagus. N Engl J Med 1990; 323:637-40.

goscopy I am puzzled about why the study was conducted withthe rigid esophagoscope, and I also wonder about the quality of thevisualization obtained with a rigid esophagoscope and whether theuse of a fiberoptic flexible esophagoscope would have changedthe visualization criteria and affected the amount of injury detected.

K.T. SHAH, M.D.Houston, TX 77081 6550 Mapleridge

To the Editor: In their meticulous study, Anderson and colleaguesfound that the incidence of strictures after the swallowing of corro-sive agents was not affected by treatment with steroids started up to24 hours after injury.The mice of Spain et al.' to which they refer had little fibroblastic

reaction when given cortisone starting 24 hours before wounding,but showed a normal response when it was started at 48 hours. Inacute spinal-cord injury,2 steroids were effective only when givenwithin eight hours.

It would be interesting to know whether the children who weretreated early did better than those treated late.

Fordingbridge,Hampshire SP6 2EJ, T.H. HUGHEs-DAVIES, F.R.C.P.United Kingdom Breamore Marsh

1. Spain DM, Molomut N, Haber A. The effect of cortisone on the fonnation ofgranulation tissue in mice. Am J Pathol 1950; 26:710-1. abstract.

2. Bracken MB, Shepard MJ, Collins WF, et al. A randomized, controlled trialof methylprednisolone or naloxone in the treatment of acute spinal-cord in-jury: results of the Second National Acute Spinal Cord Injury Study. N Engl JMed 1990; 322:1405-11.

To the Editor: The study by Anderson et al. was primarily a studyof accidental ingestion of caustic substances in children. A largenumber of such children take a lick or taste' and are not found tohave esophageal burns, as was the case in 54 percent of the patientsof Anderson et al. The authors' conclusions may not apply to delib-erate ingestions, especially in adolescents and adults, who usuallyconsume larger quantities of caustic material.

Unlike the patients in other studies, all those with esophagealburns in this study had oral or pharyngeal burns. Other studiesreport that only up to 53 percent of patients with oral burns aresubsequently found to have esophageal burns.' The authors need toclarify the extent to which the pharynx (anterior and posterior) wasexamined.Of the 71 patients without esophageal burns, 57 had oral burns.

Of the 60 with esophageal burns, all had oral burns (in contrastwith other published studies). In this study, the positive predictivevalue of oral burns in predicting esophageal burns was 51 percent,whereas the negative predictive value was 100 percent.

These points should lead one to conclude that in accidental child-hood ingestions of caustic substances, the absence of oral burnsmeans that esophageal burns are not likely to be present. The pres-ence of oral burns leads to a 50 percent false positive rate in theprediction of esophageal burns.

If steroids do not prevent subsequent strictures, unlike Lovejoy inhis accompanying editorial,2 we question the value of esophagos-copy in all cases of accidental childhood ingestions of caustic sub-stances. If the treatment plan is not altered by endoscopic findings,why subject children to an unnecessary procedure? Our approach toasymptomatic children with the "lick and taste" ingestion of causticsubstances would be to do nothing immediately except begin sur-veillance for symptoms or signs of subsequent formation of stric-tures (i.e., dysphagia, vomiting, and so forth). If these occurred, wewould refer the patient to a surgeon. For the more conservativephysician, at the very least the data of Anderson et al. show thatpatients without oral burns do not require endoscopy. Of course, inthe presence of deliberate ingestions of caustic substances, wewould recommend endoscopy in patients with findings compatiblewith major caustic injury (shock, chest pain, or abdominal pain), toexclude the possibility of perforation of the viscus.

SUMAN WASON, M.D., M.B.A.MARIA STEPHAN, M.D.

Cincinnati, OH 45229 Children's Hospital Medical Center

The New England Journal of Medicine Downloaded from nejm.org on July 9, 2015. For personal use only. No other uses without permission.

Copyright © 1991 Massachusetts Medical Society. All rights reserved.