document resume ed 083 333 ud 013 853 - eric · document resume ed 083 333 ud 013 853 title...

DOCUMENT RESUME

ED 083 333 UD 013 853

TITLE Nutrition, the Nervous System, and Behavior.Proceedings of the Seminar on Malnutrition in EarlyLife and Subsequent Mental Development. (Mona,Jamaica, January 10-14, 1972).

INSTITUTION Pan American Health Organization, Washington, D.C.

SPONS AGENCY Ministry of Health, Kingston (Jamaica).; WellcomeFoundation Ltd., Beckenham (England).

PUB DATE Jan 72NOTE 150p.; Scientific Publication Number 251

EDRS PRICE MF-$0.65 HC-$6.58DESCRIPTORS Animal Behavior; *Behavior Development; *Cognitive

Development; Ecology; Health Needs; Infancy; InfantBehavior; *Language Development; Mental Health;Neurological Defects; *Neurological Organization;*Nutrition; Physical Health; Public Health; ResearchMethodology

ABSTRACTFive years have elapsed since the International

Conference on Malnutrition, Learning, and Behavior at theMassachusetts Institute of Technology. The present Seminar was heldto examine progress since then. The following papers were presentedand discussed: "Malnutrition and the Nervous System," Donald B.Cheek, A. B. Holt, and E. D. Mellits; "Lasting Deficits andDistortions of the Adult Brain Following Infantile Undernutrition,"John Dobbing; "Some Speculations on Mechanisms Involved in theEffects of Undernutrition on Cellular Growth," Myron Winick, J. A.Brasel, and Pedro Rosso; "Small-for-Dates Offspring: An AnimalModel," R. J. C. Stewart; "Empirical Findings with MethodologicImplications in the Study of Malnutrition and Mental Development,"Robert E. Klein et al.; "Malnutrition and Mental Capacity," FernandoMonckeberg; "The Influence of Malnutrition on Psychologic andNeurologic Development: Preliminary Communication," Jan Hoorweg andPaget Stanfield; "The Functioning of Jamaican School ChildrenSeverely Malnourished During the First Two Years of Life," Herbert G.Birch and Stephen A. Richardson; "Environmental Correlates of SevereClinical Malnutrition and Language Development in Survivors from.,Aawshiorkor or Marasmus," Joaquin Cravioto and Elsa DiLicardie;"Ecology of Malnutrition: Nonnutritional Factors InfluencingIntellectual and Behavioral Development," Stephen A. Richardson;"Issues of Design and Method in Studying the Effects of Malnutritionon Mental Development,"- Herbert G. Birch; and, "Nutrition, PublicHealth, and Education," Jack Tizard. (CM)

FILMED FROM BEST AVAILABLE COPY

U 5 DEPARTMENT OF HEALTH,EDUCATION & WELFARE

I r" NATIONAL INSTITUTE OFEDUCATION

I re%THIS DOCUMENT HAS BEEN REPRODUCFD EXACTLY AS kECEIVED FROM

I 144.4%THE PERSON OR ORGANIZATION ORIGINA T1NC, IT POINTS or VIEW OR OPINIONS

I reN STATED DO NOT NEC ESS.RiLY REPRESENT OF FICIAL NATIONAL INSTITUTE Of

CIO EDUCATION POSITION OR POLICY

NUTRITION, THE NERVOUS SYSTEM,

AND BEHAVIOR

Proceedings of the Seminar on

MALNUTRITION IN EARLY LIFE AND

SUBSEQUENT MENTAL DEVELOPMENT

MONA, JAMAICA

January 10-14, 1972

Scientific Publication No. 251

PAN AMERICAN HEALTH ORGANIZATIONPan American Sanitary Bureau Regional Office of the

WORLD HEALTH ORGANIZATION

525 Twenty-third Street, N.W.Washington, D.C. 20037, U.S.A.

1972

FOREWORD

Five years have elapsed since the known relationship bet wee., malnutrition andthe development of brain and behavior was reviewed at the International Conferenceon Malnutrition, Learning, and Behavior at the Massachusetts Institute of Tech-nology. Since then new studies in the laboratory and in the field have been carriedout or hegunin many countries. The present Seminar was held at the University ofthe West Indies at Mona, Jamaica, from January 10 to 14, 1972, to examine thisprogress and bring together researchers in the field to share their ideas and tech-niques. The participants were all active investigators from Latin America, theCaribbean, Europe, and the United States. In preparing their papers and discussionsto follow the presentations, they were asked to bear in mind the following purposesof the seminar:

I. To review current knowledge about malnutrition and its effect on brain andbehavior. Major attention was to be given to developments that had occurred sincethe publication in 1968 of Scrimshaw and Gordon's Malnutrition, Learning, andBehavior, the proceedings of the previous year's conference at the MassachusettsInstitute of Technology.

2. To identify and discuss issues and problems in nutrition studies that relateto mental and behavioral development.

3. To enable some of the investigators to present progress reports on their cur-rent work.

4. To identify areas and requirements for future research.

5. To consider the implications of present scientific knowledge for public policyand practice, and the role scientists may play-in communicating their findings so asto be helpful to policy makers and concerned citizens.

Dr. David Picou conceived the Seminar, and Drs. Herbert G. Birch, M. Martinsd.i Silva, Picou, Stephen A. Richardson, and John Waterlow developed its structureand program. The number of participants was small but included investigators frommany centers where research is being conducted on various aspects of the Seminar'stopic.

The Government of Jamaica, the Wellcome Trust, and the Pan American HealthOrganization generously gave financial assistance, and the University of theWest Indies and the University Hospital graciously provided the conferencefacilities.

Authors and rapporteurs revised their presentations and reports after the Seminar.The editors have tried to ensure that the rapporteurs' reports reflected the discus-sions at the end of each session.

iii

CONTENTS

Welcoming Speakers, Participants, and Invited Guests vii

Program Committee and Secretariat ix

Session I. Malnutrition and the Nervous System in Animals and Man

Malnutrition and the Nervous System Donald B. Cheek, A. 13. Holt, andE. D. Me Ibis 3

Lasting Deficits and Distortions of the Adult Brain Following Infantile Under-nutrition John Dabbing 15

Some Speculations on Mechanisms Involved in the Effects of Undernutrition onCellular Growth Myron Winick, J. A. Brasel, and Pedro Rosso 24

Small-for-Dates Offspring: An Animal Model R. J. C. Stewart 33

Summary of Session I Herbert G. Birch 38

Session II. Functioning of Children Malnourished in Infancy orfrom Disadvantaged Social Backgrounds

Empirical Findings with Methodologic Implication; in the Study of Malnutrition .indMental Development Robert E. Klein, Jean-Pierre Habicht, Charles Yar-brough, Stephen G. Sellers, and Martha Julia Sellers 43

Malnutrition and Mental Capacity Fernando Miinckeberg 48

The Influence of Malnutrition on Psychologic and Neurologic Development: Pre-liminary Communication Jan lloonveg and Paget Stanfield 55

The Functioning of Jamaican School Children Severely Malnourished During theFirst Two Years of Life Herbert G. Birch and Stephen A. Richardson 64

Environmental Correlates of Severe Clinical Malnutrition and Language Develop-ment in Survivors from Kwashiorkor or Marasmus Joaquin Cravioto andElsa DeLicardie 73

Summary of Session II Jack Tizard 95

Session HI. Ecology of Malnutrition

Ecology of Malnutrition: Nonnutritional Factors Influencing Intellectual and Be-havioral Development Stephen A. Richardson 101

Summary of Session III Joaquin Cravioto 1 1 1

Session IV. Methodologic Issues in Malnutrition Studies

Issues of Design and Method in Studying the Effects of Malnutrition on MentalDevelopment Herbert G. Birch . . . 115

Summary of Session IV Robert E. Klein 129

Session V. Future Research Directions

Summary of Session V Herbert G. Birch .

Session VI. Social and Administrative Problems

133

Nutrition, Public Health, and Education Jack Tizard 139Summary of Session VI Robert Cook 147

Prof. L. R. RobinsonPro-Vice ChancellorUniversity of the West IndiesMona, Jamaica

WELCOMING SPEAKERS

The Hon. Dr. H. EldemireMinister of HealthKingston, Jamaica

Dr. M. Martins da SilvaDepartment of Research Development

and CoordinationPan American Health OrganizationWashington, D.C., USA

PARTICIPANTS AND INVITED GUESTS

Dr. G. A. 0. AlleyneTropical Metabolism Research UnitUniversity of the West IndiesMona, Jamaica

Dr. K. AntrobusCaribbean Food and Nutrition InstituteUniversity of the West IndiesMona, Jamaica

Prof. Michael BeaubrunDepartment of PsychiatryUniversity of the West IndiesMona, Jamaica

Dr. Michael BegabNational Institute of Child Health

and Human DevelopmentNational Institutes of HealthBethesda, Maryland, USA

Prof. Herbert G. BirchDepartment of PediatricsAlbert Einstein College of MedicineBronx, New York, USA

Dr. Oliver BrookeTropical Metabolism Research UnitUniversity of the West IndiesMona, Jamaica

Prof. Donald CheekDepartment of PediatricsJohns Hopkins UniversityBaltimore, Maryland, USA

vii

Dr. Robert CookCaribbean Food and Nutrition InstituteUniversity of the West IndiesMona, Jamaica

Prof. Joaquin CravictoDivision of Scientific ResearchHospital Infantil de la IMANMexico City, Mexico

Prof. E. K. CruickshankDepartment of MedicineUniversity of the West IndiesMona, Jamaica

Dr. A. DavisEpidemiological Unit, Medical Research CouncilUniversity of the West IndiesMona, Jamaica

Dr. John DobbingDepartment of Child HealthUniversity_ of ManchesterManchester, England

Miss I. FosterCaribbean Food and Nutrition InstituteUniversity of the West IndiesMona, Jamaica

Miss Helen FoxMinistry of HealthKingston, Jamaica

Dr. Sally Grantham- McGregorEpidemiological Unit, Medical Research CouncilUniversity of the West IndiesMona, Jamaica

Dr. M. GuerneyCaribbean Food and Nutrition InstituteUniversity of the West IndiesMona, Jamaica

Dr. Ann HillTropical Metabolism Research UnitUniversity of the West IndiesMona, Jamaica

Drs. Jan HoorwegAfrican Study CentreLeiden, The Netherlands

Prof. D. B. JelliffeCaribbean Food and Nutrition InstituteUniversity of the West IndiesMona, Jamaica

Mrs. P. JelliffeCaribbean Food and Nutrition InstituteUniversity of the West IndiesMona, Jamaica

Dr. Douglas KerrTropical Metabolism Research UnitUniversity of the West IndiesMona, Jamaica

Dr. Robert KleinInstitute of Nutrition of Central America

and PanamaGuatemala City, Guatemala



Dr. Colin MillerDepartment of PediatricsUniversity of the West IndiesMona, Jamaica

via

Prof. Fernando MimckebergLaboratorio de Investigaciones PediatricasEscucla de Medicina, Universidad de ChileSantiago, Chile

Dr. David PicouTropical Metabolism Research UnitUniversity of the West IndiesMona, Jamaica

Dr. Frederick RichardsonChild Development CenterUniversity of MiamiMiami, Florida, USA

Prof. Stephen A. RichardsonDepartments of Pediatrics and Community

HealthAlbert Einstein College of MedicineBronx, New York, USA

Prof. K. StandardDepartment of Social and Preventive MedicineUniversity of the West IndiesMona, Jamaica

Dr. Paget StanfieldDepartment of PaediatricsMakerere UniversityKampala, Uganda

Dr. R. J. C. StewartDepartment of Human NutritionLondon School of Hygiene and Tropical

MedicineLondon, England

Prof. Jack TizardInstitute of EducationUniversity of LondonLondon, England

Prof. Myron WinickInstitute of Human NutritionColumbia University College of Physicians

and SurgeonsNew York, New York, USA

Dr. Leila Wynter-WedderburnKingston, Jamaica

PROGRAM COMMITTEE

Prof. Herbert G. BirchDepartment of PediatricsAlbert Einstein College of MedicineBronx, New York, USA




Dr. M. Martins da Silva (Secretary)Department of Research Development and

CoordinationPan American Health OrganizationWashington, D.C., USA

Prof. Stephen A. RichardsonDepartments of Pediatrics and Community

HealthAlbert Einstein College of MedicineBronx, New York, USA

Prof. John C. WaterlowDepartment of Human NutritionLondon School of Hygiene and Tropical

MedicineLondon, England

SECRETARIAT

ix


Session I

MALNUTRITION AND THE NERVOUS SYSTEM IN ANIMALS AND MAN

Chairman

E. K. Cruickshank

Rapporteur

Herbert G. Birch

MALNUTRITION AND THE NERVOUS SYSTEM'

Donald B. Cheek, A. B. Holt, and E. D. Mellits2

The present paper reviews the subject of mal-nutrition and the nervous system in animals andman. Our own interests over the years havefocused more on the role of hormone and nutri-tion in cellular growth of the soma (16, 17) ,and the nervous system has been of recent inter-est because of our work on fetal growth. Dur-ing the last decade, a great deal of interest inthe role of nutrition in brain growth has arisen.Three years ago, Winick, realizing that millionsof children were suffering from undcrnutritionand taking Cognizance of the results of progress,made the following prediction: "Both animalexperiments and human studies suggest that acritical period of brain growth may exist duringwhich malnutrition even in a mild form andeven for a short time, may produce irreversibledamage. This critical period appears to be

before birth and during early postnatal life"(80). Dobbing had a similar message: "Thcpossibility that undernutrition in early life maypermanently reduce the intellectual capacity ofmen and women has become increasingly recog-nized in recent years. In pediatrics we must nowbe much more concerned about undcrnutritionat certain vulnerable stages in development hav-ing long-term sequelac which may be irreversi-ble in spite of most strenuous attempts nr reha-bilitation" (29).

' This work was supported by the National Instituteof Child Health and Development Grant HD-00126-08,National Institutes of Health, Bethesda, Maryland.

2 Johns Hopkins University, Baltimore, Maryland,USA.

3

These are profound statements, which notonly make all branches of the medical worldspring to attention but lie heavily on the mindsof the administrator and politician.

The evidence for these statements arises fromwork in animals and man. It is important toreview this evidence briefly. One may considerinformation on rodents (small mammals), pigs(large mammals), and monkeys and humans(primates). We intend to deal mainly withchemical information.

Nutrition and the Rodent Brain(the small mammal)

It has been known for 50 years that restrictednutrition in postwcaning rats does diminishbrain size or brain weight (48). The findingsof Eayrs and Goodhcad (32) confirmed thethinking. The early experiments of Widdowson(75, 76) showed that more severe effects 'aroseif restriction extended from an earlier time orfrom one to 21 days.

In 1964, Dobbing (27) reported on the in-fluence of early nutrition on the developmentand myclination of the brain. Later he andWiddowson (28) studied the effects of rehabili-tation on myclination in deprived rats. Thechanges produced by restricting diet betweenthe third and 11th weeks of life were reversedat 19 weeks. In 1965, Benton and coworkers(8) showed that restricted nutrition during thepreweaning period resulted in deficits of brainweight, lipid ,:ontent, phospholipid, cholesterol,

and cerebroside. Refeeding reversed all of thesechanges, however.

Hatt and coworkers (62) drew attention tohistologic changes in the neurone and neu-roglia of protein-restricted rats. Such degenera-tive changes did not reverse after proteinfeeding.

Nucleic acid analyses by Winick and Noble(78) of brain tissue showed that restrictednutrition in the first 21 days of life caused anabrupt cessation of cell division (DNA increase).In the postweaning period, the reduction Of

DNA and the deficit in cell number was perma-nent, provided the insult had occurred duringweaning. If the insult had been inflicted after2.1 days, no changes persisted. Dickerson andWalmsley (24) found no changes in cell num-ber resulting from food deprivation after 21days. If nutritional status is restored before the10th day, however, effects can be reversed (79).

Other workers made similar findings. Cul leyand Mertz (21) found in 1965 that undernutri-tion from five to 20 postnatal days caused asustained reduction in brain weight and phos-pholipid, cholesterol, and cerebroside content.Guthrie and Brown (42) made similar findings,and Cul ley and Lineberger (22) showed that ifmalnutrition was extended to 17 days of post-natal life, irreversible changes occurred. Chaseand coworkers (11) found that the incorpora-tion of sulfatide into the myelin of rat brain wasmarkedly reduCcd by malnutrition during thefirst three weeks of postnatal life. The exten-sive studies, chemical and histologic, of Bassand coworkers (6, 7) focused on the somatosen-sory area of the rat cortex. They showed sig-nificant arrest of brain growth as a result offood restriction from one to 21 days. Not onlywas there delayed growth of the cerebral cortex,but the migration of cells was delayed and un-differentiated cells remained in the white matter.There was a great decrease in dendritic prolifer-ation.

Despite the multiple studies showing changesin the cerebrum or the whole brain during and

4

after the weaning period, the balance of evi-dence indicates that the cerebellum is the mostaffected in rats. This was the conclusion of'Culley and Lineberger (22), and was clearlyshown by Chase and coworkers (12) in their ratstudies. Winick also noted that in rats deprivedfor a short time, or for the first nine days only,the cerebellum was the organ affected (79),Howard and Granoff (47) found the greatestchange in mice deprived in the weaning periodin the cerebellum. In the rat it is the cells of thecerebellum that are most rapidly dividing atthat time (1). Hence, substrate deprivationaffects that organ most.

The question arises whether calories or pro-teins are lacking in these animals. Careful workby Miller (57) of the Massachusetts Instituteof Technology, using artificial feeding, has

shown that the central nutritional factor miss-ing is protein.

Bearing in mind the pattern of cell growth inthe rodent, as shown by Enes-.-3 and Leblond( and by Winick and Nobi: (77), namelyhyperplasia in utero, hyperplasia and hyper-trophy postnatally, and finally, hypertrophy perse, it would seem reasonable to agree withWinick (80) that not only is the extent orduration of nutrition important, but the ageof the animal at the time of insult or reha-bilitation is critical. Thus, we can look forcritical periods or times when cell multiplica-tion is greatest, and at those times we maysuspect that substance restriction will be highlydeleterious and therefore may influence subse-quent growth. Dobbing (29) has emphasizedthe vulnerable period as being the period ofthe brain growth spurt, the period during whichit is passing through the rapid phase of its

sigmoid growth trajectory. This is the periodwhen minimal food restriction might be ex-pected to produce maximal alteration in braingrowth. He has emphasized that nothing willdelay the "scheduled spurt" of growth in thebrain, though the intensity of the spurt may bediminished by delete,ious factors.

We have so far reviewed the information onrodents and mainly the effects of protein re-striction on-brain growth during the first 21days of postnatal life. One might confidentlyexpect that dramatic changes in rat braingrowth occur during this period. That is in-deed the case. Rats are born with only 20

per cent of their expected brain cells present,and a critical period may be defined at nine. to10 days.

As shown by Brasel and coworkers (9), atseven to eight days there is a marked increase inDNA polymerase activity. This increase is fol-lowed by a doubling within three to four daysof DNA and brain cells, while from 10 to 17 daysthere is a great acceleration. The full comple-ment of cells is reached. Surely this is a criticalperiod. At the same time, as shown by MillichapCS 8) and Ashby (2), the carbonic anhydrasepresent in the neuroglial cells (39, 40, 51) also

rises precipitously at nine days. Chase and co-workers (11) have shown accelerartd SO4 in-corporation, which indicates active myelination.

We can now understand why experimentsincluding our own (14) with food or proteinrestriction in rats from one to 10 days old donot produce any later changes in brain chem-istry. Indeed, it can also be pointed out that thebrain changes that result from thyroid ablationin newborn rats can be reversed if hormonetherapy is given before the 10th postnatal day(38, 52). The intensity of the insult mustalso be important, as we shall discuss later. Itis now pertinent to review the situation in ratsin the prenatal period.

Zamenhof and coworkers (84) showed thatwhen pregnant rats were fed 8 per cent protein,the offspring had a reduction in brain DNA andprotein content. Zeman and Stanbrough (85)confirmed this, while Zeman (86) later foundthat adequate nutrition in the immediate post-natal period reversed such chemical changes.

Thus, the balance of evidence indicates thatin the rat maternal food restriction limits avail-able substrate for the fetus, and though only

5

20 per cent of the brain cells are due to appearduring fetal life, the nutritional stress is suf-ficient to interfere with cell development. If

adequate nutrition is instituted postnatally, nochemical changes persist. Nutritional insult tothe weanling rat would appear to be of greaterconsequence than for the fetal rat. Studies ofbehavior in deprived rats support this thesis

(3). Conversely, as shown by Chase and co-workers (13) in the guinea pig, in which almostall cell multiplication occurs in utero (30), foodrestriction to the mother produces changes inthe cerebellar DNA, protein, and lipids of thefetus. Adequate feeding postnatally does nottotally reverse the diminution of cellularity inthe adult cerebellum.

Nutrition and the Pig Brain(large mammal)

At birth the pig weighs only 6 kg but as anadult it weighs 200 kg, a 30-fold increase.

Hence, postnatal nutritional deprivation shouldstrongly affect somatic growth. The DNA in theforebrain, cerebellum, and cord increases maxi-mally just before birth, and cell multiplicationcontinues to 12 weeks postnatallf (25). Thereis a peak in the rate of DNA increase five weeksbefore birth. Pond and coworkers (63) gavepregnant pigs a protein-free diet but offered anadequate diet postnatally. In the adult phase theoffspring showed no reduction of DNA in thecerebrum and only a borderline reduction inRNA. Psychologic and intellectual changes atfirst thought to be significant (4, 54) have sincebeen found to be heavily influenced by environ-mental factors (5, 36).

Dickerson and coworkers (26). restrictednutrition in pigs for the first year of lifeasevere stressand then. refed the animals fortwo more years. The cellular content of thebrain did not reach expected levels.

Evidence with respect to pigs is limited, butit would appear that the fetal brain is highlyresistant to maternal protein restriction and thatsevere and prolonged protein restriction is neces-

sary to produce chemical changes in postnatallife.

Nutrition and Brain Development in Man

(a) Normal brain growth

The human has one-third of his eventualnumber of brain cells at birth, which is less

than that of the guinea pig and monkey butgreater than that of the rat. The rate of brain-weight increase in rat, pig, and man has beencompared by Davison and Dobbing (23). Thepeak in nun is reached shortly before birth, butthat in the rat is postnatal. The work of Dob-bing and Sands (31) and of Dobbing (29)reveals that if the cerebral oNA is consideredfrom 10 weeks' gestation to three months post-natally, one can draw a curve to fit the pointsindicating a fourth-order relationship (a sigmoidcurve followed by a straight line). It is be-lieved that the first spurt (from 15 to 20weeks) is neuronal and the second (from 25

weeks on) is neuroglial, while the completecomplement of cells is reached two years post-natally. We have analyzed these data for wholebrain (IS) and more recently for cerebrum bypolynomial regression analysis. We obtainstrong linear fit for whole brain and a quad-ratic fit for cerebral INA. No evidence of ahigher fit up to .1 fifth-order relationship hasconic to hand. If there is some biologic reasonfor regarding the period 1(1 to 22 weeks asseparate (as Dobbing has divided his data (29))and 22 weeks to three months postnatally as asecond period, then the relationship would holdas stated. By considering the increments in

brain DNA as a percentage of the adult valueper unit of time (five-week periods), Dobbinghas shown two peaks of cell growth, the majorone being three months postnatally, which maycorrespond to a period of extensive myelinationand presumably is to be regarded as a vulnerableperiod.

We do not interpret the data as does Dobbing.If we .apply the technique of Mellits (56) to thestrong quadratic relationship obtained for cer-

6

brat DNA against time, it should be possible totit two intersecting straight lines in place ofthe quadratic. The point of intersection canhave significance insofar as it indicates a periodin which there is a change in the rate of growth(56). If we undertake this procedure, a breakis obtained at 32.8 weeks' gestation. If we alsoinspect the c..tensive data of Schultz and co-workers (66) on brain weight (1,193 brains)and plot these data as a percentage of increasein weight against conceptual age, we find twopeaks, one at 32 weeks' gestation and anotherfive months postnatally (Figure 1). The dataof Coppoletta and Wolbach (20) yield the sameinformation if plotted.

If one reassesses Dobbing and Sands' data onbrain DNA content (31), they may be expressed

150

120

100

80

60-

4C

201

Human BrainPerccrnaga incroosein weight par month

32 ..!,lekscosIaticn

St,

I

Schulz et.al.

Bith2 4 6 _AR: 12 14 20 22 240 e;

Ago from Conception (Months)

Figure 1. Brain growth is expressed as the percentageof increase per month from conception. The data aretaken from the work of Schultz and coworkers (66).Note that at eight months' gestation (32 weeks) thereis a decided peak of change in brain growth. A secondpeak is also described at five postnatal months (20weeks postnatally). The same second may be found ifthe data of Coppoletta and Wolbach (20) are similarlyexamined.

2000-WhoI3 Crain DNA (Sche natic)

1600

1200

800

40C'r : *.!-

Birth

cf. Schulz

JI

'.nick et al.

( ciritc)

20 40 CO ED 100 120 140I

Conceptual Age (Weeks)

Figure 2. Composite schematic graph showing the growth of whole brain DNAFrom 20 weeks' gestation to 15 weeks postnatally the data of Dobbing and Sandshave been used (31), and from that time until 110 weeks postnatally the recentdata of Winick and coworkers (81} have been employed. From available evidenceone would suspect that the steepest rise is at 20 weeks postnatally or at the timeshown in Figure 1.

as a linear or a quadratic function up to 55weeks (conceptual age) (18) (Figure 2 ) . At55 weeks, data from Winick and his coworkers(81) may be used. (Earlier data are different(18)). If the line of best fit is drawn subjec-tively (Figure 2 ) , it becomes clear that from15 to 20 weeks postnatally there is a period ofmaximal cell growth. This result is differentfrom the earlier thinking of Davison and Dob-bing (23). These remarks are made not to con-tradict their excellent work, but to draw atten-tion to an alternate interpretation of their data.

(b) Malnutrition and human brain growth

Both kwashiorkor and marasmus exert dele-terious effects on brain growth. Brown (10)reported a reduction in brain weight and Gar-row (37) found reduced 40K radiation from thebrain in living infants and the restoration ofthat count after rehabilitation. Winick andRosso (82) analyzed brains from marasmicchildren who died during their first year of lifeto determine RNA, DNA, and protein content.

7

The early gestate nal history of the infants wasnot known, but deficits in the chemical markerswere found and brain growth arrest was demon-strated. Fishman and coworkers (35) reporteda reduction of lipids, cholesterol, and neura-minic acid in the brains of infants who died ofmalnutrition between two and 22 months ofage.

Available data about normal postnatal valuesfor DNA, RNA, and protein in the human brainfrom three to 12 months are meager, and dis-crepancies in data between laboratories exist(18). Dobbing notes that losses of RNA andprotein occur following death (29). Winick

and his coworkers (82) showed that the DNAcontent of the normal human cerebrum reacheda constant value six months postnatally, whichis not in accord with Dobbing's data (29). Themarasmic infant had significantly less cerebralDNA. Similar findings were made in the cere-bellum, and brain lipids were reported to bereduced (81 ) . The important question is

whether this growth arrest in the living infantis reversible.

Psycho logic studies of malnutrition in in-fants are numerous and their results indicatemental impairment ( (67) for review), but theinvestigator has to separate the effects of theenvironment from the effects of nutritionalrestriction as such (5, 36).

With respect to prenatal malnutrition in thehuman, the information relating to wartimestarvation in Holland is often cited (71). Therethe intake was 400 calories per person per day.Newborn infants were decidedly retarded ingrowth. Those infants (now 27- and 28-year-old adults) have been thoroughly investigatedby Dr. Z. Stein of New York (72), who foundno deficit in intelligence. Severe placental in-sufficiency as seen in human parabiotic twins isknown to cause changes in the TQ of one twin(19).

The evidence with respect to the human thusindicates an arrest of brain growth if nutri-tional restriction occurs. Whether intelligenceis impaired is not clear. Maternal protein-calo-rie deprivation would not appear to alter theintelligence of the offspring.

Nutrition and the Brain of Macaca mulatta

We have recently inspected the changes inDNA, RNA, protein, cholesterol, ganglioside,water, and chloride in the cerebrum and cere-bellum of the rhesus monkey from midgestation(80 days) until term (160 days) and for 30days postnatally. From midgestation the cere-bral weight, and DNA, RNA, protein, and gangli-oside content rise as a slight sigmoid curve andreach adult values at or soon after 30 days post-natally. The RNA:DNA ratio rises and reaches apeak at birth. The increment in ganglioside

(NANA) is prominent from 120 days to birth,when a steady value is obtained. Cholesterolcontent increases later and continues to increasewell into postnatal life. The activity of car-bonic anhydrase activity is markedly increasedimmediately after birth.

8

The growth of the cerebellum is, of course.entirely different. The concentration of DNA in-creases while that of the cerebrum decreases.with time, as others have found (55) (Figures3 and 4). This organ continues to grow aneven four months postnatally has not reachadult levels. The nNA, RNA, and protein reachadult levels 80 days postnatally, but cholesteroldeposition must continue to increase well intopostnatal life. Carbonic anhydrase in the cere-bellum increases markedly after birth, as withthe cerebrum. These data will be reported

separately (46). For the cerebrum, the biggestsurge forward in cellular growth would ap-pear to be from 80 days and for the cerebellumfrom 120 days (Figures 1 and 2). The earlyattainment of adult levels in the cerebrum (30

DNA CONCENTRATION IN CEREBELLUMmg /9

7.0

6.0

5.

4.

3.0

2.

MacaqueI..%/.i

/I.,.

'

.

Birth

s_--.

Mature

100 150 200 days

4.0-CHICKEN

3. Margolisir/

2. tBirth

ys

0 10 20 30 40 50 daysConceptual Age

Figure 3. Changes in cerebellar DNA concentrations inthe monkey, pig (25), and chicken (55) during gestationand the postnatal period. Note the comparison.

DNA CONCENTRATION IN CEREBRUMmg/g

Birth4Mature

Ntilbeelpi,..- -sr-00 100 150 200 250 Days

Binh

- - t

--Whole broin

10 20 30 40 50 DaysConceptual Age

Figure 4. Changes in DNA concentration in the cere-brum of the monkey, pig (23), and chicken (55) duringgestation and the-postnatal period.

postnatal days) would indicate that criticalperiodsif they existwould be in intrauterinelife.

Two laboratories have studied the effects ofpostnatal protein restriction in monkeys, andour own laboratory is now studying the effectsof maternal nutritional deprivation on themonkey fetus.

Ordy and coworkers (60) gave a 3.5 per centprotein diet to the infant monkey from three tonine months postnatally without producing anychange in brain weight. Kerr and coworkers(50) produced growth arrest in monkeys fromone to seven months of age with a protein-deficient diet. Many of the infants died. Brain'

weight was not reduced in general, but changesin the liver, myocardium, and intestines weresignificant. The infants were then refed thecontrol diet from seven months to one year.Catch-up growth occurred and was eventuallycomplete anthropometrically. Head circumfer-ence was normal. Drs. A. Deets and P. Harlow

9

of the University of Wisconsin have found nosubsequent behavioral or psychologic deficits inthese primates (personal communication).

In the rhesus monkey, cerebral cell numberis almost complete at one month, and certainlyso by three months. Myelination is still in

progress, as is the growth of the cerebellum.It can be argued that nutritional restriction inthe prenatal period will alone be the time to pro-duce permanent cerebral changes in the monkey.

By removing the secondary placenta of thefetal monkey at 100 days' gestation, we did pro-duce a half-sized term fetus (44, 59). Sincethe placenta is responsible for the transfer ofsubst aces, we believe that this represents a

good example of fetal malnutrition. We havebeen able to show in the cerebellum, but notin the cerebrum, a reduction in protein andDNA content and a borderline change in RNA,but we do not know whether these changes arereversible or not

Our second approach is to study brain growthafter restricted maternal nutrition during preg-nancy. This work is ;i1 collaboration with theNational Institutes of Health, and in particularwith Drs. London, Weiss, Ellenberg, and Bieri.We are studying the effects of severe protein andcalorie restriction on the pregnant rhesus

monkey. The intake of calories and proteinhas been kept to about one-third of normal.The weight loss of these pregnant animalsranges from 0.5 to 1.0 kg. A few control andexperimental fetuses have come to hand andhave shown us little chemical change. A de-crease in cholesterol, which could well be re-versible (28), would appear to exist, and alsoa decrease in protein:DNA, especially in the-cerebellum. More work is necessary before thesefindings can be substantiated.

Implications

It has been shown that development of therat brain is highly susceptible to nutritionaldeprivation between seven and 17 days, whencellular growth is remarkable. The abnormal

changes persist. Protein restriction during preg-nancy does cause changes in cell number andin protein accretion in the fetal rat brain, butthese changes are reversible (86). This is notso for the guinea pig, in which cell multiplica-tion is almost completed during intrauterine lifeNutritional depravation during pregnancy inter-feres with subsequent growth of the cerebellumpostnatally (13 ).

By implication, one might anticipate thatinsults would be significant for the subhumanprimate during pregnancy and would also

mainly affect the cerebellum, since most of thebrain cell growth is in titer° and cerebellargrowth again is still in progress during thepostnatal period. It is probable that for thehuman the period around five months post-natally is critical. No evidence exists, however,to show whether brain changes do or do notpersist in the primate, although it is clear thatsome growth arrest occurs. We do not knowwhether refeeding will reverse the changes, butthe observations of Stein and coworkers (72)would certainly suggest that the cerebral cor-tex does not suffer any changes because of nu-tritional deprivation during pregnancy. More-over, one must keep in mind that primate cellgrowth is much slower than in nonprimates(61), so that the insult would have to be dis-proportionately greater to produce a changecomparable to that found in rats. It is also truethat the smaller the mammal, the greater thelitter size (53) and the faster the rate of fetaltissue accretion (18). It is for these reasonsthat we suggested that the syndrome of growtharrest in the brain without reversal is mainly acondition present in small mammals, in whichmetabolic rate, protein turnover, and nutri-tional requirements are disproportionatelyhigher ( 18).

While obvious differences do exist in primategrowth relative to nonprimate growth, it is

also true that events in brain growth are verysimilar. To illustrate this point we have drawnFigures 5, 6, and 7. The percentage increases inthe DNA content and brain weight for the rat,

10

guinea pig, and man are shown. The time atwhich stability of brain weight occurs is taken.This is plotted as 100 per cent, and on theabscissa 0 to 100 per cent represents time fromconception to stability of brain weight. Theordinate represents the percentage of weight orof DNA reached relative to maturity. For thehuman brain, the assumption was made that cellgrowth occurs as shown in Figure 2. We assumethat the mature level of DNA in the humanwhole brain is 1,500 mg (81). The similarityof Figures 3, 4, and 5 is obvious, and the maxi-mal periods of cell growth may be seen. Forthe rat, however, the spurt in cell growth(from 50 to 90 per cent of expected) is indeedrapid. This would support the inference thatthe critical period in the rat is more remarkable.

RAT BRAIN

% DNA joiesent relative

% Brainto 7u days

ain wt. relative towt. at 70 days A's

a DNAWt

20 40 60 80 100% Time

Conception to brain wt. stability .100% (70 days)

Figure 5. The period from conception until the brainreaches a stable weight is documented as 100 per cent.In the rat this period is 70 days from conception. Theordinate shows the percentage of DNA or percentageof weight relative to the stable level reached at anyparticular time. Note the steep climb in the rat thatoccurs halfway through development. Points are calcu-lated from available data (9, 70).

GUINEA PIG BRAIN

%DNA present relativeleu to 110 days

100

To Brain wt. relative towt. at 110 days

dI

/

DNAWt

20 40 60 80 100% Time

Conception to brain wt stability = I00%(110 days)

Figure 6. Figure 6 is similar to Figure 5 but plottedfor the guinea pig, whose brain weight stability is

reached at 110 days from conception. The rate of DNAaccumulation is not as fast as the rat's. Points have beencalculated from available data (30).

120

10

HUMAN BRAIN

%DNA present relativeto 150 weeks

%Brain wt. relative towt. at 150 weeks

II

DNAwt.

20 40 60 80 100% Time

Conception age to brain wt. stability .100% (150 weeks)

Figure 7. This figure is again similar to Figure 5 butis plotted for the human. Stability of brain weight istaken as 150 weeks from conception. 'Note that DNAincrease rate is again not as rapid as for the rat.Points have been calculated from available data (20, 31,65, 66, 81).

140

11

Our own work has shown us that the cere-bellum is the organ mainly affected in the pri-mate. Although this work is incomplete, it

could be that no lasting cerebral or cerebellardefects will unfold because of protein restric-Lion in the prenatal period.

Workers studying psychologic and behavioralchanges due to undernutrition are becomingmore convinced that environmental deprivation(stimulation) is the key factor. Undoubtedly,altered reaction to stimuli, emotional instability,and 'withdrawal are characteristic of nutri-tionally deprived subjects (5). If lack of stimu-lation is added to protein deprivation, thenmental retardation may be expected. Whethersuch is reversible is not known.

Sereni and coworkers (68) have shown thatnorepinephrine and serotonin levels are reducedin the brain of rats deprived nutritionally fromone to eight days. Shoemaker and Wurtman(69) find that the brains of the offspring con-tain less dopamine and norepinephrine if thelactating mother receives a diet restricted inprotein. Food intake stimulates insulin secre-tion, which in turn elevates amino acid levelsincluding that of tryptophan, which sequen-tially elevates the level of serotonin.

Our earlier work drew attention to the failureof proper insulin release in protein-deficient ratsor infants (41, 43). It is possible that insulinplays some role in brain growth (45).

Whether growth hormone performs any rolein brain growth is also not clear. Some evi-dence exists to show that it does ( ) 5, 38, 52,83). Observations by Taplin (74) indicatedthat the number of acidophil cells in thepituitary and the weight of the gland are

linearly related to the weight and size of themature rat following restricted feeding early inpostnatal life. Stephan and coworkers (73)found that the offspring of underfed dams havesmaller pituitaries containing lower concentra-tions of growth hormone. In other words, itwould appear that hormonal mechanisms adjustto growth processes but nutritional intake is the

ke' or leading factor (49, 64). It is conceiv-able that an imbalance of the sympathetic nerv-ous system and reductions of hormones im-portant to growth arc also crucialdevelopment of these brain changes.these mechanisms arc just as important

to thePerhapsas criti-

cal periods since, for example, we find thatexposure of rats to 12 per cent oxygen from oneto seven days postnatally later produces exten-

sive changes to the brain, while nutritional dep-rivation during the same, noncritical period hasno effect (14). Moreover, we have found inunpublished work that thyroid hormone is fun-damental to brain growth during fetal life. (Itis known to be important for postnatal braingrowth.) The role of this hormone may well beimportant with respect to brain growth andprotein deprivation.

REFERENCES

1. AtrmAN, J. Proliferation and migration of un-differentiated precursor cells in the rat during postnatalgliogenesis. Exp Nettro/ 16:263-78, 1966.

2. Astuty, W., and E. M. SIICSTER. Carbonic an-hydrase in the brain of the newborn in relation to

functional maturity. / Bwl Chem 184:109-16, 1950.

3. BAIN), A., it al. Effects of calorie and proteindeficiencies early in lift: on the subsequent learningability of rats. Br 1 Altar 25:391-403, 1971.

4. BARNES, R. H. Experimental animal approaches tothe study of early malnutrition and mental develop-ment. Fed Proc 26:144-47, 1967.

5. . Nutrition and man's intellect and be-havior. Fed Pro,- 30:1429-33, 1971.

6. BASS, N. H., et al., Effect of neonatal malnutritionon developing cerebrum. 1. Microchemical and histo-logic study of cellular differentiation in the rat. ArchNewt): 23:289-302, 1970.

7. et al. Effect of neonatal malnutrition ondeveloping cerebrum. 2. Microchemical and histologicstudy of myelin formation in the rat. Areb Nenrol23:303-13, 1970.

8. BEN-roN, J. W., et al. Modification of the scheduleof myelination in the rat by early nutritional deprivation.Pediatrics 38:801-07, 1966.

9. BRASEL, J. A., et al. DNA polymeiase activity inrat brain during ontogeny. Dev Biol 23:424-32, 1970.

10. BROWN, R. E. Decreased brain weight in malnu-trition and its implications. East Afr Med 1 42:584-95,1965.

II. CHASE, H. P., et al. The effect of malnutrition onthe synthesis of a myelin lipid. Pediatrics 40:551-59,1967.

12. et al. Undernutrition and cerebellar devel-opment. Nature (Lund) 221:554-55, 1969.

13. et al. Intra-uterine undernutrition andbrain development. Pediatrics 47:491-500, 1971.

14. CHEEK, D. B., et al. Hypoxia and malnutrition innewborn rats: effects on RNA, DNA, and protein in tissues.Am 1 Physiol 217:642-45, 1969.

12

15. and J. E. GRAVSTONE. The action of

insulin, growth hormone, and epinephrine on cell

growth in liver, muscle, and brain of the hypophysecto-mizecl rat. Pethatr Res 3:77-88, 1969.

16. and D. E. 11111.. Muscle and liver cell

growth: roll of hormones and nutritional factors. FedPro, 29:1503-09, 1970.

17. and . The effect of growth hor-mone on cell and somatic growth: a review. In: ErnestKnouil (ed.) , Handbook of physiology, Washington,American Physiological Society, in press.

18. Brain growth and nucleic acids: effectof nutritional deprivation. In: Frederick Richardson(ed.) , Ecology of child human development, Baltimore,National Educational consultants, in press.

19. CHURCII11.1, J. A. The relationship between in-telligence and birth weight in twins. Neurology 15:341-47, 1965.

20. COPPOI.IXTA, J. M., and S. B. WOLBACII. Bodylength and organ weights of infants and children: studyof body length and normal weights of more importantvital organs of body between birth and 12 years of age.Am 1 Pathol 9:55-70, 1933.

21. Cm.t.1tY, W. J., and E. T. MERTZ. Effect of re-stricted food intake on growth and composition of pre-wearding rat brain. Prot- .Sue Exp Biol Med 118:233-35,1965.

22. and R. 0. LINEBERGER. Effect of under-nutrition on the size and composition of the rat brain.

Nntr 96:375-81, 1968.23. DAVISON, A. N., and JOHN DOBBING. Myelination

as a vulnerable period in brain development. Br MedBull 22:40-44, 1966.

24. DtCKERSON, J. W., and A. L. WALMSLEY. Theeffect of undernutrition and subsequent rehabilitation onthe growth and composition of the central nervous sys-tem of the rat. Brain 90:897-906, 1967.

25. and JOHN DOBBING. Prenatal and post-natal growth and development of the central nervous

system of the pig. Proc R Sot. Lond [Biol] 166:384-95,1967.

26. tt al. The effect of undernutrition on thepostnatal development of brain and cord in pigs. ProcR Sot Lond [Biol] 166:396-407, 1967.

27. DOBBING, JOHN. The influence of early nutritionon the development and myelination of the brain. ProcR Soc Lond [Biol] 159:503-09, 1964.

28. and E. M. WIDDOWSON. The effect ofundernutritiod and subsequent rehabilitation on mveli-nation of rat brain as measured by its composition.Brain 88:357-66, 1965.

29. . Undernuttition in the developing brain:the releVance of animal models to the human problem.Am J Dis Child 120:411-15, 1970.

30. and JEAN SANDS. Growth and develop-ment of the brain and spinal cord of the guinea pig.Brain Res 17: 115 -23, 1970.

31. and . Timing of ncuroblast multi-plication in developing human brain. Nature (Land)226:639-40, 1970.

32. EAYRS, J. T., and B. GOODHEAD. Postnatal devel-opment of the cerebral cortex in the rat. / Anat 93:385-402, 1959.

33. ENESCO, M., and C. P. LEBI.OND. Increase in cellnumber as a factor in the growth of the organs andtissues of the young male rat. / Embryo! Exp Morphol10:530-62, 1962.

34. FERNSTROM, J. D., and R. J. WURTMAN. Brain

serotonin content: physiological depedence on plasmatryptophan levels. Science 173:149-52, 1971.

35. FISHMAN, M. A., et al. Low content of cerebrallipids in infants suffering from malnutrition. Nature(Lond) 221:552-53, 1969.

36. FRISCH, R. E. Present status of the suppositionthat malnutrition causes permanent mental retardation.Am J Clin Nutr 23:189-95, 1970.

37. GARROW, J. S. Loss of brain potassium in kwa-shiorkor. Lancet 2:643-45, 1967

38. Gr.: Ei., S. E., and P. S. TIMIRAS. Influence ofgrowth hormone on cerebral cortical RNA metabolismin immature hypothyroid rats. Brain Res 22:63-72,1970.

39. GIACOBINI, E. A cytochemical study of the locali-zation of carbonic anhydrase in the nervous system.

Neurochem 9:169-77, 1962.40. . Metabolic relations between glia and

neurons studied in single cells. In; M. M. Cohen andR. S. Snider (eds.), Morphological and biochemicalcorrelates of neural activity, New York, Harper & Row,1964, pp. 15-38.

41. GRAHAM, G. G., et al. Infantile malnutrition:changes in body composition- during rehabilitation. Pe-diatr Res 3:579-89, 1969.

13

42. GUTHRIE, H. A., and M. L. BROWN. Effect ofsevere undernutrition in early life on growth, brain sizeand composition in adult rats. / Niter 94:419-26, 1968.

43. Hit.'., D. E., ct al. The influence of protein.calorieversus calorie restriction on the body composition andcellular growth of muscle and liver in weanling rats.Johns Hopkins Med / 127: 146 -63, 1970.

44. et al. Fetal growth retardation producedby experimental placental insufficiency in the rhesusmonkey. 2. Chemical composition of the brain, liver,muscle and carcass. Thol Neonate 19:68-82, 1971.

45. et al. Brain growth in the fetal primate:brain growth during hormonal and nutritional imbal-ance In: D. B. Cheek (ed,), Pre- and postnatal cellu-lar growth: role of hormones and nutrition. Inpreparation.

46. HOI.T, A. B., et al. Brain growth in the fetalprimate: significance of brain determinants and thevalue of cortical slice analyses. In: D. B. Cheek (ed.),op. cit.

47. HOWARD, E., and D. M. GRANOFF. Effect ofneonatal food restriction in mice on brain growth, DNA

and cholesterol, and on adult delayed response learning.Nutr 95:111-21, 1968.48. JACKSON, C. M. Changes in the relative weights

of the various parts, systems, and organs of young albinorats held at constant body weight by under-feeding forvarious periods. J Exp Zool 19:99-156, 1915.

49, KENNEDY, G. C. The development with agc ofhypothalmic restraint upon the appetite of the rat.Endocrinol 16:9-17, 1957.

50. KERR, G. R., et al. Malnutrition studies h. therhesus monkey. 1. Effect of physical growth. Am I ClinNutr 23:739-48, 1970.

51. KORliONEN, E., and L. K. KORHONEN. Histo-chemical demonstration of- carbonic anhydrase activityin the eyes of rat and mouse. Acta Ophtba Imo! 43:475-81, 1965.

52. KRAWIEC, L., ct al. Hormonal regulation of braindevelopment. 3. Effects of triiodithyronine and growthhormone on the biochemical changes in the cerebralcortex and cerebellum of neonatally thyroidectomizedrats. Brain Res 15:209-18, 1969.

53. LEITCH, 1., et al. The maternal and neonatalweights of some mammalia. Proc Zool Soc Lond 113:11-28, 1959.

54. LOWREY, R. S., et al. Influence of caloric leveland protein quality on the manifestations of proteindeficiency in the young pig. J Nutt. 78:245-53, 1962.

55. MARGOLIS, F. L. DNA and DNA- polymerase ac-tivity in chicken brain regions during ontogeny. l Neu-rochem 16:447-56, 1969.

56. MELLITS, E. D. Statistical methods: estimationfor intersecting regressions. In: D. B. Cheek (ed.),

Human growth, Philadelphia, Lea 6: Febiger, 1968, pp.19-38.

57. Mu LER, S. A. Nutrition in the neonatal develop-ment of protein metabolism. Fed Proc 29:1497-1502,1970.

58. Mtl LICHAP, J. G. Development of seizure patternsin newborn animals: significance of brain carbonic an-hvclrase. Prot Soc Exp Biol Med 96: 125 -29, 1957.

59. MYERS, R. E., et al. Fetal growth retardation pro-duced by experimental placental insufficiency in therhesus monkey. 1. Body weight, organ size. Biol

Neonate 18:379-94, 1971.60. ORnx, J. M., et al. Brain vulnerability to postnatal

protein calorie deficiency in infant rhesus monkeys. ProcSoc Exp BR)! Med 135:680-84, 1970.

61. PAYNE, P. R., and E. F. WHEELER. Comparativenutrition in pregnancy. Nature (Lon(?) 215:1134-36,1967.

62. PLATT, B. S., et al. Experimental protein-caloriedeficiency. In: H. N. Munro and J. B. Allison (eds.),Mammalian protein metabolism, New York, AcademicPress, 1964, vol. 2, pp, 445-521.

63. POND, W. G., et al. Effect of protein deprivationof swine during all or part of gestation on birth weight,postnatal growth rate and nucleic add content of brainand muscle of progeny. I Nutr 99:61-67, 1969.

64. Ross, M. H., et al. Influence of protein andcaloric intake upon spontaneous tumor incidence of theanterior pituitary gland of the rat. I Nutr 100:177-89,1970.

65. SCAMMON, R. F.., and H. L. DUNN. Empiricalformulae for the postnatal growth of human brain andits major divisions. Proc Soc Exp Biol Med 20: 114 -17,1922.

66. SCHULZ, D. M., et al. Weights of organs offetuses and infants. Arch Pathol 74:244-50, 1962.

67. SCRIMSHAW, N. S., and J. E. GORDON, eds. Mal-nutrition, learning, and behavior. Proceedings of anInternational Conference cosponsored by The NutritionFoundation, Inc., and The Massachusetts Institute ofTechnology held at Cambridge, Mass., March I to 3,1967. Cambridge, Massachusetts, M.I.T. Press, 1968.

68. SERENI, F., et al. Undernutrition and the devel-oping rat brain. 1. Influence on acetylcholinesterase andsuccinic acid dehydrogenase activities and on norepi-nephrine and 5.0H-tryptamine tissue concentrations.Biol Neonate 10:254-65, 1966.

69. SHOEMAKER, W. J., and R. J. WURTMAN. Peri-natal undernutrition: accumulation of catecholamines" inrat brain. Science 171:1017-19, 1971,

70. Si Kov, M. R., and J. M. THOMAS. Prenatal growthof the rat. Growth 34:1-14, 1970.

14

71. SMITH, C. A. Effect of wartime starvation inHolland upon pregnancy and its product. Am ObstetGynnol 599-608, 1947.

72. STEIN, Z. A.. et al. Sequel to the Dutch famine:effects on mental performance. Ann Meeting, Am PubHealth Assoc, Minneapolis, October 1971.

73. STEPHAN, J. K., et al. Relationship of growthhormone to the growth retardation associated with ma-ternal dietary restriction. I Nutr 101: 1453 -58, 1971.

74. TAPLIN, D. E. The pituitary acidophils of nutri-tionally stunted rats. Aust I Exp Biol Med Sri 45:P-6,1967, abstract.

75. WIDDOWSON, E. M., et al. Severe undernutritionin growing and adult animals. 4. The impact of severeundernutrition on the chemical composition of the softtissues of the pig. Br / Nntr 14:457-71, 1960.

76. and R. A. MCCANCE. The effect of finiteperiods of undernutrition at different ages on the com-position and subsequent development of the rat. Prot-

1? Sin Lond lj 158 : 329-42, 1963.

77. WINICK, MYRON, and ADELE NOBLE. Quantita-dye changes in DNA, RNA, and protein during prenataland postnatal growth in the rat. Dev Biol 12:451-66,1965.

78. and . Cellular response in ratsduring malnutrition at various ages. I Nutr 89:300-06,1966.

79. et al. Cellular recovery in rat tissues aftera brief period of neonatal malnutrition. Nntr 95:623-26, 1968.

80. . Malnutrition and brain development. IPediatr 74: 667-79, 1969.

81. and PEDRO Rosso. The effect of severeearly malnutrition on cellular growth of human brain.Pediatr Res 3:181 -84, 1969.

82. et al. Cellular growth of cerebrum, cere-bellum, and brain stem in normal and marasmic chil-dren. Exp Nemo( 26:393-400, 1970.

83. ZAMENHOF, STEPHEN, et al. Stimulation of theproliferation of cortical neurons by prenatal treatmentwith growth hormone. Science 152: 1396 -97, 1966.

84. et al. DNA (cell number) and protein inneonatal brain: alteration by maternal dietary protemrestriction. Science 160:322-23, 1968.

85. ZEMAN, F. J., and E. C. STANBROUGH. Effect ofmaternal protein deficiency on cellular development inthe fetal rat. I Mar 99:274-82, 1969.

86. . Effect of protein deficiency during gesta-tion on postnatal cellular development in the youngrat. I Nutr 100:530-38, 1970.

LASTING DEFICITS AND DISTORTIONS OF THE ADULT BRAIN

FOLLOWING INFANTILE UNDERNUTRITION

John Dobbing'

The question whether undernutrition duringcertain stages of brain development contributesto lasting behavioral changes is now attractingincreasing attention from research workers inmany disciplines. Among those primarily inter-ested in the human implications of the problem,inquiry has centered largely on severely mal-nourished babies in starkly poor communities,usually but not always in developing countries.It has also impinged on the general pediatricconcern for the fate of low-birth-weight babies,especially those of retarded intrauterine growth,and any other babies whose growth may beimpaired, including those with metabolic errors,congenital anomalies, endocrine disturbances,chronic hypoxia, long-term exposure to growth-retarding drugs, or any syndrome that includesgrowth retardation in its consequences.

The question being asked is deceptively sim-ple for the unwary, and many of the moreresponsible workers in the field are dismayed atthe readiness with which it is sometimes over-dramatized. The question whether early mal-nutrition causes mental subnormality or mentalretardation or brain damage is, in their view,wrongly posed. Rather it should be askedwhether undernutrition or growth retardation,among the multitude of other important earlyenvironmental factors, can be identified as acontributor to the algebraic sum of those influ-

Department of Child Health, University of Man-chester, England.

15

ences that determine adult "attainment."Efforts to answer this more realistic, but muchmore difficult question must attempt to isolatethe variable of nutrition experimentally, andthis of course is strictly impossible.

The present paper will examine the proposi-tion that there are periods of heightened vul-nerability in the physical development of thebrain, during which growth retardation resultsin long-lasting and detectable distortions anddeficits in adult brain structure. This will neces-sarily be established in animal species before anattempt is made to extrapolate the idea to ourown. Sonic evidence for the validity of suchextrapolation will be discussed.

"Vulnerable-Period" Pathology

Developmental neuropathology has hithertobeen exclusively concerned with lesions, struc-tural or biochemical. The former have beeneither first-trimester teratologic malformationsor last-trimester focal destruction through suchagencies as hypoxia, hypoglycemia, hyperbiliru-binemia, focal hemorrhage, or trauma. Bio-

chemical "lesions" have by contrast been dif-fuse and identifiable either histochernically (thelipidoses) or not (the inborn errors).

The new pathology proposed here shows noneof these features, and will often not even showrelated physical signs. It consists of quantita-tive disorders of the brain's growth program

resulting in easily detectable differences in adultbrain structure, but none of these would berevealed by orthodox neuropathology. Neitherthe very rare nutritional condition of Wernicke'sencephalopathy nor such lesions as those ofvitamin A deficiency form part of the newpathology, though it is possible that some formsof microcephaly (e.g., in rubella) may be alliedwith it. An important possibilitythough it is

no more than thatis that some cases of

hitherto unclassified mental retardation mayshow such quantitative, rather than "lesion,"pathology.

The science of developmental nutrition recog-nizes three important parameters of undernu-trition apart from its specific nature. These areits severity, its duration, and its timing, andthey are interrelated. The importance of thethird factor (the age at which undernutritionoccurs) has given rise to the idea of transientperiods of heightened sensitivity which havesome apparent resemblance to sensitive periodsduring the development of behavior. Largelybecause of the need to imply both lasting dis-tortion and lasting deficit, the term "vulnera-ble" has been used rather than the moreacademic terms "critical" or "sensitive" period.

The basis of the vulnerable-period hypothesis(11) is best illustrated by a clear-cut finding inrats that if bodily growth is retarded at the timeof the brain growth spurt, there is a resultinggrowth deficit that resists subsequent nutritionalrehabilitation. There are permanent deficits ofboth brain and bodily growth attainment, andin the brain at least there is some permanentstructural distortion as well as deficit. There isgood emerging evidence that the principle is

valid for other species, including man.There is an apparent dependence of good

bodily growth on ,:he achievement of a satis-factory brain growth spurt. The bodily growthspurt is later in all species than that of the brain,and is not itself "vulnerable" in the above, last-ing sense. Nutritional retardation of bodilygrowth at this later stage is fully recoverable

16

on restoration of a good diet (22), and evensevere starvation in the adult produces no last-ing detectable effects on the brain (12) .Whether growth retardation before the braingrowth spurt has lasting structural effects is lessclear, although by gross measurement these havebeen too small to detect in the brain (at leastwhen growth is retarded to the extent com-monly occurring in humans), just as there areno obvious changes after severe adult under-nutrition.

In its originally proposed form ( I I ), theperiod of vulnerability was considered to be thewhole period of the brain growth spurt to bedescribed below. A second and overlappinghypothesis (23) can also be applied to all othertissues. It was discovered that during their post-organogenetic growth all tissues pass through aperiod of cell division followed by one of growthin cell size. Undernutrition during the formerphase, but not during the latter, permanentlyreduces growth attainment in all tissues.

There can be no question that this latteris a basic law for catch-up potential throughoutthe body, but it may nevertheless be unsatis-factory when applied to those aspects of braindevelopment that may be functionally impor-tant. For example, glial mitosis occurs laterthan neuronal mitosis, and glia eventually out-number neurons heavily. Most experimentalproof of the vulnerability of "brain cell" mitosishas been confined to the later phase of glialmultiplication, which happens to occupy thefirst part of the brain growth spurt. Most ofthe lasting "brain cell" deficit in these experi-ments must therefore be glial, and it seems

unlikely that a numerical glial deficit will befunctionally important. It is even possible thata numerical neuronal cell deficit would not bevery significant for brain function comparedwith a deficit, for example, in subsequent

dendritic branching and in the establishment ofsynaptic connections. These later and probablymore important features of neuronal growth arenot mitotic events. They are the brain's equiva-

lent to growth in cell size, and it is possible thatsuch postmitotic events within the brain growthspurt are also vulnerable. Finally, the techniqueof expressing average cell size as a protein:nNA ratio must surely be less meaningful forneurons, with their enormously long cytoplas-mic extensions, than for a homogeneous mass ofliver cells.

The phrase "growth spurt" is derived fromthe sigmoid trajectory of most organ or wholebody growth. It is simply the transient periodof high growth velocity. In the brain it is

found in all mammalian species examined andrepresents phases of brain development similarfrom one species to another. Only the oc-currence of birth shows interspecies variation(11): in the rat the growth spurt is a post-natal event (17), in the guinea pig it is pre-natal (14), and in pigs (10) and people (13)it is both pre- and postnatal. Norm-,i birth hasno significance for most growth programs in-cluding that of the brain. Thus there are obvi-ous pitfalls here for those wishing to use animalsto investigate "fetal" or "postnatal" braingrowth unless the species is carefully chosenand the timing of the brain growth spurt in thatspecks known.

The brain growth spurt begins at about thetime neuroblast multiplication ends and theadult number of neurons has already been al-most achieved. This is toward the end of thesecond human (fetal) trimester (15) and in thefirst postnatal days of the rat. It ends with theend of the major period of rapid myelination, atabout two years of human postnatal age (18)and at about 25 days in the rat (17). In verygeneral terms the velocity curve of incrementsin brain wet weight encompasses the wholeperiod, except perhaps for the later stages ofrapid myelination. The major easily detectableevents, apart from an increase in size, are analmost explosive multiplication of oligo-

dendroglial cells (5) followed by a period ofintense lipid synthesis related to myelination

(7). There are large and sometimes sudden

changes of enzyme activity (1, 3 ). Tissue waterfalls reciprocally with the rise in brain lipid(17). Sodium and potassium move rapidlytoward their adult values (8). Perhaps moreimportantly, but less easily measured, the

growth spurt includes the growth and branch-ing of dendritic processes and the establishmentof interneuronal connections. These are held tobe plastic to some extent in the adult, and itwould be surprising if they were not even moreso during development.

Of course it is gro.....ly facile to attach func-tional significance to the vulnerability of thegrowth-spurt complex as described above. Someof the oversimplifications will now be

enumerated.First, neurobiologists have only measured

parameters that are comparatively easy tomeasure. No one knows what to measure as aphysical index of important aspects of highermental function. Certainly cell number, brainsize, degree of myelination, and so forth are nomore than tangible examples of structural char-acteristics that, by analogy, may possibly reactin a manner similar to whatever structuresactually do matter.

Further, developmental processes in the brainoccur at different times in different regiOns andat different rates. There is a highly organizedtemporal and spatial sequence, and it is likelythat there is a differential regional vulnerabilityrelated to the normal rate and timing in anyparticular region. One clear example is the

differential vulnerability of the cerebellum,which will be described later. This part of thebrain grows more rapidly than other parts andis therefore more specifically affected by growthretardation (9). Some of its neurons dividelater than most (4), and they are thereforedifferentially selected by later growth retarda-tion (16).

17

Experimental Designs

Experimental studies of nutritional depriva-tion early in life have often been on rats. It has

been convenient that the brain growth spurt inthis species is largely confined to the sucklingperiod. One of the following four nutritionalprocedures has usually been employed: maternalundernutrition or malnutrition, restriction ofsuckling time, and rearing in large litters. Ma-ternal undernutrition is achieved by feeding themother a restricted quantity of a good-qualityfood daily, whereas in the case of maternal mal-nutrition, the rat has access to an unlimitedsupply of an unbalanced diet, often low in

protein and consequently high in sonic otherconstituent, usually carbohydrate. Restrictionof suckling time necessitates regular separationof mother and young so that normal feedingopportunities are curtailed. Th,:se three tech-niques usually involve standardization of littersize at birth. The large-litter method requiresthat two or more mothers give birth within ashort space of time. Young are removed fromtheir mothers and randomly assigned to an ab-normally large or small litter. Mother rats ormice would be given large litters of 15 to 20young.

The desired timing of the nutritional re-striction to some extent determines the methodof deprivation to be used. Obviously the re-stricted suckling and large-litter procedures areapplicable only to the suckling period, whereasdietary restriction of the putative mother maybegin even before she has conceived.

Information on the effects of diet on milkcomposition is scanty for both humans (20)and laboratory animals (21). What evidencethere is suggests that both protein and calorierestriction suppress milk, yield without appreci-able alteration in quality. Hence the nutri-tional effect of the four methods of deprivationon the suckling animal is probably similar.

Some Lasting Effects of EarlyGrowth Retardation

Provided the brain growth-spurt period is

carefully selected, it is only necessary to retardbodily growth rates toward the lower limits of

18

the "normal" range to produce permanentchanges. No nutritional disease nor any obviousill health is required. Changes have been foundin the brains of adult rats whose only nutritionalhandicap has been to be suckled in larger-than-normal families during the brain growth spurt.They are fed a highly nutritious diet ad libitumfrom weaning at three weeks of age until theyreach maturity. The permanent residual physi-cal effects in adults of such mild growth re-tardation, when looked for in previously under-fed adults, represent both deficits of and dis-tortions from the normal, and include the fol-lowing:

(a) Small brain size

This is a true microcephaly, the brain beingpermanently smaller than is appropriate for thebody weight, in spite of having shown the tra-ditional characteristics of "brain sparing" dur-ing the growth -period (17). The brain is no',uniformly small, being more so in the cerebel-lum than elsewhere. For both reasons, therefore,the small brain is distorted rather than merelydeficient.

(b) Fewer cells

When measured by DNA analysis these smallerbrains have fewer cells (17). There are dis-proportionately fewer in the cerebellum, andamong the cerebellar neurons the later-dividinggranular neurons are disproportionately reduced.There are also indications that some cerebralneurons are also disproportionately reduced(16). Most ofglial, however.brain structure

the total cell deficit is probablyHere again there is distortion ofin addition to simple deficiency.

(c) Less lipid

Many of the brain lipids are permanentlyreduced to a greater extent than would be pre-dicted from the smaller brain size. Such defi-ciencies per unit fresh weight are selectivelyfound in those lipids most characteristic ofmyelin (12), and this is a further example ofdistortion.

(d) Altered enzyme activit y

Assessments of enzyme activities are bedeviledby argument about their significance and the-ignificance of the various ways of expressingthem. Should activity be expressed per cell(DNA), per unit protein, per unit fresh weight,per brain region, or how? The conceptual diffi-culties are greatly magnified in a tissue whosearchitecture and cell type are as heterogeneous as

are those of the brain compared, for example, tothe more homogeneous liver. Such differences as

have been found in the present context prob-ably represent structural changes. For example,the activity of .acetylcholinesterase per unitfresh weight in previously undernourished adultrats is much higher than in controls (2), inspite of having been lower during the earlyperiod of restriction (1). It seems very likelythat the enzyme is related to structures (per-haps cholinergic nerve endings) whose concen-tration has been increased as a result of a dif-ferentially greater deficit of other structures.Interpretation can be extremely tortuous, butat least here is another distortion.

In summary, the evidence for the vulner-ability of the brain growth spurt is very good.Certainly, however, it could never be claimedthat it was the only vulnerable period of braingrowth, even in the present restricted sense,without a much more careful analysis of thevulnerability of other stages, especially theearlier and quantitatively smaller phase ofneuronal multiplication corresponding to thehuman second trimester. Nevertheless, the hy-pothesis remains a useful one and in generalterms is well supported by the evidence.

Brain Growth Opportunity

It has recently been shown that the braingrowth spurt is obliged to occur at a prede-termined chronologic age, even when condi-tions are unfavorable and growth has been nu-tritionally retarded (17). The effect of such"retardation" is to reduce the extent of braingrowth processes, not to delay their occurrence.

19

This has been demonstrated by constructingreliable velocity curves of the accumulation offresh weight, DNA, and cholesterol (a majorbrain lipid) in undernourished animals andnormal controls. The smaller area under thevelocity curves of the undernourished animals,together with their failure to exhibit "catch up"on restoration to a normal diet, accounts for theultiMa te deficit. An example of this is il-lustrated in Figure 1 for the phase of (glial)cell division.

This introduces a principle of possibly greatpractical importance to human babies. If thereis a once-only opportunity to grow the brain.properly, it is presumably important that thebest conditions should be provided at this time.There-. may even be something to be said fordirecting nutritional aid toward the relevantage-group in times of severe shortage at theexpense of older individuals whose brain growthspurt is over.

The Extrapolation of the Vulnerable-Period Hypothesis to Man

It has been shown that the changes describedin previously growth-retarded adult animals arerelated to the timing of the restriction to coin-cide with the period of the brain growth spurt.Leaving aside any consideration of whether suchchanges matter, it ought to be' possible to pre-dict which human babies are at risk in thissense. All that is required is to identify thehuman brain growth spurt, and this is underway.

A study of nearly 200 complete human'brainsreveals the following relevant facts:

(1) Human neuronal multiplication occursmainly in the second trimester of gestation(15). The subsequent growth of dendrites andthe establishment of synaptic connections is,

by inference, later. Probably it occupies theremainder of the growth spurt.

(2) Thus the period corresponding to thevulnerable one discussed above in animals prob-ably begins with the third trimester.

6

4

2

0

BRAIN CNA P RATE CURVES

SMALL LITTERS

LARGE LITTERS

0 4 8 10

BIRTH WEANING

AGE ( WEEKS)

Figure 1. Velocity curves of DNA increment in the brains of rats suckle.:.. smalland large litters during the brain growth spurt. Growth retardation has not delayedthe timing of the peak velocity (17).

(3) The end of the (glial) cell multiplica-tion phase is at about 18 postnatal months (13),and the whole growth spurt is virtually over bytwo postnatal year (18).

Thus it can be seen that only about one-eighth of the human vulnerable period is fetalin term babies, a situation much closer to thatof the infant rat than has hitherto been recog-nized. In this light, the eventual outcome ofrestricting human fetal growth would largelydepend on the following theoretical considera-tion: Is good catch-up growth being institutedfrom birth onwards, and is there successfulgrowth promotion during the last seven-eighthsof the brain growth spurt that is postnatal inhumans? How much recovery of physical braingrowth is possible if good bodily growth is in-stituted after restriction occupying only thefirst one-eighth of the vulnerable period? Thechances are that recovery would be virtuallycomplete. This would convert the first two

20

years of postnatal life from a period of vul-nerability to an important one of opportunity.There is sonic evidence from experimental ani-mal undcrnutrition that if rehabilitation is

instituted well before the end of the braingrowth spurt, there is apparent recovery (24).Recovery here was measured in terms of wholebrain amounts, however, and it is not yet clearwhether there are persisting regional deficits.It could be that restriction on early growthprocesses in those regions of the brain that de-velop early may produce lasting regional deficitswhich would be masked by analysis of wholebrain. The nutter is open to experiment.

It is therefore likely that any baby whoseproper, lean-body-mass growth is seriously re-tarded for the whole, or for a substantial pro-portion of the period from* 30 weeks' gestationto two full years of postnatal age may emergewith changes in the physical state of its braincomparable with those outlined above in rats.

Whether these will matter is an entirely openquestion, since it is completely unknown whichparts of the physical brain are related to "highermental function." It is also not known howmuch compensation may be available in the

multiplicity of other developmental, environ-mental factors bearing upon the ultimate out-come. The most that can be suggested in prac-tical, management terms is that, among all theother steps taken to promote baby welfare, thebest true growth of babies should be ensuredat this particular period. This applies equallyto the baby in an underprivileged developingcommunity, to the prematurely born, the "small -for- dates," and the normal term baby.

Finally it may be worth recording a per-

sonal note. In my view serious errors of thoughtare currently arising from a failure, mainly ofpediatricians, to consider this subject biologi-cally. It is nonsense to talk of the "fetal brain"or "neonatal brain" without taking into ac-count the growth characteristics of the brainin the particular species under discussion. Thereis no demonstrable difference in principle be-tween the human brain during the first one-eighth of its growth spurt (which is fetal) andthe rat brain at the same developmental stage,just because in this species it is postnatal. Simi-larly, in considering the relevance to man offindings in animals, it is a travesty to ignore thethree fundamentals of all studies of develop-mental undernutrition: its degree, its duration,and particularly its timing in relation to thewhole developmental program. It seems to beunpalatable to some that man can often re-semble rats or pigs very closely. Valid extra-polations are possible, but they must be verycarefully made on a sound basis of knowledgeof comparative development. This is especiallyso for those who would use subhuman primates.It is a fatally erroneous assumption that monkeybrain development so closely resembles that ofman, particularly in its timing, that such nice-ties of extrapolation can be ignored.

The lasting effects of early undernutrition onthe structural development of the brain have

21

necessarily been investigated in animals. The

question may now be asked whether there areany behavioral consequences in those animalsWhose brains have been permanently altered?And can the animal model be used to helpanswer the all-important question whetherhumans suffer any lasting intellectual deficitfollowing early growth retardation? There aresome formidable difficulties in extrapolatingfrom one species' behavior to another's, eventhough their patterns of physical brain growthare so similar. The suggestion that the humanproblem, inextricably entangled as it is witha multitude of socioeconomic and other particu-larly human factors. can be elucidated throughanimal studic:. in simple laboratory situations isprobably realistic. A discussion of the oftenconflicting evidence from relevant experimentsin animal behavior will be found elsewhere(19).

Summary

The evidence for the lasting structuralchanges in the brain related to the timing ofearly undernutrition is clear-cut. The braingrowth spurt is by far the most vulnerableperiod in crude terms of quantitative ultimateachievement and measurable physical distor-tion. There is some evidence even from studiesof human growth and development in privi-ledged communities that seems to resemble thephysical findings in animals (6), and some ofthe as-yet-scanty behavioral data from poorcountries do support the idea that children canbe permanently affected by malnutrition in theirearly years. Indeed, the correspondence is closebetween the vulnerable period for human be-havioral development and the human braingrowth spurt. It must be remembered, however,that in all developmental findings such cor-respondence can as easily be coincidental as

meaningful.It would not be stretching the evidence too

far to say that in the light of present knowledgethere is a period of human development ex-

/

tending from the second trimester of gestationwell into the second postnatal year, duringwhich the brain appears to have a once-onlyopportunity to grow properly. It is at this

time especially important that children should

grow at a proper rate and under the best en-vironmental conditions. Among these nutritionis central to proper growth, a restriction ofwhich may well have lasting behavioral con -seq uences.

ACKNOWLEDGMENT

This work is supported by a grant from the Medical Research Council.I am als'o grateful to the National Fund for Research into CripplingDiseases and the Spastics Society for their help. I especially thank mycolleagues, Dr. B. P. F. Adlard, Miss J. Sands, and Dr. J. L. Smart fortheir continuing collaboration.

REFERENCES

AnLAkn, B. P., and JoHN Donnt8c. Vulnerabilityof developing brain. 3. Development of four enzymesin the brains of normal and umlernourisbed rats. Brain

Res 28:97 -107, 1971.

2. and . Elevated acetylcholinesterase

activity in adult rat brain after undernotritinn in earlylife. Brain Res 30:198-99, 1971.

3. and . Phosphofructokinase and

fumarate hydratase in developing rat brain. I Nenro-hem 18: 1299-1303, 1971.

4. ALTMAN:, J. Postnatal neurogenesis and the prob-lem of neural plasticity. In W. A. Himwich (ed.),Developmental neurobiology, Springfield, Ill., Thomas,

1970, pp. 197-240.

5. BENsTEAD, J. P., et al. Neurological developmentand myelination in the spinal cord of the chick embryo.I Erobryol Exp Morphol 5:428-37. 1957.

6. DAVIES, P. A., and J. P. DAvis. Very low birth-weight and subsequent nead growth. Lancet 2: 1216 -19.

1970.

7. DAvisoN, A. N., and A. PETERS. Mvelination.Springfield, Ill., Thomas, 1970.

8. DE SOUZA, S. W., and j0/IN DOBBING. Cerebraledema in developing brain. I. Normal water and cationcontent in developing rat brain and postmortem changes.Exp Nenrol 32:431-38, 1971.

9. DICKERSON, J. W., and JOHN DOBBING. Some

Fieettliarities of cerebellar growth in pigs. Proc R Soc ?lied59:1088, 1966.

10. and . Prenatal and postnatal

growth and development of the central nervous systemof the pig. Proc R Soc Lord [Btoll 166:384-95, 1967.

11. DOBBING, JOHN, Vulnerable periods in developingbrain. In: A. N. Davison and John Dobbing (eds.),

22

Applied neurochemistry, Oxford, Blackwell Scientific

Publications. 1968. "p. 287-.316.I2. . Effects of experimental undernutrition

on development of the nervous system. In N. S.Scrimshaw and J. E. Gordon (eds.) , Malnutrition.learning, and behavior, Cambridge. Massachusetts,

MAX. Press, 1968, pp. 181-202.13. . Undernutrition and the developing

brain: the relevance of animal models to the humanproblem. Am I Dis Child 120:411-15. 1970.

14. and JEAN SANDS. Growth and develop-ment of the brain and spinal cord of the guinea pig.Brain Res 17:115-23, 1970.

I 5. and . Timing of neuroblast multi-plication in developing human brain, Nature (Loud)226:639-40, 1970.

16. et al. Vulnerability of developing brain.7. Permanent deficit of neurons in cerebral and cere-bellar cortex following early mild undermitrition. Exp

Neorol 32:439-47, 1971.

17. and JEAN SANDs. Vulnerability of develop-ing brain. 9. The effect of nutritional growth retarda-don on the timing of the brain growth-spurt. Dial

Neonate 19:363-78, 1972.

IS. and Growth and development ofthe human brain. In preparation.

19. and J. L. Si ART. Early undernutrition,brain development and behaviour. Clinics in Develop-mental Medicine, in press.

20. GUNTHER, M. Diet and milk secretion in women.Proc Notr Soc 27:77 -82, 1968.

21, Ni E SON, M. M., and H. M. EVANS. Diet:Iry re-quirements for lactation in the rat and other laboratoryanimals. In: S. K. Kon and A. T. Cowie (eds.), Milk:

the mammary gland and its secretion, New York Aca- 23. WINICK, Is.,IsitoN, and April, Nolu F. Cellular re

(Ionic Press, 1961, p. 137. sponse in rats during malnutrition at various ages. /

22. Winnowsos, E. M., and R. A. Mr:CANcv. The Notr 80:300-06, 1066.

effect of finite periods of undernutrition at different ages 24. ct al. Cellular recovery in rat tissues after

on the composition and subsequent development of the a brief period of neonatal malnutrition. / Note 94:

rat. Pros R Lund [1011 158:329.-42, 1963. 623-26, 1068.

23

SOME SPECULATIONS ON MECHANISMS INVOLVED IN

THE EFFECTS OF UNDERNUTRITION ON CELLULAR GROWTH

Myron Winick, J. A. Brasel, and Pedro Rosso'

During the past It) years a great deal ofquantitative data has been accumulated on thenormal cellular growth of various organs, re-gions of organs, and in some cases discreteareas. Growth has been viewed as a smoothprogression from pure proliferation throughcombined hyperplasia and hypertrophy andfinally to hypertrophy alone. Malnutrition hasbeen shown to retard the rate of cell divisionduring proliferative growth without changingthe time at which proliferation cea: .s. Thus theresultant organ is smaller and contains fewercells. This change is permanent. In contrast,the same degree of undernutrition during hyper-trophic growth will inhibit the normal in-crease in cell size. This change, however, is

reversible with rehabilitation.It becomes evident in focusing on the de-

veloping brain that permanent cellular changes,if they arc to occur, must come about as a con-sequence of early malnutrition. In the rat,cell division is over in brain by 21 days of ageand only before that age will malnutrition in-duce a reduction in brain cell number. Variousbrain regions have individual patterns of cellgrowth. The most rapid rate of cell divisionin rat brain is in cerebellum and this continuesto 17 days of life. In cerebrum, cell division isslower but continues longer, whereas in brain

' Institute of Human Nutrition, Columbia UniversitcCollege of Physicians and Surgeons, New York, NewYork, USA.

24

stem proliferation it continues at a very slowrate only until the 14th day of life. In cere-bellum both neurons and glia continue to di-vide postnatally, whereas in cerebrum only gliadivide at this time of life.

Neonatal malnutrition retards the rate ofcell division in all areas studied. The earliest andseverest effects are on those areas where cell di-

vision is most rapid. For example, a significantdecrease in cerebellar cell number is noted aftereight days of malnutrition beginning at birth.Changes in cerebrum occur later and are lesssevere, and there are only very few chaiiges inbrain stern. This reduction in the rate of celldivision affects any cell type undergoing pro-liferative growth. In addition, the migration ofcells from under the lateral ventricle to thehippocampus, which usually occurs on the 15thday of life, is curtailedprobably because therate of cell division under the lateral ventricleis reducedresulting in fewer cells able to

migrate.

During the past few years these principleshave been partially confirmed in man. Cell di-vision normally ceases in human brain aroundthe end of the first year of life. Proliferationin all three regions so far studied (cerebrum,cerebellum, and brain stem) stops at the sametime. Malnutrition during the first year of lifewill reduce the rate of cell division and resultin fewer cells in whole brain and in all threebrain regions.

There may be questions about the magnitudeof this reduction, about the types of cells in-volved, and about the severity of the mal-nutrition necessary to produce an effect. Thereare certainly questions about the meaning ofthese changes, but the fundamental principlethat undernutrition during proliferative growthwill retard the rate of cell division in humanbrain appears well established.

How then does malnutrition exert its effecton cell division? During the past few yearsa number of laboratories have begun to addressthemselves to this problem.

During cellular growth and cell division weknow that amino acids from the general bodypool are supplied directly for protein synthesisand for the synthesis of nucleotides, which thenenter the general body of nucleotide pool. Thenucleotides arc then used for either DNA or RNAsynthesis. Newly-formed DNA is quite stable,whereas the newly formed RNA turns over andthe rate of synthesis is in equilibrium with therate of degradation. A number of enzymes areinvolved in all these processes and their activitydepends, in part at least, on the quantitative andqualitative nature of protein synthesis. .1* hus theavailability of amino acids will affect bothnucleic acid and protein synthesis, and theamount and quality of the protein synthesizedwill in turn affect the synthesis of nucleic acids.To further complete the cycle, since RNA is anessential element in protein synthesis, changes inthe quantity and quality of RNA will also affectthe quantity and quality of the proteinsynthesized.

Where in this cycle does protein-calorie mal-nutrition exert its effect? Although the dataarc not yet complete, they certainly point to thefact that limitation of protein in the diet willdecrease the availability of amino acids for pro-tein synthesis. The work of Munro and morerecently that of Miller has demonstrated re-duced protein synthesis both in vivo and invitro when either protein or amino acids in

the diet arc limited. Further, Munro has sug-

25

gested that although any amino acid may belimiting experimentally, tryptophan appears tobe the limiting amino acid in most physiologicsituations. Thus, although definitive proofawaits measurement of amino acid pool size andtracking of amino acid pathways under condi-tions of protein-calorie malnutrition, presentevidence strongly indicates that the amino acidpool size is reduced.

How is this reduced quantity of amino aciddistributed? Amino acids for building blocksof protein are less available, as shown by thedecreased incorporation of labeled amino acidsinto total protein. In contrast, there is sonic

evidence indicating that amino acids are prefer-entially converted to nucleotides. For example,total nucleotide pool size in brains of neonatallymalnourished animals is unaffected. Incorpora-don of labeled orotic acid into the nucleotidepool is significantly increased, however, andpreliminary data suggest that .incorporation oflabeled amino acids into the nucleotide pool is

also increased. We may postulate then that thesynthesis of nucleotides is increased in brainsof neonatally malnourished rats. If the nucleo-tide pool size does not change in brain oractually decreases in liver, where arc these

nucleotides being distributed? Incorporation oflabeled nucleotides into DNA is markedly re-duced in the brains of malnourished animals,demonstrating that there is a decreased distribu-tion into DNA synthesis. In contrast, incorpora-tion of labeled precursor into RNA is markedlyincreased in brains of neonatally malnourishedanimals. Thus the rate of DNA synthesis is

reduced, whereas the rate of RNA synthesis isincreased in neonatal protein-calorie restriction.One effect, then, of early malnutrition is a re-distribution of available nucleotides. How isthis redistribution controlled? What are themechanisms involved in the decrease I DNA syn-

thesis, and how can we explain an ,ncrease inRNA synthesis in the face of descriptive datathat conclusively demonstrate a reduced RNAcontent per cell?

We can answer the second question by ex-amining the rate of RNA degradation afterearly malnutrition. There is a marked increasein the rate of decay of previously labeled RNAin liver of malnourished rats, and in brain thereis a drop in the RNA/DNA ratio at a time whenRNA synthesis is increased. Thus although therate of RNA synthesis is increased, the rate ofdegradation must also be increased. The latterincrease is presumably greater than the former,resulting in a net loss of RNA, which explainsthe drop in the RNA DNA ratio or RNA contentper cell.

At this point, we may summarize the effectsof early malnutrition on some of the syntheticand degradative pathways involved in cellulargrowth. Incorporation of amino acids into totalprotein is reduced as is net protein synthesis.Incorporation of labeled precursors into thenucleotide pool is increased in the face of anunchanging or dropping total nucleotide poolsize. Incorporation of nucleotides into DNA isdecreased as is net DNA synthesis. Incorpora-tion of nucleotides into RNA is increased, indi-cating an increased rate of RNA synthesis in theface of a decrease in cellular RNA content. Therate of RNA degradation is markedly increased.

In an attempt to explore some of the mecha-nisms by which these dynamic changes occur,the activity of certain of the enzymes involvedin the regulation of the steps just described hasbeen measured. DNA polymcrase is an enzymeinvolved in the terminal phase of DNA synthesis.In a number of nonphysiologic situations,activity of this enzyme has been shown toincrease under conditions that stimulate DNAsynthesis. Moreover, the increase has been

shown to precede the increase in DNA synthesis.Recent experiments in our laboratory havedemonstrated that the response to unilateralnephreCtomy in the opposite kidney is hyper-plastic in infant rats and hypertrophic in adultrats. This hyperplastic response is preceded by aburst of DNA polymerase activity in the remain-ing kidney of the infant animal, whereas the

26

polymerase response in the adult is much less.More recently it has been shown that activity

of this enzyme in liver of growing rats is in

part under the control of pituitary growthhormone. Hypophysectomy reduces enzymeactivity and growth hormone replacement ele-vates the activity before any increase in DNAsynthesis can be shown.

We have demonstrated in brain that theactivity of DNA polymerase parallels the rate ofcell division during normal growth, and thatthis parallel relationship holds when variousbrain regions are studied. Our data show thatthe activity of this enzyme is an excellent indi-cator of the rate of cell division and suggestthat during normal growth DNA polymerase mayplay a role in regulating the rate of DNA syn-thesis. We believe that neonatal malnutritionwill reduce the activity of this enzyme in ratliver.

Thus, the evidence at this stage indicatesthat one way by which early malnutrition maycurtail the rate of DNA synthesis and perhapsindirectly regulate the distribution of availablenucleotides is by reducing the activity Of DNApolymerase.

While we have not investigated the enzymesinvolved in RNA synthesis, others have begunsuch investigations. For example, Metcoff andcoworkers have demonstrated an increase in

RNA polymcrase activity in leukocytes ofundernourished children and in placentas frommothers who were malnourished and whoseinfants demonstrated intrauterine growth fail-ure. Indicative as these data may be, experi-ments are lacking that reveal an elevation inthe activity of RNA polymerase in tissues ex-posed to neonatal malnutrition and correlatethis increased activity with an increase in therate of RNA synthesis.

The degradative phase of RNA metabolismis of particular interest since it is the increasedrate of degradation that accounts for the netloss of RNA per cell. The enzyme alkalineRNase has been presumed to be involved in RNA

catabolism. Available data indicate that undermost circumstances activity of this enzyme isinversely related to cellular RNA content. Wehave, therefore, undertaken a series of investi-gations designed to explore the role of thisenzyme during normal growth and the altera-tions that may occur in early malnutrition.

During normal brain development the ac-tivity' of this enzyme per cell increases, butthis increase is exactly proportional to theincrease in cellular RNA content and the rela-tionship between enzyme and substrate remainsconstant throughout development. Total ac-tivity of alkaline RNase increases per milligramof DNA per cell throughout development exceptfor a drop at birth and again at 17 days of age.By contrast. specific activity per milligram ofprotein declines during development. Whenactivity is expressed per milligram of RNA permilligram of substrate, there is no change dur-ing development. These data emphasize theimportance of enzyme activity's mode of ex-pression and point out clearly that interpreta-tion of changes in such activity may dependa great deal on the reference point employed.By using these three reference points, a clearerpicture can be drawn. Activity per cell in-creases during brain development; this increaseis less than the increase in other cellular proteins

and hence the specific activity drops. The in-creased activity per cell, however, is directlyproportional to the increase in RNA content percell that occurs during development, and there-for the activity per milligram of RNA does notchange. This constant relationship betweenRNpe activity and RNA content under circum-stances in which both are changing again indi-cates a role for this enzyme in the regulationof RNA metabolism. In the studies noted above,only total cellular alkaline RNase activity was

examined because of our initial findings of verysmall amounts of activity in the nuclear frac-tion and the absence of free activity (non--inhibitor-bound) in the cytoplasmic fraction.

27

In other tissues, however, alkaline RNase isactive intranuclearly as well as in the cyto-plasm, and is present as both bound and freeenzyme. Our own preliminary laboratory dataindicate that the nuclear activity is in somemanner related to the rate of RNA synthesiswithin the nucleus, whereas the cytoplasmicenzyme is involved in the regulation of RNAcatabolism in the cytoplasm. For example, ifone examines the activity of cytoplasmicRNase in three different adult tissues thatcatabolize RNA at three different rates, a directcorrelation between enzyme activity and rateof catabolism is immediately apparent. Themost rapid rate of RNA degradation occurs inkidney, with liver and brain following indescending order. The highest activity ofRNase per cell.also occurs in kidney, with liverand brain, respectively, following. Moreover,the differences in enzyme activity are directlyproportional to the differences in catabolic rate.

These data have convinced us that alkalineRNase must play a role in the regulation ofRNA metabolism and, indirectly, protein syn-thesis by influencing the rate of RNA catabo-lism. We have accordingly 'studied the effectof such malnutrition on the activity of alkalineRNase since, as previously pointed out, RNAcatabolism is increased in brains of animalsexposed to early malnutrition. Malnutritionimposed at birth will elevate the activity ofthis enzyme in rat brain, and this elevationprogressively increases as the duration of mal-nutrition increases. Although these data areexpressed per milligram of DNA per cell, thesame findings are present if the data are ex-pressed per milligram of protein or per milli-gram of RNA. Neonatal malnutrition thusselectively elevates the activity of this enzyniein the face of a decrease in both overall proteinsynthesis and a fall in tissue RNA content.

At this point we can summarize what isknown about the effects of early malnutrition'nit cellular growth of the brain and presentour hypothesis of the effects of malnutrition

on the regulatory mechanisms involved in thecontrol of cellular growth. Malnutrition dur-ing proliferative growth will curtail net pro-tein, RNA, and DNA synthesis and result in anorgan with a reduced number of cells. It wouldappear that any cell type in any region wherecells are dividing is vulnerable to the effects ofnutritional deprivation. Early malnutritionlimits the availability of amino acids for incor-poration into protein, thereby reducing therate of protein synthesis. In contrast, an in-creased flow of amino acids enters the nucleo-tide pool, which presumably limits even furtherthe availability of these amino acids for proteinsynthesis. Nucleotides are removed from thepool much more slowly along the pathway toDNA synthesis, since the synthesis of DNA isoccurring at a, much slower rate than normal.

This reduced rate of INA synthesis is atleast in part a consequence of the reducedactivity of certain enzymes involved in thatsynthesis, for example, DNA polymerase In

contrast, nucleotides are removed much morequickly along the pathway to RNA synthesis,since synthesis of RNA is more rapid than nor-mal. This results in either a depletion in thesize of the nucleotide pool, as occurs in liver,or a maintenance of normal amounts, as occursin brain. Regardless of the effect on pool size,however, the distribution from the pool hasbeen altered by early undernutrition. The in-crease in RNA synthesis may direct alterationsin protein synthesis that result in selectiveelevations in certain proteins, for example,RNA polymerase and alkaline RNase. The in-creased RNA polymerase activity may furtherstimulate the enhanced RNA synthesis, whereasthe increased activity of alkaline RNase mayinitiate the increased rate of RNA degradationthat has been demonstrated. This increasedcatabolic phase may in turn be responsible forthe shift in polysomc pattern described byMunro, and through this mechanism may con-tribute even further to the selective reductionin protein synthesis.

28

In addition to the opening of new areas ofresearch aimed at further exploring the mecha-nisms by which early malnutrition exerts its

effects on cellular growth, the data just re-

viewed have opened two avenues of researchfor reexamination. The entire problem of pre-natal malnutrition can be reexamined usingthese more sensitive markers of tissue nutri-tional status. This approach seems warranted,especially in view of the less marked quantita-tive changes in weight, and protein, RNA, andDNA content that have been reported in certainkinds of intrauterine malnutrition. Second,some of these tissue changes themselves andthe reflection of these changes in body fluidscould provide a clinically sensitive index ofnutritional status and recovery from malnu-trition.

Let us address ourselves to the first problem,"intrauterine malnutrition." If one examinesall the available data from all species of experi-mental animals studied, it becomes quite clearthat at least two types of intrauterine malnu-trition exist. One type produces obviouschanges in cellular growth of the fetal brain,whereas the other produces little if any changein fetal brain when measurements of weight,and protein, RNA, and DNA content are used.

As we shall sec, however, more subtle changesas measured by some of the parameters just de-scribed do occur. These two types of fetal mal-nutrition do not, as far as we can tell at present,depend on the species employed but rather onthe method by which the malnutrition is in-duced. For example, "placental insufficiency"will result in retarded fetal growth. This con-dition has been produced by ligating the uterineartery in rats and more recently by ablating aportion of the placenta in monkeys. In the ratsuch a ligation produces a reduction in weight,and protein, RNA, and DNA content in placentaand most fetal tissues. Brain, however, is un-affected. These data have been substantiatedby Minkowski and colleagues, and more re-cently by Oh. A similar situation apparently

occurs in the monkey following ablation of aportion of the placenta. Cheek and coworkershave demonstrated, using such a model, thatalthough profound fetal growth failure will

occur, brain weight and protein and DNA con-tent are unaffected, and RNA content is re-duced only about 5 per cent in cerebellum.This relatively small reduction may becomemore significant when considered together withmore recent data about other tissue parameters.Thus, both in the rat and in the monkey thistype of placental insufficiency results in a

markedly disproportionate type of growth fail-ure in which brain is relatively spa'red andorgans such as liver markedly affected.

If we now reexamine this model using our"newer markers" of tissue nutritional status,certain changes not previously seen becomeobvious. Within 24 hours after ligation a

marked elevation of placental RNase activitycan be demonstrated. Enzyme activity in-creases in inverse proportion to the distancefrom the ligation. There is a marked elevationin the proximal placenta, a milder elevation inintermediately located placentas, and minimalchanges in those located distally. If we limitour observations to the proximal placenta, alka-line RNase activity becomes elevated within 24hours and remains elevated for at least 96

hours. This increase in activity precedes anychanges in placental weight, or protein, DNA,or RNA content. In the fetal organs, there is amarked increase in liver RNase within 24 hours,with brain unaffected. By 48 hours, however,both brain and liver RNase.are distinctly ele-vated. Thus, by examining activity of alkalineRNase, changes in brain can be demonstratedafter 48 hours of ligation. in the absence of anyreduction in weight, or protein, DNA, or RNAcontent.

These data become even more interestingwhen coupled with the observations of Cheekand coworkers in the monkey. Although thoseinvestigators did not measure alkaline RNaseactivity, their results demonstrating a slight

29

reduction in cerebellar RNA content may bemore meaningful in this context. At present,then, the available data indicate that this typeof placental insufficiency will produce certainchanges in RNA metabolism of fetal brain with-out permanently affecting cellular growth asmeasured by brain weight, protein, or nucleicacid content. What the significance, if any, ofthis increased enzyme activity is remains to bedetermined.

In contrast to the placental insufficiencymodel, maternal protein restriction in rats, pigs,and guinea pigs produces a symmetrical reduc-tion in weight, and protein, RNA, and DNAcontent of all organs including brain. Brainis affected by the 16th day of gestation, andall areas examined, including white and graymatter, demonstrate a marked reduction in

the number of dividing cells. By birth thereis an approximately 15 per cent reduction incell number when this model is used. Thisreduction agrees with the findings of Zamen-hof and colleagues, and is roughly of similarmagnitude to the reduction in cell numberfound in the other fetal organs. Thus thistype of fetal malnutrition is quite differentfrom the placental insufficiency type. Here wesee symmetrical growth retardation in whichhe brain is not spared. Combining maternal

restriction with postnatal restriction, we canproduce even more marked effects on cellulargrowth of the developing brain. Although ineither prenata1 or postnatal malnutrition thereis an approximately 15 per cent reduction inbrain cell number, the combination of bothproduces a 60 per cent reduction.

If we now turn to measurement of our morerecently described markers of tissue nutritionalstatus, we can demonstrate first that activityof DNA poIymerase parallels the rate of celldivision during normal placental growth andthat this type of maternal malnutrition willresult in reduced activity of this enzyme inplacenta by 12 days of gestation. Moreover,

alkaline RNase activity is markedly elevated in

such placentas. Changes in other fetal organsincluding brain are currently being investi-gated. Activity of both DNA polymerase andalkaline itNase has thus proved a .sensitive, indexof nutritional statusat least in placentainrats exposed to maternal protein restriction.

Flow can these changes provide us withclinically useful tools for the assessment ofnutritional status? DNA polymerase activityhas recently been measured in human leuko-cytes and has been shown to increase markedlyin situations of abnormally rapid cell divisionsuch as leukemia. This increased activity more-over precedes by several days any increase inleukocyte cell number and has been used topredict recurrence of disease after drug - inducedremission. The possibility exists that enzymeactivity may be reduced in leukocytes of mal-nourished children and that activity might in-crease with therapy. We intend to explore thispossibility. In addition, since it has been shownthat DNA polymerase activity parallels the rateof cell division in tissues other than brain,activity in a muscle biopsy could give us a

quantitative measure of the rate of cell divisionin muscle of malnourished children. This is

only speculation, but we intend to investigatethe possibility. Finally, since activity of thisenzyme is reduced in placenta in experimentalmalnutrition, we intend to measure its activityin placentas of malnourished women. To thisend we have worked out the requirements andmeasured activity in normal human placenta.

As previously mentioned, Metcoff and co-workers are actively examining leukocytes andplacenta for RNA polymerase activity in popu-lations of malnourished women. We may hopetheir investigations will also sharpen our toolsfor assessing nutritional status.

We have recently focused our attention onalkaline RNase as a clinically useful marker ofnutritional status. Examination of placentasof a group of malnourished mothers in Quito,Ecuador, reveals an elevated RNase activitywhen they are compared to placentas from

30

normally nourished women in the same city.This enzyme is present not only in all tissues

but in normal plasma and urine, and we havemeasured it in normal amniotic fluid. Obvi-ously changes in the activity of this enzymein any of these fluids would provide a practicalapproach to the monitoring of either postnatalor prenatal nutritional status. Plasma RNaseexists almost entirely in the free form, andduring development the activity of this enzymeper milliliter of plasma decreases. There wasa significant increase in enzyme activity in

plasma from 14 malnourished children whencompared to that from age-matched controls.Moreover, plasma activity returned to normalin all cases after only two weeks of therapy.Sometimes this return to normal preceded anynoticeable weight gain in these children. By

contrast, activity of alkaline RN3SC in the urineof marasmic children is reduced, indicating areduced clearance of the enzyme. Therefore,when the urine and plasma data are combinedand expressed as a u/p ratio, the most markedchanges are noted. These data demonstratethat the activity of alkaline RNase in plasmaand urine of marasmic infants is a sensitiveindex of nutritional status. The data haveencouraged us to investigate this enzyme inless severe malnutrition and in amniotic fluidduring prenatal undernutrition, both in ouranimal models and in human populations.

In an attempt to uncover other useful bio-chemical indices of fetal growth by monitoringchanges in amniotic fluid, we have examinedthe concentrations of urea in one of our animalmodels and nondialyzable hydroxyproline in

the human.It has been shown that urea excretion can

be used as a rough measure of the rate of pro-tein synthesis and degradation. Studies in

Jamaica and Africa have indicated that ureaexcretion in severely malnourished children isreduced. Since amniotic fluid is composed

largely of fetal urine, we have studied ureaconcentration in rat amniotic fluid during nor-

mal development and after uterine artery liga-tion. The data indicate an increase in ureaconcentration during normal development anda marked reduction following uterine arteryclamping in fetuses showing 25 per cent orgreater reduction in body weight. Althoughthese data are still in the early stages of collec-tion, we intend to pursue these findings bothin rats and in the human.

Urinary hydroxyproline has been shown tobe elevated under conditions of rapid collagensynthesis. More recently it has been demon-strated that under these conditions, it is thelarge molecular weight nondialyzable fraction(1.MN Ftypro) that increases. We have mea-sured t.N1N tlypro concentration and contentin human amniotic fluid at various stages ofpregnancy. We have found that the concen-tration of LMN Hypro remains relatively con-stant during the first 28 weeks of gestationand then decreases. This drop in concentrationis not due to any dilutional effect, since amni-otic fluid volume increases more slowly and insome cases even decreases toward the end ofpregnancy. At present the best workinghypothesis seems to be that at around 28 weeksfetal swallowing occurs, LMN Hypro is de-graded, and smaller peptides are either utilizedor reexcreted. The smaller peptides then passinto the maternal circulation and are excretedin maternal urine. We have recently been ableto document an increase in total hydroxypro-line in maternal urine during the last tri-mester. We are now planning to fractionatethe urine and see if it is a small polypeptidemolecule that increases. Again, although onlyin the early stages of investigation, determina-tion of LMN Hypro in amniotic fluid mightprovide a means for monitoring early fetalgrowth, and determination of small hydroxy-proline polypeptides in maternal urine duringthe last trimester might be a way to monitorlate fetal growth.

Finally, the previously described evidencein animals shows that maternal protein restric-

31

Lion will affect fetal growth and brain devel-opment. Data in man are less clear and mostlyindirect. The available data do, however, indi-cate that this may occur. For example, ex-amination of brains of children who died ofmalnutrition reveals three distinct groups.Children who died of kwashiorkor after 18

months of age had normal brain oNA contents,the protein /DNA ratio being reduced. By con-trast, children who died of marasmus duringthe first year of life could be divided into twogroups. In the first there was a 15 per centreduction in cell number, but in the secondthe reduction was 60 per cent. The differencebetween these groups was that in the formerall the infants weighed more than 2,500 g atbirth while in the latter they were all below2,000 g at birth. These data indicate thateither they were true prematures and the pre-mature infant is more susceptible to postnatalmalnutrition, or that they were infants whowere already malnourished in utero and repre-sent the clinical counterpart of the doubly de-prived rat previously described. More recentlyChase has reported reduced DNA' content inbrains of infants who were "fetally malnour-ished" and who died shortly after birth.Although the data are extremely limited, wemust continue to investigate the possibilitythat fetal malnutrition may affect brain growthin the human. To this end the techniquesdescribed above for monitoring human fetalgrowth may prove exceedingly useful.

In summary, during the past five or six yearsa number of descriptive observations on growthof various organs have been made using mea-surements of weight, and protein, RNA, andDNA content. In rat brain, using these measure-ments, it has been shown that early malnutri-tion will retard the rate of cell division in anyregion undergoing proliferative growth andresult in a permanent deficit in cell number.Any cell type dividing will be affected. In thehuman, a few studies have led to similar find-ings. More recently, experiments have been

undertaken that are designed to explore themechanisms by which these changes in cellulargrowth occur. These studies, while still in

progress, have already revealed that certainalterations in protein synthesis, distribution ofamino acids and nucleotides, RNA synthesis,RNA degradation, and uNA synthesis take place.The activity of enzymes controlling these proc-esses is also altered. Specifically, DNA polymer-ase activity is reduced and alkaline RNase

activity elevated. These tissue changes pro-duced by malnutrition have allowed us to

explore more deeply the effect of other typesof nutritional deprivation. Two distinct typesof fetal malnutrition can be described. One isa placental insufficiency that produces onlysubtle changes in RNA metabolism in the fetalbrain, as demonstrated by a delayed increasein alkaline RNase activity, and the other is dueto maternal protein deficiency that producesall the changes in fetal brain described withpostnatal malnutrition.

These tissue changes have also supplied clini-cal markers for determining nutritional status

32

and response to therapy. Alkaline RNase is

thus elevated in placentas of undernourishedwomen and in serum of undernourished chil-dren. Moreover, it It decreased in urine of thesechildren, causing a marked reduction in theu 'p ratio. With therapy it rapidly returns tonormal.

Urea and Lxi N tiypro concentration in

amniotic fluid and smaller peptide hydroxy-proline concentration in maternal urine are

also being investigated as possible markers forfetal growth and nutritional status.

The problems we must yet solve are enor-mous. Do the changes described here have anyfunctional significance? Does prenatal mal-nutrition in the human affect the fetal brain?If so, what kind of malnutrition? How arethese effects manifested? What are their sig-nificance? Perhaps we do not yet have themeans to answer these questions, but the datapresented at least demonstrate that we aresharpening our tools, refilling our measure-ments, and applying the new findings to someof these perplexing problems.

SMALL-FOR-DATES OFFSPRING: AN ANIMAL MODEL

R. J. C. Stewart'

It is generally accepted that the earlier inpostnatal life a dietary deficiency can be estab-lished, the more severe are its effects. This isparticularly true with respect to protein-caloricdeficiency and it seemed probable that a pre-natal deficiency might produce very severe

changes.

Intrauterine malnutrition was produced indogs whose mothers were fed a protein-deficientdiet (NDpCal,4 =-- 6.8) from weaning (12).The dietary regimen caused no changes in ges-tation times, no difficulties in parturition, andno significant increase, in the number of still-born, but mean birth weights were 20 per centlower than those of controls. This was not,however, a consistent reduction, some animalsweighing as much as the heaviest controls(> 350 g) and others being markedly small-for-dates (si,D) and weighing as little as 140 g.When the congenitally malnourished pups

were suckled by their own mothers, there wasa high death rate (41 per cent) during suck-ling; after weaning they grew less rapidly thannormal pups, irrespective of the quality oftheir diet (Figure 2, (12)). Those given dietsof low protein value (NDpCal') = 5.0)developed some incoordination, a proportionshowed athetoid movements of the head andlimbs, and some suffered severe and repeatedconvulsive seizures.

Department of Human Nutrition, London School ofHygiene and Tropical Medicine, London, England.

33

The congenitally malnourished dogs exhibitedalterations in the electrical activity of theirbrains (9), modifications of their carbohydratemetabolism (7), and changes in the weightand histologic appearance of the central nerv-ous system, endocrine glands, bones, and manyother tissues ( I I). Animals allowed to reachadulthood could always be differentiated fromwell-fed controls by their short legs, unusualbehavior, and tendency to obesity.

It was clearly important to determinewhether members of a colony that had beenmaintained for several generations on the de-ficient regimen could adapt their metabolismand regain the original size, stabilize at a

smaller body size, or progressively deteriorate.Owing to a shortage of accommodation thework could not be continued with dogs and,as it was known that restrictions in the quality(3, 4, 5) or quantity (2) of a rat's diet led tometabolic, physical, and behavioral alterationsin her offspring, the tests were continued withrats.

From the results of the studies mentionedabove and other work (1, 13, 14, 16) we

concluded that a more severe dietary restric-tion would be necessary in rats than in dogs,and for the first experiment reduced the pro-tein value of the rats' diet to NDpCalV, = 5.0.In the first generation there was a large numberof SFD offspring, the mean litter weight fell by21 per cent (p < 0.01), and the animals grewless rapidly than the controls both before and

after weaning (I 5). But there was a veryhigh loss during suckling (63 per cent), manyof the survivors failed to rear a litter, and thefeeding of a powdered diet proved wasteful inboth materials and labor.

Efforts were then directed to the productionof diets with higher protein values (NDpCall,

6.8 and 10.0) which could be "cubed" (10).Colonies have now been maintained on cubed

diets for eight generations. The reduction inthe mean birth weights of the offspring of mar-ginally malnourished mothers has been con-firmed, but there is no statistical evidence ofany change in the average values from genera-tion to generation (Table 1). The number ofyoung per litter has risen slightly in the well-fed and declined in the underfed colony, sothat the average weight per litter is markedlydifferent. Birth weights vary widely in bothcolonies-from 3.5 g to 7.0 g in the controland 2.8 g to 6.3 g in the deficient group-butin the latter colony 75 per cent of the seventhgeneration were born weighing less than 5.0 g.SFD offspring have h:_en defined in these experi-ments as term ms that had birth weights morethan 2 SD hzlow the mean of the well-fedcolony. The mean birth weight of pups fromthe control mothers was 5.6 g and those bornat less than 4.5 g are considered SFD. The

proportion of SFD varied between 2 per cent inthe fourth and 4 per cent in the seventhgeneration, with a mean of 3 per cent for the

Table 1. Birth weights of rats maintained on diets ofdifferent protein values.

Diet 13NDpCaI% = 10.0

Mean and SD

Diet ANDpCaI% = 6.8

Mcan and SD

Generation 0 5.6 ± 0.46 5.6 ± 0.46Generation 1 5.3 ± 0.39 4.9 ± 0.56Generation 2 5.5 ± 0.36 4.6 ± 0.45Generation 3 5.4 ± 0.58 5.0 ± 0.48Generation 4 5.6 ± 0.46 4.8 ± 0.65Generation 5 5.6 ± 0.60 4.7 ± 0.6YGeneration 6 5.6 ± 0.48 4.8 ± 0.60Generation 7 5.5 ± 0.58 4.3 ± 0.75Generations 0-7 5.57 ± 0.51

34

whole colony. As the colonies were developedfrom litter mates, tliere are minimal **geneticor race differences," and the sun criterion ofless than 4.5 g in the normal group can alsobe used to define st o in the malnourishedcolony. In the latter the proportion of suchyoung is much higher, varying from 11 per

cent in the third to 43 per cent in the seventhgeneration, with a mean of 28 per cent. Both

before and after weaning the congenitally mal-nourished offspring grow and develop slowly,so that at four weeks of age the young haveonly 50 per cent of the weight of the age-matched controls. Hair growth is obviouslyaffected, and at 10 days of age some pups arepractically bald. On the other hand, eye open-ing does not appear to be delayed. Weightdifferences between the sexes appear later thanexpected, and in the latest (eighth) genrationsom delay in vaginal opening times is sus-

pected (only a small number of animals arenow available).

Much of the slow growth in early life maybe attributed to the poor mammary develop-ment of the mothers (see Plate 11, (1 1)) , anda deficient supply of milk undoubtedly con-tributes to the high preweaning losses (45per cent). Some of the losses are less easilyexplained. Occasionally what appear to be

reasonably efficient mothers will scatter theirnests around the cage, kill the young, and,although the walls and floor are spattered withblood, not cat them. Can this be an episodesimilar to the convulsive seizures observed inthe dogs? Convulsive seizures have never beenseen in these rats. Behavioral changes are re-stricted to a slightly enhanced activity withoccasional tremors in early life, followed bya prolonged sluggish period during which theanimal reacts more violently than normally tounexpected noise or mild\ electric shock.

Preweaning food restriction is not the onlyfactor in poor growth. Animals of the fifthgeneration fostered at birth to normal mothersplateau at weights below those of the control

colony, although during early life they appearto grow normally.

It must be reiterated that, as with the dogs,the adequacy of the poor diet varies with thephysiologietilate of the recipient, being defi-cient during early growth, gestation, and lac-tation but adequate for adult maintenance. Itis not surprising, t herefore, that the very highpercentage deficits observed in body weightsat four week' of age became proportionatelyless by six months (Table 2 ) . The effectsof protein-calorie deficiency on other values-body, head, and femur lengths-are also shownin Table 2. Of the organs weighed-liver, kid-

ney, endocrine glands, and brain-the last wasthe least affected. Tables 3 and 4 show theweights of the whole and various parts atthree and six months of age in both male andfemale rats. The differences between the colo-nies are small, but just as averages of the wholecolony did not give a complete picture of birthweights, they do not reveal all the distribu-tional changes in brain size. Brain weightsvaried between 1.47 g and 2.02 g in the con-trol males but only 9 per cent were below1.60 g, whereas in the deficient males, althoughthe range was only slightly lower (1.36 g to1.99 g), 46 per cent were below 1.60 g. A

Table 2. Measurements" of six-month-old rats from colonies maintained on diets of different protein values.

DietNDpCal%

Means

B

= 10and SD

DietNDpCal%

Means

A= 6.8

and SD

9 8 9

Body weight (g) 270 ± 31.7 399 ± 72.6 191 ± 26.7 290 ± 35.5Body lengths' (mm) 190 ± 7.0 213 ± 12.1 175 ± 7.0 187 ± 6.7Head length (mm) 51.1 ± 1.9 52.9 ± 2.5 48.0 ± 2.3 50.1 ± 3.5Femur length (mm) 34.7 ± 0.61 39.3 ± 0.63 31.7 ± 1.7 34.7 ± 1.5Brain (mg) 1639 ± 159 1772 ± 178 1583 ± 153 1625 .± 128

" After formalin perfusion except for body weight.6 Excluding tail.

Table 3. Brain weights" of male rats from colonies maintained on diets of different protein values.

Diet BNDpCal% = 10.0

Means and SD

Diet ANDpCal% = 6.8Means and SD

3 months 6 months 3 months 6 months

Brain stem (mg) 163 ± 33 173 ± 21 127 ± 12 160 ± 27Cerebellum (mg) 254 ± 28 270 ± 30 207 ± 25 238 ± 24Forebrain (mg) 1285 ± 147 1323 ± 140 1124 ± 111 1229 ± 89Whole brains (mg) 1720 ± 185 1772 ± 178 1459 ± 135 1625 ± 128

" After formalin perfusion.

Table 4. Brain weights" of female rats from colonies maintained on diets of different protein values.

Diet BNDpCal% = 10.0

Means and SD

Diet ANDpCal% = 6.8Means and SD

3 months 6 months 3 months 6 months

Brain stem (mg) 138 ± 20 162 ± 18 122 ± 21 157 ± 31Cerebellum (mg) 229 ± 18 244 ± 23 195 ± 28 223 ± 23Forebrain (mg) 1154 ± 126 1233 ± 127 1058 ± 123 1202 ± 110Whole brains (mg) 1520 ± 159 1639 ± 159 1375 ± 163 1583 ± 153

" After formalin perfusion.

35

similar but less marked situation occurs in theslower growing females.

Clearly, many of these animals have brainsand other organs that, while small for theirgenetic background, are not small relative tobody size.

Is there any evidence that the changes de-scribed in the malnourished colony vary fromgeneration to generation? Total litter weightsare, and average birth weights may be declining.There is an increase in the number of verysmall individuals, and although no statisticsare available, there is a firm impression amongthose handling the rats that a rat of generation7 weighing 3 g has a better chance of survivalthan did one of similar weight in generation 1.

Table 5 attempts to summarize the availableinformation. It will be seen that all the valueslisted are lower for the sixth-generation malesthan for the total male population. If con-firmed, the alterations in brain weight in latergenerations will be of particular interest. Theaverage weight in well-fed males at six monthsof age was 1.77 g and there was no differencein the first malnourished generations. Afterfour generations of malnourishment, however,the weight fell to 1.66 g, in the fifth genera-tion to 1.60 g, and in the sixth to 1.52 g. Itmust be emphasized, though, that the brainsare still large relative to body weight (see per-centages in Table 5). When the brains weredivided (Tables 3 and 4), it became clearthat the cerebellum was the most consistentlyand severely affected. This is in keeping withthe histologic observations in rats aged one,three, and six months.

The granular and molecular layers of thecerebellum are reduced in size and the formerin density. The Purkinje cells are poor in

chromatin and there is an increased number ofBergmann cells. Areas of the forebrain andbrain stem are also modified, there appears tobe a reduced number of cells in the cerebralcortex, the Nissl granules are reduced in somemotor nuclei, and there is an increased number

36

Table 5. Measurements of protein-calorie-deficient rats,expressed as percentages of normal values (6 monthsof age).

Females

Generations1-6

Males

Generations1-6 6

Body weight 71 73 66Body length 95 S8 85

Head length 98 94 87Femur length 91 SS 84Brain weight 97 92 86

of active neuroglial cells. Similar but moremarked changes were seen in the brains of the15- and 21-day-old offspring of mothers fed alow-protein diet during gestation (8). Theseanimals also exhibited deficits in deoxyribo-nucleic acid and gangliosides, but the work ofWinick and Noble ( /7) and Dobbing (6)shows that some of this change must be

attributed to poor postnatal development. Pre-liminary tests (six brains only) indicate that inthe newborn of the sixth generation of themalnourished colony there may be a reductionin the arnoui.t but an increase in the concen-tration of deoxyribonucleic acid.

The important and outstanding questionsare whether or not the brains of adult animalsshow chemical changes that lead to the histo-logic modifications and if so, whether the modi-fications adversely affect behavior and learning?The simple answer is that we do not know.Tests, using animals of generations 8 and 9,are planned with Drs. Dickerson and Birch,and we hope these aspects will then becomeclear.

Our present experiments show that intra-uterine or congenital malnutrition occurs inanimal colonies reared on inadequate diets;that, after the first-generation, reductions inbody weight continue but are small, andthat there may be later adjustment in brainweight. Even after six generations the brains,though small for genetic background, are largerelative to body weight. While their qualitymay be suspect, there is no proof that they areabnormal.

REFERENCES

1. CHANDRAN, K., and S. D. AMISEGAOKAR. Physio-

logical effects of low protein diets. Ind Med Res47:539-62, 1959.

2. Qum, B. F., et al. Maternal nutrition and metabo-lism of the offspring: studies in rats and man. AmPublic Health 58:668-77, 1968.

3. COWLEY, J. J and R. D. GRIESEL. Some effectsof a low protein diet on a first filial generation of whiterats. / Genet Psychol 95: 187 -201, 1959.

4. and . The development of secondgeneration low protein rats. / Genet Psychol 103: 233-42, 1963.

5. and . Low protein diet and emo-tionality in the albino rat, I Genet Psychnl 104:89-98,1964.

6. DOMING, 101IN. Effects of experimental under-nutrition on development of the nervous system. In:N. S. Scrimshaw and J. E. Gordon (eds.) , Malnutrition,learning, and behavior, Cambridge, MasSachusetts, M.I.T.Press, 1968, pp. 181-201

7. HFARD, C. R. The effects of severe protein-caloriedeficiency on the endocrine control of carbohydrate

metabolism. Diabetes 15:78-89, 1966.8. MI:RAT, A. Effects of protein-calorie malnutrition

on brain gangliosides. Ph.D. thesis, University ofSurrey, England, 1971.

9. MEYER, A., et al. EEG and neuropathological

changes in dogs with experimental malnutrition. ExcerptaMed 1961. (International Congress Series, No. 39, Ab-stract 17.)

37

10. PAYNF, P. R., and R. J, STLWART. Cubed diets ofhigh and low protein values. Lab Arlin, 6:135-40,1972.

11. Pt ATT, B. S., et al. Experimental protein-calorie.deficiency. In: H. N. Munro and J. B. Allison (eds.),Mammalian protein metabolism. New York AcademicPress, 1964, vol. 2, pp. 445-521.

12, and R. J. STEWART. Effects of protein-calorie deficiency on dogs. I. Reproduction. growth andbehaviour. Det' Med Child Nenrol 10:3-24, 1968.

13. RA I Al.AKSI1X11, R., ct al. Effect of dietary proteincontent on visual discrimination learning and brain

biochemistry in the albino rat. / Nenrochem 12:261-71.1965.

14. ct al, Effect of inanition during the neo-natal period on discrimination, learning and brain bio-chemistry in the albino rat. / Nenrochem 14: 29-34,

1967.

15. STEwART, R. J., and H., G. Si EPPARD. Protein-

calorie deficiency in rats: growth and reproduction. Br

Nntr 25: 175 -80, 1971.

16. VliNKATACUIALANI. P. S., and K. S. RAMANATIIAN.

Severe protein deficiency during gestation in rats onbirth weight and growth of offspring. Ind l Med Res54:402-09, 1966.

17. WINICK, MYRON, and Ain i.e Nom.E. Cellular

response in rats during malnutrition at various ages.

Nntr 89 : 300-06, 1966.

SUMMARY OF SESSION I

Herbert G. Birch'

We have listened to four extremely interest-ing papers. The speakers all agree that thereare indeed changes in physical structure, in cellnumber, and in certain enzymatic characteris-tics when malnutrition occurs at particularpoints in the developmental course. They dis-agree about the specific age levels at whichorganisms may be most sensitive to the influ-ence of malnutrition and about the ubiquity ofthe phenomena for all species of organisms.

When Dr. Cheek began his paper andpresented a comparative view of mammaliansubspecies, he sought to argue that nutritionalinsults at particular points of development areimportant for small mammals, less importantfor larger mammals, and probably still less

important for subhuman primates and man.Though this was his thesis, he neither developedit with enough system to be convincing norsufficiently analyzed explicit problems andissues that must be considered if any cross-species comparisons are ever validly to be made.One problem that must always be dealt withacross species is the developmental rate thatcharacterizes a species and, in consequence,the durations of insult that would be neededto produce equivalent insults. The secondpoint that was not systematically consideredis the degree to which the ages at which insultsoccur correspond in fact as homologous pointsin the developmental courses of the targetorganin this case, the brain. For example,one has great difficulty in comparing anddefining equivalent developmental time pointsin precocial species such as the guinea pig ascontrasted to altricial species such as the rat.Given gross differences in developmental cal-

' Department of Pediatrics, Albert Einstein College ofMedicine, Bronx, New York, USA.

38

endars, it is essential in interspecies compari-sons to define the equivalent time points forcomparisons of effects.

Whether primates and particularly man areimmune from the effects of malnutrition can-not be concluded from the data presented.The models used are limited and produce par-ticular pieces of evidence that, though inter-esting to consider as special questions, do notcontribute firm evidence that other methodsof inducing fetal malnutrition would be in-effective in limiting growth and development,e.g., ligation versus maternal malnutrition.Though this finding opens up enormously im-portant questions as to the types of change inmaternal-fetal transport that result under thesedifferent conditions of interference, one clearlycannot accept the particular procedure of liga-tion as the experimental analogue of develop-mental inadequacy. These issues resulted in aconsiderable amount of parallel play among thevarious speakers as to whether their findingsagreed with the particular picture that hadbeen presented. No real agreement was

apparent.Another set of issues raised was about a

point of method. There are indeed very goodreasons for wanting to know what the specificbreak points in a developmental course arc.

For example, if one were dealing with the prob-lems of prevention, one would be concernedwith the particular points at which vulner-ability existed in any individual's developmentalcourse. All the speakers agreed that there areindeed such critical points in development.What they disagreed about is where those

points arc.It was argued that we know that neuronal

replication is completed in various organismsbefore the peak of glial replication occurs. As

a consequence, one would hope that it wouldsomehow be possible to define in developmenta number of kinds of points in time when neu-ronal replication is at its peak and other pointsin time when glial replication would occur. Itwould be delightful if these phenomena did notoverlap. From the point of view of analyzingbiologic systems, they unfortunately do. Ifthese phenomena did not overlap, we wouldhave alternating peaks and plateaus, and no onewould be fighting about anything. What wehave, in fact, is a pattern of glial replicationproceeding at a relatively low rate and increas-ing in its replication rate in the later period,neuronal replication proceeding at a more rapidrate in the earlier period, and intermediateperiods reflecting some kind of mix. The ques-tion then becomes, How can we sort out thiskind of mix in relation to the problem?

Considerable discussion took place in whichDr. Dobbing gave evidence to support the hy-potheses that have guided his work most pro-ductively for many years: permanent changesoccur in brain because of nutritional restrictionduring appropriate periods of brain growth,and these changes in brain are not unique tobrain itself since alteration of brain structurereflects general restriction of growth. Heargued that no basic species differences exist ifrelevant time dimensions are considered, andthat if one applies insults appropriately withrespect to rates and course of development,common outcomes are to be anticipated in anyspecies under consideration. He has furtherasked us to consider two different kinds ofmodel. One may be viewed as a crisis modelin which extremely severe insults are appliedto organisms for very brief periods. This iscontrasted to mild and chronic models of in-sults that may be applied over longer periods.

A number of variations of these models havebeen discussed, any of which could have dif-ferent consequences for development. Mr.

Stewart, for example, has used a quasigenetic,intergenerational chronic model to study the

39

cumulative consequences of adaptation to mar-ginally adequate nutritional inputs. Dr. Dobb-ing has described modest restrictions for periodsin which particular features of growth aretaking place. In some of his studies Dr. Winickwas concerned with acute nutritional insult; inother studies, with insults similar to Dr. Dobb-ing's, and in still others at the human levelin particular, with chronic conditions followedby acute, severe exacerbation. Whether allthese conditions should result in identical out-comes is, in my opinion. is. I think thatif we begin to examine .)blems in termsof the actual conditions .,,,ultits timing,severity, and durationwe cannot treat all

kinds of malnutrition as alike because the sameword, "malnutrition," is used to describe a

variety of conditions. We may indeed be ableto define specific conditions that have particu-lar consequences for the development of orga-nism. Within this framework, then, it seemsto me that Dr. Winick has turned our atten-tion in one direction and Dr. Dobbing inanother. Dr. Dobbing has suggested that thenumber of cells that are in brain may not bethe critical factor in defining the functionalintegrity of the system. He and others haveargued that such things as dendritic prolifera-tion, normal branching, the development ofsynaptic organization, and the properties offiring in the brain all make particularly impor-tant contributions to integrated organization.

Lashley (1) and others showed that manyfunctions may still be completely unimpairedas a consequence of the destruction of braintissue, amounting to as much as 20 per centof cells in animal brains. This is not true ofall functions, but for many general functionsof adaptation and learning it certainly is the

case. If there are functional consequences forbehavior that are a consequence of malnutri-tion, I would be quite doubtful about the like-lihood of 10 to 15 per cent cell reduction beingthe mechanism for the production of behavioralalteration. This is not true when there are 50

and 60 per cent reductions such as those Dr.Winick and others have described.

It was argued that the Lashley evidence wasirrelevant because ablation is quite differentfrom growth failure or disturbance in braincaused by malnutrition. I would agree withDr. Dobbing that what we need to know ismore about the functional attributes of thenervous system, more about its refined struc-ture, more about its intimate organization.His plea for the expansion of investigation ofmore complex integrative systems and inter-cellular organization is very well put, I think.

Dr. Winick has turned our attention in

exactly the opposite direction. He wants tolearn more and more about underlying mecha-nisms. What he is saying is that we can eithermove from the phenomenon of reduced cellnumber to a consideration of its consequences,or to a consideration of the mechanisms thatproduce fewer cells. He has directed his atten-tion to the latter, and has explored a variety ofenzymatic systems from particular features ofprotein synthesis to the relation of alteredamino acid pools to such synthetic and degra-dation processes. His attention is therefore di-rected not at the consequences of malnutritionbut at the subcellular-level mechanisms thatmay contribute to the organization of cellularsystem. This is important in and of itself. Buthe has a social conscience and therefore tries toextend its importance elsewhere, including diag-nosis and the potential usefulness of particularproportion levels among enzymes as definers ofnutritional status. I can only wish him well.At present I do not think that he has pre-sented a convincing argument that they willbe such useful indicators in defining nutri-tional status. We need such better indicators:

there is every reason to want to be able todefine the nutritional status of populations andindividuals who are under stress. This may bea step in that direction, but I think only thefuture will tell us whether it is.

Dr. Stewart presented a most fascinatingmodel of animal growth which to a very con-siderable extent parallels certain features of thehuman condition. He reports on generations ofrats living at marginal subsistence levels, andhis work makes it possible to analyze the inter-generational consequences of subsistence under-nutrition. He has produced two populations ofrats derived from a single stock, the first nor-mal and the second the result of generationsof undernutrition. These populations providenumerous exciting research opportunities. Wecan now explore the behavioral consequencesof such different patterns of development in agenetically homogenous species. Since he doeshave selective survivals in his litters of orga-nisms under stress, however, he must determinethe extent to which genetically selective mecha-nisms have been operating. To do so he mustcarry out rehabilitation studies over severalgenerations to define the degree to which themalnourished survivors are representative of theparental strain. That would also provide usefulinformation that would permit us to determinewhether the animals will return to previouslevels of growth and function after severalgenerations of improved nutrition. Compara-tive work with such recovered groups of ani-mals would also help to define whether geneticheterogeneity had in fact been introduced as aconsequence of selective survival. But themodels themselves are engrossing. He can onlybe congratulated for his patience and skill indeveloping them.

REFERENCE

1. LASHLEY, K. S. Brain mechanisms and intelligence.Chicago, University of Chicago Press, 1929.

40

Session II

FUNCTIONING OF CHILDREN MALNOURISHED IN INFANCY

OR FROM DISADVANTAGED SOCIAL BACKGROUNDS

Chairman

h 'Ilael Beaubrun

Rapporteur

Jack Tizard

EMPIRICAL FINDINGS WITH METHODOLOGIC IMPLICATIONS IN THE

STUDY OF MALNUTRITION AND MENTAL DEVELOPMENT'

Robert E. Klein, Jean-Pierre Habicht, Charles Yarbrough, Stephen G. Sellers, and

Martha Julia Sellers'

The study of the effects of protein-caloriemalnutrition on intellectual development is a

chronicle of investigators' attempts to estimatethe effect of malnutrition per se, apart from thevariety of other biologic and social factors thatare also known to influence intellectual devel-opment. Our report describes a continuation ofthis line of research and presents data on psy-chologic test performance with independentestimates of variance contributions from socialand biologic factors. Data recently collectedin a longitudinal study of children sufferingfrom mild to moderate protein-calorie malnu-trition are presented and discussed.

MethodSubjects

The subjects of this study were childrenthree through six years of age living in isolated,

' This research was supported by Contract PH43-65640 from the National Institute of Child Health andHuman Development, National Institutes of Health,Bethesda, Maryland, USA.

2 Institute of Nutrition of Central America and Pan-ama, Guatemala City, Guatemala.

rural, subsistence-farming communities in east-ern Guatemala. The numbers of subjects in-cluded in the analysis vary slightly with eachpsychologic test and are reported in Tables I

through 4. The subjects are part of a con-tinuing longitudinal study of the relationamong malnutrition, physical growth, andcognitive development,

Variables measured

Three types of data were collected and arereported here: indices of physical growth,sociocultural information on the family andthe environment, and psychologic test per-formance. The physical growth and psycho-logic test performance data were collected onthe child's birthday plus or minus 15 days. Thesociocultural characteristics of the family weremeasured over two years and represent a com-posite picture of the family during that period.

Physical growth. Two measures of physicalgrowth are included in the present analysis,total height and head circumference. These

Table 1. Multiple correlations between height, head circumference, and psychologic variables.

Boys Girls

Variable N r HeightHead

circum. HeightHead

circum.

Verbal Analogies 138 .19 .08 .11 131 .18 .06 .12Memory for Designs 51 .38 .00 .35 63 .40 .30 .14Reversal Learning:Trials to Criterion 42 .20 .13 .11 37 .30 .03 ,26Response Time 42 .42 .10 .30 37 .26 .12 .17

43

Table 2. Multiple correlations between six social-classindices and psychologic variables.

Boys Girls

Variable N N r

Verbal Analogies 138 .25 131 .37Memory for Designs 51 .36 63 .43Reversal Learning:Trials to Criterion 42 .55 37 .35Response Time 42 .70 37 .42

household tasks, and accompanying him to anearby town) ; (3) occupation (father's occu-pation, scored on a two-point scale) ; (4)mother's clothing (regular use of a sweater andshoes); (5) mother's hygienic practices (f!quency with which the mother washeshands and face), and ( 6 ) sociability (numi,of visits both made and received by the family

Table 3. Multiple correlations between height, head circumference, six social-class indices, and psychologic variables.

Variable

Boys Girls

6SESN

HeadHeight circum.

6SES Height

Headcircum.

Verbal Analogies 138 .28 .17 .15 131 .37 .04 .36(.08) (.11) (.25) (.06) (.12) (.37)

Memory for Designs' 51 .43 .35 .32 63 .52 .27 .25(.35) (.36) (.30) (.14) (.43)

Reversal Learning:Trials to Criterion 42 .54 .54 37 .36 .21

(.13) (.11) (.55) (.03) (.26) (.35)Response Time 42 .73 .07 .72 37 .42 .25 .28

(.10) (.30) (.70) (.12) (.17) (.42)

measures were chosen because they are fre-quently reported in the literature as indices ofnutritional history.

Sociocultural family status. Six socioculturalindices were derived from a body of interviewand observational data collected about thefamily of each subject. These indices reflectthe material status, social interaction, andreported child-rearing practices and parentalbehavior. The specific indices and their com-ponent elements are: (1) housing (quality ofthe kitchen, bedroom, and dining area, numberof rooms in the house, and the materials fromwhich the roof and walls are constructed); (2)teaching (interaction of the mother, father, andsiblings with the child in three activities:teaching the child to count, teaching him to do

Table 4. Projected increases in growth due tosupplementation.

Projected increase Age-sex-specificVariable at 7 years of age standard deviation

Height 5.0 cm 4.0 cmHead circum-

ference 0.9 cm 1.3 cm

44

during the previous year). Except for thefather's occupation, the indices were con-structed by bimodal coding of the elementsand summing.

Psychologic tests. The psychologic tests re-ported here are part of a large psychologic testbattery currently being applied in the longi-tudinal study of malnutrition and mental de-velopment. The four psychologic variablesemployed were chosen because they are com-monly used tests and measure different aspectsof psychologic functioning. They were VerbalAnalogies, Memory for Designs, ReversalLearning, and the subject's response time in theReversal Learning task. The first was a simpleverbal analogies task cast in the form of a sen-tence-completion problem. In this 15-itemtest the examiner would read a sentence lack-ing the final word and the subject had to com-plete the sentence (e.g., "The skin on thepineapple is rough; the skin on the banana is

") . This task was given to childrenfrom three through six years. In the Memoryfor Designs task the child had to reconstruct

a design from memory with multicoloredblocks after having inspected the design forfive seconds. Three consecutive trials weregiven on each design and scoring was by thenumber of blocks correctly placed with respectto the original design. This task was used withchildren five and six years. The ReversalLearning test was a two-choice, shape-discrimi-nation learning task employing four blocks of20 trials each. The child was rewarded forchoosing one shape over another for blocks I

and 2. In blocks 3 and 4, the other shape wasarbitrarily designated "correct" and the childhad to learn to reverse the previously madediscrimination. Data on blocks 3 and 4 follow-ing the reversal of the discrimination are pre-sented here.

Data analysis

The psychologic and physical growth vari-ables were normalized within age-sex groups,and for those variables in which more than oneage was represented, ages were combined andsexes were kept separate for all analysis.

The sociocultural data were treated in raw-score form and all indices were the combina-tions of simple raw scores. In all cases, theindices characterize the subject's family ratherthan himself.

For the multiple correlations involving bothmeasures of attained .size and social-class in-dices, only those variables that were statisticallysignificant and accounted for 2 per cent ormore of the variance in psychologic test per-formance were included.

Results

Table I presents the multiple correlationsbetween total height, head circumference, andthe four psychologic variables under considera-tion. For Verbal Analogies, the multiple corre-lations are essentially similar for boys and girls.Moreover, approximately equal amounts of thevariance are contributed by height and headcircumference, respectively, for both boys and

45

girls. For the Memory for Designs test, thetotal multiple correlations between perform-ance on this task and the two measures ofphysical growth are also essentially similar forboys and for girls, but there are differences inthe amount of variance contributed by heightand head .circumference, respectively. Heightcontributes none of the variance to perform-ance of the boys, whereas for girls it predictsperformance more strongly than does head

circumference. In the case of Reversal Learn-ing, trials to criterion arc better predicted forgirls than for boys. On this measure, headcircumference and height contribute aboutequally for boys, whereas for girls head cir-cumference predicts performance better thandoes height. For response time on ReversalLearning, boys' performance is better predictedby the two physical growth measures than isgirls', and head circumference predicts more ofthis performance for boys than does height.In contrast, both height and head circumfer-ence contribute about equally for girls.

The multiple correlations between the fourpsychologic variables and six social-class in-dices are presented in Table 2. Looking firstat Verbal Analogies, we find girls' performanceis better predicted than boys' by these indices.The same is true for Memory for Designs. Forthe Reversal Learning task, however, the tend-ency for girls' performance to be better pre-dicted by social-class variables is reversed. Inthe last task both trials to criterion and re-sponse time are better predicted for boys thanfor girls by the six social-class indices.

To see the relative contribution of the twomeasures of physical growth and the six social -class indices, multiple correlations were com-puted between the four psychologic variablesand height, head circumference, and the six

sociocultural variables.The results of these analyses are presented

for boys and girls separately in Table 3. Thenumbers in parenthesis are the beta weightsfor the variables as they appeared in TablesI and 2. Looking first at the Verbal Analogies

test, we note that the multiple correlations fortwo physical growth variables and the six

social-class indices arc higher for girls than forboys. Of particular interest is the relative con-tribution of physical growth versus sociocul-tural factors. For boys, head circumferenceand the six sociocultural indices contributeabout equally to the multiple correlation of .28,whereas for girls, the six sociocultural variablescontribute the bulk of the variance to themultiple correlation of .37. Head circumfer-ence contributes relatively little and heightcontributes nothing.

For Memory for Designs, we sec again thatgirls' performance is somewhat better pre-dicted by the combination of physical growthand sociocultural indices than is boys' per-formance. The social-class indices contributestrongly to the multiple correlations for bothsexes. In the case of boys' performance, how-ever, head circumference contributes about thesame as do the social-class indices, whereas forgirls' performance height rather than head

circumference contributes heavily to the vari-ance in performance.

Turning now to the Reversal Learning task,we find that, as was the case for the physicalgrowth and sociocultural variables consideredseparately, boys' performance is better pre-dicted by the combination of physical growthand the social-class indices than is girls' per-formance. This is true both for trials to cri-terion and response time. For trials to criterion,the social-class indices contribute the majorportion of the variance for both boys and girls.In the case of response time, however, thesocial-class indices contribute the bulk of thevariance to performance for boys, whereas forgirls both the social-class in lices and head cir-cumference contribute about equally to re-sponse time.

Discussion

There are three major points to be madefrom these data. First, there are differences

46

associated with attained size, but they are com-plicated by sex differences in the patterns ofperformance and in the manner in which at-tained-size and social-class indices predict psy-chologic performance for boys and girls. None-theless, it is obvious that both attained-size andsocial-class indices used here arc powerful pre-dictors of psychologic performance.

The second point to be made from thesedata is the necessity for careful considerationof the type, range, and focus of the psycho-logic test to be employed in studies of this sort.That the obvious variability among the inde-pendent variables regressed against psychologicperformance through their contribution to thevariance of psychologic performance underlinesthe need to use a variety of psychologic indices.By using a variety of measures, the investigatorcan try to replicate his findings not only forsingle variables but also in the profile across abattery of measures. This becomes extremelyimportant when we consider the complicatedinteractions such as those observed for sex ofthe child, sociocultural status, and the type ofpsychologic test. The great danger in usingsingle or relatively few psychologic measuresis that it is entirely possible to have chosen onewith unusual interactions, leading to an errone-ous interpretation.

For example, if we consider only VerbalAnalogies and Memory for Designs, we wouldconclude that girls' performance is better pre-dicted than boys' and that the social-class in-dices do not generally predict a dramaticallygreater amount of the variance in psychologictest performance than do the measures ofattained size. When we include ReversalLearning in the analysis, however, we find thatthese generalizations do not obtain. Boys' per-formance is better predicted than girls' and thesocial-class indices contribute the bulk of thevariance to boys' performance.

The third point deals with the nature of themeasures of attained size and social class. Since

the elaboration of these sociocultural indices is

essentially an empirical operation and is onlyvery generally guided by theoretical considera-tions, it simply is not clear to what extentmeasures of attained size are also relativelygood indices of social class and to what degreesociocultural indices are themselves reasonablyeffective estimates of nutritional status. Thus,although the isolation and partitioning of vari-ance yields interesting and informative infor-mation about psychologic performance and itsrelationship to both attained size and socialclass, an experimental treatment is ultimatelynecessary if we are to understand the impactof malnutrition on mental development.

A successful experimental treatment, suchas nutritional supplementation, must produceindependently of social-class indices a suffi-

ciently large change in growth to measure thecontributions of these two variables with psy-chologic performance. In the longitudinalstudy discussed here, such comparisons will bepossible given the present differences in growthvelocity between supplemented and unsupple-mented children. Table 4 presents the expecteddifferences at seven years of age due to nutri-tional supplementation. Thus, with continuedsupplementation an answer to this problem willbe forthcoming.

47

MALNUTRITION AND MENTAL CAPACITY

Fernando MOnckeberi

Numelou., data in the literature, based onanimal experimentation and observations inhumans, confirm that severe undernutrition inearly life delays brain development (/-9).The occurrence of many biochemical altera-tions and clear retardation of mental develop-ment have been described during this period.Follow-up studies of animals or children whohave suffered ffom severe early malnutritionhave demonstrated that brain metabolism andmental performance are affected, and that hinall probability this damage is definite (3, 5, 9).

In recent years marked interest has arisen inwhether chronic undernutrition during the pre-school or school years modifies behavior andmental capacity. This problem is of extra-ordinary importance because almost 70 per centof the world's population now suffers fromchronic undernutrition of varying degrees (10).If these statements are valid, it would be obvi-ous that undernutrition constitutes one of themajor obstacles to the progress of underdevel-oped countries since for their advancement theyneed highly qualified individuals at all levels.

Different investigators have noted the factthat poor preschool children have reducedgrowth and a slower maturation rate (3, 5).A higher frequency of retardation of mentaland motor development has also been observedat the same time. From current experimentaldata, it may safely be assumed that retardation

1 Laboratorio de Investigaciones Pedi;tricas, Escuelade Medicina, Universiclaci de Chle, Santiago.

48

of growth and maturation is the consequenceof undernutrition. We cannot be so certainas regards mental and motor impairment.There arc many environmental factors thatmay negatively influence intellectual capacity.Unfortunately, the social groups that suffer

from malnutrition are precisely those that arcoutside the mainstream of society because theyhave very low educational, cultural, and sani-tary levels. All these factors contribute toinadequate stimulation for mental development.

When studying groups of low socioeconomicstatus and intellectual performance, as in slumareas, a significant relationship between growthretardation and low intellectual capacity maybe observed (Figure 1) . A similar correlationmay be seen between animal protein intake andintelligence quotient (Figure 2), so that thosechildren consuming a smaller proportion ofanimal proteins have significantly lower intelli-gence quotients. This correlation is not ob-served when analyzing the total amount ofcaloric intake. In children under three yearsof age a significant correlation can also be

observed between cranial growth retardationand intelligence quotient (Figure 3). Thiscorrelation ceases to be significant when cranialgrowth is within normal limits (7).

All these correlations, even though very

suggestive, do not allow us to conclude thatmalnutrition per se is the cause of mental im-pairment. This is because even in slum areasof more or less homogeneous low-level popula-tion, it is to be assumed that children found to

110

100

90

80

70

60

50

1 to 3 pears of

r =0.56pr,,- 0 001

ap %

IP

8.r

to 10.

50110

10C

80

70

60

50

60 7.0

3 to 5 oars

so

of

r =O 412p=c0.001

KN.

$

Oa

120

50 60 70 804.1 growth deficit norm' (growth

growthHeight Deficit roil g 100xpcted growth

Figure 1. Height and mental development quotient in preschool childrenfrom 1 to 5 years of age.

be nutritionally normal come from bettereducated and better -off families. It is there-fore possible that malnutrition could be a

concomitant and not the cause of the impaireddevelopmental quotient. On the other hand,even children with acceptable nutritive statushave lower IQS than preschool children belong-ing to groups of a higher socioeconomic level.

49

At any rate, nutritional factors cannot bethe only cause conditioning the high incidenceof mental capacity deficit among the poor.

The high frequency of mental impairmentis not only observed in preschool children, butit also affects school-age children in the sameproportion. One of the main problems in LatinAmerica during school age is the high per-

0

st

DIFFERENT AMMAL PROTEIN INTAKE.

100

50-

0

100

50

0

Children 1-3 years of ageI Gesell test)

EINORMAL.

SUB NORMAL

DEFFICIENT.

0 -5 5.110.0 101-15.0 15.1-20 0

Children 3-5 years of ago( Terr n an - Merril test).

0 -5 5.1-10.0 10.1-15.0 15.1-20.0

Animal Protein Intake lg. per day).

Figure 2. Percentage of mental normality, subnormality,and deficiency in preschool children with different animalprotein intakes.

centage of dropouts; of 100 children who beginprimary school, only 20 finish. It is probablethat this higher dropout rate is due in part tolearning difficulties and low mental perform-ance.

During the past year we planned and carriedout a study in a rural mblic school nearSantiago, Chile. It was located in a poor, rela-tively homogeneous agricultural zone in whichalmost every father was a tenant farmer and

50

not a land owner. One hundred sixty-six chil-dren were studied. They constituted a sampletaken at random from the school's 800 stu-dents, whose ages ranged from six to 13 years.The following parameters were studied: (1)anthropometry and clinical signs of nutritionalstatus; (2) the nutritional characteristics ofeach child and his family; (3) socioeconomiccharacteristics of the family that defined lifeconditions and the sociocultural environment;(4) the child's school performance, for anestimate of the amount of knowledge he hadassimilated; and (5) the child's intelligencequotient, through the infantile Wechsler intel-ligence test.

The complete data are now being processed,but preliminary results indicate a high inci-dence of undernutrition and low mental per-formance. Table 1 shows some of the datafrom the different years of primary education.The number of students enrolled is high duringthe first and second years but decreases there-after because of dropping out. During the firsttwo years children had an average height deficit,in relation to age, of 10 per cent compared tothe Iowa Scale's 50th percentile. In later schoolyears this deficit was only 3 per cent. The: ime thing can be seen as regards weightthefirst years showed a deficit .f 15 per cent andthe later ones only 3 per cent. The intelligencequotient was low during the first and secondyears and increased during the following yearsuntil reaching an average value of 101 in theseventh and eighth years. Something similarhappened with school performance, which in-creased significantly during the later years.

For the same group of children we devel-oped a "cultural deprivation" index in whichthe following parameters were considered:eduoation of the mother and father, IQ of themother, and the family's access to mass com-munications media (television, radio, news-papers, and magazines). Significant correlationswere observed between the intelligence quotientand the degree of cultural deprivation (p <

110

100

90

r = 060p=(0001 .

Sit

40,80

70

60

50

50 60 70 80as growth deficit

90 100 110

normal growth

Head Circumference Deficitexpected growth

Figure 3. Head circumference and mental development quotient of preschool children1 to 3 years old.

rest growth

120 150

100

.01), growth retardation (p < .05) , and con-sumption of minimal proteins (p < .01).

We are now studying the group of childrenwho left school prematurely and even thoughwe have no definite results, it seems clear thatthese dropouts have a lower intelligence quo-tient than children of the same age who con-tinue in school.

All these observations seem to indicate thatthe low intellectual performance so frequentlyobserved in the poor seriously interferes withschool performance and probably determinesdropping out. From the data so far analyzedwe cannot conclude the degree to which under-nutrition or cultural deprivation influencesmental development.

wo years ago, we tried to see if a schoolchild's low intellectual performance could bemodified by appropriate feeding. With thisobject we chose 60 poor children aged sevento nine years who were small for age. Theyreceived three meals at school adequate fortotal food requirements. The trial lasted thenine school months, and the children's intelli-gence quotients were measured at the begin-ning, fourth month, and end of the trial(Table 2). During this period a normal in-crease in weight and height could be observed,but there was no change in the high frequencyof intellectual deficiency. Even though theobservation period was relatively short, itwould appear that adequate nutrition during

Table 1. Nutritional, physical, and intellectual correlates of children in the Alto Jahuel Primary School nearflag°, Chile.

School year 1st/2nd 3rd/4th 5th/6th 7th /8th

Number of studentsHeight deficit for age.(avg.)Weight deficit for age (avg.)Caloric deficit for age (avg.)Animal protein deficit for age (avg.)IQ (Wisconsin) (avg.)School performance

403

10%15%16%.

32%81

50

51

260

7%

160 80

3%8% 9% 3%

10% 10%. +2%20% 17% 6%87 92 101

57 60 66

Table 2. Intelligence quotient, weight7 and height of 60 schoolperiod.

children (7 to 9 years old) during nine-month

Period of study Month 0 Month 4 Month 9

to Normal (>91) 9% 11%

(Gillc) Subnormal (80-89) 50% 44% 48%Deficient (<79) 41% 45% 44%

Weight increase (avg.) 0 1.8 kg 2.7 kgHeight increase (avg.) 0 2.3 cm 4.1 cm

the school years does not improve intellectualcapacity. It may be postulated that psychologicdeficit is due plainly to factors other thannutritional ones (e.g., sociocultural factors),or that the damage produced by undernutritionbefore this timeis irreversible.

A very interesting experiment with preschoolchildren is now taking place in Cali, Colombia,under the supervision of Drs. Leonardo Sinis-terra, and Harrison and Arlene McKay. I haveparticipated in it as a consultant. Three-year-old undernourished children from a slum area ofthe city have been selected and entered in aprogram of physical and cognitive stimulationwhile simultaneously being adequately fed. Asa control, another group of children of thesame age and nutritive and socioeconomicstatus have been receiving an adequate feedingat home but no stimulation. The program hasso far been in operation only one year, but theresults seem to be clear. While both groupshave gained physically according to norm, onlythe stimulated group has improved significantlyin intellectual development, achieving after ayear values very similar to those of Colombianmiddle-class preschool children. This indicatesthat the impairment of mental developmentmay be reversible at this age, though feedingalone is not sufficient to increase mental devel-opmentat least in a period of only one year.

Of all the factors analyzed, it seems obviousthat the poor intellectual performance ob-served in low socioeconomic groups cannot beattributed only to malnutrition, but must alsoand in a very important way be ascribed tosociocultural factors that interfere with ade-quate stimulation during the first years of life.

52

Malnutrition is never an isolated phenomenon,and it is almost impossible to analyze sepa-rately the importance that each factor con-tributes to intellectual retardation and thusobtain a definite answer. The important thingto know is not so much the mechanism thatproduces mental impairment but the fact thatthis really exists and that it interferes veryseverely with the individual's and thereforewith society's development. Poor intellectualperformance interferes with learning processesand education in general, preventing childrenfrom reaching adequate educational and cul-tural levels and thus being wholly incorporatedin the socioeconomic development of thecountry.

It may also be concluded that it is too lateto correct mental impairment after seven yearsof age. Neither adequate feeding nor conven-tional education improves mental performancein any important way. Instead, very goodresults seem to be obtained if stimulation andadequate nutrition start during the first threeyears of life.

An inadequate environment, from a socio-cultural as well as a nutritional viewpoint,evidently lessens the child's developmental pos-sibilities when he enters school and good per-formances during the school years are verydifficult to obtain.

It seems evident that programs of stimula-tion and education should start during the firstyears of life so that the child may later per-form adequately in school. Nevertheless, wemust recognize that an adverse environment isa very difficult obstacle to overcome. We havemeasured the intellectual quotient of mothers

in different slum areas in Santiago and havefound a high frequency of mental retardation(Table 3). Seventy per cent of the mothershad an IQ under 79 when studied with theTerman-Merril test, while middle-class moth-ers showed much higher IQs, none of themscoring below 79.

A very close relationship was observed be-tween the mother's t,Q, and the growth statusof her children (p < .001) (Figure 4): the

Table 3. Intelligence quotients of slum-area and middle-class mothers.

Slum area (%) Middle class (%)

Normal (>91) 6 96

Subnormal (80-89) 17 4'Deficient (<79) 77

(HEIGHT) OF CHiLOREN .

110

/00

90

80

70

60,

50

lower her IQ, the worse the nutritional statusof the child. This fact is extremely importantand shows that a correlation exists betweeninadequate cultural environment and nutri-tion. This is a vicious circle that explains whymental impairment persists from one genera-tion to another and offers very few possibilitiesfor the individual to escape from it. The onlysolution is to create a new and different envi-ronment for the child in which he may receiveappropriate stimulation, education, and nutri-tion from the first years of life. More researchis needed in this field, but given present knowl-edge it seems that this is the only way to reallyget to the heart of the problem and obtainrapid changes from one generation to another.

I 55 1.5 s

0.

Aes

Se

r 0.71

p < 0.001

60 70 80 90 100 110

real growthHeight Deficit(child) 100

expected growthFigure 4. Intellectual quotient of mothers (Wechsler-Bellevue scale) and growth(height) of children (r=0.71, p =-- < 0.001).

REFERENCES

I. CRAVIOTO, JOAQUIN, et al. Nutrition, growth andneurointegrative development: an experimental and eco-lOgic study. Pediatrics 38:319-72, 1966.

53

2. DOBBING, JOHN. Effects of experimental under-nutrition on development of the nervous system. In:N. S. Scrimshaw and J. E. Gordon (eds.), Malnutri-

tion, learning, and behavior, Cambridge, Massachusetts,M.I.T. Press, 1968, pp. 181-202.

3. FOOD AND AGRICULTURE ORGANIZATION. Produc-tion year book. Rome, Food and Agriculture Organiza-tion, 1964.

4. MENEGHEI.I.0 RIVERA, JULIO. Desnutricio'n en el

lactante mayor, distrofia policarencial. Santiago, Chile,Central de Publicaciones, 1949.

5. meiNCKEBERG, FERNANDO. Effect of early, marasmicmalnutrition on subsequent physical and psychologicaldevelopment. In: N. S. Scrimshaw and J. E. Gordon(eds.) , op. cit., pp. 269-78.

6. Nutrition and mental development. Confnutrition and human dev, East Lansing, Mich., 1969.

54

7. et al. Malnutrition and mental develop-ment. Am / Clin Nutt. 25:766-72, 1972.

8. STEWART, R. J., and B. S. Pt ATT. Nervous systemdamage in experimental protein-calorie deficiency. InN. S. Scrimshaw and J. E. Gordon (eds.), op. cit.,pp. 168-80.

9. Srocuf, M. B., and P. M. SMYTHE. Undernutritionduring infancy, and subsequent brain growth and in-tellectual development. In: N. S. Scrimshaw and J. E.Gordon (eds.), op. cit., pp. 278-88.

10. WIN:CK, MYRON, and Piluto Russo. The effectof severe early malnutrition on cellular growth of humanbrain. Pedintr Res 3: 181-84, 1969.

THE INFLUENCE OF MALNUTRITION ON PSYCHOLOGIC AND

NEUROLOGIC DEVELOPMENT: PRELIMINARY COMMUNICATION'

Jan Hoorweg" and Paget Stanfield'

The long-term effects of malnutrition onhuman and animal development have capturedconsiderable attention over the last few years.Although much is written and a number ofcongresses have been held on this subject, factsare comparatively scarce, particularly in re-spect to psychologic development. The moststriking effects have been found in animalexperiments but the results with human sub-jects are hardly as impressive. The indicationsthat a period of malnutrition hampers a child'sdevelopment are so strong, 11,wever, that manyresearchers no longer pose the question whethermalnutrition influences development but lookinto more specific questions such as how, when,and under what conditions malnutrition affectsdevelopment.

Retrospective and Prospective Studies

In its simplest form the problem requiresthe comparison of malnourished and well-nourished subjects. Two major research strate-gies are possibleprospective and retrospective

' The work reported herein was supported by the ChildNutrition Unit, Medical Research Council, Kampala,Uganda.

Department of Social Psychology, Makcrcrc Uni-versity, Kampala, Uganda. Drs. Hoorweg is now withthe Africa Study Centre, Leiden, The Netherlands.

"Child Nutrition Unit. Medical Research Council,Kampala, Uganda. Prof. Stanfield is now with the De-partment of Paediatrics, Makerere University, Kampala,Uganda.

55

studies. A prospective or longitudinal study isoften considered preferable because it is ex-pected that other variables influencing develop-ment can be better controlled than in a retro-spective study. Huge efforts may be spent inobtaining results that arc outdated by thetime of publication, however. There is also

the difficulty that because of the very observa-tions that have to go into a longitudinal studyto control other variables, cases of malnutri-tion are much earlier detected and consequentlyearlier treated. Before starting on a longitudinalstudy baseline knowledge is needed. This base-line can be obtained in retrospective studies.

As malnutrition most often occurs in areaswhere records of cascs are hard to obtain, eventhe collection of subjects is difficult. To takea known result of malnutrition such as .short-ness of stature as indicative of it is a verydangerous procedure, as Richardson (9) has

convincingly argued. It is at least necessary.that the group3 to be compared differ decisivelyon the variable of nutrition. This not onlyposes the problem of obtaining a group that hassuffered from malnutrition, but also a groupknown not to have suffered from malnutrition,which is perhaps even more difficult.

Since 1953 the British Medical ResearchCouncil has run a research unit, until recentlycalled the Infantile Malnutrition Unit, inKampala, Uganda. Up to 1962 some 1,000patients were treated there. Most of them were

drawn from a rural clinic at Namulonge, some19 km from Kampala, where later they werealso followed. At the same clinic many reason-ably normal children were regularly seen. Theclinic kept and carefully preserved records ofall persons seen, and from the records a groupof patients and another of controls wereselected.

Design

Selection of malnourished patients(Groups 1-3)

It has been strongly suggested that the effectsof malnutrition are greater the more severe theillness and the younger the age of the childwhen malnutrition occurs. Following are thestudy's design features:

(a) Age at occurrence of malnutrition.Three groups of 20 patients each who had beenadmitted to the clinic at different ages wereselected and matched (see below) with each

Table 1. Composition of groups.

other and a control group. Any doubt ordiscrepancy about the child's date of birththat could not be clarified by checking thebirth registers excluded 3 child from selection.Group 1 consisted of patients who had beenadmitted before 16 months of age (averageadmission age, 12.6 months). Group 2 patientshad been admitted between the ages of 16 and21 months (average, 18.7 months). Group 3

patients had been admitted between 22 and 27months of age (average, 24.1 months).

(b) Severity of malnutrition. Details ofclinical admission and progress were scoredaccording to (1) overall clinical impression,made and recorded at the actual time of dis-charge; (2) the admission weight as a per-centage of the Baganda mean; (3) the amountof edema; (4) skin changes, and (5) the totalserum protein level. Although severity of mal-nutrition was not taken into account in match-ing, the average severity scores of the threemalnourished groups do not differ (Table 1).

Malnutrition Controls

15 months 16-21 months 22-27 months

Group 1 2 3 4

SubjectsN=

Age (average in years)Educational level (average in years)Severity of malnutrition

2014.1

4.617.1

2014.0

4.516.9

2013.8

4.417.5

2013.9

4.4__

Male Education:guardian reached senior or primary 5,6, or 7 level 41% 56% 4756 56%

Occupation:farmer 65% 35% 89°,-: 5556

trader or semiskilled 35% 60% 30%white-collar worker 5% 11% 1556

Female Education:guardian reached senior or primary 5,6, or 7 level 40% 34% 12% 35%

Family Lives in house:with mud walls and Nom 5056 75% 80% 60%no. of rooms (a.g.) 5 4.4 3.7 4.0

Has:dining table 85% 89% 70% 60%radio 65% 79% 5556 5556

bicycle 5556 58% 45% 35%

56

(c) Age group. Children were selected whowere separated from their episode of malnutri-tion by the longest possible period of time.The age of the subjects ranged from 11 to 17years. They had suffered from malnutritionnine to 15 years previously.

(d) Type of malnutrition. Malnutritionmay be either edematous (kwashiorkor) ormarasmic. In Buganda the clinical presentationis nearly always a mixture of 'both.

(e) Duration of malnutrition. Histories arenotoriously unreliable and would have been oflittle use in assessing the period over whichthe child had suffered from malnutrition be-fore admission. Almost all the children recov-ered fairly rapidly. Follow-up continued for afew weeks or months after discharge. Eachchild had one attack of clinical malnutrition.

(f) Associated conditions occurring duringthe period of acute malnutrition. Childrenwho had had any possible acute brain insultsuch as convulsions or severe respiratory infec-tion with anoxia were excluded, as were thosewith long-term chronic infections such as

tuberculosis. Severe anemia did not exclude achild unless a disturbance of consciousness orconvulsion was recorded.

(g) Genetic factors could not be controlled.Sib controls were too few, and too different inage, education, environment, and upbringing.Besides, the family and social structure aresuch that uncertainty always exists as to thetrue genetic constitution of the sibs.

Selection of controls (Group 4)

The controls were children who had attendedthe outpatient clinic and had not suffered aclinical attack of malnutrition. The criteriafor their acceptance in regard to the criticalvariable of nutrition were a recorded follow-upof at least two years beginning in the first yearof life and a weight curve that did not fallbelow the Boston tenth percentile during therecorded period. In fact, three children ulti-mately chosen did fall below the tenth per-centile on one occasion each and one of them

57

fell below the third percentile. Notes recordedat the times of these falterings did not mentionany signs of clinical malnutrition.

As in the groups of malnourished children,those who had had severe illness that mighthave adversely affected mental developmentwere excluded.

Matching

Numerous variables affect physical and men-tal development. In a study such as this, caremust be taken that The experimental variable(malnutrition) is not confounded with othervariabk. The cultural background of the sub-jects in this study was kept constantonlyB.,ganda children were included. Patientstinier the age of 16 months were relativelyfew; individuals in the other groups werematched with individuals in this group, as

nearly as possible, on the following variables(Table I):

(a) Age. The average age for groups 1,

2, 3, and 4 was 14.1, 14.0, 13.8, .nd 13.9 years,respectively.

(h) Sex. Each group consisted of 11 boysand nine girls.

(c) Environment. Stimulation. provided bythe environment is a major factor in psycho-logic development. The following aspects weretaken into account:

(1) The child's educational level. Theaverage education of the four groups based onthe primary grade reached by each child was4.6, 4.5, 4.4, and 4.4, respectively.

(2) Socioeconomic status. Informationon education guardians, male guardian occupa-tion (when there was no male guardian livingwith the child, the female guardian was re- ,garded as such), and several indicators of eco-nomic prosperity are presented in Table 1.

Some of these indicators show considerablevariation among groups, but there are no sys-tematic differences. The average scores over thethree malnourished groups are not noticeablydifferent from those of the controls.

Examinations

Five groups of examinations were carried out,the tirst a battery of psychologic tests givenby one of the authors, usually outside thechild's home in a Volkswagen bus convertedinto a test vehicle. Later the child was broughtto the clinic for clinical and neurologic exami-nation, an thropometric measurements, hemo-globin and total serum protein estimation,motor development assessment.

Psychologic examination

A variety of tests thought suitable for thisparticular group of subjects and this particularproblem was used:

The Raven Matrices is generally consideredone of the tests coming closest to measuring ageneral intelligence factor (10). The test hasproved valuable in different African cultures(5, 14). Subjects have to choose one of sixpieces to complete a pattern, and in this casethe board form of the Coloured Matrices wasused (7). Vernon (12) factorized scores onsonic 20 tests covering a wide range of abilitiesamong 12-year-old Ugandan schoolboys. Hefound a strong verbal factor, a general intelli-gence factor or induction, and a practical orperformance factor composed of spatial andperceptual abilities.

Two tests were used for assessing verbal abili-ties. The arithmetic test of the wIsc w.is ad-apted and a test of Luganda words and idioms(vocabulary) was developed.

Block ( \x'Als) and Porteus Mazes (1)were selected for measurements of spatial andperceptual abilities. In the Block Design test,subjects have to copy a diagram with coloredblocks. Time limits for this test were doubled.In the Porteus Maze test subjects have to drawthe correct way out of e:ch of a number ofmazes. This test was given in an abbreviatedform.

The Memory for Designs test (4) consistsof 15 cards with simple patterns, each of whichis shown the subject for five seconds, after

58

which they have to draw the patterns frommemory. This test of visual memory has beendeveloped for diagnosis of organic impairmentand has been standardized on normal and brain-damaged groups in the United States.

Klein (6) has reported that formerly mal-nourished patients sometimes did poorer thancontrols in a short-term memory task, theKnox Cubes ( / ), if the presentation was per-formed quickly. Both normal and quick pres-entation were used in this study. Subjectshave to repeat the order and number of tapson four cubes.

To gauge learning and incidental learning,subjects were instructed to memorize associa-tions between six animals and six colors. Twoseries of colors were used, colors having differ-ent shapes in each series. Therefore each colorwas presented on two shapes, there being 12shapes in all. The instruction did not mentionshapes, but concentrated instead on the associ-'tion color - animal that had to be memorized.Subjects were given four trials (one trial con-sists of both series), errors were corrected, andresponses were recorded. On completion of thispart subjects were presented with the 12 shapesand asked to remember the color of each shape.This thel is a measure of incidental learning.

For m)tor development assessment, use wasmade of the Lincoln-Oseretsky Motor Devel-opment scale (//). The test consists of 36increasingly difficult motor tasks involvingbalance, dexterity, coordination, and rhythm.Scores are based on accuracy of accomplish-ment and often on speed of completion.

Results

It must be ei4hasized that this presentationof results is preliminary, and that there is a

considerable amount of detailed analysis still tobe done. Result have yet to be compared withother findings in the literature.

The clinical examination of all the childrenrevealed no serious disease. One child had anacute pyomyositis, which delayed the assess-

ment of motor development for two weeks.The spleen was palpable in 11 children scat-tered among all four groups. There was a con-siderable amount of tineal skin infection andresulting slight lymphadenopathy. Two chil-dren had old poliomyelitis, one involving theright shoulder girdle and arm, and the otherthe right leg. Hemoglobin and scrum proteinlevels were all within normal limits and themean values were similar in all four groups.An estimate of the stage of each child's pubertywas attempted, and the scatter from prepuber-tal to postpubertal was similar in the fourgroups. Two of the oldest girls were pregnantand one in the control group was nursing athree-month-old baby.

Tables 2 and 3 give the means per test foreach of the malnourished groups and for thecontrols. Results of the analysis of variance,randomized block design (2), are presentedin Tables 4 and 5.

Table 2. Anthropometry: means per group.

To have independent comparisons, orthogo-nal contrasts were constructed. As it was

hypothesized that the effects of malnutritionwould be inversely related to the age of occur-rence, the following comparisons were chosen:controls (Group 4) were compared with thethree malnourished groups taken together; thegroup that had suffered after 21 months (3)was compared with the two earlier groups (1and 2); and the middle group (2) was com-pared with the group that had suffered before16 months ( 1 ) .

Malnutrition(controls versus three malnourished groups)

Height, weight, and head circumference weresignificantly greater in the controls (Table 4).Differences on the external interiliac diametercome close to being significant .(F = 3.81,p < .10). Mid-upper arm circumference and


-15 months 16-21 months 22-27 months

Group 1 2 3 4

Height (cm) 146.7 145.6 14 i.4 150.1

Weight (kg) 38.5 39.4 38.4 43.8

Head circumference (cm) 53.3 53.5 53.3 54.2

Mid-upper arm circumference (cm) 21.7 22.0 21.7 22.4

Triceps skinfold thickness (cm) 6.8 7.3 6.4 7.2

Subscapular skinfold thickness (cm) 6.5 7.5 7.0 7.0

External interiliac diameter (cm) 21.6 21.4 21.3 22.3

Table 3. Psycho logic tests: means per group.


-15 months 16-21 months 22-27 months

Group 1 2 3 4

Raven 24.5 23.5 23,1 25.9Vocabulary 8.4 9.3 9.7 9.7Arithmetic 8.8 9.7 9.6 9.3Block Design 17.1 15.7 16.4 21.7

Portcus Mazes 6.4 6.3 6.2 7.0Knox Cubes

slow 10.1 10.4 10.9 10.6quick 8.4 8.9 9.5 9.0

Memory for Designs (errors) 7.1 8.7 4.4 4.7Incidental Learning 4.0 5.1 4.7 5.9Lincoln-Oseretsky 104.2 102.5 107.1 114.0

59

Table 4. Anthropometry: results of independent corn- dental Learning, and Lincoln-Oseretsky (Tableparisons, analysis of variance, randomized block design 5). The results of the learning task at each(F 1,57).

Groups

Comparison

4vs.

2vs.

3

vs.1 1,2 1,2,3

Height F=7.07p< .02

Weight F=8.85p< .01

Head circumference F=4.72.p< .05

Mid-upper armcircumference

Triceps skinfoldthickness

___

Subscapular skinfoldthickness

External interiliac F= 3.81

diameter p< .10

trial are presented in Figure I. Although dif-ferences are in the expected direction, neitheron overall score nor score at each trial do theyreach significance. It is only on the first trialthat there is an indication of differences be-tween Group 4 and Groups I, 2, and 3 (F3.56, p < .10).

On the Memory for Designs test comparisonbetween Group 3 and Groups I and 2 does

reach statistical significance (F = 5.28, p <.05), but the comparison controls versus mal-nourished groups does not. Inspection of themeans (Table 3) for each group shows thatthis can be explained by the fact that Group 3is doing very well, which reduces the differencesbetween controls and malnourished patients.From the means it appears that Groups I and 2

Table 5. Psychologic tests: results of independent corn- are doing less well than the controls. Differ-parisons, analysis of variance, randomized block design ences were not significant on Vocabulary,(F 1,57). Arithmetic, Porteus Mazes, and Knox Cubes.

Groups

Comparison

2

vs.3

vs.4

vs.1 1,2 1,2,3

Raven F=4.53p< .05

Vocabulary.ArithmeticBlock Design F=6.98

p< .02Portcus MazcsKnox Cubes

slow F=3.44quick p< .10

Memory for Designs F =5.28p< .05

Learning F=4.22 F =3.561st Trial p< .05 p< .10

Incidental Learning F= 3.20 F= 5.83p< .10 p< .02

Lincoln- Oscretskv F=7.86p< .01

triceps and subscapular skinfold thicknessshowed no noticeable differences.

On the psychologic tests controls did sig-nificantly better on Raven, Block Design, Inci-

6 0

Age of admission

Comparison of the three groups that suf-fered from malnutrition at different ages

100

00

IA

/0

60

SO

.........

,1Ion1,01,

IS 00111

1161001,11100 11 /I 0100,11s- I/ /I 010011s

Figure 1. Correct percentage of responses per trial fordifferent groups.

yielded no significant differences except forthe Memory for Designs test, on which Group3 did better than Groups I and 2, and the learn-ing task (first trial), on which Group 2 didbetter than Group 1. Two other psychologictests, Incidental Learning and Knox Cubes.(quick), yielded differences that reached a sig-nificance level of .10. It should be noted thatthese four tests are all concerned with differentaspects of memory and learning.

For the psychologic tests average scores pertest for each of the three groups were ranked1, 2, and 3. Inspection of Table 6 shows thatoverall this ranking distribution can be at-tributed to chance; but that on .she tests con-cerned with memory and learning the averagescore of Group 3 is on all four tests higherthan that of Group 1. These differences are inthe expected direction in that those who suf-fered younger do poorer.

Severity of malnutrition

Because the three malnourished groups arenot independent but matched, they cannot becombined for a comparison of severity of themalnutrition episode. Comparisons can onlybe carried out within each malnourished group.In each of the three groups cases were dividedinto two groups according to the severity ofthe malnutrition episode, thus yielding threesevere groups and three less severe groups of 10subjects each. Although detailed analysis has

Table 6. Ranks of means for malnourished groups pertest.

15months

16-21months

22-27months

Group 1 2 3

Raven 1 2 3

Vocabulary 3 2 1

Arithmetic 3 1 2

Block Design 1 3 2

Portals Mazes 1 2 3

Memory for Designs 2 3 1

Knox Cubes 3 2 1

Learning 3 2 1

Incidental Learning 3 1 2

61

still to be done, there appear to be no differ-ences between severe and less severe subjects.

Discussion

Differences in height, weight, and headcircumference have repeatedly been demon-strated in follow-up studies of malnourishedpatients. These retardations confirm that thecontrol and experimental groups are differentin nutritional history.

It appears then that the children who suf-fered an episode of protein-calorie malnutritionare lower in general intelligence than the con-trol children. If there is a relationship betweenmalnutrition, brain growth, and mental devel-opment, one would expect, on the basis of theliterature on European subjects, that tests forspatial and visual abilities would show greaterdifferences than tests for verbal abilities (10,13). The Block Design especially is oftenregarded as a very sensitive test in this regard(8). Results are in line with these expectations..

'Tests for verbal abilities show no differences,but the Block Design test does. That the secondperformance test, the Porteus Mazes, does notyield any differences may possibly be attributedto the fact that an abbreviated form of this testwas used that gay.2 a very small range in scores.It does not appear, however, that these differ-ences can be attributed to severe brain damage.Although Groups 1 and 2 score higher in errorsthan Groups 3 and 4 on the Memory for De-signs test, the average score of Groups 1 and 2still falls below the average score reported inthe United State for various groups of patientssuffering from brain disorder (4). It is mostlikely that differences on the Block Design andMemory for Designs tests have to be attributedto poorer spatial and perceptual abilities; onthe Memory for Designs, poorer memory orattention or both could also play a role.

Differences on memory and learning tasksseem to appear only with slightly complexstimuli. The Knox Cubes test uses simplestimuli (taps on blocks), and there are no dif-ference's between controls and malnourished

groups. Although differences on the learningtask were not significant, differences betweencontrols and malnourished patients tended tobe largest on the first trials. This could indi-cate difficulties in attention rather than mem-ory. The significant differences on incidentallearning also indicate a limited field ofattention.

The Lincoln-Oseretsky Motor DevelopmentScale samples a variety of motor skills. Itemscan easily be divided into groups of balancingand dexterity skills. When the total score is

broken down into subscores for balance anddexterity, it is only on the latter score thatcontrols and malnourished groups differ sig-nificantly (F = 7.92, p < .01).

Defects in attention, spatial and perceptualabilities, andhampered fine-motor developmentare also features of the "minimal brain dys-function syndrome" or the "cerebral minimalsyndrome" (3), and as such could be the resultof minimal, widespread cerebral cortical dam-age at the time of the attack of malnutrition.

That little difference was found betweendifferent groups of patients indicates that anattack of malnutrition in itself is so seriousthat age of occurrence or severity of the dis-ease were of little importance for furtherdevelopment.

Summary

Three groups of 20 children each who hadsuffered an acute episode of malnutrition atthree age periods, up to 15 months or from16 to 21 months or 22 to 27 months were com-pared with a group of 20 control children fromthe same area of Uganda and of the same tribe,sex, age, education and socioeconomic statuswho, from evidence of clinical and weight

records, had not passed through any overtperiod of malnutrition in early childhood.

All the children were between the ages of11 and 17 and were examined clinically andneurologically, measured anthropometrically,and subjected to a battery of psychologic testsand a motor development test.

An analysis of variance between the groupsof previously malnourished children revealedno significant differences in anthropornetry,and on only two tests (Incidental Learning andMemory for Designs) did children who suf-fered young do significantly poorer than thosewho suffered from malnutrition at a later age.

Weight, height, and skull circumferencewere significantly greater in the control groupthan in the three malnourished groups together.Five of the 10 psychologic tests revealed sig-nificantly better performance in the controlgroup as compared with the three malnourishedgroups taken together. These were the RavenMatrices as a measure of general intelligence,the Block Design test, Memory for Designs,incidental learning, and the dexterity part ofthe Lincoln-Oseretsky motor developmentscale. No differences were found on verbalabilities, a maze test, and a short-term memorytest.

These results suggest an impairment in gen-eral intelligence, spatial abilities, memory, andlearning arising from lack of attention and indexterity in the subjects who had had an acuteepisode of malnutrition in early childhood, ascompared to a group who had escaped such anincident.

This impairment is not of the magnitudethat occurs in known syndromes associatedwith severe brain damage, but there is a re-semblance to the so-called "cerebral minimal"synd romc.

ACKNOWLEDGMENT

We wish to express our gratitude to Miss I. Rutishauscr of theMedical Research Council, Kampala, who made most of the anthro-pometric measurements, and to Mr. P. Sorlie, biostatistician with theMaternal and Child Health Project, Makerere University, Kampala,who advised and assisted us with the analysis.

62

REFERENCES

I. An Turn, G. A point scale of performance tests.Ness York, Psychological Corporation. 1947.

2. linowsw:. K. A. Statistical theory and method-ology in science and engineering. New York, John

Wiley & Sons, 1965.3. DENuou Ii, E. The cerebral minimal dysfunctions.

Proc int cong pediatrics, Vienna. 1971, vol. 3. 63-67.4. GRAIIANI. F. K., and B. S. Krsaivi t . Memory-for-

designs test: revised general manual. P, r, ept Mot SkillsI : 147-85, 1960.

5. InviNE. S. H. Figural tests of reasoning in Africa.Int I Psy, Ind 4:217-28, 1969.

6. K1 vas, R. E., it al. Performance of malnourishedin comparison with adequately nourished children. AIMmeeting, Ant Assoc Adv Sci, Boston. December 1969.

7. RAYEN. I. C. Guide to the coloured progressivematrices. London. H. K. Lewis & Co., 1965.

8. REISCI:NWI.DFR, MARION. The use of modified blockdesigns in the evaluation and training of the brain-in-jured. Pay, 1,0! Monogr. vol. 67, no. 21, 1953.

63

9. Ricumuisos. S. A. 'Die influence of social-

environmental and nutritional factors on mental ability.In: N. S. Scrimshaw and J. E. Gordon (eds.), Malnu-trition. learning, and behavior, Cambridge. Massachusetts. M.I.T, Press, 1965, pp. 346-61.

10. Svvvo . R. D. Psychometric assessment of theindividual child. Harmondsworth. England. Penguin

Books, 1965,

11. St nvis. W. The Lincoln-Oseretsky motor develop-meat scale. Genet Pay. bid Monogr 51:153-252, 1955.

12. VITNON. P. Abilities and educational attainmentsin an East African environment. / Spec Filn, 1: 335-45,

1967.

13. WI,cusult, Dvvin, The measurement and ap-praisal of adult intelligence. 4 ed, London,Tindall & Cox. 1958,

14, Womt, M. The meaning and stability of Raven'smatrices test among Africans. Ins l Psydnd 4:229 -35.1969.

THE FUNCTIONING OF JAMAICAN SCHOOL CHILDREN SEVERELY

MALNOURISHED DURING THE FIRST TWO YEARS OF LIFE"'

Herbert G. Birch and Stephen A. Richardson'

Introduction

This is a report of a study conducted inJamaica, West Indies, on the long-term conse-quences of severe malnutrition during the firsttwo years of life. Our subjects were school chil-dren hospitalized during their first 24 monthsfor severe clinical malnutrition (marasmus,kwashiorkor, or marasmic kwashiorkor). Theyhave been compared with two groups of school

' The studies described here were carried out in

Jamaica in the Tropical Metabolism Research Unit ofthe Medical Research Council. The investigations alsoreceived active support from the Epidemiology Re-

search Unit of the Medical Research Council in Jamaica.Though most patients had been treated initially in theTropical Metabolism Research Unit, additional patients,particularly in the youngest age group at hospitalization,were obtained with the help of the Department ofPaediatrics, University of the West Indies, Follow-up

studies included anthropometric, pediatric, neurologic,

psychologi. educational, and social evaluations. Differentaspects of the inquiry will be considered in separate

reports.

Support for this study was provided by the Asso-ciation for the Aid of Crippled Children; The Nutri-tion Foundation, Inc., New York; Grants HD-00719and NIH-71-2081 from the National Institute of ChildHealth and Human Development, National Institutesof Health, Bethesda, Maryland; the Medical ResearchCouncil. London, and from program support to the

Child Development Research Unit, Institute of Educa-tion. University of London,

"Departments of Pediatrics and Community Health,Albert Einstein College of Medicine, Bronx, New York,USA.

64

children of like sex. The first comparisongroup consisted of the patients' sibs closest inage to the index patient. The second compari-son group was composed of unrelated class-mates or neighbors closest in age to the indexchild or the sib.

The concerns and design of the study stemfrom problems in interpretation and questionsthat arose in previous investigations of theeffects of early malnutrition upon physical andmental development. Until recently few chil-dren who were severely malnourished in infancysurvived to reach school age. Early research inmalnutrition quite properly was focused pri-marily on its physiologic effects and on thedevelopment of treatment and managementprocedures thartirould prevent death and maxi-mize recovery. As Champakam and coworkers(12) have put it: "Survival was the main con-cern. Awareness and knowledge of the bio-chemical pathology of malnutrition and theavailability of more efficient means for betterdiagnosis and treatment have reduced the im-mediate mortality among malnourished chil-dren. In direct proportion to the success in

this regard is the clear possibility of increasingpools of survivors who may be handicapped ina variety of ways and for variable periods oftime" (page 844).

These changed circun.5tances have, over thepast decade, resulted in a growing number ofinvestigations exploring the long-term conse-

quences of both severe acute malnutrition and-chronic subnutrition" (10) for growth andintellectual development. In addition, numer-ous experimental animal models for studyingsuch consequences ha..., been employed (.30).In studies of human subjects where controlleddeprivation experiments could not be under-taken, follow-up studies have been conductedon children about whom both direct or indirectevidence of malnutrition has been available.

Two types of study designs have been usedto explore the consequences of malnutritionfor cognitive functioning in children. In somestudies the presence of antecedent malnutritionin school-aged and preschool children has beeninferred from differences in height for age. Incommunities where the risk of malnutrition isendemic, short children have been judged tohave been at greater nutritional risk than tallerage-mates. Groups of tall and short childrenhave then been compared as to IQ level andintersensory competence (17, 19). The absenceof direct evidence of severe malnutrition andthe wide range of variables other than nutri-tional ones that differentiate the families of tallchildren from those of short ones (18, 26)have made it difficult directly to attribute thedifferences between the tall and short groupsto malnutrition per se.

In the second type of study children' withknown histories of severe malnutrition havebeen compared with children in their commu-nities without such histories (5, 8, 11, 12, 14,16, 23, 25, 2%, 31, 32). While the fact of ante-cedent malnutrition has been clearly establishedin such studies, other factors limit the degreeto which they can be interpreted as indicatinga direct association between antecedent malnu-trition and intellectual outcome. Children whoare severely malnourished early in life comefrom families that are at greater risk for dis-organization, cultural limitation, social and

economic disadvantage, poor housing, exces-sively frequent and short-spaced reproduction,maternal and child ill-health, and the like, thanare other families of even the same social class

65

(IN). These additional factors themselves arecapable of influencing intellectual development(6, 24, 29). In all instances the children withwhom index cases have been compared couldnot be sufficiently well-matched on these non-nutitional variables to meet the requirementsof rigorous experimental design. The mostthorough attempt has been that of Champakamand coworkers (12) in which age, sex, religion,caste, socioeconomic status, family size, birthorder, and parental education were taken intoconsideration. In that study, however, it waspossible that the families of children hospital-ized for kwashiorkor were either geneticallydifferent, or provided less adequate experientialopportunities for intellectual development thanthe comparison families (6, IN).

Another tactic is to use sibs of the malnour-ished children as comparisons: Such sib studies,though admittedly imperfect, provide differentopportunities for interpreting findings. Onlytwo such sib studies hale thus far been carriedout (7, 22). In one of these (7), sibs withouta history of hospitalization for severe nutri-tional illness were found to be significantlysuperior in IQ to their hospitalized brothers orsisters. The conclusions from this study arelimited, however, by the facts that the samplestudied was relatively small, the sibs were notlike-sexed, and no general population compari-son group was examined. In the other sibstudy (22), the age at which malnutrition wasexperienced ranged from 10 months to fouryears, and only 10 of the children in the groupwere under 18 months of age. Moreover, grossdifferences in IQ in different segments of thekwashiorkor sample as well as in their siblingsmake the data difficult to interpret.

In conducting a sibling comparison study itmust be recognized that while sibs may havemany backgrourvi. characteristics in common,a number of differences must be .ansidered inthe analysis of findings. Sibs, unless twins, arenecessarily different in age and ordinal position,both of which factors may affect child-rearingpractices as well as intelleCtual outcome (3).

Moreover, in a community such as Jamaicasibs frequently have different fathers and ex-perience different child-rearing conditions.These factors must all be taken into accountin designing and int-rnrcting a sib comparisonstudy. This has been attempted in the presentstudy.

A second question with which the presentstudy is concerned has emerged from bothreports of animal investigation and studies ofrecovery in children who were hospitalized forsevere malnutrition at different ages in earlylife. The eviden,.., of the animal investigationsindicates that the risks of dcfcctivc myclina-tion, reduced cell replication, and delayed bio-chemical maturation are greatest when malnu-trition coincides with particular periods ofrapid brain growth (13, 20, 21, 34). In thehuman organism the period of most rapid braingrowth extends from about the beginning ofthe last trimester of pregnancy to the lastquarter of the first postnatal year. Moreover,during this period cellular replication in thecentral nervous system is completed.

These facts have led to speculation thatsevere malnutrition during the first nine monthsafter birth would have more severe conse-quences than severe malnutrition later in in-fancy and early childhood. Some support forthese expectations has derived from a study byCravioto and Robles (16) in which childrenhospitalized with severe malnutrition in thefirst six months of life were found, duringconvalescence, to recover behavioral compe-tence less fully than children hospitalized forthe same illness later in infancy. For severalreasons, however, these data are by no meansconclusive. Although cell replication is com-pleted at an early age, other aspects of growthand differentiation in the nervous system suchas dendritic proliferation, axonal branching,and synapse formation, all of which may be

important for integrative organizationmore

than cell number, continue to develop at veryrapid rates throughout early childhood. More-over, the Cravioto and Robles study (16) did

66

not follow up the children in later childhoodto evaluate later aspects of development, andtheir conclusions cannot be extended for out-comes beyond the immediate recovery period.

Though animal experimental data have indi-cated that particular vulnerabilities to nutri-tional stress exist in brain in relation to the ageat which the stress is experienced, our knowl-edge is incomplete. We cannot now identifywith certainty a specific age after which suchvulnerability ceases to exist. Since many aspectsof neuronal growth and differentiation continueto occur throughout the first years of life (28,29), however, it is unlikely that vulnerabilityis restricted to the prenatal period and earlyi nf a nc y.

Subjects and Study Design

The 74 severely malnourished children (here-after referred to as index patients) were allboys who had been hospitalized for treatment ofsevere infantile malnutrition during the firsttwo years of life. Sixty of these children hadbeen inpatients the metabolic ward of theTropical Metal Research Unit of theBritish Medic ,earth Council, Mona,Jamaica, W. I. The remaining 14 patients wereadded to have fuller representation over the en-tire first two years of life. The latter were ob-tained from the pediatrics ward of the UniversityHospital, University of the West Indies, Mona.At the time of hospitalization all the childrenwere suffering from severe malnutrition, re-flected variously in syndromes of marasmus,kwashiorkor, or marasmic kwashiorkor. De-tailed clinical and metabolic records were avail-able in all cases. The children received anaverage of eight weeks of inpatient care. In

general, follow-up visits in the homes wereconducted for two years following discharge.

Sixty-four male infants had been treated inthe metabolic ward of the Tropical MetabolimiResearch Unit in the years relevant to thestudy. At follow-up two of these children andtheir families had moved and could not be

traces... Two additional children were foundbut not included in the study, one because hewas a mongoloid and the other because of thepresence of infantile hemiplegia, most probablyof perinatal origin. The remaining 60 childrenwere all included in the follow-up study. The14 patients from the pediatrics ward were thefirst 14 to be located from an initial pool of50 patients. Approximately half of them livedin the city of Kingston, and the remainder livedin other towns, villages, and rural districts,some of which were more than 160 km fromKingston.

At the time of the study the index patientswere 5 years 11 months through 11 years old.These ages were selected in order to be farenough removed from the time of acute illnessto eliminate the effects of immediate sequelaeand for the children to be at arr age at whichintelligence testing has predictive validity forlater life.

Whenever there was a male sibling of theindex patient between six and 12 years of ageand nearest in age to the index patient, andwithout a history of severe clinical malnutri-tion, he was included in the study. Because ofwidely varying patterns of family compositionin Jamaica, a sib was defined as a child havingthe same biologic mother as the index patientand having shared a home residence with theindex child for most of his life. Thirty-eightsuch sibs were available and studied. Because

of the selection criteria used, the sibs wcrcsomewhat older than the index patients.

In addition to the sibs, a classmate or neigh-bor comparison was selected for each indexpatient and for each sib. For index childrenand sibs attending school, two classmates ofthe same sex closest in age to the index patientand the sib were selected. On evaluation, ifthe first Comparison child was not availablefor examination, the second comparison wasused. Some of the index boys and sibs, thoughof school age, were not going to school. In

such instances, a comparison case was chosenby finding the nearest neighboring child who

67

was not a relative and who was within six

months of the index patient's age. For sonicindex patients at small schools no , classmatewas within six months of age. Then neighborchildren were also used as comparisons. In

three cases comparison children could not bebrought in for:, study. This resulted in 71

matched pairs of index and comparison chil-dren. Table 1 summarizes the ages at testingof the children studied. The comparison groupshould not be considered as controls but ratheras children who were identified because of theirgeographic proximity and their closeness in ageto the index children.

Detailed interviews were conducted to se-cure the background histories of all children.These indicated that eight of the comparisonchildren had been sick between the ages of onemonth and two years, either from malnutritionor with symptoms thac could have been asso-ciated with malnutrition. Only one of themwas hospitalized during that 'age interval, fortwo weeks because of diarrhea and vomiting.

At the time of follow-up these data wcrcobtained about each of the boys in the studyto assess their current physique and level offunctioning: (1) height, weight, and headcircumference; (2) a detailed assessment ofpediatric neurologic status; (3) intelligencelevel using the wise and the verbal portion ofthe WPPSI; (4) estimates of intersensory devel-opment, perception, laterality, lateral aware-

Table 1. Ages of index, sib, and comparison boys attime of IQ testing.

Age inyears/months Index

Numher of

Si h Comparison

5-0 to 5-11 1 5

6-0 to 6-11 19 3 16

7-0 to 7-11 16 6 14

8-0 to 8-11 15 10 13

9-0 to 9-11 19 4 20

10-0 to 10-11 4 9 3

11-0 to 11-11 4

12-0 to 12-11 2

Total 74 38 71

ness, and motor coordination; (5) an estimateof educational achievement using the WideRange Achievement Test; (6) teachers' assess-ments of school performance and behavior inthe school setting; (7) sociometric data toobtain a measure of the child's social accepto-bility by his peers, and (8) mothers' assess-ments of behavior at home.

To obtain a detailed social and biologichistory of each subject and information aboutthe general ecologic conditions that mightinfluence the children's development, a detailedinterview was conducted with each boy'smother or guardian.

Results

Thus far, the major focus of analysis hasbeen directed toward determining whether theindex children show impairment in physiqueand functional development compared with theother boys studied.

The characteristics of the groups selectedwould lead us to anticipate that the indexpatients who suffered acute malnutrition wouldbe most severely impaired, that their sibs, someof whom would have experienced chronic sub-nutrition but not an acute severe episode,would have been intermediately impaired, andthat comparison children, who did not belongto a family having a child with a history ofsevere clinical malnutrition, would be leastimpaired. The findings bear out these expecta-tion, but they will only be summarized herebecause they are to be reported in detailelsewhere.'

Hcrtzig, M. E., et al. Intellectual levels of schoolchildren severely malnourished during the first twoyears of life. Pediatrics, 49:814-24, 1972.

Richardson, S. A et al. The behavior of children inschool who were severely malnourished in the first twoyears of life, I Health Soc Ramo 13:276-84, 1972.

ct al. School performance of children who wereseverely malnourished in infancy. Am J Meru Defic,in press.

68

Pb) sica/ a/Minnie/as

Index boys were significantly shorter in

height, weight, and head circumference thantheir classmate or yardmate comparisons. Indexboys were significantly lighter and had smallerhead circumferences for age than their sibs,but were not significantly shorter. The sibsdid not differ from their classmate or yardmatecomparisons in either height or weight.

To determine whether physical growth re-covery of the index boys might be different indifferent environments, boys who had grownup in urban and rural surroundings were com-pared, It was found that index-sib differencesheld for rural but not for urban environments.

Jnie11(e/nal f undioning

All IQ measurements were significantly lowerfor the index boys, For the Full Scale, Verbal,and Performance IQ measures the index patientshave the lowest mean scores, the sibs occupyan intermediate position, and the comparisonchildren have the highest scores. These differ-ences are in accordance with exp.eetation. Thethree IQ measures are seven to nine points lowerfor the index children than for the compari-sons. Each of the differences is statisticallysignificant at less than the 0.001 level ofconfidence.

That these mean differences in IQ score arenot the result of a few cases of extreme compe-tence in the comparison group or of extremeincompetence in the index patients but repre-sent differences over the whole range is clearlyindicated in Figure I. As may be seen fromthis figure, the groups, in general, differedfrom one another in Full Scale 1Qs over thewhole range of measures. The cumulative fre-quency percentage curves for Performance andVerbal IQS in the two groups are almost identi-cal in form, and indicate that the comparisongroup is also superior on these measures.

The index children also have lower scoreson the three IQ measures than their sibs. TheFull Scale and Verbal IQ differences are statis-

12n-124

115- 119

110- 114

105- 109

100. 104

95- 99

0 90- 94L. 85- 89u 80- 84

75- 79

5' 70- 74LL

65- 69

60- 64

55- 59

50- 34

45- 49

40- 44

rr

Index Group

Companson Group

I L 1 I J L 1 J10 20 30 40 50 60 70 80 90 100

CUMULATIVE PERCENT OF SUBJECTS

Figure 1. IQ distribution of index and comparison patients.

tically significant (p < 0.025). Differences

in Performance IQ arc not significant. Whenthe sibs are contrasted with the comparisonchildren, they are found to differ from themsignificantly only in Performance IQ.

Before the results can be interpreted, twofurther analyses are required. In the first placesibs tend to be older than index and compari-son subjects, and the differences between thesibs and the other two groups could have de-rived from this age difference. Moreover, sinceordinal position can affect IQ within a sibship,differences between sibs and index patients

could have been the result of differences inordinal position. To deal with the first ques-tion, the relation of IQ level to age was analyzedin each of the groups. No age trends for IQ

were found. To deal with the second question,older sibs were separately compared with

younger index patients and younger sibs witholder index patients. Identical trends and

directions of difference were found when thesibs were respectively in lower or higher ordinalpositions in the sibships. These analyses indicateclearly that the differences among groups werenot artifacts either of age or of ordinal position.

69

School aChiel'0111CIll

In the reading, spelling, and arithmetic testsof the Wide Range Achievemcr.;. Test, theindex boys were on average 7 to 9 points lowerin their quotients than their classmate com-parisons. In the same three tests the sibs wereon average 4 to 9 points lower than theirclassmate comparisons. The index and sib

average quotients on the three tests were almostidentical in each case, being less than one pointapart. Index children had significantly lowerscores than their classmate comparisons on allthree tests. Sibs were significantly lower thantheir comparisons on reading (p < 0.05 one-tail), but the differences on the spelling andarithmetic tests were only trends.

Teachers' evaluations represent judgments offunctioning based on continuing contact withthe child in a school setting. Such judgmertsprovide a different perspective than that de-rived from an achievement test. Because teach-ers cannot be expecte?, to have uniform stand-ards for evaluating children, comparisons weremade only between pairs of children judged bythe same teacher. On the overall evaluationof school performance, the teachers' rating forthe index boys were lower than their classroom

comparisons (p < 0.005). Twenty-four percent of the index children were rated in thelowest category of quality of overall schoolwork"Iie is severely backward"comparedwith S per cent of the index comparisonchildren. No significant differences were foundbetween sibs and their comparisons.

Bcharior in the school scithig

Index boys were judged by their teachers tohave worse attention and memory', and to bemore easily distracted and less spontaneous andcontributive in classroom discussion than class-mate comparisons. In their social relations theindex children get along less well with theirpders and are less cooperative with their teach-ers. They had more frequent problems in class-work and more often manifested behavior andconduct problems. When sibs were comparedwith their classmates,' the only behavioral dif-ference found was greater distractibility in thesibs.

Age of seterc malnutrition

The age distribution of the index childrenwhen they were hospitalized for the treatmentof severe malnutrition ranged from three to24 .months, with admissions at almost everymonth between these ages.

If the age at which a child is hospitalizedfor severe malnutrition is system.tically relatedto the severity of his physical and functionalimpairment, we would expect a significantpositive correlation to exist between age atadmission and attributes at school age. Whenphysical growth, IQ, and school achievements

and behavior are correlated with age at hos-pitalization,. there is no significant associationwith earliness of illness. The data indicate arandom relationship between the age of theseverely malnourished child at hospitalizationand function or growth attainments at schoolage.

Because the replication of cells in the centralnervous system tends to be completed beforethe age of nine months, it has been suggestedthat children experiencing severe malnutritionbefore that age are at particular risk of braindamage and lowered intellectual level. Wetherefore considered it desirable to comparethose children who had been hospitalized beforethe age of eight months with those admittedto hospital for severe malnutrition at later ages.When children admitted before eight monthsof age, between eight and 12 months, andbetween 13 and 24 months are compared, nosignificant diffeiences are found among groupsfor Full Scale, Verbal, and Performance IQ,growth attainment, school achievement, orschool behavior. The data thus provide nosupport for the hypothesis that systematic dif-ferences in outcome at school age are attrib-utable to severe malnutrition at different agesduring the first two years of life.

The second aspect of the study deals withthe ecology of growth and functional develop-ment of these children. In this approach nutri-tional history is considered as one variable inthe context of the wide range of influences thataffect the development of children. This strat-egy is described in. part in the paper by S. A.Richardson in this monograph.

ACKNOWLEGMENT

Many persons contributed to this study. We thank in particularProf. J. C. Waterlow, former Director of the Tropical MetabolismResearch Unit, University of the West Indies, Mona, Jamaica; Dr.David Picou, the present Director; Dr. W. E. Miall, Director of theEpidemiology Research Unit at the same university; Prof. Eric Back,Department of Paediatrics there, and Mrs. Caroline Ragbeer, Mrs.Patricia Desai, and Mr. V. A. Morris. Members of the examining team

70

included Dr. Ira Belmont, Dr. Morton Bonner, Mrs. Juliet- Bonner,and Dr. Barbara Tizard. We are most indebted to Prof. Jack Tizardand Dr. Margaret E. lienzig for their role in data gathering, planning,and analysis. Dr. R. S. Schoenfeld, Miss F.: Grabie, and Miss K. Yoderhave provided invaluable statistical and analytic assistance,

REFERENCES

I. A INS WORT] I, M. (Cd.) . Deprivation of ma-ternal care; a reassessment of its effects. Geneva.. World

Health Organization, 1962. p. 165. (wilts Public HealthPapers No. 14.)

2. AI .TNIAN, J., et al. The influence of nutrition .011neural and behaviorial development. Deo Psycliabial3:281 -301, 1970.

3. ANAst-Asi, A. Intelligence andPsycho( Ball 53:157-209, 1956.

4. Bsswiz, H.. and R. 13:00vt8. Clinical managementof the behavioral disorders in children. Philadelphia.

W. B. Saunders, 1960.

5. iImittERA-MoNcAoA, G. Estudios sobre alteraciones

del crechniento v del desarrollo p.sicoldgico dcl sindrome:41tricarencial (kwashiorkor) . Caracas, Editora Grafos.

1963.

6. Buic:th H. G., and J. D. Gussow. Disadvantagedchildren: health, nutrition and school failure. NewYork, Harcourt Brace and World and Grime and Stratton,1970, p. 322.

7. it al. Relation of kwashiorkor in early child-hood and intelligence at school age. Pediatr Res 5:579-55. 1971.

S. BoTitA-ANToloN, E., et al. Intellectual developmentrebated to nutritional status. Trop Pedioir 14: 112 -15,

1965.

9. BOW1.15Y, J. Separation anxiety. Int Psyclioanal,41:89-113, 1960.

10, BROCK, J. Recent advances in human nutrition.London, J. & A. Churchill, 1961.

11. GARAI:, V., and R. NA JDANVIC. Effect of under-nutrition in early life on physical and mental develop-ment. Arch Ms Child 40:532-34, 1965.

12. CI !ANWAR ANL S., et al. Kwashiorkor and mentaldevelopment. Am I Clin Nate, 21:544-52, 1968.

13. CI IA SE, H. P., et al. The effect of malmuritionon the synthesis of a myelin lipid. Pediatrics 40:551 -59,1967.

14. and H. P. MARTIN. undernutrition andchild development. New Eng! I Med 282:933-76, 1970.

15. Cowl, J. L. The postnatal development of the

family

71

human cerebral cortex. Cambridge. Harvard UniversityPre.ss, 1963.

16. Citivlo.r0, JomituN, and BEATRIY. Rein ES. -

lion of adaptive and motor behavior during rehabilitationfrom kwashiorkor. Ain / Orthopsy, Worry 35:449-64,1965.

17. rt al. Nutrition. growth and neuroin-tegrative development: an experimental and ecologic

study. Pediatrics 38:319-72. 1966.

18. It al. The ecology of infant weight gainin a pre-industrial society. Arlo Paediatr Seance 56:

71 -54. 1967.

19. and E. R. DELInAitniE. Intersensory de-velopment of school-age children. In: N. S. Scrimshawand J. E. Gordon (eds.), Malnutrition, learning, andbehavior, Cambridge. Massachusetts. M.I.T. Press, 1968,pp. 252-69.

20. DAVISON, A. N., and )01IN DOBBIN:C.. Myelinationas a vulnerable period in brain development.- Br MedBall 22:10-44, 1966.

21. DOBRING, JOHN. The influence of early nutritionon the development and myelination of the brain. ProfR SocLond J13in11, 159:503 -09, 1964.

. 22. HANSEN, J. D., it al. What does nutritionalgrowth retardation really imply? Pediatrics47:299-313,1971.

23. IVIiim:Knisiatn, FERNANDO. Effect of early maras-nsic malnutrition on subsequent physical and psycho-logical development. In: N. S. Scrimshaw and J. E.

Gordon (eds.) , op. cit., pp. 269-77.

24. Perspectives on human deprivation: biological,psychological and sociological. Bethesda, Maryland, Na-tional Institutes of Health, 196$.

25. Ponirr, E., and D. M. GRANDEE. Mental andmotor development of Peruvian children treated forsevere malnutrition. Rev Interim? Psicol 1:93-102, 1967.

26. and H. N. RICCIUTI. Biological and

social correlates of stature among children living in the

slums of lima, Peru. Am Ortbopsychiatry 39:735 -47,1969. .

27. Ecology, malnutrition. and mental de-velopment.. Psychosom Med 31: 193 -200, 1969.

28. Riccit:Ti, LI. N. Theoretical and methodologicalissues in the stuck nutrition turd mental development:malnutrition in infants. Unpuhlkhed manuscript. 1970.

29. Ric; iAli DSON, S. A., and A. F. Gttrist Acitmt(eds.) . Childbearing: its social and psychological..aspects. Baltimore, Williams and Wilkins, 1967.

30. SCRINISHAW, N. S., and ). Ii. Gouttost (eds.).Malnutrition, learning, and behavior. Proceedings ofan International Conference cosponsored .by the Nutri-tion Foundation, Inc., and The Massachusetts Instituteof Technology held at Cambridge, Mass., March 1-3,1967. Cambridge, Massachusetts, M.I.T. Press, 1968,

pp. 165-248.

77

31. S-rocti, M. B., and P. M. SvmTim. Does under-nutrition during infancy inhibit brain growth and sub-sequent intellectual develop menC Art b Pta Chdel 38:546-52. 1963.

32. and . The effect of under-nutritionduring infancy on subsecittent brain growth and intel-lectual development. S'Afr Med I 41:1027-30, 1067.

33. \'11:1iNoN. P. E. Intelligence and cultural environ-mcnt. London. Methuen, 1969.

34. WI NICK, MYRON. Nutrition and tell growth.

Nut?. Rev 26: 195 -97, 1968.

ENVIRONMENTAL CORRELATES OF SEVERE CLINICAL MALNUTRITION

AND LANGUAGE DEVELOPMENT IN SURVIVORS FROM KWASHIOR-

KOR OR MARASMUS'

Joaquin Cravioto and Elsa Delicardie"

At the community level, malnutrition, ormore specifically protein-calorie malnutrition,is a man-made disorder characteristic of thepoorer segments of society, particularly in pre-industrial societies. In the latter the socialsystem consciously or unconsciously createsmalnourished individuals, generation after gen-eration,' through a series of social mechanismsthat includes limited access to goods and serv-ices, limited social mobility, and restricted ex-periential opportunities at crucial points in life.

When applied to an individual, the term"protein- calorie malnutrition" is a genericname used to group the whole range of mildto severe clinical and biochemical signs presentin children as a consequence of a deficientintake or utilization of foods of animal origin,accompanied b) :ariable intakes of rich carbo-hydrate foods. Kwashiorkor and marasmus arethe names given in the U.S. and Common-wealth literature to the two extreme clinicalvarieties of the syndrome that occur in infantsand children. The appearance of marasmus orkwashiorkor is related to the age of the child,

'This work was supported by The Nutrition Founda-tion, Inc., the Association for the Aid of Crippled,Children, the Van Ameringen Foundation, the MoncllFoundation, and the Hospital Infantil de MA;co.

Division of Scientific Research, Hospital Infantil de laMAN, Mexico City, Mexico,

73

the time of full weaning, the time of introduc-tion of food supplements to breast milk, thecaloric density and protein concentration ofthe supplements the child actually ingests, andto the frequency and severity of infectiousdiseases present during weaning.

Research in the field of malnutrition hasrecently centered on the question of the laterfunctioning of individuals who suffered fromit in early life. A series of studies done invarious parts of the world have shown thatsurvivors of early severe malnutrition differfrom well-nourished children in a great varietyof functional ways ranging from psychomotorbehavior to intersensory organization (2, 4, 5,6, 7, 10, 12, 13, 15, 16, 17, 18). The problemnow is to separate the specific role that thedeficient food intake may have from the con-tribution of other factors that interfere withthe correct functioning of the individual. Thereason for this rests on the fact that malnutri-tion in humans is an ecologic outcome (11),with many of the factors that either cause oraccompany malnutrition being in themselves

capable of negatively influencing mental andbehaA;ioral development. It is obvious that thistype of research question can only be ap-

proached through the longitudinal observationof children at risk of appropriate controls.With this idea in mind, we have been engaged

since March 1966 in an ecologic study of acohort of children born in a community wherepreschool maim trition is highly prevalent, andwhere other factors related to the life of thechildren have variations of sufficient range topermit associative analyses to be carried out.

In brief, the project is the study of a totalone-year cohort of children, all born in a ruralvillage between March 1, 1966, and February28, 1967. The children and their families havebeen closely observed from the nutritional,pediatric, socioeconomic, and developmentalpoints of view in a coordinated manlier, withgreat attention to detail, and so far as feasibleusing validated research instruments. A goodnumber of the instruments were devised andtested by the project staff during the 10 yearsbefore the start of the cohort's induction.

The objective of the study is to analyze therelationship between the conditions surround-ing care of the child, especially as they affecthis nutrition, and the course of his physicalgrowth, mental development, and learning.

The main hypothesis to be tested is thatintellectual growth at all stages and school-ageperformance are related to the nutritional andhealth conditions to which the child hadbeen exposed.

The children first brought into the studyduring the prenatal period have been followedfor five years and will continue to be observeduntil they have completed their first sevenyears (March 1974), the earliest time at whichcertain crucial mental examinations can be

meaningfully applied. At that time the chil-dren can all be assessed in the relatively uni-form environment of a primary school.

While the particular focus of the project ison the relationship between nutrition and men-tal development, by the nature of both variablesthe design is that of an ecologic study of youngchildren in their family and social environments.

The ecologic approach was chosen becauseit constitutes a particular form of the naturalhistory method, which seeks to determine thenature of effective variables through a con-

74

sideration of their interrelations in a singlepopulation. When applied to the problem ofmalnutrition, it attempts to determine pat-terns of cause and consequence by consideringthe interrelations among nutritional, health,and social factors. Moreover, by orientingitself longitudinally the ecologic approach canidentify age-specific conditions at risk, relateantecedents to consequences at different devel-opmental stages, and integrate biologic andsocial time scales. It can take into accountboth the general and the microenvironment ofthe developing individual and deal with theinteraction of biologic and social variables.Perhaps most important for its usefulness is

the fact that it employs uncontrolled variationas the fact of study. A basic requirement forthe use of the ecologic method is, therefore,sufficient variation in the attributes to be con-sidered in the population studied. If suchvariation is present, associative analysis canidentify, segregate, and interrelate the factorsinfluential in affecting the consequences withwhich one is concerned.

Through the use of the ecologic method weplan to analyze: (1) the influence of social,economic, and familial conditions on the devel-opment of malnutrition; (2) the effect ofmalnutrition on physical growth, mental de-velopment, and learning, and (3) the inter-action of nutritional factors with infectiousdisease, family circumstances, and social cir-cumstances on the processes of growth anddevelopment.

The intensive analysis to be done can beunderstood only through a detailed acquain-tance with the setting of the study, the cohortunder observation, and the measures used.

The Setting of the StudyTo examine the complex set of variables with

which our investigation is concerned, we havehad to study a large group of children. Esti-mates of sample needs required the selection ofa community of sufficient size to provide atleast 250 births in the annual cohort to be fol-

lowed longitudinally. The community chosenalso had to contain a considerable range ofsocial and nutritional conditions as well as beone in which the population was likely to becooperative, willing to enroll, and remain in thestudy. For a longitudinal study, it was clearlyessential that the population be relatively stableand that a high proportion of the families andinfants enrolled at birth for study be likely tocontinue to live in the community for theduration of the inquiry period.

The community chosen met these require-ments. Selection was based on previous experi-ence with rural communities and field studies.In one of these studies the present village wasa participating community and had shown ahigh level of population stability over time,excellent cooperation, and a wide range ofvariation in social, economic, familial, andhealth attributes. Furthermore, its annual birthexpectancy was 300.

The village is located in a semitropical, sub-humid zone in southwestern Mexico in a pri-marily agricultural region where arid hillsides

alternate with fertile valleys and meadows. Itis at an altitude of between 900 and 950 mabove sea level and has a hot, subtropical cli-mate modified by its altitude. The medianannual temperature is between 23° and 25°Cin the shade, and the climate ranges from chillywinters to very hot summers in which dailyhigh temperatures of 40 °C are common. Asmall river whose waters are used for irriga-tion, laundering, and other general purposesruns through the village.

As is characteristic of Mexican rural villagesin this region, the town radiates outward froma shaded central plaza along a series of unpavedand rutted dirt streets, which are related to oneanother to form' roughly quadrangular blocks.A street map of the village is presented in

Figure 1. The small inset at the upper rightindicates the arrangement of households withintwo specific blocks.

The area surrounding the townsitc was en-tirely agricultural, with sugar cane constitutingthe major commercial crop. Small amounts ofcotton and rice were also grown commercially.

64 511 6

10/92191 Aoe

n

itfaT7' I 64

DE-4i La U L

o.

65

Figure 1. Plan of study village; inset at upper right shows arrangement of house-holds in two specific blocks.

7 75

Interspersed among the large commerciallyorganized fields were small family parcels andrentable areas that the villagers used for theproduction of food crops principally corn,chilis. tomatoes, and other garden crops andfruitsfor their own consumption.

In 1965 we carried out a census of all villagehouseholds. Our findings indicated the presenceof 5,637 persons from 0 through 85 years Ifage, organized in 1.041 families. The numbersof men and women were insignificantly dif-ferent: 2.830 men and 2,807 women. Thedistribution by age and sex in the populationare presented in Figures 2 and 3. As may be seenfrom both figures, the population was relativelyyoung: fully 50 per cen' of the inhabitantswere under 15 years of age, and 80 per centwere under 35. In a stable community such asthis, these age ratios naturally reflect a reducedlife expectancy calculated from birth. Nonotable sex differences in age-specific frequencywere found.

80

u_

60u.1

iY

uj 40

2

Birth rates over the previous 20 years hadbeen at a mean of approximately 55 per thou-sand and led to an annual expectancy of ap-proximately 300. In the 12-month series ofbirths upon which the current study is based,the predicted expectation was fulfilled.

The principal occupation of the villagerswas agriculture, although relatively small num-bers of people wer.r employed as workers andartisans, and a still smaller number were !n-gaged in commerce or in the practice of a

profession. The occupations, as reflected by themain sources of family income, are presentedin Figure 4. Approximately 65 per cent of thepopulation gained its livelihood directly fromagriculture; 12.5 per cent worked at a varietyof jobs, including transport work, labor in thesmall cotton gin or mattress factory, carpentry,masonry, or other skilled and semiskilled crafts;and 9 per cent re small tradesmen, shop-keepers, or teachers. Included in the last groupwere one architect, one engineer, and two

5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85AGE IN YEARS

(FIVE YEAR INTERVAL)

Figure 2. Cumulative percentage of villages population distribution by age.

76

AGE IN YEARS

80 - 85

75-79

70 -74

65-69

60-64 MALES FEMALES55-59

50-54

45-49

40-44

35-39

30-34

25-29

20-24

15-19

10-14

5-90-4

10

PERCENT IN POPULATION

Figure 3. Age pyramid by sex.

6

LLI

1.0ILO 4

0 -J0

Z

CCwa-

64.97

12.5% 19.8%1 12.8%

AGRICULTURAL WORKERS CCMMF_RCE ATand and H

ARTISAN' PROFESSIONSOME

Figure 4. Main sources of family income in village."At home" families are those in which mothers andchildren were mainly dependent on some other familialor social source of support.

physicians who served both the village properand the surrounding communities.

The social and economic status of peoplewho worked in agriculture is by no meanshomogeneous, as can be seen in Figure 5. Mostworkers were agricultural da3 laborers (jorna-

77

60-

40

20J

1

69 2%

WAGELABOR

234

I -1 1j-2MFAMILY RENTEC SMALL

PLOT PLOT OWNER

TYPE OF AGRICULTURAL WORKFigure 5. Percentages of people engaged in differenttypes of agricultural work.

!ems), whose employment is seasonal andwhose income is marginal. They supplementedtheir income by the seasonal planting of foodcrops on communal la.ids at sonic distancefrom the village. A second group comprisingsome 23 per cent of those engaged in agricul-ture had family plots that arc cultivated forboth small-scale commerce and self-use, Thebest-situated cultivators were either small c n-ers or renters of relatively high-grade farm-lands; taken together, these two groups con-stituted approximately 7 per cent of thosewho drew the main source of their familyincome directly from agriculture.

A few words are required to describe thedifferent types of land-holding if the differentkinds of agricultural work are to be under-stood. The jornalero owns no land and worksentirely for wages. He is, in the main, notinsured for social and health services and rep-resents the most economically deprived andunstable segment of the population. He maybe described as an agricultural proletarian.

The ownership of land by individuals is oftwo kinds, the first of which is the family plot(ejido). These are small parcels of land, rang-ing generally from 0.5 to 1.25 hectares, that

were distributed to landless peasants (Peones)at the time of the agrarian reform (1917).Such land is held by the family in perpetuity solong as it is worked, but it may not be sold,traded, divided, or attached for debt. All theejidos owned by the villagers had been amalga-mated into a sugar cooperative. The secondtype of individual land ownership is representedby the small proprietors whose plots are gen-erally larger than the ejidos. Their plots may bebartered, sold, mortgaged, attached for debt asprivate property, and divided or sold with nospecial limitations. The rented lands are gener-ally larger, and thoseeconomically better offtural groups.

Until 30 years ago

who operate them arethan the other, agricul-

the village was almosttotally agricultural in character. Since then ithas been in a state of slow transition to a mixedeconomy representing more advanced levels ofboth industrial technology and agriculturalorganization. The beginning of the transitionwas marked by a national law authorizing andfacilitating the development of agriculturalcombines and cooperatives. Shortly after thislaw was enat..ed, a large cane-growing coop-erative and a sugar refinery 17 km from thevillage were established. Much of the landsurrounding the village w .s assimilated intothis large cooperative. A mattress factory em-ploying 20 workers was established 12 yearsago, and this was followed live years later bythe construction and operation of a smallcotton gin.

These economic changes have been accom-panied by greater availability of transportation,the improvement and construction of roads,and a considerable increase in interaction be-tween people in the village and more advancedurban and semiurban centers. The technologicadvances have been accompanied, too, by theimprovement of a variety of village ser-..cesincluding school, a central water supply, anda social welfare and health center. The villagenow has one kindergarten, two primary schools,and one school accepting pupils to the ninth

grade. Attendance at school is compulsorystarting at seven years of age and continuingfor six years; beyond this, study may be vol-untarily continued for three more years insidethe village and on higher levels outside it.

Variation in health conditions in the com-munity and in the relative opportunities avail-able for the growth and development of

children are reflected in two sets of data. Thefirs,. of these deals with the weights and heightsof school children. A comparison of the growthattainment of children attending school in 1965with the growth of children who were at schoola generation earlier suggests that the techno-logic changes were not accompanied by eithera change in mean height or weight for age orany diminution in the variability of these char-acteristics in the population. The Wetzel Gridspresented in Figure 6 illustrate this lack ofchange.

The second set of data uses tuberculin reac-tivity as an index of health risk. It reflects thehigh frequency of the children's exposure tosources of infection. Age-specific values forpositive reaction to tuberculin show a con-tinuing rise of positive reactors with age, as

can be seen in Table 1.

110 120 BO 140 1E0

HEIGHT

78

AGE IN YEARS

Figure 6. Relation between height and weight in twogenerations of school children.

Table 1. Distribution by age of positive PPD reactorsamong village children.

Age (years) Positive reactors t,Yo'

3

4 0

5 3.7

6 2.1

7.S

12.4

9 S.7

10 7.6

11 2.3

12 2.6

13 1

14

15 20.0

16 37.5

Average 11.2

As a final step in assessing the suitability ofthe village for our purposes, we made a sam-pling survey of households with children ofpreschool age for clinical evaluation of chil-dren and the assessment of the incidence ofsubnutrition as indicated by a depression ofweight. for age (Table 2).

The figures indicate that at all levels in thepreschool period, mild, moderate, and severedegrees of malnutrition were present in thecommunity, with a frequency that made itfairly representative of communities in whichchronic subnutrition is prevalent. All the dataconsidered provided a basis for choosing thecommunity as a suitable site in which to con-duct a longitudinal study of the ecology ofgrowth and development in a total annual

Table 2.

village.Nutritional status of preschool children in

Age(months)

No. ofchildren

Malnutrition (%)

Normal(%)

ThirdDegree

Mild-Moderate

0-5 132 0 5 95

6-11 112 3 9 CS

12-23 210 5 18 77

24-35 207 9 21 7'd

36-47 212 8 8 84

48-59 201 5 11 84

79

cohort of births over the preschool and schoolyears.

The Birth Cohort

The ;en

Of the 300 infants born during the 1 2-

month period in -xhich the cohort of births wascollected, there were equal numbers of boysand girls. The social origins of these childrenclosely matched the distribution of principalsources of income and occupation in the villageas a whole. Minor diffuences from the overalldistribution may be attributed wholly to theconstriction in age range associated withreproduction.

The manner in which principal soun:e. offamily income. were represented in the cohortare presented in Figure 7. The great majorityof families were agricUltural, and 66 per centof all children were from such families. Sixteenper cent of the children were born to familiesof workers or artisans, with four times as

many of the former type of family representedas of the latter. Equal numbers of tradesmenand professional families were present in the4.6 per cent of the births among these social

13.37

AGRrtILTURAL WORKERS COMMEALE AT

ARTISANS PROaPEISSIONS"O"

Figure 7. Main sources of family income in the cohort."At home" families are those in which mothers andchildren were mainly dependent on some other familialor social source of support.

groups. Sonic 13 per cent of the births occurredin families in which it was difficult or impos-sible to define an intrafamilial source of in-come. These families were ones in which themother and children were in the main de-pendent upon some other familial or socialsource of support; they are labeled a/ homein Figure 7.

The distribution of birth weights in theannual cohort is presented in Figure 8. Amongthe 291 infants who lived long enough to beweighed, there were eight weighing 1,500 to1,999 g, 28. weighing 2,000 to 2,499 g, 147weighing 2,500 to 2,999 g, 86. weighing 3,000to 3,499 g, and 22 weighing 3,500 to 3 999 g.Mean -birth weight was 2,898 444 g. Abirth weight of less than 2,500 g was thusobtained for 12.3 per cent of the cohort,whereas only 7.6 per cent of the childrenweighed more than 3,500 g. As would haveb -in expected from other bodies of data, them::an birth weigl.t of boys was significantlyhigher than that of girls (Figure 9). The meanbirth weight of boys was 2,977 394 g andthat ol girls 2,860 408 g. The mean differ-ence if 117 g was significant at the 0.02 levelof onfidcnce.

The distribution of total body length atbirth for the children in the cohort is pre-sented in Figure 10. The median body lengthwas 48.5 cm, with 25 per cent of the children

1500.1999

7.6% I

2003-2499 2500-22:6 30004493 3500-3999

BOCY WEIGHT IN G

Figure 8. Birth weight of children in annual cohort ofbirths.

01= 2625Md: 287803. 3185

80t

;.;

I

_ __. 1 1roo xioo n lvf. 27`t 3411

WEIGHT IN GM

Figure 9. Sex distribution of body weight at birth.

40

*3' 30

26

u.

40 42 44 46 45

01, 47.0cr,Mcl: 455

,.95

50 52 54

HEIGHT IN G

Figure 10. Distribution of total body length in 'cohort atbirth.

having a bcdy length of less than 47 cm andan equal number having body lengths between49.5 and 53 cm. As was the case for birthweight, mean body length at birth was signifi-cantly higher in boys than in girls. As maybe seen from Figure 11, boys had a meanlength of 48.7 cm and girls one of 48.0

cm. This difference, though small abso-lutely, is statistically significant (t = 3.3;p < ).

Mortality and emigration

Of the 300 children delivered, 296 werealive at birth and four were dead at delivery.The stillborn were born into no particularoccupational group: one child was born to afamily whose principal source of income wasagricultural day-labor, one to a cultivator of

20

Boys

Girls

40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55

HEIGHT IN CM

Figure 11. Distribution of body length at birth by sex.

rented lands, and two others to owners offamily plots. Of the 296 live-born infants,seven died during the first week of life (23.5per 1,000) and three others during the re-mainder of the first month. If the stillbornsand those who died during the first week oflife are combined and considered as perinataldeaths, the rate for such deaths is 36.6 perthousand. Combilooti all deaths of live-borninfants during the first month results in a

neonatal death rate of 33.6 per thousand.In the main, the infants who were stillborn

or who died during the first month of life wereof low birth weight. Because of custom, unfor-tunately, those who were stillborn or who dieda few hours after birth could not be weighed.The remaining seven children had birth weightsbelow 2,300 g.

Nine infants died during the remainder ofthe first year of life, seven between the secondand sixth months of life. It is of interest thatone of the two infants whose dcatia occurredfter six months of life was mortally bitten

by a scorpion. Taken as a whole, these dataindicate that 19 live-born infants in the cohortdied in the first year of life, with a resultantinfant mortality rate of 63.3 per thousand.

The children who died were not random!)distributed in the cohort with respect to weightor body length at birth. The mean birth weightin the cohort was 2,898 g, and the mean birth

81

weight of the children who died during thefirst year of life was 2,536 g. Similar, meanbody length of the deceased was 45.8 cm, ascontrasted to a mean body length of almost 49cm in the cohort as a whole. Both of thesemean differences were significant at less thanthe 0.01 level of confidence. It was of coursepossible that these mean differencCs were con-tributed by an excess of premature infantsamong the infants who were stillborn or whodied in the first year of H.!. Such a possibilityis supported by the fact that six of the infantswho died were prernatures. That number re-sults in a prematurity of 260 per thousanddeliveries, which is more than twice the rateof 124 per thousand deliveries present or inthe cohort as a whole. If one considers onlyinfants above 2,500 g at birth among the still-born and infant deaths, however, one still notesa preponderance of poorly developed children,with 63 per cent of them having birth weightsbelow the median of the cohort and the re-mainder with weights only slightly higher thanthe median birth weight. Only three of the 23infants had body lengths equal to or greaterthan the median of the cohort as a whole.

Two were exactly at the median and one was1 cm higher. All other infants were between1.5 and 7.0 cm shorter than the median bodylength for the whole cohort. It appears, there-fore, that those who were either stillborn orwho dial during the first year of life were lesswell developed in utero than the survivors.

After the cohort's first year of life 10 morechildren died. With the exception of two chil-dren whose deaths were due to accidents (severeextensive burns, and bronchoaspiration) anda child who died of hypoprothrombinemia andpurpura, all other deaths in this age period canbe directly related to infectious disease accom-panied in most cases by severe malm.:1-Tition.

Table 3 shows the sex, age, and probablecause of each death that occurred during thecohort's first five years.

Besides the 300 children born in the villageduring the calendar year 1966-67, twelve fami-

Table 3. Mortality in cohort during first five years of study.

Period Number Code Scx

1 13.12 M

Still- 2 13.5 F

born 3 13.26 M

4 4.22 F

1 12.24 F

2 13.25 F

First 3 7.12 F

week of 4 9.18 M

life 5 6.11 NI

6 8.24 M

7 13.2 M

Days 1 4.23 F

8-30 2 2.3 F

3 12.15 F

1 13.29 M2 13.16 M

3 12.14 F

Days 4 10.13 F

30-360 5 11.2 F

6 11.2.7 .F

7 12.17 F

8 7.11 F

9 5.24 M

1 3.22 M

2 13.6 F

3 1.21 M

4 11.6 F

One to 5 4.19 Mfive years 6 2.22. F

7 11.20 M8 3.1 M

9 9.13 M

10 8.22 F

Age at death (days)

0

0

00

1 hour1 hour5 hours5 hours6 hours2 days3 days

unknown.unknownunknownunknown

Cause of death

unknownunknownbronchoaspirationprematurityh ;aline membranecc,;enita: malformations and brouchopneumoniabronchopneconi

9

17

21

ADO incompatibilityelectrolyte disturbanceprematurity

40 electrolyte disturbance due to diarrhea43 pyclonephritis76 electrolyte disturbance due to diarrhea78 septicemia85 bronchopneumonia

130 electrolyte disturbance due to diarrhea139 electrolyte disturbance due to diarrhea268 scorpion bite312 electrolyte disturbance due to diarrhea

375 electrolyte disturbance due to diarrhea400 bronchopneumonia and malnutrition405 bronchoaspiration435 bronchopneumonia and malnutrition654 acute miliary tuberculosis737 kwashiorkor amebiasis

1036 kwashiorkor diarrhea1229 electrolyte disturbance due to diarrhea1240 third-degree burns1413 kwashiorkor purpura

lies came to live in the village during the timeof induction of the birth cohort. These familieshad children less than one year old, and theywere included in the longitudinal study.

In the course of the last five years a total of50 families have left the village. Reasons foremigration are mainly related to better work-ing conditions in another, larger town orMexico City itself. Table 4 shows the timespent by the children in the study.

As may be seen from the cumulative fre-quency figures, all 50 children who left thevillage were examined at least during the new-born period. We have data about the first

82

three months of life for 44, about the first sixmonths for 34, about the first year for 22,

about the first two years for seven, and aboutthe first three years for two. Family data wereavailable for all 50 children who emigrated.

Familial and social background

Broadly considered, background factors inthe cohort were of three kinds: first, the motheras a biologic and social organism; second, thefamily structure; and third, objective circum-stances of life such as sources of family incomeand the family's housing conditions. Each kindof factor will be considered it turn, with the

Table 4. Time spent in study by children whose families and the median age was 24 years, with 75 percent of mothers under 30 years of age. Theemigrated from village.

Days tinderobservation

Number ofchildren

Lanni 'Rivefrequency

0

1-3

0

0

4- I 5 I 50

16-10 1 49

31-90 4 48

91-180 10 44

181-270 7 34

271-360 5 27

361-420 5 22

421-540 3 17

541-610 4 14

611-720 3 10

721-810 1 7

811-900 2 6

901-990 1 4

991-1080 1 3

1081-1170 0 01171-1260 1 2

1261-1350 0 1

1351-1440 1 1

exception of sources of income, which hasalready been discussed.

We shall begin our consi&ration of the back-ground characteristics of the children by view-ing some features of the mother as a biolbgicorganismher age, height, weight, and previ-ous pregnancies. The age distribution ofmothers in the cohort is presented in Figure 12.Ages ranged over 30 years: the mothers in-cluded two girls who had their first childrenat age 13, and one woman who had a child atage 43. The mean age was 25.6 ± 6.8 years,

cc

w 15

zz

_r

Q1.20 yrs.Md. 24Q3.30

Mean: 25.6S.D. 6.8

15 2 3

AGE IN YEARS

Figure 12. Age distribution of mothers. in cohort.

83

distribution of maternal age tends toward bi-modality and, as a whole, contains sufficientvariability to permit. associative analysis.

The height, and eights of the mothers alsovaried widely (figures I3 and 14 ) . Themothers' heights ranged from 133 to 165 cm,with a mean value of 148.2 ± 2.8 cm. Theheights were relatively normally distributed,with a slight tendency toward increased fre-quency in the lower range of stature. Themedian value was 147.5 cm, and 75 per centof the women were less than 153 cm tall.

Weight of the mothers ranged from 32 to86 kg. This upper weight value was ma,;cdlyatypical and is so indicated by a break in tht.distribution of values on the x axis of Figure14. Mean weight was 53.0 ± 4.8 kg with amedian value of 51 a slight

80

60-cr

4

a

kg, suggesting

Q1.145 5 CmMd. 14703, 52 5

Meant 148 7SD. 2 8

133 1

Figure

HEIGHT154 157 160 163 166

IN CM

13. Height distribution of mothers in cohort.

ce

20

z1

01.47 0 KgMd t 51 0Q3.560

Mean. 53.1SD. t4 8

32 343636 2 46 Si) .54 58 62 66 70 74 66

WE 'GHT IN KG

Figure 14. Weight of mothers in cohort.

excess in frequencies at the lower end of thedistribution. l3oth height and weight exhibitsufficient variability to make associative analy-sis feasible.

The distribution of pregnancy numbers inthe cohort is presented in Figure 15. Preg-nancy number ranged from one to more than11, with the highest being 16. Relatively equalnumbers of births occurred at pregnancy num-bers ore through four. Higher pregnancy num-bers in the cohort tended to Occur withdiminishing frequency. From the viewpointof epidemiologic evidence in obstetrics, it is

not meaningful to consider the distribution ofparities in terms of mean value. Available evi-dence suggests that special conditions of riskattend first births and births that occur afterthe fifth pregnancy; enough cases in each ofthese parity ranges exist to permit analysiswithin the cohort of births.

In addition to the foregoing biologic charac-teristics of the mothers, at least three othercharacteristicsthe mothers' personal hygiene,literacy and educational level, and contact withinfOrmation mass mediamay be considered inrelation to the child's condition at birth, Rat-ings of personal hygiene ranged from a low of20 per cent to a high of 100 per cent (Figure16). The median score was 56.6 per cent, withthree-fourths of the mothers scoring below anestimated cleanliness level of 76 per cent. Widevariation in personal cleanliness existed among

5

14.1 14.112.7 13.0

100

627.7

3.9 59 42 62

2 3 4 5 6 7 8 9

PREGNANCY NUMBER

10 11

Figure 15. Distribution of pregnancy numbers in cohort.

84

20-

7-11

'45

1 14 1

0 29 33 40

1

232: ET:16

ti

:5:131

II

. l _L-69 70 80 90

to 107

a,50 5`4

Q3 75:7".

PERCENTILE SCORE

Figure 16. Personal hygiene of mothers in cohort (basedon 289 cases).

the mothers, and personal hygiene subgroup,Could readily be defined.

The mothers' literacy and school attainmentsare presented in Figure 17. Almost half themothers were completely illiterate. Another10 per cent had either become literate throughadult literacy campaigns or acquired basic lit-eracy by completing one school grade. Only6.3 per cent had completed the full primaryschool curriculum of six years, and the remain-ing 1.5 per cent had schooling beyond primaryschool. From a functional point of view itwas possible to group the mothers into fourclauses: (1) the illiterates, (2) the adult liter-ates and those who had completed the first andsecond grades, (3) those who had completedthe third, fourth, or fifth grades, and (4)those who had completed primary school. These

AMOUNT C= SCHOOLING

Figure 17. Literary and school grade completed bymothers in cohort.

functional groupings have been used in laterassociative analyses.

As may be seen from. Figure 18, very fewof the women had any exposure to television.Half had little or no regular contact with theradio. :Ind almost 70 per cent read no news-papers except episodically. If consideration ofnewspaper reading is limited only to the literatesegment of the population, slightly -fewer than50 per cent of the mothers actually did so.

Thus, in the present study we have viewedradio as the most effective medium of dissemi-nating mass information and have restrictedour consideration of relation between thechild's characteristics and the mother's contactwith mass media to this medium.

Family size and the sanitary characteristicsof the household are presented in Figures 19and 20, respectively. There was considerablevariation in both of these attributes. Familysize ranged from a single-child family of chreeindividuals to both nuclear and extended fami-lies of a dozen or more members.

The sanitary characteristics of the householdvaried widely. Most households were substand-ard, but a considerable number exhibited goodto excellent conditions and facilities. Sufficientvariation existed to permit the sanitary charac-teristics of the household to be related, as a

cecew1--0

6

zwce 2wo.

5Z9327

144

NO YES NR NO YES NR

NEWSPAPER RADIO

750

14.498

NO Yt:c:NR

T V

Figure 18. Contaz :' mothers in cohort with ;nformationmass media.

85

20-

LI4 5 6 7 8 9 10

NUMBER OF FAMILY MEMBERS

Figure 19. Size of families in cohort.

20-

10 20 30 40 50 60PERCENTILE SCORE

Q1.137%Md C12.240%

Q3z 391%

70 80 90 100

Figure 20. Sanitary characteristics of households incohort.

differential background factor, to the child'scharacteristics.

The Variables

In the course of the study we have beenconcerned with two sets of variables: those

relating to the child's family circumstancesand background environment, and those relat-ing to characteristics of the child himself. Wehave avoided the terms "antecedent" and "out-come variables," 1>ecause among each of thegeneral sets of factors certain characteristicsmay with equal justification be considered asantecedents or consequences of others withinthe same set. Thus, the mother's educationmay well affect her patterns of personal hygieneand preferred mode of child care.among the variables that characterize the child,a factor such as birth weight may well be animportant antecedent for outcomes at laterages.

In general, three types of kckground fac-tors have been considered: fam:.y pattern, 'o-

logic anu psychosocial .characterist: of the

parents and caretakers, and macroenVironmen-tal characteristics of the child ranging- fromsurvival to physical and behavioral changes

with age.Data for ratings of variables were obtained

by interview, by direct observa.:..1 and mea-surement, by clinical assessments, and by theapplication of special tests.

The reliability of all measures and ratingshave been assessed, and any measures withexcessive intra- or interexaminer, or score:-

restore error levels were either improved bymodification or eliminated. For somatometricmeasurements including weight and height,scales were regularly recalibrated and exam-ined. Errors beyond the 2 per cent level wereconsidered excessive. For rating scales, relia-

Table 5. Studies ,nade at childi 3n in cohort.

Variablz Measurement

1. Growtha. Physical

b. Mental

bility levels e..pressed as retest or rescore corre-

lation cot h.:icats, or both, of less than 0.82were considered unacceptable.

The director of the study, the senior psy-chologist, the senior pediatrician, and the socialworker-nu:,-Rionist. acting as a research com-

mittee, shared the responsibilities of qualitycontrol and maintenance of strict standardiza-tion and completoless in data collection. Thesenior biostatistic of the project has been amember of this committee since the time hejc,ined the team.

The types of studies already made of eachof the 229 children remaining in the cohort atfive years of age (300 births plus 12 arrivalsduring the first year of study, minus 33 deathsand 50 emigrations), as well as the tests to begiven those remaining in the study in their

fifth and sixth years, are shown :n Table 5:

Body weightHead, chest, height, and arm circumference

Skinfold thickness

Bone age

Psychomotor, adaptive, language, and social-personal development

Bipolar concept formation

Finger-thumb oppositFinger localizationWei li.Irr preschool and primary scale of intelligence

response to cognitive 'demands-rotor and movement skills: static ! lance;

dy balance; visual-motor coordination involv-ing aiming and accuracy, motor control, and in-hibitionSpontaneous languagePsycholinguistic abilities: language decoding (mean-ing); language encoding (expression); grammaticallevels of decoding and encoding; development andgrowth of grammatical languageVisual perception and form recognition

86

Age period

0-5 years fortnightly0-3 years monthly3-4 years bimonthly4-5 years thrice yearly0-3 years monthly3-4 years bimonthly4-5 years thrice yearly4-5 years once yearly

0-2 years monthly ..

2-3 years quarterly3-4 y --irs thrice yearly2-3 years thrice yearly3-4 years twice yearly4-5 years twice yearly3-5 years Ivce yearly4-5 years twice yearly4-5 years twice yearly4-5 years twice. yearly

4-5 years twice yearly5-7 years twice yearly

5-7 years twice yearly5-7 years twice yeari.:

Frequency

Table 5. (cont.)

Variable

2. Nu trition

3. Heal th

NIcasurement

Intersensory integration: visual-haptic: visual-kines-theric, kinestheric-haptic, auditor'- visual

Analysis and synthesis of geomr ic formsLearning strategies: Concrete ooeratauis iPiaget"'

Food consumption (child)

Age period Frequency

Family food consumption

Clinical signs of child's malnutrition

Prenatal and delivery historyPediatric examination; morbidity history of .nildand of each member of the household

4. Mother-child interactionTime-sample observatio., and inventory of homestimulation

5. Sociocultural characteristicsa. Familial

Family composition and Organization; family type;family size:main source of income and annual income;sanitary facilides; social morbidity; social change;informal communication net; mortality record

h. Parentali. Biological

ii. Sociocultural

Age, weight, heightParity and reproductive history"

Personal cleanliness

Formal education and contact with mass mediaNutritional knowledge"; health education"; pro-clivity toward change"; cultural mobility"Psychologic profile"; maternal attitudes; intelligenceperformance"

Mother only.

Language Developmery: and Malnutrition

Having described in detail the setting of thestudy and the birth cohort, we would now liketo present the preliminary results of the studyof certain language eatures in a group ofchildren who developed severe clinical mal-nutrition.

During the first five years of life of thecohort, 22 children-14 girls and eight boys-were found to be suffering. from severe clinicalmalnutrition. It must be said that such cases

87

5-7 years twice yea,ly-7' years twice yearly

0-7 ye.,', twice yearly0-7 yeirs once Year'0-7 years once %early

0-2 years inontb'y2-3 years lluarterly3 - -i ..ears twice ye y4-5 years once ye ..!3-4 years twice yearly4-5 years once Yearly0-5 years fortnight' }'

On cc

0-5 years fortnightly

0-3 years .011C111:

0-3 years quarterly3-4 years twice yearly4-5 years once yearly

0.5 rears once Yearly0-5 years once

0-5 years for t y

0-5 years once0-5 years once Yearly

0-3 years monthly3-5 .ears thrice yearly0-5 years once yearly

0-5 year. once

0-5 years once

appeared despite all medical efforts to preventthem. The patients' age at the time of diagno-sis ranged from four to 53 months, with onlyone child below one year of age, nine childrenbetween one and two years, eight between twoand three years, three between three and fouryears, and one child 53 months of age.

Fifteen of the 22 cases matched the clinicalpicture of kwashiorkor; the other seven caseswere of the marasmic variety (/ ) . The pro-portion of marasmus in females and males was4:3, but the number of females with kwashior-

kor was twice the number of buys. Because ofthe small number of cases, these differences arenot of statistical significance.

Ten children, six with kwashiorkor and fourwith marasmus, were treated at home, and ninechildren with kwashiorkor and three withmarasmus were treated in the hospital. Theaverage duration of hospital stay was 30. days,.laiJ none of- the children stayed longer than 60

days. No deaths occurred in the hospital-treated group. In contrast, three of the 10children treated at home died. Of the latter,two had kwashiorkor and one had marasmus;their respective ages at the time of diagnosiswere 12, 14, and 22 months. All tlree patientswho died did so within 15 to 60 days of diag-

nosis. Of the 19 survivors, one child emigratedfrom the village after his discharge from thehospital, leaving a total of 18 cases for study.

In the present report, perceived languagedevelopment in the 19 Addl. ., who developedclinical severe malnutrition before the age of39 months has been compared with the lan-guage development of a group of children ofthe same birth cohort who were never con-sidered as severely malnourished and who werematched at birth for gestational age, bodyweight, and total body length.

As may be seen in Table 6 and Figure 21,

mean language development, as measured bythe Gesell method (14) , is very similar in

index patients and controls during the first

'ruble 6. Language development scores of severely malnourished children and matched controls (days' equivalent).

Age (days) Birth 180 360 540 720 900 1080

Pot or present severerfalnutrition.

27 ± 3.2 167 ± 14.4 289 47.2 385 ± 86.0 467 ± 102.7 534 ± 103.0 657 ± 119.5

Cof;Prols 28 ± I 177 ± 21.2 334 ± 55.4 490 ± 73.3 633 ± 93.4 785 ± 143.1 947 ± 135.2"t" test 1.37 1.69 2.69, 3.904 4.80 5.804 6.53'

° Significant at less than 0.01.

a, 9-ro

d

450

. I 2

aottf

p

0

0

t tt

M.,InconshoCapos.s ofSever,.MainOntly,

15 18 21 24 27 30 33 19

CHRONOLOGIC ALE (Maths)

Figure 21. Mean language development as age function in severely malnourishedchildren and matched controls.

88

year of life when only one case of severemalnutrition had bee, diagnosed. As timeelapsed and more children came down withsevere mainutt'itiont difference in languagepe:iormance favorable to the matched controlsb:can evident. Th_ difference was more pro-nounced at each successive age tested.

s:ot only were tnear values significantlylower in the index cases, the distribution ofindividual scores was also markedly differentfrom that obtained in the control group. Thusfor example, at three vears of age (Figure 22) ,

11 children of the control group had languagesc, es above 1,021 days' equivalent and onlyone child scored below 720 days. In contrast,while none of the children with past malnutri-tion scored above 960 days' equivalent, 12

children had values below 720 chys and threeof these children had language performancessix months below th shown by the controlchildren with the loss,. scores.

Concept development and particularly theemergence of verbal conceptions have longbeen viewed as a bask factor in the develop-ment of human intelligence. The emergenceof the conception of opposites and with itbipolar labelling represents an early and readilymeasured aspect of the development of con-cepts in young children. As a consequence of

6

4

2

0--oConlrols

541-540 -600 'EEO -720 -7E0 -540 -9C0 -960 -IWO IOW

LANGUAGE SCOPE days)

Figure 22. Distribution of language scores at 1,080days of life of index patients and controls matched atbirth for sex and body length.

89

their concern with improving the school per-formance of disadvantaged children, FrancisFl. Palmer and his colleagues at the Institutefor Child Development and Experimental Edu-cation of the City L'inversity of New York,formed the view th it the develoment of a

progressively more difficult series of bipolarconcepts could be used for the systematictraining and enrichment of language experi-ence. Accordingly, as part of their studies onthe effects of intervention programs startedat age two, they developed a test coveringboth "poles" of 23 concepts (e.g., big-little,long-short, in-out) in two different situations.Most of the items included require the childto select an object representing a given polefrom two objects differing only with respectto their position on one of the concepts' con-tinua. The items are grouped into two formsso that each form contains items covering bothpoles of each concept. The forms differ onlywith respect to the setting in which the con-cepts are placed. The score derived from thetest provides a measure of the child's knowl-edge of various categories that are commonlyused in organizing sensory experience.

Although Palmer and his associates have notviewed the series of bipolar concepts that weredeveloped to be a language test, it is implicit intheir protocols that the progressively more diffi-calt training series could indeed be used in

itself without training . a measure for assess-ing the natural acquisition of bipolar conceptsin young children. To test this hypothesis,we administered 22 of the 23 concepts selectedby Paln.er as a repeated test of bipolar conceptacquisition at ages 26, 31, 34, and 38 monthsto a total cohort of children living in a pre-industrial society (9). All items were pre-sented to the 229 children at all ages inde-pendently of the number of successes or

failures. In all instances the order of presenta-tion was the same, beginning with the firstitem contained in Form 1. Data obtained atthe successive ages tested clearly demonstrateda developmental course of competence in re-

sponse to tasks involving the utilization ofoipolar concepts.

The competence in bipolar concept acquisi-tion in children with past or present severemalnutrition and matched controls at successiveages is presented in Table 7. and illustrated inFigure 23. As may be seen, the mean numberof bipolar concepts present in the index chil-dren is significant' lower than the meannumber of concepts shown by the controlgroup. It is important to remember that after40 months all the children included in she

malnourished group actually represent cases

rehabilitated from sevt e clinical malnutrition.i.e., survivors considered as cured of the disease.It may be noted in this respect that the meanvalue found in the index cases at 46 monthsof age is almost -twice the value obtained at38 months. Nonetheless, the increment is notenough to bring the index children to thevalue shown by the controls. In other words,the lag in language development found in

severely malnourished children continued to bepresent after clinical recovery had taken place.

It has been repeatedly stated that humanmalnutrition does not occur in a vacuum, butthat malnutrition is the outcome of an ecologicsituation characteristic: of the preindustrialsocieties (8). Because we are confronted by aphenomenon with multiple causation, beforeinterpreting our findings as due to the ante-cedent of severe malnutrition it is necessary totry to sort out what other factors beside thenutritional deficiency may be operating tointerfere with the normal development of thesechildren. In trying to answer this question

it; 3C 74 A.E1 fiC .+A trc,

A G E MorthS)

Figure 23. Number of bipolar language concepts presentin children with and without antecedents cic severe clin-ical malnutrition.

we have compared the macroenvironment andsome features of the microenvironment of -hefamilies of the severely malnourished childrenand of the matched control group.

Broadly considered, the macroenvironmentalfactor, to of three kinds relating, first, to theparent.) as biologic and social organisms; second,to the family structure; and third, to objectivecircumstances of life such as sources of familyincome, income per capita, and sanitary facili-ties present in the household. Since a detaileddescription of these factors has already beenpresented, we will consider now the associationof each one with the presence or absence ofsevere malnutrition.

Biologic characteristics of the parents

The variables in age, height, and weight ofeither parent, number of pregnancies, and num-ber of live children in the family failed to dis-crimi.late between families with and withoutseverely malnourished children.

Table 7. Mean number of bipolar concepts in children with past or present severe malnutrition and in matchedcontrols.

Age (months) 26 31 34 38 46 52 58

Past or presentmalnutrition

1.61 ± 1.26 3.92 ± 2.56 4.85 ± 3.15 6.07 ± 2.94 12.16 ± 4.13 1535 ± 3.05 17.21 ± 2.60

Controls 3.54 ± 2.11 5.46 ± 2.96 8.92 ± 3.26 13.42 ± 3.56 16.92'± 3.26 18.42 ± 3.29 20.07 ± 1.38"d" test 2.68' 1.42 3.366 5.97" 3.236 2.57" 3.661'

Significant at less than 0.05.h Significant at less than 0.01.

90

Socioeu ' cbtaacteristics

No s ificant relationship was found be-mom the presence or absence of severe clinicalmalnutrition and the variables of personalcleanliness, literacy, and educational level.

Contact with information mass media wasexplored through literate parents' newspaperreading and radio listening. The number ofmothers or fathers of malnourished childrenwho were regular newspaper re 'lers was notdifferent from the number who were in thematched control group. Similarly, the numberof fathers who listened regularly to the radiowas the same in both malnourished and controlgroups. As may be seen in Table 8, the caseof the mothers was different. Then: werealmost equal numbers of radio listeners andnonlisteners in the malnourished group, but thenumber of listener, among the matched con-trol group was more than three times thenumber of nonlisteners. The difference is sig-nificant at the 0M5 level of statistical confi-dence (Chi square = 4.20; Df = 1; p <0.0 5 ).

Family structure

The variables of family size and type offamily (nuclear or extended) were not foundto differentiate between the malnourished andcontrol groups.

Family economic status

The socioeconomic status of the lamiliesWas estimated using four indicators: mainsource of family income, sanitary facilities inthe household, annual income per capita, andpercentage of total expriditures spent on food.

Table 8. Radio listening by mothers of severely mal-nourished chileLen and matched controls.

.Mothers ofRadio Listening

Yes No Total

Severely malnourished children 8 10 18

Matched controls 14 4 18

Total 22 14 36

X2=4.20; Df =1; p<0.05.

91

No significant associations were found be-tween any of these four indicators and thepresence or absence of severe malnutrition.

In summary, of all the features of themacroenyironment considered, the only differ-ential between severely malnourished childrenand controls, matched at birth for gestationalage, bady weight, and total body length, wasthe contact with the world outsidethe village through regular radio listening.None of the other characteristics of the parents(biologic, socio!, or cultural) or familial cir-cumstances, including income per capita, mainource of income, and family sin, were signifi-

.antly associated with the csence or absence.of severely malnourished children.

Since the features of the macroenvironmentcould not explain severe malnutrition, ouratention has turned toward the analysis of themicroenvironment of the two sets of children.We hav :elected the potential stimulation ofthe home our first general imaicator of thequality of child care, and of the mother asdie principal stimulating agent for youngchildren.

The instrument used for estimating horrestimulation was the inventory developed byBeattye Caldwell (3), This inventory wasdesigned to sample certain aspects of the quan-tity and, in some ways, the quality of social,emotional, and cognitive stimulation availableto a young child within his home. Two formsof the inventory were used, one designed forinfants up to three years of age, and the otherfor children three to six years old. In both'version; the selection of items included hasbeen guided by a set of assumptions aboutconditions that foster development. Accord-ingly, the inventory dese.ibes and quantitateseight areas of the home environment (1) fre-quency and stability of adult contact; (2)vocal stinnlation; (3) need gratification; (4)emotional climate; (5) avoidance of restric-tion; (6) breadth of experience; (7) aspectsof the physical environment, and (8) availableFlay materials. In each of these areas almost all

.1

items receive binary scores and no attempt ismade in rate finer gradations. The total scoreis the number of items recorded as positive forthe child's development. If it is desirable, eacharca may he scored separately and related to

specific features of development.A trained psychologist recorded the inven-

tory of home stimulation in every child in thecohort at six-month intervals during the firstthree years of life and at yearly intervals ther:-after. At the time of data collection and scor-ing, the psychologist was unaware of the nutri-tional antecedents of the children.

Figures 24 n d 25 show the distribution ofhome stimulation total scores obtained in mal-nourished and control children at six and 48months of age. As may be noted, even at sixmonths, whcn only one case of severe clinicalmalnutrition was present, the control childrenhad significantly higher home stimulationscores. Thus, while none of these children hadhomes with less than 30 points, almost one-fourth of the homes of the future malnourishedchildren scored below 30 arid almost one-halfhad scores below 32 points. Similarly, at 48

1 00_

0.80

4. 060

o5 040

020

24 26 28 30 32 34 36. 38

Home Stimulation Score

Figure 24. Proportion of malnourished and control chil-dren showing different home stimulation scores at sixmonths of age.

92

100

060.

C 040.oQ

o -

320.

60 80 160 120

Home Stimulation Score

Figure 25. Proportion of malnourished and control chil-dren showing different home stimulation scores at 48months of age.

months of age children who had recovered

from malnutrition were living in homes whosescores in home stimulation were well belowthose in which the control children were living.In a range of scores from 60 to 120, aboutone -half the survivors from severe clinicalmalnutrition had home stimulation scores below94 points and only one home had a score be-tween 105 and 109. This distribution of scoresis markedly different from that shown by thehomes of the control children, among whichonly one had a score below 95 while fourreached values between 110 and 120. These

differences are statistically significant at the

0.01 level of confidence.The difference found in the quality of home

environment between severely malnourished

children and matched controls points towardthe value of analyzing other features of themicroenvironment. The 'association between the

presence of severe clinical malnutrition and themother's psychologic profile, maternal attitudes,proclivity toward change, concepts of healthand disease, concepts of food and feeding, andfamily visiting pattern is now under analysis.

Since on the one hand the presence of severemalnutrition was significantly associated withhome stimulation, and on the ocher survivorsof severe malnutrition showed a significant lagin language bipolar concept formation, it

seemed logical to investigate the interrelationsamong these three factors in order to estimate-their possible role. As a first approach to thisissue a technique of partial correlation wasused to look at the degree of association be-tween two variables -holding constant" theinfluence of the third variable. Since the num-ber of cases of malnutrition was rather small,we decided to test for interrelations in thetotal birth cohort.

The coefficients of correlation productmoment among home stimulation scores, totalbody height, and number of bipolar conceptspresent at 46 months of age in the total cohort(229 children) were:

Home stimulation score;Number of bipolar concepts

Home stimulation score:Total body height = 0.23

Total body height:Number of bipolar concepts .---- 0.26

When the relation between home stimulationand number of bipolar concepts v-s partialedout for body height, the coefficient of correla-tion dropped from 0.20 to 0.15. When therelation between body height and number ofbipolar concepts was partialed out for homestimulation, the coefficient changed from 0.26to 0.23. Finally, when the number of bipolarconcepts was "held constant," the coefficient ofcorrelation between home stimulation and body

=0.20

REFER

I. Atmthr, M., and MOISES Beards, Sindrome

pluricarencial infantil (kwashiorkor) and its preven-

tion in Central America. Rome. Food and AgricultureOrganization, 1954. ( rAo Nutrition Series No. 13.)

2. BoThA-AN-rot:N, E.. et al. Intellectual develop-

ment related to nutritional status. J Trop

14:112-15. 1965.

height changed from 0.23 to 0.19. These re-sults suggest that the association between homestimulation and number of bipolar conceptsis mediated to a good extent through bodyheight, which in turn holds a significant de-glee of association with the number of bipolarconcepts independentlyto a large extentofhome stimulation. Within the limits of theprobabilities gi, en by the magnitude of thecoefficients, home stimulation contributes rela-tively more to body height than to number ofbipolar concepts, while body height contributesmore than home stimulation to the variance ofbipolar concepts. Other forms of statisticalapproach to the problem of the relative contri-bution of several variables to both the presenceof malnutrition and the presence of somaticand mental lags in survivors, such as regressionanalyses and multivariate analysis of variance,should be tried to obtain a more quantitative.answer to this subject.

With the moults now available, one can fairlystate that (1) susceptible infants cannot beidentified before the development of severe

clinical malnutrition since they do not differfrom the rest of their birth cohort somaticallyor behaviorally; (2) the appearance of severeclinical malnutrition seems to be associated, inpreindustrial communities, with features of themicroenvironment; (3) children who have re-covered from severe clinical malnutrition lagbehind controls in language development, butpoor microenvironmental conditions are notsufficient to fully explain the behavioral lag,and (4) how long the survivors will performbelow the matched controls' values still lacksan adequate answer.

ENCES

3. Cm.nwFt.i., B. M. Descriptive evaluations of childdevelopment and of developmental settings. Pediatths40 : 46-54, 19(7.

4. CUIANWAKANI, S., et al. Kwashiorkor and mentaldevelopment. Am Chn Nntr 21:844 -52, 1968.

5. CHASE, H. P., and H. P. MARTIN. Undernutritionand child development. New Eng! J ,pled 282 : 933-39.1970.

93

6. CRAVIOTO, JOAQU1N, and BEATRIZ ROBLES. The in-

fluence of protein-calorie malnutrition on psychologicaltest behavior. Proceedings of the First Symposium of theSwedish Nutrition Foundation on Mild Moderate Formsof Protein-Calorie Malnutrition, B;stad and G;;teborg,August 1962, pp. 115-25.

7. et al. Nutrition, growth and neurointe-grative development: an experimental and ecologicstudy. Pediatrics 38:319-72, 1966.

S. cr al. The ecology of infant weight gainin a pre-industrial society. Arta Raedtatr Scam( 56:71-54, 1967.

9. et al. The ecology of growth and develop-ment in a Mexican preindustrial community. 1. Methodand findings from birth to one month of age. MonogrSoc Res CIA( Div 34:1-76, 1969.

10. and E. R. DI:LICARI/IF. Mental perform-ance in school age children: findings after recoveryfrom early severe malnutrition. rim / Dis Child 120:404-10, 1970.

. The complexity of factors involved in

protein calorie malnutrition. 13thliotbeca Nutritto et

Dicta 14:7-22, 1970.12. and E. R: DELicARmE. The long-term

consequences of protein-caloric malnutrition. Nutt. ReV29:107-11, 1971.

94

13. and . Infant malnutrition and

later learning. In: S. Margen and N. L. Wilson (eds.),Progress in human nntrition, Westport, Conn., AviPublishing Co., 1971. vol. I, pp. SO-96.

14. GEREu., A. L., and C. S. AmATRyDA. Develop-mental diagnosis; normal and abnormal child develop-ment, clinical methods and practical applications. 2 ed.

New York, Hoeber, 1947.15. LtA No., P. H., et al. livaltiation of mental devel-

opment in relation to early malnutrition. Am Chit

Nutr 20: 1290 -94, 1967.

16. 14NEREDERG, FERNAND°. Effect of early maras-mic malnutrition on subsequent physical and psycho-logical development. In: N. S. Scrimshaw and J. E.

Gordon (eds.) , Malnutrition, learning, and behavior,Cambridge, Massachusetts, M.I.T. Press, 1965, pp. 269-

75.

17. Poi.t.m., E., and D. M. GRAN0EE. Mental andmotor development of Peruvian children treated forsevere malnutrition. Rev /nteram Psicol 1:93-102,

1967.

IS. YARTIN, U. S., anti D. S. Me:LAREN. The be-havioural development of infants recovering from severe

malnutrition. / Merit Delft. Res 14:25-32, 1970.

SUMMARY OF SESSION II

Jack Tizard'

The five papers in our second session had cer-tain common elements. Dr. Klein and his col-leagues in Guatemala have been studying chil-dren born in four Guatemalan villages duringthe last five years, and he has presented someof his preliminary findings on psychologic testperformance and its implications. The objectwas to elucidate the relations among variousmeasures used to assess the effects of dietarysupplementation upon growth and develop-ment. Like the other speakers today, Dr. Kleinwas concerned with the development of mea-nures as well as the development of children.Among the measures he has used are indicatorsof socioeconomic status and psychologic func-tioning, as well as indicators of nutrition andof the child's physical growth. The messagereported was a twofold one: first, we cannotsay that "performance" is determined by oneset of factors rather than another, and second,for predictive purposes we need to adopt morecomplex models than the simple additive orcorrelative ones that are commonly used.

Dr. Klein's paper raised a number of ques-tions. One is the definition of an adequate psy-chologic testsomething that came up in oneway or another in all the papersand whichindeed constantly arises in work with childrenwho differ from the samples of North Ameri-can, British, or other European populations thathave been used to standardize most of our tests.To simplify the presentation of his findings,Dr. Klein picked out a number of particularsubtestsfour psychologic subtests out of 40which he used as the basis for his arguments.I think that this was probably a mistake. Withmental test data one finds that the validity

' Institute of Education, University of London, Lon.don, England.

95

of individual items and subscales is almost uni-versally low, even when the validity of scalesmade up of numbers of items is very high.Thus, correlations between scores on individualitems in a Binet test, for example, may be aslow as 0.3, but when you add a number of thesecarefully selected items together you get highlyreliable test scores that also have a substantialface validity. I should think that when onehas given a number of ad hoc tests of a generalkind, the first and most successful next stepwould be to examine the way in which scoreson different subtests are related one tc another.One might be able to add items or subscales,and so derive indicators that were of high re-liability and evident face validity. Once onehad such indicators, the way would be openfor more sophisticated analyses of the signifi-cance of differences in scores obtained on thevarious factors believed to exert an influenceon performance.

Perhaps the most interesting aspects of Dr.Klein's presentation came out during the fol-lowing discussion rather than in the paper itself.He told us that the growth rates of childrenwhore diets had been adequately supplementedwere not significantly different from those ofwell-fed North American children, but theywere vastly different from those of children incontrol villages and in the rest of Guatemala.We would all have liked to examine these datain detail.

Dr. Klein also told us in the discussion aboutobservational studies which he has been carry-ing out. Data collection on the rate and fre-quency of social and physical contact betweenmother and child is laborious, and the data aredifficult to score, he reports. I do lope, how-ever, that he will not too readily give up thefurther collection and analysis of this kind of

material. Inasmuch as it bears on the qualityof life of children, it is likely .to be most im-portant to an understanding of the outcome ofsevere malnutrition, just as it is likely to CJI1-tribute greatly to our knowledge of the con-comitants and effects of chronic sn!liiutrition.

The Mexican studies reported on by Dr.Cravioto in his masterly pallet...represent thework of a large team of careful investigatorswho in 15 years or so have developed indicatorsthey are now able to use very effectively toanswer specific questions. The careful descrip-tion of a particular villagethe "Village of theWhite Dust," which must by now be the mostfamous village in Mexicois, I think, a neces-sary beginning to longitudinal study of thesort in which Dr. Cravioto is engaged.

Dr. Cravioto arrested our attention by tell-ing us that despite the presence in the villageof three research pediatricians, four psycholo-gists, 10 social workers, two administrators,and several others on the research team, therewere still 22 ci!m!s of malnutrition there over aperiod of time. By asking why this shouldoccur, he began to raise issues of the most pro-found clinical as well as social importance, Heanalyzed in detail the macroenvironment of thefamilies of these children and of a matchedcontrol group, and showed them to be similar.Then he went on to analyze the microenviron-ment of the two sets of families to show whysome children living in very poor circumstancesdeveloped clinical malnutrition, whereas othersdid not. The answer seemed to be very clearlyrelated to differences in the qualities of themicroenvironment. It follows that if one is

going to intervene to cut down the incidenceof severe malnutrition, one has to find ways toinfluence the microenvironment. A carefuldescription of the qualities of this environmentis an essential first step toward this goal.

Dr. Monckeberg's paper again reminded usof the recurring cycle of malnutrition, poverty,and poor health and environmental circum-stances, and of their blighting effects on thedevelopment of children. He pointed out that

96

Santiago children who are not dropouts or earlyleavers but remain at school are those who are

heavier, and better fed and who havehigher and better school performance. In 3tantalizingly brief account, he reported on astudy in Colombia in which a systematic at-!e'npt is being made to vary physical and cog-nitive stimulation and to supplement the feed-ing of children of different ages in order toexamine differences in outcome. It was impos-sible from his brief description to get a clearpicture of this study, however. It is fairlyelaborate and very ambitious, but if it can becarried through it seems likely to yield resultsof profound importance for our understandingof recovery from malnutrition.

I would like to take issue with one of theconclusions Dr. Miinckeberg drew from his

Chilean studies of the effects of interventionwith a group of children aged seven to nineyears. A nine-month dietary supplement re-

sulted in an increase in the height and weightof the children, but no increase in intelligence.From this Dr. Miinckeberg appears to haveconcluded that by the age of seven it was alla bit too late, that no intervention after theage of five or six was likely to be very effectivein shifting 1Q. It is not legitimate to assumethis from one particular study, in which inany case no special attempt was made to pro-vide mental Stimulation. Furthermore, thereare other studies which show that changes inlife circumstances can bring about substantialincreases in intellectual functioning even if

they occur during late adolescence. Husen, forexample, in a study of young men entering theSwedish Army, compared the cis they achievedat call-up with their 1Q5 at 11 years of age.Those who had left school early had markedlylower IQS at the age of 18 than those of identi-cal 11-year-old IQ who had continued longerin school. SimilarI7,, Alan Clarke has shownthat mildly retarded young men and womenwho in adolescence enter a mental subnormalityhospital (in itself a pretty unstimulating envi-ronment, but one perhaps more stimulating

. _

.than the environments in which they were liv-ing before admission) are likely to increase inintelligence.

The point is, we simply cannot afford to

give up in adolescence, nor indeed does theevidence suggest that mental performance is

immutable by then. What we must do is to

learn from what I understand the Cubans andthe Chinese arc doing, namely to think aboutthe wholesale education of adults as well as

children. It is perfectly true, as Dr. Moneke-berg has said, that if we are to solve the prob-

., lents of malnutrition on a world scale, some-thing like a new environment will have to hecreated, but to say this is not in itself veryhelpful. We have also to look at all stages andages to see what possibilities are open to us inan imperfect world.

The studies in Uganda and Jamaica weresimilar in many respects: they both in%:!stigatedthe long-term effects of malnutrition in infantsand young children admitted to research wardsin a university hospital. They also inquiredinto the age of acute illness and its relationshipto outcome. Interestingly, though these twostudies were both carried out at about thesame time and by people working in BritishMedical Research Council units, none of thosewho took part in either study was more thanbarely aware that the other study was beingcarried out. This raises the question of howcommunication can be improved. I do notthink the answer is through newsletters in

which people are asked to write about projectsthey are just about to undertake.

Despite a lack of communication betweenthose taking part in the Kampala and ,Jamaicanstudies, the two investigations had much incommon, both in methodology and he con-clusions that stemmed from them. The Jamai-can study showed in a population at risk thatchildren hospitalized because of an acute epi-sode of severe mainutrition are likely in laterchildhood to be shorter, less intelligent, poorerin school performance, and less socially compe-

97

tent and to weigh less than their classmates.Differences between the severely malnourishedchildren and their sibs are less clear. Further-more, the study produced no evidence to sug-gest that children admitted to hospital afterthe first nine months of life were less likely tobe physically or mentally stunted in later child-hood than those admitted before nine months.It -did suggest, however, that when the post-hospital environment was more in ellectuallystimulating, the outcome was likely to be betterthan among children growing up with poorerintellectual stimulation. This is the only cheer-ful feature in a set of otherwise melancholyconclusions which are of great social impor-tance.

In the Kampala study it was found thatpreviously malnourished children were stuntedrelative to their comparisons. Using a widebattery of psychologic tests, the investigatorswere able to show that certain functions in themalnourished children were significantly de-

pressed whereas other capacities were relativelyspared. They also report that motor coordina-tive activity and intellectual level were signifi-cantly lowered in the previously malnourishedchildren. The studies' main difference lies in

the Jamaican attempt to explore the back-ground histories of the children to determinewhat social and biologic factors besides nutri-tion influence the children's current level offunctioning.

All the work we have described clearlypoints up the need to develop better attainmentand outcome measures for children of differentcultures. Measures of the performance of indi-vidual children are perhaps better developedthan measures of home stimulation. Caldwell'sIndex is probably the first not wholly unsatis-factory measure that we have of home stimula-tion, but it seems evident that we have a longway to gO before we can really assess the

quality of the child's microenvironment in a

satisfactory way. Both Dr. Cravioto's and Dr.Richardson's papers give us sonic idea of theways iri which we can approach this task.

Even with tile measures we already bye,comparisons can be made betweeii subgroupsliving in particular cultures. It is much moredifficult, and indeed it may be impossible inthe short term at any ;ate, to carry one's com-parisons across cultures. But if the same kindof functional relationships are found to obtainfirst in one culture and then in another, we dogain confidence in their generality. Thus,though different operational measures of intel-lectual stimulation were employed in theJamaican and Mexican studies, both revealedthat the severely malnourished children hadexperienced less intellectual stimulation thantheir comparisons.

It should also be noted that the studies todayare showing the need to ask more, and moresophisticated questions. We arc already lookingmuch more analytically than in the past at thenature of psychologic deficit and at the inter-actions among sociocultural and nutritional

variables that arc responsible for poor perform-ance, for example. I hop.: that we will soon behearing reports of other studies in which theefficacy of different types of treatment regimenis compared. The stage will then 1>e set for usto move, in different fields, from epiderniologicand clinical studies which arc largely descrip-tive, to studies that enable us to bring clinicaland socially relevant factors under some kindof experimental control.

My final point concerns the relevance ofanimal data to human epidemiologic and clini-cal study. Dr. Stewart's research and some ofthe other work mentioned yesterday indicatesthat future studies arc likely to be much morethan biochemical. Theoretical and methodo-logic problems exist that arc greatly elucidatedby animal experiments and models, and com-parative psychologists have in recent years beenpursuing them with great sophistication andconspicuous success.

REFERENCES

1. CLARKE, A. D., and A. M. CA.ARKE. Consistencyand variability in the growth of human characteristics.In: W. D. Wall (ed.), Advances in educational psy-

98

chology, London, University of London Press, 1972.2. HttsuN, T. Influence of schooling upon IQ.

Theoriti 17:61 -88, 1951.

Session III

ECOLOGY OF MALNUTRITION

Chairman

Fernando Monckeberg

Rapporteur

Joaquin Cravioto

ECOLOGY OF MALNUTRITION: NONNUTRITIONAL FACTORS

INFLUENCING INTELLECTUAL AND BEHAVIORAL DEVELOPMENT

Stephen A. Richardson'

Introduction and Review

In the report of the International Conferenceon Malnutrition, Learning, and Behavior heldin 1 967 at the Massachusetts Institute of Tech-nology, the editors made the following com-ment on the ecology of malnutrition:

With the wide variety of determinants alreadyidentified,. the approach becomes clear; it is eco-logicalthe interplay between man as a biologi-cal organism and as a human being, and the wholeof the many-sided environment in which he lives( 22 ).

In his introduction to the session, the moder-ator, Dr. William Darby, stated:

Of necessity, the test of early malnutrition as adeterminant of learning and behavior in man isthrough field study of people in their naturalsetting, exposed to all the influences that deter-mine intellectual capacity and behavior patterns.

. . The aim of the field study is to sort out therelative contribution of malnutrition among avariety of other factors (22).

The purpose of many studies presented atthat conference was to investigate the long-term consequences of malnutrition duringpregnancy and the early life of the offspring.In almost all the studies a quasi-experimental

' Departments of Pediatrics and Community Health,Albert Einstein College of Medicine, Bronx, New York,USA.

design was used in which the attempt was madeto hold constant all ecologic variables exceptnutrition that might impair functional devel-opment. One design was an experimental inter-vention model using two human populationsthat were as nearly as possible equivalent in allrespects and in which severe malnutrition wasprevalent during pregnancy and early child-hood. Food supplementation was given one ofthe populations without any other change.The later intellectual and behavioral function-ing of the children in the two populations wassubsequently measured. A study using thisapproach was outlined by Canosa (4).

To meet the requirements of this design, itis necessary to determine that severe malnutri-tion does occur in pregnant women and infantsin the two populations; that supplementaryfeeding occurs in the experimental group, andthat the food given is used as a supplementand not as a substitute; that the supplementaryfeeding does not change other factors (or thatthese other factors are changed in the controlgroup) and that all factors that influence intel-lectual and behavioral development in the chil-dren studied are the same in the experimentaland control populations. A satisfactory studymust identify and measure the factors otherthan nutrition that affect the functional devel-opment of children and not merely assume theirequivalence in the experimental and controlgroups. Such a study must begin before the

101

mother's pregnancy and continue during theinfancy of the offspring and into school ageuntil the behavioral and intellectual level ofchildren can be tested with measures that havepredictive value. To meet these experimentalrequirements, a number of very difficult prac-tical and ethical problems must be overcome.

A. second quasi-experimental approach de-pends on "natural" intervention (3, 5, 6, 21).In this type of approach a group of childrenwho were known to have been severely mal-nourished in infancy are identified, and anattempt is made to find a matched controlchild for each malnourished child. The con-trols must not have been severely malnour-ished, and ideally must be matched for all

biologic and social factors that contribute tolater intellectual and behavioral functioning.In practice the cases called "controls" havebeen matched with the experimental cases onsuch variables as age, sex, socioeconomic status,

and being in the same classroom at school. Analternative has been to use the malnourishedchild's sib as a comparison (1, 14). Here theassumption is made thar by living in the samefamily the index child and sib will have ex-per;vnced very similar ecologic conditions. Theywill necessariiy differ in age, pregnancy num-ber, and ordinal position within the sibship, andthey may differ by sex unless this is used as acontrol variable. In these studies the assump-tion is made that all factors other than thecontrol variables are held approximately con-stant, or that what variation does exist is ran-domly distributed between cases.

The quasi-experimental approaches have in-herent limitations. One is that if all variablesexcept nutrition that influence the functionaldew4opment of the child are controlled, itwould prevent the study of variations in all

variables except, malnutrition. It limits the

study of interaction effects between nutritionand the other ecologic factors, or the varyingcircumstances and conditions under which mal-

nutrition has differential effects on later func-tioning. It is reasonable to expect that theconsequences of severe malnutrition v.nry

widely depending on the ways in which otherbiologic and social factors influence :1 particularchild. The interaction of background factorsrelated to the intelligence of children may beillustrated from the following report of a studyof school children in Scotland:

The purpose was to examine the associationbetween combinations of birth weight and gesta-tional age, and later intellectual functioning ofthe children. Further, we wished to sec whetherlow birth weight and gestational age would havedi ff...ent associations with the child's intelligenceat age 7 depending n,:on the general life stylesin which the child lived. We therefore lookedat the associations for each social-class categoryseparately. In order to avoid the effects of obstet-rical and neonatal complications, any cases withsuch complications were removed and analyzedseparately . . . !The analysis was also restrictedto children who had not been administrativelyclassified as mentally subnormal. We had foundseparately that low birth weight] . . and lowgestational age were more frequent for mentallysubnormal children than for children who werenot mentally subnormal.

Figure 1 shows the average IQ of children withdi fferent birth weights and gestational ages

within each social c' .gs category. For each socialclass, children w; Th-th weights of less than5 pounds and g: .clonal ag,.. of less than 37weeks have lower IQ scores than the average IQ

for all children within the same social classes

[Table 1]. With the exception of the uppersocial class I-111a, when children have experi-enced both low birth weight and low gestationalage, they have lower average intelligence scoresthan when low gestational age was present, butnot low birth weight. The size of differencein average IQ between all children in a socialclass and the children with low birth weightand gestational age is larger in the lower thanin the upper social classes. This suggests that thereis a biosocial interaction, with children of lowersocial class families at greater risk of intellec-tual impairment from low birth weight and

102

10Social

110 I-ClassV 1-11Ia

100

90

80

Illb

Inc

IV

V

Total inSocial Class

GA 37 wks.8CE 5 lbs.

`.\

GA tw 37 wks.8W. 5 lbs.

Figure 1. IQ test scores and social class of seven-year-old children with different gestational age (GA) andbirth weight (BW) in Aberdeen, Scotland, 1962. Caseswith evidence of obstetric complications or low physi-cal grade at birth were excluded.

gestational age than children from upper classfamilies (20).

This study shows the way in which low birthweight and shorter gestation appear to have

Table 1. IQ test scores and social class of seven-year-old children with different gestational age (GA) andbirth weight (BW) in Aberdeen, Scotland, 1962.

Social class Number Average IQ

/-///aTotal in social class 1,556 110GA < 37 wks., BW > 5 lir. 37 105GA < 37 wks., BW < 5 lbs. 9 107

Total in social class 1,490 105GA < 37 wks., BW > 5 lbs. 18 99GA < 37 wks., BW < 5 lbs. 13 93

IIkTotal in social class 1,310 102GA < 37 wks., BW > 5 lbs. 29 99GA < 37 wks., BW < 5 lbs. 16 94

IVTotal in social class 939 98GA < 37 wks., BW > 5 lbs. 22 .95GA < 37 wks., BW < 5 lbs. 7 <80

VTotal in social class 864 96GA < 37 wks., BW > 5 lbs. 42 92GA < 37 wks., BW < 5 lbs. 10 <80

more severe affects on intelligence in childrenwho live in a lower-social-class family.

A second limitation of quasi-experimentaldesigns in the study of severe malnutrition is

the highly questionable assumption that throughcontrolling variables such as socioeconomicstatus, however defined, all factors that influ-ence the intellectual and behavioral functioningof children are also controlled. This is recog-nized in the discussion of most of the studies.

The literature on malnutrition in childhoodsuggests some of the ecologic factors that needto be considered. Craviotu Aid his coworkers(7) give a schematic presentation of the inter-relation among biosocial factors and low weightgain. Richardson (17) discusses the influenceof social, environmental, and nutritional factorson mental ability, and Ka llen (II) provides aschema of interdependent, interacting factorsdetermining learning. Hansen and coworkers(9) suggest that retardation of growth anddevelopment occurs with ". . . poverty, in-adequate maternal care, disturbed family rela-tionships, overworking, lack of educational andplay facilities, illness, scant medical facilitiesand general paucity of the cultural environ-ment." These presentations are suggestive anddescriptive. There is need for fuller and moresystematic consideration of factors other thannutrition that influence the functional devel-opment of children, however. There is also

need to develop malnutrition research that in-corporates these ecologic factors into the re-search design, data gathering, and analysis.The remainder of this paper is a considerationof these two issues.

Identification of Ecologic FactorsInfluencing Functional Development

Even a cursory review of the social andbiologic research literature dealing with factorsthat influence the intellectual and social devel-opment of children reveals a diverse and for-midable array of theory, concepts, researchreports, inferences, and speculations. ..ertainly

103

no investigation can deal systematically withall the known and suspected factors that influ-ence the functional development of children,but the study of selected background factorsjudged to be salient is posAble and provides anessential basis for examining the relative impor-tance of the influence of malnutl ition comparedto other ecologic factors.

The ecologic factors shown in Figure 2 areillustrative of what may be derived from areview of the general topics of social and humandeprivation, child development, pediatrics, andobstetrics (8, 12, 13, 15, 16, 21, 24, 25). Indeveloping the list of factors, the backgroundof the malnourished children was consideredin terms of factors that influence the func-tional development of children and may bebiologic, social, physical, or technologic in

nature. One useful way of looking at a widevariety of background factors is to considerthe material and human resources that are

brought to bear in the care and upbringing ofthe child. Sonic ecologic factors may have apervasive effect that influences many of a

child's functional modalities. Other factorsmay have more limited influence and affect asingle function o.( the child.

The outcome measures of functioning ofchildren should include not only anthropometryand IQ, which have been the major preoccupa-tion of investigators studying the long-termeffects of malnutrition, but also measures ofschool performance,- cognitive functioning, be-havior, level of health, and indicators of centralnervous 53 stern impairment.

The list of factors presented in Figure 2 wasdeveloped for a study in Jamaica of the long-term consequences of severe malnutrition inthe first two years of life. Because data fromthat study will be used to illustrate points inthis paper, a brief description of its researchdesign is necessary.

Boys who had been hospitalized with theprimary diagnosis of severe malnutrition some-time during their first two years of life wereselected for study. They were traced when

Figure 2. Ecologic factors.

Child's Biologic M itherReproductive historyHealth history

Mother's or Caretaker's Capabilities and ActivitiesVerbal abilityValuesExposure to ideasActivities and affiliationsHuman resourcesTraining and upbringing of the childAspirations for the childMother as teacher of the child

Father or Adult MalePresence of adult male role model in the householdExistence of affectionate ongoing relationship with the

childJoint activities of husband and wife with the childDegree of conflict or cooperation between husband and

wife in childrearingFamily

CompositionStabilityExtended familySocial relations between family, friends, and neighborsAlternative caretakers available for the child

Physical and Economic Resources of the FamilyIncome in cash and kindType and size of dwellingWater supplyAvailability of electricityAppliancesType of fuel usedTransportation

Area of ResidenceSpectrum ranging from isolated rural location to large

urban centerChild's Background History

Pregnancy number and ordinal positionBirth weight and general health history

(other than malnutrition)Feeding during the first two years of lifeContinuity in prime caretakerContinuity in composition of familyRelationships with adultsRelationships with other childrenExposure to ideas and languageActivities and experiences outside the homeSchool history

they were between six and 10 years of age.For each malnourished boy (n = 74 indexcases), two boys were also chosen for studywho had not been severely malnourished: thesib of the index case of the same sex andnearest in age (n -= 38 sibs) and a classmateof the index case of the same sex and nearestin age to the index case (n = 71 comparisons).

104

For each boy in the study a detailed psycho-logicmthropometric, and pediatric neurologic.assessment was made. Their schools were visitedto obtain their teachers' assessments of the boys'school performance and behavior. Home visitswere made to obtain comprehensive informationOn the child's curial and biologic history and arange of ecologic factors that may have influ-enced his functional development. The factorsin Isigure 2 vary widely in levels of abstraction,and the translation of these factors into datacollection procedures is a major undertaking.The procedure must be partly inductive be-cause the specific questions, observations, oruse of documentary sources must be meaning-ful and relevant to the particular society andculture in which the study is undertaken. Theprocedure must also be partly founded on pre-vious research or theory to provide a basis forselection among the almost infinite number ofvariables that might be identified.

Translation of Concepts into aData-Gathering Procedure

There is a conceptual procedure that mustbe undertaken regardless of where a study is

conducted. This involves the translation ofeach of the general factors that may influencethe child's functional development into thespecific units of information to be obtained indata ccillection. The procedure can best be

explained by means of an illustration from theJamaican study. Conceptually, an importantfactor in the development of a child is "intel-lectual stimulation." This is an abstractionthat needs to be reduced to the various specificways in which a child receives intellectualstimulation: for example, being read to aloudor being told stories, having pictures to lookat and materials for drawing and painting,contact with mother or principle caretakerhaving a good command of language, convers-ing with and encouraging the child to talk andthink about things. Observations or questionsthat will get information on these indicator

variables have to be developed in careful pi;-testing. Seven questions used in Jamaica asindicators of intellectual stimulation will serveas an illustration. .The questions were:

1. Does he have any toys you or anyoneelse has given him?

2. Does he have. any books or magazines?3. Does he listen to the radio?4. Does (child's name' watch TV?5. Have you or anyone else taken [child's

name] on trips to other places besides thevisits to relatives and friends you havetold me about?

6. Does anyone tell him stories?7. Does anyone read to him?Questions were asked of the child's mother or

principle caretaker and answers were coded as"yes," or ''no," or "do not know."

Data Analysis

In undertaking the analysis of functionaloutcomes, we already knew that the malnour-ished children were significantly smaller, hadlower IQ's and school performance, and weremore impaired in behavior in the school setting(2, 10, I8, /9).

Each question in the intellectual stimulationindex was based on the theoretical predictionthat those children for whom negative responseswere obtained would receive less stimulation,and that those with less stimulation wouldhave lower IQ, poorer school performance, andimpaired social behavior. To test these assump-,4ons, the associations between the answers toeas:h of the questions and measures of thechild's functioning were examined separatelyfor the index and comparison children. Thestrongest evidence in support of the predictionswould be a positive association between nega-tive responses to the questions on intellectualstimulation and the presence of impaired orpoorer functioning for both index and com-parison sets of children. If a positive associa-tion is found for the comparison but not theindex children, the result might suggest that

105

malnutrition is sufficiently powerful in itsinfluence to mask the effect of intellectualstimulation. On the other hand, if an associa-tion is found for the index children but notfor the comparison, the finding might suggestthat intellectual stimulation is more influentialfor malnourished children. If no association isfound for either set of children, it suggeststhese particular indicators of intellectual stimu-lation are unrelated to functional competence

or are not appropriate indicators of intellectualstimulation.

Table 2 shows the associations between theseven indicators of intellectual stimni,,ion andfull-scale IQ for the index and co:- -1 sets

of children.' For the index child II the

' Consideration of other functional 1110dalitieti such asschool performance and behavior have been omitted tosimplify the illustration. Consideration of the sib is

omitted for the same reason.

Table 2. Relationship between indicators of intellectual stimulation and full-scale 10 for malnourished and non-malnourished sets of boys.

Full-scale IQMalnourished

(index)Nonmalnourished

(comparison)

X= p'

Toys given child

X2 P'

Toys given child

Yes NoYes No

Below median 22 16 0.891 N.S. 23 13 5.937 <0.01Ahove median 24 11 31 4

Has hooks or magazines Has hooks or magazines

Yes No X2 p Yes No X'

Below median 7 31 6.998 <0.005 13 23 3.156 <0.05Above median 17 19 20 15

Listens to radio Listens to radio

Yes No X2 p Yes No X2

Below median 28 10 1.016 N.S. 30 6 0.430 N.S.Above median 30 6 27 8

Watches TV Watches TV

Yes No X2 p Yes No X2 p


Taken on trips Taken on trips



Told stories Told stories


Below median 19 19 4.912 <0.025 24 12 0.113 N.S.Above median 27 9 22 13

Stories read to child Stories read to child

Yes No X2 p Yes No X2

Below median 26 12 1.426 N.S. 27 9 0.2541 N.S.Above median 29 7 28 7

'All probability levels are one-tail due to the directional nature of the hypothesis.

106

associations are in the predicted direction andfour are statistically significant.2 For the com-parison children, four are in the expected direc-tion and all four are statistically significant.

Having examined the indicators of intellec-tual stimulation separately, they may now becombined into an overall index. This providessome indication of the cumulative effect ofresponses to the seven separate indicators. Eachresponse is given equal weight because there isno obvious conceptual basis for differentialweighting. The overall score ranges from 0-7,a score of 7 indicating maximal stimulation,and a score of 0 minimal stimulation. Table 3shows a significant positive associ:, tion betweenintellectual stimulation and IQ for index andcomparison sets (p < 0.005 and p < 0.025,respectively). This result does not establishthat a higher level of intellectual stimulationin childhood causes better intellectual func-tioning. It only shows an association, and causecan be attributed only through "guilt by asso-

All probability levels included in the test and tablesarc one-tail due to the directional nature of thehypothesis.

ciation." Nevertheless, the analysis does pro-vide useful evidence.

Having found evidence in support of thehypothesis that children who receive less intel-lectual stimulation will have lower lQs, it is

now worth testing the prediction that themalnourished boys have received less intellec-tual stimulation than the comparisons whowere not malnourished. No significant differ-ences were found between the index and com-parison sets of children for any of the sevenindicators of intellectual stimulation, althoughall show a trend in the expected direction. Forthe overall index of intellectual stimulation(Table 4), however, the association is in thepredicted direction at the p < 0.025 level ofsignificance.

So far the associations between malnutritionand the children's lQs, and between intellectualstimulation and IQ have been considered sepa-rately. To examine the combinations of nutri-tion and intellectual stimulation, the followingfour subsets of children were chosen.

1. Malnourished boys low on intellectualstimulation.

Table 3. Relationship between index of intellectual stimulation and full-scale IQ for malnourished and non-malnourished sets of boys.

Full -scale IQMalnourished

(index)Nonmalnourished

(comparison)

Index ofintellectual stimulation

Below Abovemedian median

X2 P"

Index ofintellectual stimulation

Belowmedian

Above X2 Pamedian

Below median 27 11 9.242 <0.005Above median 12 22

22 13 4.629 <0.02513 22

" All probability levels are one-tail due to the directional nature of the hypothesis.

Table 4. Differences between malnourished and nonmalnourished sets of boys on index of intellectual stimulation.

Index of intellectualstimulation (number of

yes responses toindicator questions)

Malnourished Non malnourished(index) (comparison) X2

0-3

4-56-7

28

30

14

24

20

28

6.974 <0.025

" The probability level is one-tail due to the directional nature of the hypothesis.

107

2. Malnourished boys high on intellectualstimulation.

3. Nonmalnourished boys low on intellec-tual stimulation. .

4. Nonmalnourished boys high on intellec-tual stimulation.

Based on the previous analyses we wouldexpect subset 1 to have the lowest level .1 andsubset 4 to have the highest level. Of particu-lar interest will be the distribution of IQ withinsubsets 2 and 3 as an indication of the relativesalience of malnutrition and intellectual stimu-lation for IQ. The results of this analysis (Table5) for the two extreme subsets arc as expected,The average IQ for the comparison boys withhigh intellectual stimulation is 18 points higherthan the index boys with low intellectual stimu-lation. No significant difference is found be-tween the two intermediate subsets (Table 6),and there is an average IQ difference of onlytwo points. Considering the interaction be-

tween nutrition and intellectual stimulation,these results suggest that malnourished childrenwith high intellectual stimulation function aswell at school age, as measured by IQ as childrenwho have not been malnourished but havereceived low intellectual stimulation.

To determine whether these results have beenobtained largely because of the contributionof a few extreme cases, cumulative percentagedistributions for the four subsets of childrenwere prepared (Figure 3). The results showthat this is not the case.

The steps outlined for examining the variableof intellectual stimulation may be applied toeach of the ecologic factorsboth social and

100

90

80

70

60I!

1 /50

//

!

1

/

40/

7.

30

//20 - /

/

ID

0 ( I

KEY:

INDEX, BELOW MEDIAN

---- INDEX, ABOVE MEDIAN-- ......... COMPARISON. BELOW MEDIAN

------ COMPARISON, ABOVE MEOIAN

I

446 46-50 51.54 55-59 60-64 65.69 70.74 75.7980.84 85-89 90.94 >94

FULL -SCALE 10

Figure 3. Cumulative percentage distribution of full-scale 10 for malnourished (index) and nonmalnourished(comparison) sets of boys grouped by above and belowmedian of index of intellectual stimulation.

biologicincluded in a study. A further stepwould be consideration of various combina-tions of ecologic factors using different formsof multivariate analysis. As the first step foreach child, a profile could be prepared of eachecologic factor and whether it is more or lesslikely to contribute a positive influence on the

Table 5. Mean full-scale 10s and standard deviation grouped by index of intellectual stimulation for malnourishedand nonmalnourished sets of boys.

SubgroupNumber

(n)?Sean(X)

Standard Deviation(s)

Malnourished children (index)Below median of index of intellectual stimulation 41 52.9 7.7Above median of index of intellectual stimulation 33 62.7 11.6

Nonmalnourished children (comparison)Below median of index of intellectual stimulation 35 60.5 11.1

Above median of index of intellectual stimulation 36 71.4 13.9

108

Table 6. "t" test results of differences of full-scale IQmeans for malnourished and nonmalnourished sets ofboys divided by above and below median of index ofintellectual stimulation.

Subsets of childrenProbability

level

Index', below median vs-.index, above median -

Index, below median vs.comparisonh, below medianIndex, below median vs.comparison, above medianIndex, above median vs.comparison, below medianIndex, above median vs.comparison, above medianComparison, below median vs.comparison, above median

4.17 p <0.001

3.41 p <0.005

7.09 p <0.001

0.80 n.s.

2.83 p <0.01

3.66 p <0.001

" Index (malnourished).I' Comparison (nonmalnourished).

child's functional development. The cumula-tion of these factors provides an overall measureof the identified elements of the ecologic situa-tion that has influenced functional develop-ment. The simplest measure of the overall arrayof ecologic factors would be an above-and-below median split with a score of 1 repre-senting the more favorable end of the indexand 0 the less favorable. A high score acrossthe factors would thus represent a favorableecologic environment. At this stage of theanalysis there is need for complex statisticaltreatment. Of particular interest will be the

examination of whether there are combinationsof favorable ecologic factors that will outweighthe unfavorable factor of severe malnutrition.

From studies employing the design outlined,greater understanding can emerge of childdevelopment in an ecologic setting, but it is

necessarily based on inferences from associa-tional analysis. Based on the results, a differentkind of research design may then be used totest the causal effects of various combinationsof ecologic factors. One such design may bebased on an intervention model in which twosets of children from comparable backgroundswould be selected, one set having been mal-nourished and the other not. The malnour-ished children and their families would be pro-vided with a set of opportunities for gainingthe conditions and experiences shown from the-earlier studies to be favorable for functionaldevelopment. The children who were not mal-nourished would not receive any special treat-ment. At various ages the functioning of thechildren in the two sets would be measuredand compared to see if the malnourished chil-dren's level of functioning became equal to orhigher than that of comparison children. Fromstudies of this kind valuable information couldbe gained to determine what would be the bestforms of intervention to help severely malnour-ished children to more fully realize their poten-tial functional capacities.

REFERENCES

1. BIRCH, H. G., et al. Relation of kwashiorkor inearly childhood and intelligence at school age. PediatrRes 5:579-85, 1971.

2. et al. Growth sequelae of severe infantilemalnutrition. In preparation.

3. BOTHA-ANTOUN, E., et al. Intellectual developmentrelated to nutritional status. / Trop Pediatr 14: 112 -15,1968.

4. CANOSA, C. A. Ecological approach to the prob-lems of malnutrition, learning, and behavior. In: N. S.Scrimshaw and J. E. Gordon (eds.), Malnutrition, learn-ing, and behavior, Cambridge, Massachusetts, M.I.T.Press, 1968, pp. 389-97.

5. CHAMPAKAM, S., et al. Kwashiorkor and mentaldevelopment. Am J Clin Near 21:844-52, 1968.

6. CHASE, H. P., and H. P. MARTIN. Undernutritionand child development. New Engl I Med 282:933-76,1970.

7. CRAVIOTO, JOAQUIN, et al. The ecology of infantweight gain in a pre-industrial society. Acta PaediatrStand 56:71-84, 1967.

8. GLASS, D. C. (ed.) . Environmental influences.

New York, Rockefeller University Press, 1968.

9. HANSEN, J. D. What does nutritional growth

retardation imply? Pediatrics 47:299-313, 1971.

109

10. Htuir/AG, M. E., et al. Intellectual levels of

school children severely malnourished during the firsttwo years of life. Pediatrics 49:814-24, 1972.

I I. Km LEN, D. J. Comment. In: N. S. Scrimshawand J. E. Gordon (eds.), op. cit., pp. 376-79.

12. MINUCIIIN. S., et al. Families of the slums: anexploration of their structure and treatment. New York,Basic Books, 1967.

13. PAN AMERICAN HEALTH ORGANIZATION. Depriva-

tion in psychobiological development. Washington.D.C., Pan American Health Organization, 1966. (rAtioScientific Publication No. 134.)

Id. POI.I.ITT, E., and D. M. GRANOFF. Mental andmotor development of Peruvian children treated forsevere malnutrition. Rev Interam Psicol 1:93-102,1967.

15. PRINGLE, M. L. Deprivation and education.

London, Longmans, Green & Co., 1965.16. RICHARDSON, S. A., and A. GUTTIAACHER (eds.).

Childbearing: its. social and psychobiological aspects.

Baltimore, Williams and Wilkins, 1967.17. . The influence of social-environmental

and nutritional factors on mental ability. In: N. S.Scrimshaw and J. E. Gordon (eds.), op. cit., pp. 346-61.

18. et al. The behavior of children in schoolwho were severely malnourished in the first two yearsof life. I Health Soc Rehm., 13:276-84, 1972.

19. et al. School performance of childrenwho were severely malnourished in infancy. Am I 'WentDefic. in press.

20. . The reduction of stress for persons withhandicaps. In: Lennart Levi (ed.), Society, stress anddisease: childhood and adolescence, Oxford. OxfordUniversity Press, in press.

21. RIESSMAN, FRANK. The culturally deprived child.New York, Harper & Row, 1962.

22. SCRIMSHAW, N. S., and J. E. Gordon (eds.).Malnutrition, learning, and behavior. Proceedings ofan international conference cosponsored by The Nutri-tion Foundation, Inc., and The Massachusetts Instituteof Technology. Cambridge, Massachusetts, M.I.T.Press, 1968, p. 388.

23. STOCH, M. B., and P. M. SMYTHE. Does under-nutrition during infancy inhibit brain growth and sub-sequent intellectual development? Arch Dis Child38:546-52, 1963.

24. U.S. NATIONAL. INSTITUTE OF CHILD HEALTHAND HUMAN DEVELOPMENT. Perspectives on humandeprivation: biological, psychological, and sociological.Washington, D.C., 1968.

25. WEBSTER , S. W. (ed.) . The disadvantaged

learner: knowing, understanding, educating. San Fran-cisco, Chandler Publishing Co., 1966

110

SUMMARY OF SESSION III

Joaquin Craviotoi

After an introduction dealing with the gen-eral design of previous investigations, excludinglongitudinal ecologic studies, Dr. Richardsonqualified them as quasi-experimental in natureand pointed out their limitations. He dwelton the various types of controls or comparisongroups that have been used in different studies,and pointed out that siblings are mainly usednot to sort out the genetic component thatmight be present but rather to provide com-parison children whose life experiences aresomewhat similar to the malnourished child's.Clearly, only macroenvironmental variablescan be reduced by this procedure.

I think we should mention that statisticaltechniques have recently been used to avoid theneed for matched controls. One can use ran-dom controls if techniques of covariance analy-sis are applied. Dr. Richardson gave an exampleof this approach to illustrate the need for con-sidering the effects of biosocial interaction.This example showed that low birth weight andgestational age have differing consequences forintellectual functioning, depending upon thesocial class of the child's family, He then pro-ceeded to show us a model for the ecologicstudy, which I have interpreted as two-phased.The first phase was the identification of theimportant intervening variables other thanmalnutrition or nutritional status in some non-intervention studies. Results of this analysiscould then be tested by intervention types ofresearch.

In the first part of the model used for non-intervention studies, consideration starts withthe question of what are important variables

' Division of Scientific Research, Hospital Infantil de laimaN;Mexico City, Mexico.

other than nutrition in mental performance orintellectual development. These were derivedmainly from the literature about maternalleprivation or environmental deprivation in

which important intervening variables were

considered. The difficulty of separating dif-ferent types of variables came out very clearlyin the discussion. When one has a chain *ofevents that goes all the way from the societyto the individual, such as in the case of mentalfunctioning, it is difficult to know what is anintervening variable and what is the actualetiologic agent.

With an ecologic model, I think that one canget away from this difficulty because if we con-sider the epidemiologic triad of agent, host, andenvironment, we do not imply that one ofthree factors is more important than the othertwo. For example, malaria can be eradicatedby getting rid of swamps, or the mosquito, orthe Plasmodium, or the sick man. In any case,the chain of events leading to disease wouldbe altered and malaria would disappear. One

does not need then to know precisely which isthe last link in the chain of events to modifythe outcome efficiently.

Dr. Richardson next gave us a very niceexample of the different steps involved in theecologic type of research. These steps beginwith the concept, which is then translated intooperational definitions, after which the investi-gator proceeds to the devising of instruments,the testing of the instrument as such, andfinally to the proof of the association betweenwhat the instrument is gathering and the con-cept that is supposed to be tapped by the in-strument. I think that the last two steps arejust a way of rephrasing the classical conceptsof reliability and validity of a method.

III

There was very little discussion of the out-come variables, but it was pointed out thatone has to be careful in choosing them, particu-larly because of different sensitivities at differ-ent times in the life of the individual. Dr.Richardson proposed that once the associationsamong variables have been identified, a sort ofmatrix should be ana:yzed by methods such asmultifactorial analysis. I think that if youhave defined your questions carefully on reach-ing the analysis phase, then what sort of statis-tical technique is used is really a matter oftaste. I am not speaking of statistical method,which is part of the design. I mean the statisti-cal technique is factorial or discriminativeanalysis. It is a matter of choice or a matter ofconvenience depending on the resources at one'sdisposal. 'With the use of factorial analysis onecan identify the variables that are so highlycorrelated among, themselves that they actuallymeasure practically the same thing, and separatethem from the ones that can be regarded asindependent factors.

In the intervention phase of Dr. Richard-son's model, one selects some of the variablesthat earlier studies have indicated are signifi-cant. The selection is also guided by the easeof modifying the variable with available re-sources, particularly staff. One alternative is

to proceed to the construction of a theoreticalmodel that will explain the influence of the

different sets of variables, either alone or incombination. From this model the quantitativecontribution of each variable is theoreticallyestimated, and the model is tested as a wholeif one can devise a common measure that willrepresent the additive interaction or action ofthe several variables invoked. Alternatively,one could pick certain variables and test themif one wished to determine their specificcontribution.

Following the paper there was a discussionof intervention for research and social actionto improve the children's mental conditionthrough use of a combination of different vari-ables, including food. Whether interventionshould occur in the family, in institutions likeschools, or at the community or national levelwas also debated. This discussion dealt not onlywith administrative procedures,. but also withconception. If a group or a person feels thatthe highest risk of brain damage is before theage of three years, basing his reasoning on theresults of the animal experiments and the infer-ences he can draw from the circumstantialevidence available about the human, he will

have to intervene in the home. If on the otherhand he believes that the main damage has notbeen done by that age, then administrative unitssuch as kindergartens or schools may be thebest place for intervention.

Session IV

METHODOLOGIC ISSUES IN MALNUTRITION STUDIES

Chairman

Stephen A. Richardson

Rapporteur

Robert Klein

ISSUES OF DESIGN AND METHOD IN STUDYING THE EFFECTS

OF MALNUTRITION ON MENTAL DEVELOPMENT

Herbert G. Birch'

Research on the relation of nutritional fac-tors to intelligence and learning has grownextensively in the past decade. Attention hasbeen directed at different components of thecombined factors of social disadvantage andpoverty in an effort to define reasons for theassociation of malnutrition with reduced intel-lectual level. Sociologists, psychologists, andeducators have advanced reasons relevant totheir particular disciplines for intellectual back-wardness and school failure. They have pointedto particular patterns of child care, styles ofplay, depressed motivation, particular valuesystems, and deficient educational settings andinstruction as factors that contribute to low-ered intellectual level and poor academic per-formance in disadvantaged children. Theimportance of such variables for achievementcannot be disputed. It would be most unfor-tunate, however, if in recognizing the impor-tance of these situational components we wereto neglect poor nutrition and, in general, defec-tive circumstances in the development of theindividual as a biologic organism who interactswith social, cultural; and educational circum-stances in the production of dysfunction.

It has long been recognized that the indi-vidual's nutrition affects his growth, health,and development. Inadequate nutrition resultsin stunting, reduced resistance to infectious

1 Department of Pediatrics, Albert Einstein College ofMedicine, Bronx, New York, USA.

disease, apathy, and general behavioral unre-sponsivenessall factors affecting the child'sdevelopment and functional capacity. It is

therefore essential to explore the relation ofnutrition to intelligence and learning ability.

Confusion has resulted from extravagantclaims as to the unique contribution of mal-nutrition to brain impairment and intellectualdeficit, as well as from efforts to minimize theimportance of nutritional factors and to arguefor the primacy of social, genetic, cultural, orfamilial variables in the production of deficit.Little of use emerges from such controversy.Clearly, malnutrition occurs most frequentlyin those segments of the population that areeconomically, socially, and culturally disad-vantaged. In discussing lowered intellect inmalnourished children from such groups, .emust deal with the interaction of nutritionaland social or familial factors by considering thepossibilities that affected children are dull be-cause they are the offspring of dull parents, orthat malnutrition and the general impoverish-ment of their environments contribute toreducing intellectual function. The task is todisentLngle the particular and interactive con-tributions that different factors make to thedevelopment of depressed functional outcomes.The real problem is therefore to define the par-ticular role that nutritional factors may playin the development of malfunction and theinteraction of this influence with others thataffect the child's development.

115

Before considering the ways in which avail-able research permits us to achieve this objec-tive, it is important to clarify the term "mal-nutrition." When we think of malnutritionour imaginations conjure up im-ges of maras-mus and kwashiorkor. These images reflectonly the highly visible tip of an iceberg. Inter-mittent and marginal incomes as well as back-ward technology result less often in the symp-toms characteristic of starvation than whatBrock (19) has called "dietary subnutrition, defined as any impairment of functionalefficiency of body systems which can be cor-rected by better feeding." When present inpopulations. such subnutrition may be mani-fested by stunting, disproportion, and a varietyof anatomic, physiologic, and behavioral ab-normalities ( /2).

Chronic subnutrition is not infrequentlyaccompanied by dramatic manifestations ofacute and severe malnutrition in infants andyoung children. These illnesses, variously re-flected in the syndromes of marasmus, kwash-iorkor, and marasmic kwashiorkor, may beconditions deriving from acute exacerbationsof chronic subnutrition, which in differentdegrees reflect caloric deficiency, inadequacy ofprotein in the diet, or both. Acute malnutritionmay also be brought about by particular com-binations of circumstances such as an episodeof malnutrition between past and future coursesof relatively adequate nutrition. Studies ofchildren with chronic subnutrition as well asthose who recover from acute severe malnutri-tion provide opportunities to study the effectsof nutritional inadequacy on behavioraldevelopment.

A number of model systems have been usedto explore the relationship of malnutrition tobehavior. At the human level these have con-sisted of (1) comparative studies of segmentsof well- and poorly-grown children in popula.:tions at risk of malnutrition in infancy; (2)retrospective follow-up studies of the ante-cedent nutritional experiences of well-func-tioning and poorly-functioning children in

such populations; (3) intervention studies inwhich children in the poor-risk populationwere selectively supplemented or left unsupple-mented during infancy and a comparativeevaluation made of functioning in the supple-mented, and unsupplemented groups; (4)follow-up studies of patients hospitalized forsevere malnutrition in early childhood; and(5) intergenciational studies seeking to relatethe degree to which conditions of malnutritionrisk in the present. generation of children de-rived from the malnutrition or subnutritionexperienced by their mothers when the latterwere children.

Studies of humans have been supplementedby a variety of animal models. The availableevidence will be considered in relation to theseinvestigative models. Perhaps the most com-plete study of the relation of growth achieve-ment to neurointegrative competence in chil-dren living in an environment of endemicsevere malnutrition and chronic subnutritionis a study of rural Guatemalan children (28).During their infancy and the preschool .yearsthe children lived in a village with a significantincidence of prolOnged subnUtrition. At schoolage relatively well-nourished children wereidentified as the better grown, and childrenwith the highest antecedent risk of exposure tomalnutrition identified as those with the lowestgrowth achievements for age. On the basis ofthis reasoning two groups of children werechosen from all village children aged six to 11years. These groups encompassed the tallestand shortest quartiles of height distribution ateach age in the total population of village chil-dren. To avoid problems associated with theuse of intelligence tests as measures of func-tioning in preindustrial communities, levels ofdevelopment in the tall and short groups werecompared by evaluating intersensory integrativecompetence through a method developed byBirch and Lefford (9).

At all ages taller children exhibited higherlevels of neurointegrative competence than didthe shorter group. Overall, the shorter children

116

were two years behind their taller agemates inthe competence they showed in processing in-formation across sensory systems.

To control for the possibility that heightdifferences were reflecting differences in ante-cedent nutritional status rather than familialdifferences in stature, the child's height wascorrelated with that of the parents. The re-sulting correlation was low and insignificant.In contrast, when the same ethnic group livedin more adequate nutritional circumstances theheight of children correlated significantly withthat of the parents.

Secondly, it was possible that the shorterchildren in the community at risk as well as incommunities not at risk of malnutrition weremerely exhibiting g,meralized developmental

lag for both stature and ncurointegrative matu-ration. No differences in neurointegrativecompetence attached to differences in staturein the children not exposed to endemic mal-nutrition, however.

Finally, it was possible that the shorter chil-dren came from home environments signifi-candy lower in socioeconomic status, housing,and parental education, and that both the mal-nutrition and the reduced neurointegrativecompetence stemmed independently from theseenvironmental deficits. When differences inthese factors were controlled for they did noterase the differences in interscnsory .integrativecompetence between children of differentgrowth achievements for age in the communityat nutritional risk.

Over the past several years replications ofthis study have been conducted in Mexico byCravioto and DeLicardie (30). In addition,Cravioto and his coworkers (29) have exam-ined another aspect of neurointegrative com-petence and auditory-visual integration in

Mexican school children. Once again, in chil-dren in communities at risk of malnutrition,differences in growth achievement at school agewere reflected in differences in auditory-visualintegration that favored the taller children.The latter findings are of particular importance

because of the demonstrated associ;Ition between

such competence and the ability to acquireprimary reading skill ( 10, II, 49).

A major consideration in interpreting thefindings of all these studies is the fact thatantecedent malnutrition is being inferred fromdifferences in height rather than by direct ob-servation of dietary intakes during the growingyears. A multitude of data from earlier studies,beginning with those of Boas ( 16) on growthdifferences in successive generations of childrenof Jewish immigrants and continuing throughthe work of Boyd Orr (61) on secular trendsin the height of British children, of .Greulich(40) on the height of Japanese immigrants, ofMitchell (58, 59) on the relation of nutritionto stature, and of Bouterline-Young (IX) onItalian children, as well as the recent study ofheights of 12-year-old Puerto Rican boys inNew York City by Abramowicz (I), all sup-port the validity of that inference.

Pek Hien Liang and coworkers (54) in

Indonesia, and Stoch and Smythe (66, 67) inSouth Africa have studied IQ levels in childrenexposed to chronic subnutrition. In the Indo-nesian study, 107 children between five and 12years of age were studied. All came from lowersocioeconomic groups. Forty-six of the chil-dren had been classified as malnourished duringan investigation carried out some years earlierinto nutritional status in the area. All childrenwere tested on the Wechsler Intelligence Scalefor Children (wisc)" and Goodenough tests.

The scores showed a clear advantage for thebetter grown and currently better nourishedchildren. Moreover, the data indicated that theshortest children were markedly overrepresentedin the group that had been found malnourishedin the earlier survey, and the largest deficits intQ were found associated with the poorest priornutritional status.

Stoch and Smythe have carried out a semi-longitudinal study of two groups of SouthAfrican Negro children, one judged in earlychildhood to be grossly underweight becauseof malnutrition, and the other considered ade-

117

quately nourished. At school age, the mal-nourished children as a group had a mean IQ22.6 points lower than that of the comparisongroup. Morecler, these relative differenceswere sustained through adolescence. Unfortu-nately, interpretation of the findings in this

study is made difficult because the better nour-ished children came from better families andhad a variety of nursery and school experiencesthe poorly grown children had not shared.

A number of follow-up studies of childrenseverely malnourished in early life have beenundertaken. As early as 1960, Wa terlow,Cravioto and Stephan (72) reported that chil-dren who suffered from such severe nutritionalillnesses exhibited delays in language acquisi-tion. In Yugoslavia, Cabak and Najdanvic(20) compared the IQ levels of children hos-pitalized for malnutrition at less than 12

months of age with those of healthy childrenof the same social stratum and reported a

reduced IQ in the previously hospitalized group.Of perhaps greater interest was their report ofa significant correlation between the severity ofthe child's illness on admissionas estimatedby his deficit in expected weight for ageandIQ depression in the school years. Indian work-ers (21) studied many variables in a group of19 children who had been hospitalized andtreated for kwashiorkor between 18 and 36months of age. When compared at school agewith a well-matched control group, signifi-cantly depressed IQ was found in the onceseverely malnourished children.

To more fully control for differences in thechild's family antecedents and microenviron-ment that may still exist even when more gen-eral controls for social class and general cir-cumstances are used in the selection of a

comparison group, we (14, 44) have comparedchildren malnourished in infancy with theirsiblings, as well as with children of similarsocial background. In the first of these studies,intelligence at school age was compared in 37previously malnourished Mexican children andtheir siblings. The malnourished children had

all been hospitalized for kwashiorkor betweenthe ages of six and 30 months. The siblingshad never experienced a bout of severe mal-nutrition requiring hospitalization. Siblingcontrols were all within three yearr of the indexpatients' age. Full-scale wisc. IQ of the indexpatients was 13 points lower than that of thesibling controls. Verbal and performance dif-ferences were of si.nilal magnitude and in thesame direction. All differences were significantat less than the 0.01 per cent level of confi-dence. These findings are in agreement withthose of the Yugoslav and Indian workers, andthe use of sibling controls removes at least

one potential complication for interpretation.In a second study (44), a large sample of 74

Jamaican boys who LA been hospitalized forsevere malnutrition before they were two yearsof age were compared with their brothers near-est in age and with their classmates whosebirthdate was closest to their own. All childrenwere between six and 11 years of age at follow-up. On examination, neurologic status, inter- .sensory competence, intellectual level, and avariety of language and perceptual and motorabilities were evaluated. Intellectual level wassignificantly lower in the index children thanin either the siblings or the classmate compari,son groups. As was to be expected, the orderof competence placed the classmate comparisongroup at the highest level, the index childrenat the lowest, and the sibs at an intermediatelevel. The depressed level of the siblings in

relation to classmates suggests one disadvantagein sibling studies. Clearly, the presence of achild hospitalized for severe malnutrition mayidentify a family in which all children are ata high level of risk for significant chronicundernutrition, and the index child may repre-sent an instance of acute exacerbation of this !chronic marginal state. Therefore, the indexchildren and sibs can be similar in that theymay share a common chronic exposure to sub-nutrition and differ only in that the indexchildren have experienced a superimposed epi-sode of acute nutritional illness as well. The

118

use of sibling controls, in any event, does :lotcompare malnourished with nonmalnourishedchildren. Instead, it determines whether sib-lings who differ in their degree of exposure toacute severe nutritional risk differ in intel-lectual outcomes.

Other follow-up studies of acutely maIn-,.:-ished children such as those of Cravioto andRobles (27) in Mexico, Pollitt and Granoff(64) in Peru, Botha-Antoun and coworkers(17) in Lebanon, and Chase and Martin (23)

in Denver, have all been shorter-term follow-ups of cognitive development in youngerchildren.

Cravioto and Robles (27) studied the devel-opment of returning competence in childrenhospitalized for malnutrition during their treat-ment and recovery while in hospital. Theirfindings, which indicated that behavioral re-covery was less complete in the youngest chil-dren (hospitalized before six months of age)than in older children, posed the possibilitythat this earliest period of infancy was theone most critical for insult to developing brainand thus to eventual intellectual outcome. Theabove-mentioned study of Jamaican childrendid not yield findings that supported this pos-sibility, however. In that study approximatelyequal numbers of children who had experiencedan acute episode of malnutrition in each of thefour semesters of the first two years of lifewere examined. Equivalent depression of iowas found to characterize each of the groupswhen they were separated by age at hospitaliza-tion (44). One long-term study of sibs andrelatives by Hansen (42) revealed no differ-ences in IQ between malnourished propositi andsibs. lo levels at different ages are so markedlydiscrepant from one another, however, as tonake interpretation of his report impossible.

In the Lebanese and Peruvian short-term.ollow-up studies noted above, depression inintellectual level tended to be found in theindex children. In the American study (23)

and in a Chilean study (60), the findings haveshown depression in intellectual function in

the preschool years in children hospitalized formalnutrition during the first year of life. TheAmerican investigators, working in Colorado,found that 20 children who had been hos-pitalized for malnutrition before the age of oneyear had a mean developmental quotient on theYale Revised Developmental Examination 17

points lower than that achieved by a matchedcontrol group of children who had not beenmalnourished. All ihese studies strongly indi-cate that severe malnutrition in early life tendsto depress later intellectual functioning.

In summary, the follow-un studies of childrenwho have been exposed to hospitalization for about of severe acute malnutrition in infancyindicate an association of significant degreebetween such exposure and reduced intellectuallevel at school age. The studies, involvingsocial-class controls and sibship comparisons,suggest that it is not merely general environ-mental deprivation but malnutrition as well

that contribute to a depression in intellectualoutcome.

The fact of these associations provides

strongly suggestive but by no means definitiveevidence that malnutrition directly affectsintellectual competence. As Cravioto and hiscoworkers (28) have pointed out, at least threepossibilities must be considered in the effort todefine a causal linkage. The simplest hypothe-sis would be that malnutrition directly affectsintellect by producing central nervous systemdamage. It may also contribute to intellectualinadequacies as a consequence of the child's lossin learning time when ill, influences of hos-pitalization, and prolonged reduced responsive-ness after recovery. Moreover, it is possible

that specific exposures to malnutrition at par-ticular ages may interfere with developmentat critical points in the child's growth and socreate either abnormalities in the sequentialemergence of competence or a redirection ofdevelopment in undesired directions. Althoughcertain of these possibilities (such as hospitali-zation and postillness opportunities for re-covery) can be explored in children, others for

119

moral and ethical reasons cannot. Thus, it is

impermissible to establish appropriate experi-mental models either for interfering with devel-opment at critical periods or for inducing braindamage. The approach to these problems re-quires either detailed analyses of naturallyoccurring clinical models or the developmentof appropriate animal investigations.

Animal models of the effects of malnutritionon brain and behavior have been used to studythe issue with a degree of control that is quiteimpossible in human investigation. A series ofpioneering investigations (31, 32, 74) havedemonstrated that both severe and modestdegrees of nutritional deprivation experiencedby the animal at a time when its nervous sys-tem was developing most rapidly results in

reduced brain size and deficient myelination.These deficits are not made up in later life,even when the animal has been placed on anexcellent diet after the period of nutritionaldeprivation.

More recent studies by Zamenhof and co-workers (79) and by Winick (77) have dem-onstrated that nutritional deprivation is also

accompanied by a reduction in brain cell num-ber. The latter effect has also been demon-strated in the brains of human infants whohave died of severe early malnutrition (78).

Enzymatic maturation and development inbrain is also affected, and Chase and coworkers(22, 23) have demonstrated defective enzymeorganization in the brains of malnourished

organisms.The evidence in all these studies indicates

that the effects of malnutrition vary in accord-ance with the time in the organism's life atwhich the deprivation is experienced. In someorganisms the effects are most severe if thenutritional insult occurs in the prenatal periodwhen as in others this phenomenon occurs dur-ing early postnatal life.

Some confusion in the interpretation of evi-dence has occurred because of the use of dif-ferent species, since in different organisms theso-called critical periods occur at different

points in the developmental course. In pigsbrain growth and differentiation occur mostrapidly in the period before birth, whereas inthe rat the most rapid growth occurs when theanimal is a nursling. In human beings theperiod of rapid growth is relatively extendedand .lasts from mid-gestation through the firstsix to nine months of postnatal life. In man,the brain is adding weight at the rate of 1 to 2mg/minute at birth and goes from 25 percent of its adult weight at birth to 70 per centof its adult weight at one year of age. Afterthat age, growth continues more slowly untilfinal size is achieved. Differentiation as well

as growth occur rapidly during the criticalperiods, with myelination and cellular differ-entiation tending to parallel changes in size.

Since brain growth in different species occursat different points in the life course, it is appar-ent that deprivations experienced at the samechronologic ages and life stages will have dif-ferent effects in different species. Deprivationduring early postnatal life will thus have littleor no effect upon brain size and structure inan organism whose brain growth has largelybeen completed during gestation. Conversely,intrauterine malnutrition is likely to have onlytrivial effects on the growth of the brain inspecies in which the most rapid period of braindevelopment occurs postnatally. When thesefactors are taken into account, the data leaveno doubt that the coincidence of malnutritionwith rapid brain growth results in decreasedbrain size and in altered brain composition.

It would be unfortunate if brain growth interms of cell number were to be viewed as theonly definer of rapid change and thus of criti-cal periodicity. In the human infant neuronalcell number is probably largely defined beforethe end of intrauterine life. Thereafter, throughthe first nine months of postnatal life, cell

replication is that of glial cells, a process thatterminates by the end of the first year. Myeli-nation continues for many years thereafter,however, as does the proliferation of dendritebranchings and other features of brain organi-

120

zation. It is most probable, therefore, that inman the period of vulnerability extends vel!

beyond the first year of life and into the pre-school period. Such a position is supported bythe findings of Champakam and coworkers(21), who it will be recalled, found significanteffects on intellect in their group of malnour-ished children who had experienced severe mal-nutrition when they were between 18 and 36months of age.

Other workers who have used animal modelshave sought to study the effects of malnutritionon behavioral outcomes rather than on brainstructure and biochemical organization. Thetypical design of these studies has been inves-tigations in which animals have been raked ondiets inadequate in certain food substances orin which general caloric intake has been re-duced without an alteration in the quality ofthe nutriments. Such animals have then beencompared with normally nourished members ofthe species as to maze learning, avoidance con-ditioning, and open field behavior. Unfortu-nately, most of the investigations have sufferedfrom one or another defect in design that makesit difficult to interpret the findings. Thoughin general the nutritionally deprived organismshave tended to be disadvantaged as learners,it is not at all clear whether this is the resultof their food lacks at critical points in devel-opment or whether the differences observedsten, from the different handling, caging, andlitter experiences to which the well and poorlynourished animals were exposed. In a consid-erable number of studies, however, food oravoidance motivation have been used as thereinforcers of learning. There is abundant evi-dence (6, 37, 52, 56) that nutritional defi-ciency in early life affects later feeding be-havior. Consequently, it is difficult to knowwhether the early deprivation has affected foodmotivation or whether it has affected learningcapacity. The use of learning situations thatdo not involve food but are based upon aversivereinforcement do not remove interpretativedifficulties since early malnutrition modifies

121

sensitivity to such negative stimuli (53).One must recognize that although the ani-

mal evidence suggests that early malnutritionmay influence later learning and behavior, ittoo is by no means conclusive. Moreover, whenlearning has been deleteriously affected, themechanism through which this effect has beenmediated is by no means dear. What is requiredis a- systematic series of experiments in whichbehavioral effects are more clearly defined, andin which the use of proper experimental de-signs, accompanied by appropriate controls,permits the nature of the mechanisms affectedto be better delineated.

Thus far, in our discussion of both the humanand animal evidence, we have been consideringthe direct effects of nutritional deprivation onthe developing organism. This is clearly toolim:ted a consideration of the problem. It haslong been known (61) that nutritional influ-ences may be intergenerational and that anindividual's grow th and functional capacitymay be affected by the growth experiences andnutrition of his mother. In particular, thenutritional history -of the mother and its effectupon her growth may significantly affect hercompetence as a reproducer. In turn, thisreproductive inadequacy may affect the intra-uterine and birth experiences of the offspring.

Bernard (8) has clearly demonstrated theassociation between a woman's nutritional his-tory and her pelvic type. He compared onegroup of stunted women in Aberdeen, Scotland,with well-grown women and found that 34per cent of the shorter women had abnormalpelvic shapes that were conducive to disorderedpregnancy and delivery as compared to 7 percent of the well-grown women. Greulichand coworkers (39) had reported still earlierthat the rounded or long oval pelvis, whichappears to be functionally superior for child-bearing, was more common in well off, well-grown women than in poorer clinic patients.They further noted, as had Bernard, that thesepelvic abnormalities were strongly associatedwith shortness.

Sir Duga Id Baird and his colleagues in Aber-deen, Scotland, have conducted a continuingseries of studies from 1947 onward on thetotal population of births in that city of200,000 to define the patterns of biologic andsocial interactions that contribute to a woman'sgroxth attainments and to her functional com-petence in childbearing. Baird (2) noted morethan 20 years ago that short stature, which wasfive times as common among lower-class thanamong upper-class women, was associated withreproductive complications. He pointed out(3), on the basis of analyzing the reproductiveperformances of more than 13,000 first-deliverywomen, that fetal mortality rates were morethan twice as high in women who were underfive feet one inch in height than in womenwhose height was five feet four inches or more.Baird ancl Isle), (4) demonstrated that pre-mature births were almost twice as commonin the shorter as in the taller group. Thomson(68) extended these observations by analyzingthe relation between maternal physique andreproductive complications in the more than26,000 births that had occurred in Aberdeenover a 10-year period and found that shortstature in the mother was strongly associatedwith high rates of prematurity, delivery com-plications, and perinatal deaths at each parityand age level. He concluded that "it is evidentthat whatever the nature of the delivery thefetus of a short woman has less vitality and isless likely to be well-grown and to survivethan that of a tall woman."

It was of course possible that these findingssimply reflected differences in social-class com-position of short and tall women, and werebased upon differences in "genetic pool" ratherthan in stunting as such. To test this hypothe-sis, the Aberdeen workers (5) reexamined theirdata for perinatal mortality and prematurityrates by height within each of the social classesfor all Aberdeen births occurring in the 10-yearperiod from 1948 to 1957. They found inevery social class that shortness was associated

with elevated rates of both prematurity andperinatal deaths. Concerned that the findingsin Aberdeen might not be representative, theyalso analyzed the data from the all-Britainperinatal mortality survey of 1958 and con-firmed their findings. Moreover, TI omson andBillewicz (69) in Hong Kong and Baird (5)have substantiated the Aberdeen findings forChinese and West African women, respectively.Other findings in a similar vein from this serieshave been summarized by Ills ley (47).

The available data therefore suggest thatwomen who are not well-grown have charac-teristics that negdtively affect them as child-bearers. In particular, short stature is associatedwith pregnancy and delivery complications andwith prematurity. Since growth achievementwithin ethnic groups is a function of healthhistory and in particular nutrition, it is clearthat the mother's antecedent nutritional historywhen she herself was a child can and does signif-icantly influence the intrauterine growth, de-velopment, and vitality of her child. Moreover,a mother's inadequate nutritional backgroundplaces her child at elevated risk of damage atdelivery.

It is instructive to consider the consequencesfor mental development and learning failurethat attach to the most frequently occurringconsequence of poor maternal growthlowbirth weight. Concern about the consequencesof this condition is hardly new, Shakespeareindicated it as one of the peculiarities of Rich-ard III, and Little (55) linked it with the dis-order we now call cerebral palsy. Benton (7)reviewed the literature up to 1940 and foundthat, though most students of the problemmaintained that prematurity was a risk to latermental development, others could find no nega-tive consequence attaching to it, At that timeno resolution of the disagreement could be

made because most of the early studies hadbeen carried out with serious deficiencies in

design and techniques of behavioral evaluation.Infants of low birth weight or early in gesta-tional age were often compared with term

122

infants who differed from them in social cir-cumstances as well as in perinatal status. Esti-mates of intellectual level were made withpoor instruments and often depended on "clini-cal impression" or testimony from parents orteachers.

Serious and detailed consideration of the

consequences of low birth weight for later

behavior can properly be said to have begunwith Pasamanick, Knobloch, and their col-

leagues shortly after World War II. These

workers were guided by a concept they re-ferred to as a "continuum of reproductivecasualty." They argued that there was a setof pregnancy and delivery complications thatresulted in death through brain damage, andhypothesized that in infants who survived ex-posure to these risks "there must remain afraction so injured who do not die, but de-pending on the degree and location of trauma,go on to develop a series of disorders extendingfrom cerebral palsy, epilepsy and mental de-ficiency, through all types of behavioral andlearning disabilities, resulting from lesser de-grees of damage sufficient to disorganize be-havioral development and lower thresholds tostress" (62). In a series of retrospective studiesthese investigators identified prematurity andlow birth weight as being among the conditionsmost frequently associated with defective be-havioral outcomes. They therefore undertooka prospective study of a balanced sample of 500

premature infants born in Baltimore in 1952

and compared them with term control infantsborn in the same hospitals who were matchedwith the prematures for race, maternal age,parity, season of birth, and socioeconomic status(50). Four hundred pairs of cases and controlswere still available for study when the childrenwere between six and seven years of age, andexamination of the sample indicated that atthis age the prematures and term children con-tinued to be matched for maternal and socialattributes (75). Findings at various ages per-sistently showed the prematures to be less intel-lectually competent than the controls. At ages

123

three to five the prematures were relativelyretarded intellectually and physically, and hada higher frequency of definable neurologic ab-normalities (43, 51). At ages sixthrough sevenu? scores on the Stanford-Binet test were ob-tained, as were wise IQs at ages eight to nine.At both age levels lower birth weights wereassociated with lower IQs (75, 76).

Although certain British studies such as thatof McDonald (57) and of Douglas (33, 34)

appear to differ somewhat from these findings,reanalysis of their data (12) indicates a similartrend. More dramatic differences between pre-matures and term infants have been reportedby Dri Hien (35, 36), but interpretation ofher data is made difficult by complexities inthe selection of the sample studied.

A number of analyses suggest that the effectsof prematurity are noc the same in differentsocial classes. Children from the lowest socialclasses appear to have subsequent lo and schoolperformances more significantly depressed bylow birth weight than is the case with infantsin superior social circumstances. This has beenreported for Aberdeen births (46, 65) and forHawaiian children in the Kauai pregnancystudy of Werner (73). There appears to bean interaction between birth weight and familysocial condition in affecting intellectual out-come, but the precise mechanisms involved inthis interaction arc as yet tticlear,

If the risk of deficient intellectual outcomein prematurity is greatest for those childrenwho are socially disadvantaged as well, ourconcern in the United States with the phe-nomenon of prematurity must increase. In

1962, more than 19 per cent of nonwhitebabies born in New York City had a gestationalage of less than 36 weeks as compared to 9.5

per cent of white babies, and in Baltimore thiscomparison was 25.3 per cent in nonwhiteinfants as compared to 10.3 per cent in whites(70). In 1967, nationally, 13.6 per cent ofnonwhite infants weighed less than 2,500 g

as compared to 7.1 per cent of white infants(71). Other relevant and more detailed analy-

ses of the social distribution of low birthweight and gestational age on both nationaland regional bases, together with an analysisof their secular trends, provide additional sup-port for these relationships (12). Thus, lowbirth weight is most frequent in the verygroups in which its depressing effects on intel-ligence are likely to be.

On the basis of the evidence so far set forth,it may be argued with considerable justifica-tion that one can reasonably construct a chainof consequences starting from the malnutritionof the mother when she was a child, to herstunting, to her reduced efficiency, as a repro-ducer, to intrauterine and perinatal risk to thechild, and to his subsequent reduction in func-tional adaptive capacity. Animal models havebeen constructed to test the hypotheses impliedin this chain of associations, most particularlyby Chow and his colleagues (24, 45) and byCowley and Griesel (25, 26). The findingsfrom these studies indicate that second andlater generation animals who derive frommothers who were nutritionally disadvantagedwhen young, are themselves less well-grownand behaviorally less competent than animalsof the same strain deriving from normalmothers. Moreover, the condition of the oa-spring is worsened if nutritional insult in itsown life is superimposed on early maternalmalnutrition.

A variety of factors would lead us to focuson the last month of intrauterine life as oneof the "critical" periods for the growth anddevelopment of the central nervous system inhumans. Both brain and body growth, togetherwith differentiation, occur at a particularlyrapid rate at this time. It has been argued,therefore, that whereas marginal maternal nu-tritional resources may be sufficiently adequateto sustain life and growth during the earlierperiods of pregnancy, the needs of the rapidlygrowing infant in the last trimester of intra-uterine existence may outstrip maternal sup-plies, The work of Gruenwald and coworkers(41), among other research, suggests that

maternal conditions during this period of theinfant's development are probably the ones thatcontribute most influentially to low birthweight and prematurity. Such concerns haveled to inquiries into the relation of the mother'snutritional status in pregnancy to the growthand development of her child. In consideringthis question, it is well to recognize that asyet we have no definite answer to the questionof the degree to which maternal nutrition dur-ing pregnancy contributes to pregnancy out-come. Clearly, whether or not nutritional lacksexperienced by the mother during pregnancywill affect fetal growth is dependent upon thesize and physical resources of the mother her-self. Well-grown women are most likely tohave tissue reserves that can be diverted to meetthe nutritional needs of the fetus, even whenpregnancy is accompanied by significant degreesof contemporary undernutrition. Conversely,under the same set of circumstances poorlygrown women with minimal tissue reservescould not be expected to provide adequatelyfor the growing infant.

The data presented reveal that the behavioraloutcomes attaching to malnutrition are at everypoint the product of both the nutritional cir-cumstances and the general environments inwhich children grow and develop. Investi-gators, many of whom have been fully awareof the varied features of environment that aresimultaneously present and capable of affectingchildren who have experienced malnutrition,have tended to use an additive model for theattribution of cause. They have argued thatthe variance in behavior outcomes may beviewed as the sum of the variances that areproduced by nutritional and nonnutritionalfactors. Expressed formally, such an additivemodel is,

3.2:c+ 72N:v.

In such a standard additive model each com-ponent can be fractionated into subcomponents,such as kinds of behavior, kinds of nutritional

124

factors, and types of nonnutritional com-ponents.

Such a model is possibly true, however, onlyif we assume we arc dealing with a quasi-experimental situation and if we ignore bothcorrelations and interactions between nutri-tional and nonnutritional influences. This canin fact be done in experimental situations inwhich randomization of both nutritional andnonnutritional components is possible. In fieldstudies in which such randomization is patentlyimpossible, the model becomes grosslyinadequate.

Malnutrition is not randomly distributedwith respect to nonnutritional factors. Rather,malnutrition occurs most frequently in certainkinds of families and environments and leastfrequently in others. A strong correlation con-sequently exists between nutritional and non-nutritional components. Our formula there-fore must be corrected to include a considera-tion of this correlation. The correction factorcould be expressed as,

(K)rN. NN 7NN.

This correction factor takes into account thecorrelative relations between nutritional andnonnutritional components.

In addition, the consequences of malnutri-tion are not separable from the nonnutritionalcircumstances in which the malnourished indi-vidual is developing. As Richardson pointedout in his paper at this conference, an inter-action exists whereby the consequences ofmalnutrition for children living in good envi-ronments are distinctly different from thecontribution malnutrition makes to develop-ment of children in poor ones. Consequently,we must introduce a second correction factorin our model, one for the interaction betweennutritional and nonnutritional components.This may be expressed as, (f) N.NN.

Our revised model, therefore, now has thefollowing form,

eb= 72N+ 0.2NK+(K) rN. NN 7NN+(f) N 'NN.

Such a model reflects the ecologic requirementsand indicates further research needed to solvethe problem of the effects of malnutrition in arepresentative sampling of environments. It isnot fundamentally different from biometricmodels applicable to genetic data (48).

A second methodologic question also requiressonic note. This is the fact that various investi-gators have studied children exposed to mal-nutrition at different points in their develop-mental course. Clearly, effects may or maynot be manifest at particular age levels, notonly for nutritional and nonnutritional fac-tors but even for genetic influences. A recentreport by El-Oksh and coworkers (38) illus-trates this point.

A parallel situation with regard to the geneticdetermination of body weight in mice (38)illustrates the point under consideration. Theresults of partitioning phenotypic variation ofweight at different ages into components areshown in Table 1:

Table 1. Phenotypic weight variation.

MaternalAge in Geneticweeks (heritability) Prenatal Postnatal

Residual(environ-mental)

0 0.00 0.52 0.14 0.62

1 0.11 0.27 030 0.31

2 0.23 0.25 0.38 0.14

3 0.30 0.12 0.30 0.29

6 0.26 0.17 0.23 0.34

As may be clearly seen, birth V, ht is inde-pendent of genotype of the young .;.id is largelydetermined by prenatal effects exercised by thedam. Their significance gradually decreases

while postnatal maternal effects and the geno-type of the offspring begin to play an increas-ingly important role. After weaning at threeweeks of age, the maternal effects begin todrop off.

If we take these genetic findings seriouslyand apply their lesson to the study of malnutri-tion, what is required is a consideration of out-comes over a wide range of ages from the

125

immediately postacute period of malnutritionto late adult life.

These models and considerations make appar-

ent the incompleteness of our present knowl-edge and, it is to be hoped, define opportunitiesfor further investigation.

REFERENCES

I. M. Heights of I2-year-old PuertoRican boys in New York Cite: origins of differences.Pediairus 43:427--29, 1969.

2. BAIRD. DUGAI.D. Social class and foetal mortality.Lan, et 2: 531-35, 1947.

3. -. Social factors in obstetrics. Lancet 1 :

1079-83,- 1949.

4. and R. 11.1.SLEY. Environment and child-bearing. Proc. R So, Mcj 46:53 -59, 1953.5. --. The epidemiology of prematurity. /

Pedtatr 65:909 -24, 1964.(i, BARNI,s, R. H., et al. Postnatal nutritional depriva-

tions as determinants of adult behavior toward food, itsconsumption and utilization. / Nutr 96:467 -76, 1968.

7. BENTON, A. L. Mental developmenr of prema-turely horn children: a critical review of the literature.Ain 1 Orthopsychiatry 10:719-46, 1940.

8. BERNARD, R. M. The shape and size of the femalepelvis. Edinburgh Med 1 59: 1-16, 1952. (In: Trans-

actions of the Edinburgh Obstetrical Society.)9. BIREI I. H. G., and A. LEE0Rn. Inter:ensory de-

velopment in children. Monttgr Soc Res Child Dry28: 1-48, 1963.

10. ---- and L. BEE stoNT. Auditory-visual ime-gra tion in normal and retarded readers. Ara 1 Orthwpsychiatry 44:852-61, 1964.

I I. and Auditory-visual integration,intelligence and reading ability in school children.Percept Mot Skills 20 : 295-305, 1965.

12. and J. 1). Gus sow. Disadvantaged chil-oren: health, nutrition, and school failure. New York,Harcourt Brace and World and Grime and Stratton, 1970,p. 322.

13. . Malnutrition and early development. InE. Grotberg (ed.) , Day care: resources for decisions,Washington. D.C., U.S. Ottice of Economic Oppor-tunity, 1971.

14. et al. The behavior of children in schoolwho were severely malnourished in the first two yearsof life. In preparation.

15. et al, School performance of children whowere severely malnourished in infancy. In preparation.

16. BOAS, F. Changes in the bodily form of descend-ents of immigrants. Washington, D.C., U.S. Immigra-tion Commission, 1910. (U.S. Immigration Commis-sion Document No. 208.)

17. Boll IA-ANfouN, E.. et al. Intellectual develop-ment related to nutritional status. / Trop Pediatr

14:112 -15, 1968.18. BouToutuf NE-YovNG, H. Epidemiology of dental

caries: results front a cross cultural studs: in adolescentsof Italian descent. New Engl 1 Med 267:843-49, 1962.

19. BRom:. J. Recent advances in human nutrition.London, J. & A. Churchill, 1961.

20. CARAK, V., and R. NAJD:m.01c. Effect of under -nutrition in early life on physical and mental develop-ment. Arch Al Child 40:532 -34, 1965.

21. CHANIPARAM, S., er al. Kwashiorkor and mentaldevelopment. Am I Om Nub. 21:544 -52, 1968.

22. Curses, H. P., et al. The effect of malnutritionon the synthesis of a myelin lipid. Pedtatrh s 40.

551-59. 1967.23. - and H. P. MARTIN. Undernutrition and

child development. Neu, Eng! / Med 282:933-76.1970.

24. CHow, B. F., et al. Maternal nutrition and

metabolism of the offspring: studies in rats and man.Ara / Pnb6c Health 58:668-77, 1968.

25. CowLEv, J. J., and R. D. GRIESEL. The develop-ment of second generation low protein rats. / GenetPsyi hol 103: 233 42, 1963.

26. ---- and . The effect on growth andbehavior of rehabilitating first and second generationlow protein rats. Alum Behan 14:506 -17, 1966.

27. CRAvurro, JoAQuIN. and BEATRIZ ROBLES. Evolu-

tion of adaptive and motor behavior during rehabilitationIron] kwashiorkor. Am 1 Ortbopsyebtatry 35:449 -64,

28. --- et al, Nutrition, growth and neurointe-grative development : an experimental and ecologic

studs. Pediatrics 38:319-72, 1966.

29. it al. Early malnutrition and auditory-visual integration in school age children. Spe, Edit,

2:75 -52. 1967.

30. and E. R. DELicARDIE. Intersensory de-velopment of school-age children, In: N. S. Scrimshawand J. E. Gordon (eds.) . Malnutrition, learning, andbehavior, Cambridge, Massachusetts, Press. 1968,

pp. 252-69.

31. DAvisoN, A. N., and Jot iN Dollars:G. 1).4yelina-

t ion as a vulnerable period in brain development. BrMed Bull 22:40-44, 1966.

126

32. DnuniNG, Jotts. The influence of early nutritionon the development and myelination of the brain.Pro, R Sec Land [Mol] 159:503 -09, 1964.

33. Douca As, J. W. Mental ability and schoolachievement of premature children at 8 years of age.Br Med I 1: 1210-14, 1956.

34. . "Premature" children at primary schools.Br ,Vied 1 1:1008-13, 1960.

35. DRILLIEN, C, M. The growth and developmentof the prematurely horn infant. Baltimore, Williamsand Wilkins. 1964, p. 376.

36. . Prematures in school. Pediatr Dig,September 1965, 75-77.

37. D.1.10-1T, 0., and J. A. KING. Effect of earlyfood deprivation upon later consummatory behavior inpuppies, Psphol Rep 6: 391-400, 1960.

38. EL-OKIsti, H. A., et al. Prenatal and postnatalmaternal influence in the growth of mice. Genetics

57:79 -94, 1967.39. GREULICIL W. W., and H. THoms. A study of

pelvis type and its relationship to body build in whitewomen. /AMA 112 : 485-92, 1939.

40. . Growth of children of the same raceunder different environmental conditions. Science 127:515-16, 195S.

41. GRUI:Nwm.n, P., et al. Chronic deprivation of

the fetus. Sinai Hasp 1 11:51 -80, 1963.

42. HANSEN, J. D., et al. What does nutritionalgrowth retardation really itnply? Pediatrics 47:299-313.1971.

43. HARPER, P. A., et al. Neurological and intel-lectual status of premattires at three to five years of age.

Pediatr 55:679 -90, 1959.44. HERTZIG, M. E., et al. Intellectnal levels of

school children severely malnourished during the first

two years of life. Pediatrics 49 : 814-24, 1972.

45. HS111.11, A. M., et al. Growth of young ratsafter differential manipulation of maternal diet. / Note91:195 -200, 1967.

46. 11.1.S1 EY, R. Early prediction of perinatal risk.

Proc R Soe Med 59: 181-84, 1966.

47. The sociological study of reproductionand its outcome. In: S. A. Richardson and A. F. Gutt-macher (eds.) , Childbearing: its social and psychologi-cal aspects, Baltimore, Williams and Wilkins, 1967, pp.75-135.

48. JINKS, J. L., and D. W. FLILKER. Comparison ofthe biometrical, genetical, MAYA, and classical approachesto the study of human behavior. Psycho( Bull 73: 311-49, 1970.

49. KAHN, D., and H. G. BIRCH. Development ofauditory-visual integration and reading achievement.Percept Mot Skills 27:459 -68, 1968.

50, KNOB! ocn, H., et al. Neuropsychiatric sequelaeof prematurity : a longitudinal study. 1.4.11.4 161:

581-85, 1956.

51. et al. The effect of prematurity on healthand growth. elm 1 Public Health 49. 1161-73, 1959.

52. LtivirsKY, D. A., and R. H. BARNES. Effects ofearly protein-calorie malnutrition of animal behavior.Read at annual nieeting. American Association for theAdvancement of Science, December 1969.

53. and . Effect of early malnutritionon reaction of adult rats to aversive stimuli. NatureILondl 225:465 -69, 1970.

54. LIANG, P. H., et al. Evaluation of mental devel-opment in relation to early malnutrition. Am / Chn'Voir 20: 1290-94, 1967.

55. Larn.E. W. J. On the influence of abnormalparturition, difficult labour, premature birth. and as-

phyxia neonat Oruro on the mental and physical condi-tions of the child, especially in relation to deformities.Trans Obstet .Sea c tond 3:293 -344. 1862.

56. MANDLER, J. M. Effects of early food deprivationon adult behavior in the rat. / Comp Physiol Psycho!51 : 513-17, 1955.

57. McD0NAt.n, A. D. Intelligence in children ofvery low birth weight. Br I Prey Soc Merl 18 : 59-74,1964.

58. MITCHELL, H. S. Nutrition in relation to stature./ A m Diet Assoc 40: 521-24. 1962.

59. . Stature changes in Japanese youth andnutritional implications. Fed Proc 28:877, 1964. (Ab-

sttact No. 27.)60. Mi)NCKEBERG, FERNANDO. Effect of early maras-

mic malnutrition on subsequent physical and psycho-logical development. In: N. S. Scrimshaw and J. E.

Gordon (eels.), op. cit., pp. 269-77.61. ORR, J. B. Food, health and income. London,

Macmillan, 1936.

62. PASAMANICK, B., and H. KNOBLOCH. Brain clam -age and reproductive casualty. Am 1 Orthopsychiatry30 : 298-305, 1960.

63. PATTERsoN, D. G. Physique and intellect. NewYork, Appleton-Century-Crofts, 1930.

64. PoDirr, E., and D. M. GRANOFF. Mental andmotor development of Peruvian children treated for

severe malnutrition. Rev lnteram P.thol 1:93-102, 1967.

65. RicHARnsoN, S. A. The influence of social, envi-ronmental and nutritional factors on mental ability. In:N. S. Scrimshaw and J. E. Gordon (eds.) , op. cit., pp.346-60.

66. STOCK M. B., and P. M. Stumm, Does under -nutrition during infancy inhibit brain and subsequentintellectual development? Arch Dis Child 38: 546-52,1963.

127

67. and . Undernutrition during in-fancy and subsequent brain growth and intellectual de-velopment. In: N. S. Scrimshaw and J. E. Gordon(eds.), op. cit., pp. 278-89.

68. THOMSON, A. M. Diet in pregnancy. 3. Diet inrelation to the course and outcome of pregnancy. Br INutr 13:509-25, 1959.

69. and W. Z. BILI.EWICZ. Nutrawnal status,physique and reproductive efficiency. Proc Mar Sot22:55-60, 1963.

70. U.S. NATIONAL CENTER FOR HEALTH STATISTICS:

Natality statistics analysis, United States, 1962. Wash-ington, 1964. (Public Health Service Publication No.1000.)

71. . Vital Statistics of the United States,1965. Washington. 1967.

72. WATERLOW, J. C., et al. Protein malnutrition inman; advances in protein chemistry. New York, Aca-demic Press, 1960, pp. 131-238.

73. WERNER, E. Cumulative effect of perinatal com-plications and deprived environment on physical, intel-

lec.tnal and social development of preschool children.Pediatrns 39:490-505, 1967.

74, WimowsoN, E. M. Nutritional deprivation in

psychobiological development: studies in animals. In:Pan American Health Organization. Deprivation in

psychobiological development, Washington, D.C., 1966,pp. 27-38. (PAHo Scientific Publication No. 134.)

75. WIENER, G., et al. Correlates of low birth weight:psychological status at 6-7 years of age. Pet atri, s35:434 -44, 1965.

76. et al. Correlates of low birth weight: psy-chological status at eight to ten years of age. PednitrRes 2:110-18, 1968.

77. FINICK, MYRON. Nutrition and cell growth.Mitt Rev 26: 195-97, 1968.

78. and Pfluxo Rosso. The effect of severeearly malnutrition in cellular growth of human brain.Pediatr Res 3: 181 -84, 1969.

79. ZAMENHOF, STEPHEN. DNA (cell number) andprotein in neonatal brain: alteration by maternal dietaryprotein restriction. Science 160:322-23, 1968.

128

SUMMARY OF SESSION IV

Robert E. Klein'

1)r. Birch addressed himself to two generalareas, the methodologic problems encounteredin human studies of the effect of malnutritionon later intellectual development, and animalexperiments and their utility in the study ofbehavioral outcomes following nutritionalinsult.

Dr. Birch noted that, although the extantdata and the published literature indicate long-lasting effects on behavior and performance (interms of IQ and achievement measures) follow-ing malnutrition, there are serious methodologicproblems in most studies. Much of this meth-odologic grief stems from the fact that mal-nutrition is not randomly distributed in thegeneral population. Thus, simple nonecologicmodels, which merely sum the environmentaland nutritional effects, are inappropriate.

Dr. Birch then discussed a simple interactionmodel that would allow the investigator toexamine the effects of nutritional insult at vari-ous levels of nonnutritional factors. He but-tressed his case for the need to use interactivemodels by pointing out that we must have alongitudinal focus in this type of study, sinceat distinct developmental stages genetic char-acteristics may inter-tct with the type of be-havior being examined so that the simple two-way matrices (although interactive) reallyrequire a third, longitudinal-developmental di-mension. I see this as an extremely useful wayto look at the problem and, once published anddisseminated, it should aid investigators inavoiding the pitfalls that have characterizedmuch of the published research in this area.

In discussing the use of animal models in thestudy of malnutrition and its effects on human

Institute of Nutrition of Central America and

Panama, Guatemala City. Guatemala.

behavioral development, Dr. Birch identifiedthree important advantages: tighte.g geneticcontrol, random assignment of subjects toexperimental treatments, and the availabilityof objective tests and outcome measures bywhich the effects of nutritional insult can beevaluated,

The discussion of animal experiments dwelton two issues, the sensitivity of behavioralmeasures and the interpretation of results. Onthe issue of sensitive behavioral measures, Dr.Birch referred to Karl Lashley's work as anexample of hov. one might use damaged ani-mals to search for evaluation procedures sensi-tive to different levels of nutritional insult.His major points were that there are differentlevels of complexity in behavior and that theevaluative technique must focus on the desiredlevel of behavioral complexity.

The focus on sensitive behavioral measures inanimal research remains an important point.Discussion of this issue was obscured becauseof the use of Lashley's work as an example.Dr. Dobbing pointed out that it is inappropri-ate to use ablation as an example of CNS insultin nutritional research, since nutritional insultfrequently occurs early, and the ablation ex-ample does not come to grips with the conceptof plasticity in central nervous system develop-ment and behavior. I believe Dr. Birch didhimself a disservice by using Lashley's work asan example, because it was clear throughout hispresentation that both developmental differ-ences and the point in development at whichevaluation is attempted are important con-siderations.

In conclusion, Dr. Birch pointed out thateven after sensitive measures have been devel-oped and successfully applied, additional ex-perimental probes are frequently necessary

129

because constructs encountered in the litera-ture (e.g., "motivational differences") are

often descriptive rather than explanatory andhave little or no predictive power.

In the discussion-following Dr. Birch's paper,Dr. Winick disputed the contention that littlelongitudinal animal research had been accom-plished. He noted research in developmentalbiochemistry as an example. A discussion ofhow to define "longitudinal" research followed,and this discussion more or less died instead ofbeing successfully resolved.

Dr. Stewart discussed his research with dogs.

130

After a careful and considered longitudinal in-vestigation, he gave up the study because hefelt it was relatively difficult to translate andextend the findings to human behavior. He

felt in general that the scientific communitywas uninterested in these results and that onecould only talk about differences in behavior,since there is relatively little to say about whythe differences occur and what they mean withrespect to human behavior.

No general agreement was reached about

extrapolating results from animal behavior

studies to human behavior.

Session V

FUTURE RESEARCH DIRECTIONS

Chairman

David Picou

Rapporteur

Herbert G. Birch

SUMMARY OF SESSION V

Herbert G. Birch'

Dr. Picot' opened the session by asking us toaddress ourselves to two issues: first, What aremore specific, revealing, and feasible studiesthat need to be conducted?, and second, Whatare potential areas for conceptual as well as

experimental and observational consideration?For example, what is the relevance of animalmodels to the human condition? What kindsof tools are required for better studies of youngchildren? What are the potential contributionsof intervention studies, and what kinds ofintervention studies arc needed? To what de-gree are there important areas of research thathave been ignored or relatively neglected, e.g.,prenatal malnutrition? How good is the evi-dence with respect to the permanence or tran-sience of nutritional effects on brain and

behavior? And. finally, do we have any sys-tematic information on the relation of recoveryrates during severe nutritional illness !-o the

permanence or transience of effects.In general, the first part of the discussion

concerned itself with the issue, Have wereached the point in investigation where fur-ther research is rela-ively uninformative withrespect to certain main questions? These "mainquestions" were identified as: Does malnutritionhave an effect on brain and behavior? Do weknow what needs to be done to prevent this?What kinds of information are valuable foranswering these questions?

The second half of the discussion focusedon the design of research. Consideration of aseries of models made it clear that there wereseveral levels at which research took place.One of these was a general descriptive level,namely a consideration of whether there are

' Department of Pediatrics, Albert Einstein College ofMedicine, Bronx, New York. USA.

consequences, and whether such consequences

are relatively permanent and quantifiable ormeasurable in any way. Second, there was theissue of the functional consequences of the

structural, biochemical, and physiologic altera-tions that had been reported. Third, there wasthe level of the general contribution to sciencethat concern with the probleMs of malnutritionmakes.

In this connection it was noted that thestudy of malnutrition or of the nutritionalrequirements for growth and development rep-resents strategic paths of entry into the generalquestions of developmental biology, and into aconsideration of questions relevant to the

growth and organization of functional systemsin developing individuals. Such concern leadsto the level of biologic mechanism and to aconsideration of cellular and subcellular proc-esses in development. Finally, it was pointedout that at a biosocIal level we must deal withthe relevance of :14 of these studies to thehuman condition, and to general social policy.It was agreed, however, that this issue wouldbe deferred for discussion in th:: last session.

In the course of considering animal investi-gations, one point in particular was made onseveral occasions and needs underscoring. Itwas noted that in animal investigations therehas been a focused concern upon brain and itsorganization, and relatively insufficient relationof changes in brain to the general bodily envi-ronment in which the changes exist. This wasa point made by Drs. Dobbing, Cheek, Stewart,and others. They emphasized that we knowlittle about the associations between the kindsof changes that are registered in brain physi-ology, and the kinds of changes that maysimultaneously be occurring in the rest of thebody and which may be as important as the

133

environment of brain. Particular emphasis wasplaced upon our lack of knowledge of endocrineorganization as it may contribute to the physi-ology of the brain, and to its alteration, per-manent or otherwise, in malnutrition.

There is a large body of data which indicatesthat under a variety of conditions of stressthere are alterations in endocrine organizationon a relatively permanent basis. Moreover, interms of a neuroendocrine model these altera-tions are important for the ack.ptive capacitiesof organisms and their behavior. Consequently,a renewed concern with the genlral physiologicconsequences of malnutrition represents a moreexpanded path for dealing with behavior

changes than the simplistic one of brain dam-age, e.g., brain is damaged, behavior is altered.

Several problems with respect to the humancondition were also brought forward. The firstdeals with the degree to which nutritional andnonnutritional components interact in the pro-duction of behavioral sequelae. Various people

said that we did have enough evidence to indi-cate that relatively permanent changes in brainderive from malnutrition. Others stated withequal firtun_ss that this question was not yetanswered.

It is clear that we have a huge body of evi-dence indicating that depressed levels of func-tioning are associated both with malnutritionand more generally with conditions of socialand stirnulational disadvantage that character-ize the life styles of families in which malnu-trition occurs. What is clear is that repetitiousinvestigation of this phenomenon is not pro-ductive, and that what is needed is the applica-tion of new models appropriate for the studyof biosocial interactions in the problem. Suchmodels have been proposed in the meeting, andmay be used as ,1 basis for subsequent inquiry.Clearly, we must move to a new level of con-cern with the problem, rather than with therestatement of the older question that domi-nated the beginning of the meeting.

The second issue was the relation of physio-logic recovery to recovery of behavioral func-

tions. We have relatively sparse information onrecovery states during acute severe malnutri-tion, and on the relation of physical and physio-logic recovery at that point in time to be-havioral recovery. Preliminary studies such asthose carried out by Cravioto and Robles inMexico and some information available fromD. S. McLaren (private communication) inLebanon suggest that some association exists be-tween rates of physiologic and behavioral re-covery. These data are difficult to interpret be-cause of the small number of cases and becauseof certain other difficulties in design, namely,insufficient specification with respect to thechronicity or acuteness of the illness, insuffi-

cient specification with respect to the age ofmalnutrition, and the like. It is clear fromthese gaps in our information that studies ofrecovery states and of the relation between

physical, physiologic, and behavioral phenomenaare very much to be desired. Further, we haveno systematic evidence on the long-term rela-tionship between behavioral recovery duringacute illness and later functioning. Such ques-tions are raised as issues for further and pro-ductive research.

A somewhat related question is whether themethod of malnutrition treatment in any de-gree affects the nature of the outcome. Wehave not had any systematic consideration ofthe relation of different kinds of treatmentofdifferent combinations of physiologic, physical,and behavioral rehabilitative proceduresto theways in which outcome may be affected. Mc-Laren in Beirut has begun such an inquiry.

On the basis of the partial information we haveabout that study, it is indeed an area thatappears profitable for further work. These

workers suggest that real differences in boththe rate at which children recover physicallyfrom the illness and in behavioral recovery

rates occur when stimulation at the play levelis introduced into hospital care, These data

must still be related to age at malnutritionand severity and duration of the illness.

134

In the discussion it was noted that we havelittle information about the present and back-ground history, health conditions, and circum-stances of individuals as they relate to whetheror not they become severely malnourished, andto whether or not such malnutrition as theymay experience is expressed in permanent ortransient depressions of behavioral competence.There is need for certain kinds of long-termfollow-up and of longitudinal studies.

A further question that could productivelybe explored was that of the time in develop-ment at which defective or deficient outcomesbecome manifest and the kinds of behavioralchange or alteration that are exhibited at suchpoints in time. For example, a language deficitis unlikely to be ianifest in the first year oflife, but may be manifest later. More impor-tantly, it may well be that particular featuresof behavior, and adaptation and resistance tostress may not manifest themselves, except assleeper effects much later than the time ofinsult and illness itself. What was suggestedin the discussion was that we pursue far morelong-term follow-up studies of stress into laterlife. Dr. Fred Richardson argued that it wasimportant for us to shift from considerationof this issue as purely a pediatric problem toone relating to adult functioning as well, in-cluding geriatric functioning. This aspect ofthe issue has indeed occupied a most prominentplace in studies of overnutrition, but has notconstituted a serious realm of inquiry in theareas of undernutrition or malnutrition. Stud-ies of overnourishment might benefit frominvestment in comparative studies of under-nourished adult populations. Questions thatmight be posed are: Flow does the incidenceof atherosclerosis differ in undernourished

versus overnourished populations? Does pre-mature senility occur in adults who have suf-fered either chronic lifelong undernutrition(or overnutrition), or prolonged severe mal-nutrition, as in wartime concentration camps?

Finally, it was pointed out that a consid-erable need exists for defining types and timesof intervention and types and styles of effectiveprocedures for preventing malnutrition. Inparticular, it was emphasized that we neededto know a good deal more about nutritionalattitudes and practices and the ways in whichfood use and a specific application to membersof family groups is reflected in the behaviors,attitude structures, and cultures of the com-munities in which man lives. These generalconsiderations were followed up by the state-ment of a series of explicit researches that vari-ous members of the -seminar felt they wantedto undertake in the near future. These studiesinclude: (1) an emphasis on reexamining newmodels for considering the interaction of nutri-tional and nonnutritional factors; (2) a focusupon early development and cognitive conse-quences as they may reflect modifications indevelopmental course and process in relation tomalnutrition; (3) emphasis on a detailed con-sideration of maternal malnutrition; (4) con-sideration in detail of prenatal malnutritionand its interaction with the nutritional fir -cumstances in postnatal life; (5) study of theintergenerational effects and chronicity of sub-nutrition and its consequences for development;and (6) a wish to define limits during recoverycourses in which full or relatively full recoveryof physiologic and anatomic as well as bio-chemical and behavioral function might bepossible.

135

Session VI

SOCIAL AND ADMINISTRATIVE PROBLEMS

Chairman

D. B. Jelliffe

Rapporteur

Robert Cook

NUTRITION, PUBLIC HEALTH, AND EDUCATION

Jack TizardI

As scientists concerned with the long-termeffects of malnutrition in early life, we shouldconsider ways in which our knowledge can beutilized by those responsible for social, educa-tional, and public health policy, but as a groupwe are not particularly well qualified to giveadvice on nutrition policy to governments orpublic administrators. Economists, anthro-

fologistsand sociologists, agriculturalists and

political scientists, and other biologic and socialscientists may have a more direct part to playas consultants to governments than those whoseprimary concern is biology and psychology asrelated to pathology. But we must try toanswer two questions: What use can be madeof our knowlddge?, and, What steps should weas scientists take to ensure that full use is

indeed made of it?The classical view of science's social impli-

cationsthat science is an activity that is

essentially a search for truth unsullied by con-siderations of utilityis not likely to be widelyheld by scientists. For Rutherford, as for mostof his colleagues and students, part of the satis-faction to be derived from the study of sub-atomic physics lay in the belief that theoryand discovery in this field had no practicalconsequences. Knowledge was pursued for itsown sake. Scientists and public alike agreedthat the pure scientist was not responsible forthe use of discoveries that he as pure scientistdid not exploit for personal gain himself.

' Institute of Education, University of London, Lon-don. England.

In 1945 this understanding between scientistsand the public, which was already under strain,was destroyed by the atomic bomb. The fulleffects of the bomb were not immediately ap-parent, however, in that after World War IIpure science prospered as never before. In

Britain as in other advanced countries, it hasbeen among the most rapidly growing of themajor industries and support for it has con-tinued to be given with surprisingly fewstrings. Pure science, no less than the exploita-tion of scientific discovery for practical, com-mercial, and military purposes, has been a

major beneficiary of the affluent society.This golden era for science is almost cer-

tainly passing. In an age that is increasinglydisillusioned and afraid of the consequences ofscientific discovery for society, the public is

beginning to demand that some priority be

given to scientific research that is relevant tothe human condition, that is, capable of beingused directly for human betterment. Scientistshave rightly pointed out, of course, that re-search often has unanticipated consequences forgood as well as for ill, and they have arguedthat a direct attack on major social problemsis often useless. Their argument is incontro-vertible, but the public remains unconvincedand uneasy. Furthermore, scientists themselvesappear increasingly to believe that the distinc-tion between "pure" and "applied" science hasall along been nothing more than a convenientfiction. Thus the pure scientist no less thanthe applied one is likely today to be concerned

139

with the social responsibility of scientists, andthose who deny that scientists have such a

responsibility find their views increasingly re-garded as old-fashioned.

Because biologic scientists have always hada close association with medicine, the notionof social responsibility is one that has beeneasier for them to accept than for their col-leagues in the physical sciences. Where does ittake us in regard to research on the effects ofmalnutrition?

The Application of ResearchFindings in Nutrition

There is a wide range of attitudes amongthose who work in the field of nutrition andaccept the view that the scientist cannot beabsolved of responsibility for his findings.Some maintain that our knowledge today is

too slight to enable us to offer any advice what-soever to those who make public policy. Theirargument is that the realities of undernutritionare exceedingly complex and only imperfectlycomprehended, and that before we speak atall about them we must do much more researchto elucidate basic mechanisms and to judge therelative importance of nutritional and otherfactors in mental and physical development. Anot dissimilar consequence follows from theview of others who at first sight appear tothink the opposite. To them our knowledgeis so great that no self-respecting scientistcan in good faith endorse the "simplistic" solu-tions likely to appeal to a politician. Sincewhatever he says will be misconstrued, runsthis argument, the best thing the scientist cando is to mind his own business. A third view,more commonly expressed by social than bymedical scientists, may be described as utopianor millennial. It holds that unless a "total"change is made in the structure of society, anypiecemeal reform is mere tinkering. The sci-entist can thus show his social concern mosteffectively by abandoning science for politics.A quite different view is propounded by others

who offer one or another panacea (breast feed-ing is the current favorite) that they believewill largely solve the major complex problemsof malnutrition and undernutrition.

All these schools of thought in their differentways counsel inactivity on the part of the sci-entist. He does not know enough or he knowstoo much to offer advice. The problem is in-soluble without revolution, or the solution is

simple and is already at hand so that onlypropaganda is required to implement it. Thedifficulties are "fundamentally" educational, orthey are economic; they are psychologic, havingto do with attitudes toward food or child-rear-ing practices, or they are "primarily" genetic,being perpetuated by succeeding generations ofpoor, ignorant, feckless, and stupid people whoshare a common heredity drawn from a stag-nant pool; the problems are largely adminis-trative; they are physical; they are merely sta-tistical artifacts; they are and have been sincethe time of Malthus the major problems thathave confronted the world.

The attitude most of us concerned withnutrition would probably hold is that whilethe problems are indeed complex and littleunderstood in detail, we already know enoughto advise that some forms of political actionbe undertaken and others not. The situationis similar to that in the nineteenth century inwhich various public health problems wereeffectively tackled long before their causes

were understood or specific forms of treatmentwere available. Mental retardation presents asimilar situation today: it is a problem no lesscomplex than malnutrition, but one to whichscientists have made major social contribu-tions. Through epidemiologic inquiry theyhave charted the frequency, severity, and prog-nosis of mental retardation, have investigatedforms of treatment and training, have calcu-lated the social and administrative facilitiesthat society requires to deal with the mentallyretarded, and have given estimates of the finan-cial and social costs of doing so. It is abun-dantly clear that left to themselves, adminis-

140

trators charged with making policy about thementally handicapped would be likely to clingto traditional wisdom. Radical changes insocial policy in this field have come becausepeople outside the system have provided leadstoward fresh ways of doing things. The samemay apply in the field of nutrition.

I suggest that there arc three ways theentist as scientist (as opposed to politician orcitizen) can make a contribution to socialpolicy on nutrition. First, he can provide expertfactual information together with an informedopinion based upon his knowledge; second, hecan assist in the utilization of his findings bymaking them readily available, or by addingan explicitly practical component to an on-going research project; third, he can aid in thedesign and evaluation of "action" projects totry out different ways of doing things.

The Provision of Material and Advice

Malnutrition and undernutrition arc publichealth problems whose study can be effectivelycarried out using epidemiologic methods. Thesurveys described at this seminar indicate thatepidemiologic and clinical research is alreadyhighly productive. Anthropometric measuresenable us to obtain valid estimates of the preva-lence of undernutrition in various populations,and for particular purposes they may be sup-plemented by biochemical and clinical mea-sures. Surveys carried out of properly selectedsamples can thus provide both adequate mea-sures of the extent of undernutrition in spe-cific populations and markers against which tojudge the success of changes in nutritionalpolicies. As a complement, dietary surveysincluding. investigations of the prevalence andconsequences of various food taboos and pref-erences can give a picture of current eatinghabits in populations and subgroups, while moreglobal surveys of food supplies and food con-sumption present crude national or regionalestimates of the amount of food available.

It is well recognized that figures on childmortality can be used to monitor changes in

the health of the population and to highlightdifferences between one population and another.In Jamaica in 1965, for example, the mortalityrate among children aged six to 24 months wasmore than eight times that in England andWales (6). Differences in mortality were muchless in other groups of children under the ageof five years. The period of six months to twoyears may be seen from these data to be aparticularly vulnerable and critical one for theJamaican child. Among children aged 12 to23 m malnutrition was the greatest singlerccor of death in Jamaica in 1963,

follow,, by gastroenteritis and pneumonia.Population mortality statistics and hospitaladmission and mortality statistics indiCate boththe relative importance of various diseases andthe age groups that arc particularly susceptibleto severe illnesses of various sorts. Epidemiolo-gists may be primarily interested in such datafor other reasons, but the point is that the datathemselves can be used to monitor the healthof a people.

Epidemiology is of course that branch ofmedicine which is particularly closely relatedto health service planning. In a more generalsense, reviews of current knowledge that in-clude a consideration of social implications alsoserve an invaluable function in directing atten-tion to matters otherwise likely to be knownonly in shadowy outline and by a small groupof specialists. Boyd Orr did this for nutritionin the 1930's, and in different ways the 1967

International Conference on Malnutrition,Learning, and Behavior (3) and the volumeDisadvantaged Children by Birch and Gussow(1) opened up whole areas of knowledge tothe educated public as well as specialists.

Just as reviews arc required to direct theattention even of specialists to a field's state ofknowledge and its social implications, however,so further popularization is likely to be re-quired to make the public aware that a problemexists. Furthermore, most academic researchcannot or will not be assimilated by professionalpractitioners unless it is reformulated so as to

141

be relevant to the problems that concern them.This is not likely to happen unless research find-ings are presented and interpreted for specificaudiences, with specific and practical objectivesin mind. A recent discussion concerned withdissemination of social science research findings(4) distinguished five different types of targetaudience, each requiring a different approachand possibly a different vehicle of communica-tion: academics, social scientist practitionersoutside the academic world, practitioners in

other disciplines, practicing managers and ad-ministrators (and presumably politicians), andthe general public. In the field of nutrition thelist of target audiences would be different butequally diverse.

Of all target audiences, government is per-haps the most important and most difficult toreach. The recent Waterlow Report preparedby the Nutrition Advisory Committee of theScientific Research Council of Jamaica onnutrition in Jamaica (6) provides a modelshowing how scientists can use their knowl-edge to make recommendations for publicpolicy. The Report, which has not yet beenpublished,' was concerned with long- andshort-term policy toward nutrition, its eco-nomic and sociologic implications, and trends,new developments, and education in nutrition.It dealt specifically with the situation inJamaica, but its approach is generally valid;it can serve as a model for research workers inother countries who wish to show the policyimplications of current knowledge.

One feature of the Waterlow Report (andthis is likely to apply also in other countries)was that it compiled a great deal of availablebut not readily accessible information on vari-ous aspects of nutrition in Jamaica that hadnot previously been pieced together in anysystematic way. Hence a factual documentthat connected and collated in an easily avail-able form the scattered literature on Jamaican

I am indebted to Prof. J. C. Waterlow for makingthe nonconfidential material in this Report available

to me and for permission to quote from his preface.

nutrition is likely to be useful for referenceand as a baseline for future studies and surveys.The Nutrition Advisory Committee also

thought it desirable to discuss explicitly theimplications of its findings and in doing so itwas faced with a dilemma:

On the one hand it is desirable that technical andprofessional people should give advice which isrealistic and practical in the light of the situationwhich exists. It is not helpful to make sweepingand generalised proposals which cannot he carried out in a country with limited resources. Onthe other hand, we would be failing in our dutyif we did not propose measures which seem neces-sary to improve the health and wellbeing of thepeople, simply because they will draw uponmoney and personnel which arc in short supply.We realise fully that for government all actionand expenditure are a matter of priorities. Wehave therefore attempted to maintain a balancebetween what is desirable and what is practicable.

In summarizing the existing situation, theWaterlow Report first considered the currentnutritional status of the Jamaican people andthen analyzed food supplies and patterns offood consumption. The Report's data providedcriteria by which to judge existing food policyand schemes for nutritional improvement.From these conclusions it was possible to go onto examine aspects of research and development,and the feasibility of various strategies to pre-vent infant malnutrition. The Report, as onewould expect of a document drawn up forpresentation to a scientific research council,made a clear distinction between facts and in-ference. Its publication will thus not only pro-vide a baseline for comparisons to be made inthe future and a powerful stimulus to otherscientific groups to interest themselves in theproblem of malnutrition, but also a spur topolitical discussion and action.

The Waterlow Report was directed to theJamaican Government and was therefore con-cerned with economic and social policy. Forother target audiences (for example generalpractitioners, infant welfare staff, health visi-

142

tors, pediatricians, or educators) different prob-lems need to be highlighted and differentstrategies employed. The task of making rele-vant knowledge available to professional peopleis one that cannot be left to the journalist; towrite a manual on infant feeding or the man-agement of the severely ill child in the hospitalis no task for a layman. The involvement ofdoctors or scientists is necessary. At all events,it is through reportorial activities such as thoseof the Nutrition Advisory Committee thatmany scientists are likely to make their mostdirect and effective contribution to sociov.

Making Full Use of Scientific Resources

A second way in which scientists could con-tribute more effectively to social policy thanthey commonly do is through the addition, toresearch units and groups, of components thatare expressly concerned with current social

problems, or with operational research, or withthe implications of research findings in par-ticular fields for education or medical practice.Sometimes extra staff or facilities may beneeded to enable those engaged in research toincrease the scope of their inquiries. They mayfor example have to collect extra data duringa survey, or undertake further analyses of datacollected for another purpose; collaborationmay need to be sought from colleagues in otherdisciplines to enable additional projects to begrafted on to a research program; representa-tions to the government and perhaps collabora-tion with government agencies may enablestatistical returns to be better designed or re-analyzed; they may find it possible to sharefacilities with or give advice to other researchgroups; they may hold interdisciplinary meet-ings, or engage in advanced teaching.

That such activities are desirable would seemtoo obvious to labor were it not that all overthe world -scientists tend to work in isolationeven from their colleagues in the same fieldand certainly from those in other fields andfrom practitioners. Collaboration between sci-entists and practitioners is naturally difficult

143

because the practitioner has constantly to makedecisions on the basis of inadequate evidence,and psychologically needs to believe in what heis doing. The scientist, on the other hand, is byprofession a doubter and a questioner of evi-dence and current beliefs. But other considera-tions also interfere. The scientist is apt to seepractical problems as mere distractions, whilehis qualifications and caveats may seem mereirrelevancies to the administrator. Even col-laboration among scientists of different disci-plines is rarely, in my experience, truly satis-factory; more commonly activities go on inparallel, or researches proceed quite independ-ently for all practical purposes and gain almostnothing from the fact that people working inthe same unit or building conduct them. inter-disciplinary seminars are also all too often aflop or a bore.

No one can say exactly why this should be,though of course it may not matter very muchin places where there is a wealth or researchfacilities and endless duplication of researcheffort. But where people and resources are

scarce it is clearly important that they be usedeffectivelyand it is by no means certain thatthe gains that accrue when scientists work"undistractedly" are always maximal.

Since scientists themselves find it difficult towork together, even whenas is usually thecasethey intend to do so and there is goodwill, those responsible for research strategymight from time to time, when they are con-sidering the funding of research, give attentionto problems of collaboration and of utilizationof research findings. They might for examplepersuade a research team to include among itsmembers a behavioral scientist or applied statis-tician to exploit certain implications of theirfindings. A subunit concerned with appliedproblems might be formed; research staff mightbe used more effectively in teaching or a systemof temporary exchanges might be introducedbetween the research and teaching staff of auniversity, or the scientific or administrativestaff in government. Not all of these actions

would be effective or indeed appropriate in anyparticular situation. But today we tend to beremarkably unadventurous in trying any ofthem, chiefly I believe because no one has aprimary responsibility to take the initiative indoing so. A recent discussion of the utilizationof social science research (4) stressed the partthat research councils and funding agenciesmight play in stimulating research units tobroaden the scope of their research if they gavethem more money. It is a matter of judgmentin particular cases how far additions to a re-search program can be made without destroyingthe independence and creativity of its scientists.There is no doubt, however, that sometimessuch additions would result in disproportionategains, and the opportunities provided are byno .means always used.

The Provision and Evaluation of Services

Medical and social scientists make their mostdirect contribution to government by advisingand assisting in the planning and evaluation ofservices. The contribution to public healthmade by epidemiology has already been

mentioned. Gruenberg (2) has pointed outthat surveys of the incidence and prevalenceof diseases or disorders in a community canprovide quantitative information to (1) esti-mate the size, nature, and location of the com-munity's problems, (2) identify the com-ponents of the problem, (3) locate populationsat special risk of being affected, and (4) iden-tify opportunities for preventive work andneeds for treatment and special services. Thus,he says, epidemiology serves as a diagnosticianfor the official or community leader who ispracticing community medicine, social medi-cine, public health, or public welfare. Thenature of the community's health problems isapproached diagnostically with epidemiologicmethods.

He adds that there are dangers in such com-munity diagnosis surveys. One danger is thatof overgeneralization. Another is that: "Themotives of those encouraging and financing

such studies are not always confined to gettinginformation so as to produce better servicesmore appropriate to the needs of the popula-tion. Expensive as surveys are, they are a goodbit cheaper than services to sick people, so thatthis type of research is particularly vulnerableto misuse on the part of those who are opposedto extensions of service and are using the 'needfor more information' as a delaying tactic inthe political arena." Gruenberg gives the in-vestigator who is caught in this trap a usefultip. He points out that there is enoughinformation already available about unmetneeds to go ahead with service planning, with-out awaiting a new study's results since anyoneacquainted with the field can roughly predictthem. However, he adds, this cannot be saidabout judgments regarding the efficacy of serv-ices when they have not been systematicallystudied previously."

Until comparatively recently the need forevaluative study was barely appreciated. In-stead, the view was taken that the provision ofservices to needy people was in itself likely tobenefit them. Given a necessary minimum ofcompetence and sophistication on the part of aservice's recipients, this view is a very reason-able one. A failure to monitor the quality of aservice can lead to disastrous consequences,

however. Thomson (J) tells us that in oneWest African village, recipients know driedmilk as the stuff which causes diarrhea," andBirch and Gussow (1) give similar examplesfrom this hemisphere. The Water low Reportmentions that skimmed milk powder is dis-

tributed in Jamaica in 2 kg packs whichcarry no instructions as to how the powdershould be used, nor indeed is there any moni-toring of the way it is used. Others have citedinstances in which relief organizations haveprovided dietary supplementation and thenwithdrawn their help suddenly, creating a situ-ation in which catastrophe is inevitable. Noform of intervention is more foolish and tragicthan the giving of tractors to people who are

144

not taught how to use and maintain them, oreven that they require maintenance,

At the most general level, however, no onecan doubt that better feeding is desirable. Un-certainties relate to the effectiveness of variousservices, the possibilities of improvement in

services, satisfactory methods of introducingnew forms of care and education, and alterna-tive ways of providing services. The researchworker can help resolve these problems in a

number of ways:He can devise ways to monitor health and

education services, developing indicatorsthrough which medical and social evaluationcan be carried out. The "centers of excellence"in which research is carried out (for examplethe Mexican village in which Dr. Cravioto'steam works, and the Tropical Metabolism Re-search Unit in the University of the WestIndies) can have their educative as well as

their research role fully utilized.Equally important, research on many differ-

ent patterns of care can be undertaken on asmall scale, but its results may lead to a choiceof one or other alternatives usable on a widerscale. Experience so gained before nationalcampaigns are feasible or undertaken may en-able great savings to be made at a later date.

The professional is often in a better positionthan others to give advice, provided he informshimself of the situation's everyday realitiesbefore doing so. The administrator is alwaysworking against pressure of time and a multi-plicity of other cormuitments, and even theprofessional adviser to a government may notbe really an expert, or may have time only todecide how the conventional wisdom may be

adapted to the idiosyncrasies of the localsituation.

In recommending the involvement of sci-entists in government, I am not suggestingthat it is the job of tl ..pert to make politicaldeci.sions. lie may I present a series ofalternative strategies .,1 arguments in favorand against each, expecting that the adminis-trator will seek alternative sources of advice.It is t be hoped that this does happen: theexpert is often wrong. Furthermore the poli-tician must come to a decision on politicalrather than on merely scientific grounds.Finally, the scientist should not allow himselfto be put in the position of being asked whethermoney spent on health services is more or lesswell spent than money spent on education, forexample. The cost-benefit approach to the pro-vision of services is always of dubious value,and in any case to counterpose health and edu-cation in this way seems politically naive. Forone thing, ministries tend to be remarkablyindependent, one not knowing what others arcup to. More important, it seems likely thatwork done in one field is more rather than lesslikely to produce activity on the part of a

ministry concerned with related fields. In theUnited States and in Britain, for example, therecent increase in activity in preschool educa-tion has brought about a growing awareness ofthe problems of child poverty and child health,and has stimulated rather than inhibited thedevelopment of health and welfare services.Scientists who investigate important socialquestions and who demonstrate that steps canbe taken to solve them can be most effectiveagents of social change.

REFERENCES

I. BIRCH, H. G., and J. D. Gussow. Disadvantagedchildren: health, nutrition and school failure. New York,Harcourt, Brace and World and Grune and Stratton, 1970.

2. GRUENBRG, E. M. Epidemiology. In: H. A.Stevens and Rick Heber (eds.), Mental retardation: a

review of research, Chicago, University of Chicago Press,1964, p. 260.

3. SCRIMSHAW, N. S., and J. E. GORDON (eds.). Mal-nuirition, !earning, and behavior. Proceedings of aninternational conference cosponsored by The NutritionFoundation. Inc., and The Massachusetts Institute ofTechnology held at Cambridge. Mass., March 1-3,1967. Cambridge, Massachusetts, Press, 1968.

4. SOCIAL SCIENCE RESEARCH COUNCII . The utiliza-

145

don of social science researe.i. Social Science Research

Council News /suer, September 1971, pp. 7-10.5. THONISON, A. M. Historical perspectives of nutri-

tion, reproduction, and growth. In: N. S. Scrimshawand J. E. Gordon (eds.), op. cit., p. 25.

6. WATurtt.ow, J. C., et al. Report on nutrition in

Jamaica. Prepared by the Nutrition Advisory Com-mittee of the Jamaica Scientific Research Council.Unpublished, 1970.

146

SUMMARY OF SESSION VI

Robert Cook'

Dr. Tizard began the session by outliningfour obstacles or difficulties in respect of thecontributions that scientists can make to rec-ommendations for practical action arising fromtheir research findings: (1) traces exist still ofan earlier attitude that science was a thing ofpurity and should not concern itself too muchas to the consequences of its discoveries: (2)some scientists are diffident, feeling that thesocial and administrative actions in such fieldsas nutrition and education are of such enor-mous complexity that they cannot contributeany useful- opinions at all; (3) other scientistsand this applies especially to social scientiststhink that a total political and social revolu-tion is required in developing countries andthat any attempt to ameliorate the social con-dition of the people without revolution is

merely tinkering with the problems, and finally(4), some scientists become, with inadequatebackground knowledge, hobby horse-ridingpurveyors of universal panaceas. Only thesecond obstacle was significant in the subse-quent discussion.

Dr. Tizard's conclusion was that the scien-tists' most useful contributions are provisionof information that influences professional andpublic opinion (the general reviews of topicsare most valuable in this respect), and theevaluation of both the baseline situation andthe results obtained by projects such as thosein nutrition. This evaluation of the effective-ness of services and of alternative methods ofproviding them was most important and shouldbe undertaken much more than now. Theprojects set up by scientists in their researchcould sometimes also be of use in providing

' Caribbean Food and Nutrition Institute, Universityof the West Indies, Mona, Jamaica.

pilot projects for wider adoption and applica-tion. Dr. Tizard warned, however, that weought never to allow the lack of field trials toobstruct the provision of services. This, onemight add, must certainly be true, for in Britainand elsewhere there might have been great delayin the establishment of child welfare clinics,a revolutionary instrument of change in theearly 1900's, had the local and -national gov-ernments of the time waited for evaluationthrough field trials.

The Chairman then opened the discussion,asking what guidance could be obtained fromthe information presented during the confer-ence. He thought that three main topics arosefrom Dr. Tizard's presentation: epidemiologicresearch, provision of factual material andadvice, and the design of action programs.

The discussion did indeed follow along thoselines:

Epidemiologic Research

Al) agreed that the diagnosis of the situationhad to be much wider than just a health diag-nosis. The need for all kinds of informationinput from agriculture, education, trade andindustry, finance, and other fields was wellrecognized.

The opinions of members of the communityitself were most important in respect of origi-nal diagnosis, but as Miss Fox pointed out, themore one probed for these opinions, the moreone realized what a lot of misunderstandingthere was between scientists and, say, the

mother of an infant.The point was raised that it was in the very

couniries where there was the most malnutri-tion that laboratory facilities to handle investi-gations were scarcest. One way round this

147

difficulty, however, would be agreements be-tween local institutions and a university depart-ment in an industrialized country.

Provision of Factual Material and Advice

The common interpretation by the profes-sions of the gist of the proceedings of the 1967M.I.T. conference on malnutrition, learning,and behavior was that a child who was severelymalnourished in early childhood might face aserious danger of losing inherited mental poten-tial. This was a fair interpretation and madequite an impact on policy makers all the wayup to Robert S. McNamara and his WorldBank group. What has this meeting to addto that? In short, it was this:

1. There has been a consistent demonstrationat the animal-experiment level that various de-grees of malnutrition applied at critical periodsresulted in permanent alterationboth deficitand distortionin brain anatomy, physiology,and biochemistry.

2. At the human level, where there wereother significant social disadvantages concur-rent with malnutrition, the consequences tothe disadvantaged were more severean im-pairment both in mental ability and in be-

havior. The precise mode and degree of inter-action of malnutrition with other social

disadvantages is not known, but this is noreason to lessen our concern about the impor-tance of improving nutrition in early child-hood. Our index of suspicion, or index ofinformed concern, about the effects of protein-calorie malnutrition in this respect should beat least as great as our index of concern aboutbirth injuries.

3. It was emphasized again that the effecton behavioral functions could be of majorimportance.

Design of Action Programs

Dr. Picou asked what means of rehabilitationwe proposed for those children who pass

through a period of severe protein-calorie mal-

nutrition? In Jamaica they number about 3

per cent of the child population, and it is

known for certain that in other countries evenpoint prevalence surpasses t!'is and that theincidence during the first two or three yearsof life is therefore much higher.

There was one aspect of development in

Jamaica that created hope. Public demand for"basic schools" (nursery schools and kinder-gartens) was very strong, and the groups pro-moting this movement very lively. It was

foreseeable that children aged four throughsix would receive more educational stimulation.Care should also be taken to meet the chil-dren's nutritional needs as much as possible

through lunch at school. The program of Drs.Leonardo Sinisterra, and Harrison and ArleneMcKay in Cali, Colombia, is an excellent ex-ample in this respect, and Chile hopes to followthis pattern on a national scale. That stillleaves children under four, who now can bereached only by the maternal and child healthservices. The findings of this conferenceunderline the tremendous need in all developingcountries for an increase in the coverage andand improvement in the quality of maternaland child health services. Dr. Jelliffe thoughtthat attempts to preserve breast-feeding in

particular cultures deserved endorsement in thecontext of this conference.

1)r. Richardson said that there were manyIdssons that could be learned about nutritionaland educational intervention programs, whetherin a community or institutional setting. HeadStart was an example. Another body of knowl-edge that tended to be neglected in the designof action programs was that provided by socialanthropologists studying the reasons for failureof programs that had at first seemed to be veryreasonably conceived.

Miss Fox concluded with the observationthat in evaluating action programs to improvedesign, we should make a practical distinctionbetween programs already existing and pro-grams set up for the specific purpose of opera-tional research.

148

FILMED FROM BEST AVAILABLE COPY

Recent Scientific Publications of PAHO in English

PriceUS $

No. 24t Manual of Norms and Procedures for Cervical Cancer Control.1972 (64 pp.) 1.50

No. 247 Reported Cases of Notifiable Diseases in the Americas, 19691072 (66 pp.) 0.50

No. 243 Workshop Symposium on Venezuelan Equine EncephalitisVirus. 1972 (430 pp.) 5.00

No. 240 Guidelines for Food Fortification in Latin America and theCaribbean AreaReport of a Technical Group Meeting.1972 (56 pp.) I 00

No. 239 Symposium on Systems Analysis Applied to Health Services1972 (60 pp.) 1.50

No. 238 Symposium on Vector Control and Recrudescence of Vector-Borne Diseases. 1972 (92 pp.) 2.00

No. 237 The National Food and Nutrition Survey of Barbados. 1972(148 pp.) 2.50

No. 236 IV Inter-American Meeting on Foot-and-Mouth Disease andZoonoses Control. 1972 ( I 78 pp.) 2.00

No. 235 11 International Symposium on Health Aspects of the Inter-national Movement of Animals. 1972 (248 Pp.) 2.50

No. 234 National Health' Systems. 1972 (48 pp.) 2.00No. 231 Environment, Ecology, and Epidemiology, by Reuel A

Stallones (Fourth PAHO/WHO Lecture on the BiomedicalSciences). 1971 (16 pp.) Gratis

No. 228 Guide for the Reports on tiie Aeries aegypti EradicationCampaign in the Americas. 1 971 (16 pp.) Gratis

No.No.

127226

Facts on Health Progress. 1971 (70 pp.)International Conference on the Application of Vaccines

against Viral, Rickettsial. and Bacterial Diseases of Man.1971 (686 pp.)

1.00

7.50No. 725 Seminar on Drug Control in the Americas. 1971 (165 pp.) 2.00-No. 721 Metabolic Adaptation and Nutrition. 1971 (I 50 pp.) 2.50No. 220 Venereal Diseases as a National and International Health

ProblemTechnical Discussions of the XVIII Pan AmericanSanitary Conference. 1971 (97 pp.) 2.00

No. 219 Descriptive Study Methods in Nursing, by Phyllis J. Verhonick.1971- (90 pp.) 2.00

No. 218 HI Inter-American Meeting on Foot-and-Mouth Disease andZoonoses Control. 1971 (161 pp.) 2.00

No. 217 Guidelines to Young Child Feeding in the ContemporaryCaribbean. 1970 (22 pp.) 0.50

No. 216 Health Agenda for the Americas, by Dr. Abraham Horwitz.1970 (26 pp..) Gratis

No. 211 Regional Advisory Committee On Computers in HealthFirstMeeting. 1970 (.33 pp.) 0.75

No. 208 Financing of the Health Sector--Technical Discussions of theXIX Meeting of the Directing Council of PAHO. 1970 (34PP.) 1.25

No. 207 Health Conditions in the Americas, 1965-1968. 1970 (196 pp.) 1.50No. 205 International Symposium on Mycoses: 1970 (282 pp.) 3.50No. 204 Maternal Nutrition and Family Planning in the Americas

Report of a PAHO Technical Group Meeting. 1970 (53 pp.) . 1.00No. 201 Study Group on the Coordination of Medical Car.e. 1910 (94

'PP-) 2.00

document resume ed 083 333 ud 013 853 - eric · document resume ed 083 333 ud 013 853 title...

Documents