Effects of early childhood supplementation with and

without stimulation on later development in stunted

Jamaican children13
Sally M Grantham-McGregor, Susan P Walker, Susan M Chang, and Christine A Powell

Downloaded from https://academic.oup.com/ajcn/article/66/2/247/4655658 by guest on 01 September 2020

ABSTRACT It is not known whether nutritional supplemen- In the 1960s, several studies of preventive nutritional sup-
tation in early childhood has long-term benefits on stunted chil- plementation were carried out in which high-risk pregnant
dren’s mental development. We followed up 127 7-8-y old chil- women (10) and their offspring were supplemented (1 1-13).
dren who had been stunted in early childhood and received Although the studies had many design problems ( I 4), they alt
supplementation, stimulation, or both. At 9-24 mo of age, the showed small concurrent benefits in children’s psychomotor
children had been randomly assigned to four treatment groups: development. In Bogota (10), the effects of stimulation were
nutritional supplementation, stimulation, both treatments, and con- also studied. Stimulation and supplementation independently
trol. After 2 y, supplementation and stimulation had independent benefited the children’s development and there was some ev-
benefits on the children’s development and the effects were addi-
idence of an interaction between the treatments with supple-
tive. The group receiving both treatments caught up to a matched
mentation benefiting most children who were also stimulated
group of 32 nonstunted children. Four years after the end of the 2-y
(15).
intervention 97% of the children were given a battery of cognitive
Fewer investigators have examined the effects of providing
function, school achievement, and fine motor tests. An additional
supplementation to children who were already undernourished.
52 nonstunted children were included. Factor analyses of the test
None, except the present Jamaican study, have focused on
scores produced three factors: general cognitive, perceptual-motor,
stunted children. In Indonesia (16), 4 mo of nutritional supple-
and memory. One, the perceptual-motor factor, showed a signifi-
cant benefit from stimulation, and supplementation benefited only mentation produced benefits to undernourished children’s mo-

those children whose mothers had higher verbal intelligence quo- tor but not mental development. In Cali, Colombia (17), un-
tients. However, each intervention group had higher scores than dernourished children did not benefit from supplementation
the control subjects on more tests than would be expected by alone but showed benefits from supplementation and education
chance (supplemented and both groups on 14 of 15 tests, P = combined. No group received stimulation only.
0.002; stimulated group in 13 of 15 tests, P 0.01), suggesting a In four of the above studies the children were reported to
very small global benefit. There was no longer an additive effect show small cognitive benefits several years after the interven-
of combined treatments at the end of the intervention. The stunted tions had ceased (CM Super, MG Herrera, JO Mora, unpub-
control group had significantly lower scores than the nonstunted lished observations, 1991 ; 1 8-20). However, two of the reports
children on most tests. Stunted children’s heights and head cir- gave little detail (CM Super, MG Herrera, JO Mora, unpub-
cumferences on enrollment significantly predicted intelligence lished observations, 1991 ; 18, 19). In the study in which only
quotient at follow-up. Am J Gun Nutr l997;66:247-53. the mothers received supplementation, no long-term benefits
were found (21). There is therefore only limited information on
KEY WORDS Nutritional supplementation, stunting, cog- the long-term effects of supplementation in pregnancy and
nition, school achievement, stimulation, intelligence quotient early childhood on cognitive development and no data on the
long-term effects of supplementing stunted children. Consid-
ering the large number of children who are stunted, this is
INTRODUCTION obviously an important public health question.

Low height-for-age (stunting) is estimated to effect 40%

of young children in developing countries (1, 2). It is therefore
I From the Tropical Metabolism Research Unit, University of the West
a serious concern that stunting is associated with poor mental
Indies, Kingston, Jamaica.
development and school achievement (3-8). It is unclear to 2 Supported by the Ford Foundation, United States, and Nutncia Re-
what extent inadequate nutrition contributes to these children’s
search Foundation, Netherlands.
poor development. Stunting usually coincides with many psy-
3 Address reprint requests to SM Grantham-McGregor. Centre for In-
chosocial and economic deprivations (9), which themselves ternational Child Health, Institute of Child Health, 30 Guilford Street.
may detrimentally affect children’s development. Nutritional London WC1N 1EH. E-mail: s.mcgregor@ich.ucl.ac.uk.
supplementation studies can help to separate the possible role Received July 9, 1996.
of nutrition from that of deprivation in general. Accepted for publication April 15, 1997.

Am J C/in Nutr l997;66:247-53. Printed in USA. tO 1997 American Society for Clinical Nutrition 247
248 GRANTHAM-MCGREGOR ET AL

The Jamaicah supplementation study lengths > - 1 SD of the NCHS reference values. This made a
total of 84 nonstunted children. The 52 added nonstunted
In 1986 we began a study of the effects of a 2-y intervention
children were similar to the original 32 in all measures of social
program of nutritional supplementation with or without psy-
background except they had been enrolled in school longer and
chosocial stimulation on the psychomotor development of
had fewer free school books. Both of these factors were con-
stunted children (22). The children were also compared with a
trolled for in the analyses. They were slightly older (mean age
matched nonstunted group. Supplementation benefited the chil-
7.9 y compared with 7.6 y in the original group, P < 0.01) and
dren’s growth whereas stimulation did not (23). Initially, the
had slightly lower heights-for-age (mean Z score of 0.68 corn-
stunted groups’ developmental quotients were lower than those
pared with 1.06, P < 0.05).
of the nonstunted group and the nonintervention group declined
further during the study. Supplementation and stimulation ben-
Measurements
efited the children’s developmental quotients (7 and 8 devel-
opmental quotient points, respectively). The effects of corn- All children were transported to the Tropical Metabolism
bined treatments were additive and the group receiving both Research Unit on two occasions, where a battery of cognitive

Downloaded from https://academic.oup.com/ajcn/article/66/2/247/4655658 by guest on 01 September 2020

was the only one to catch up to the nonstunted group. These and educational achievement tests were given by one of two
findings suggest that undernutrition contributes to at least part testers who were unaware of the children’s group assignment.
of the deficit in development shown by stunted children. It is Before the study began the tests were given to a small group of
critical to determine whether the benefits from supplementation nonstudy children on two occasions 1 wk apart. The test-
persist. retest correlations are given in Table 1. The battery of tests
In the present study, we reexamined the children when they were as follows:
were 7-8 y old, ‘‘4 y after the intervention ceased. The aims 1) The Wide Range Achievement Test (WRAT; 26) was
were to determine I) whether any benefits remained from the used to measure the children’s school achievement. This
interventions to their cognitive function, school achievement, test has been used frequently in Jamaica and has had
and motor skills, and 2) whether the stunted children remained minor modifications (27). It is reliable and valid in this
at a disadvantage compared with nonstunted ones. population. The test has three sections: reading, spell-
ing, and arithmetic.
2) The Stanford Binet Test was used to measure the chil-
SUBJECTS AND METHODS dren’s intelligence quotients. This test has also been
used before in Jamaica and the children’s scores cone-
Sample late highly with their future scores on the Weschler
One hundred twenty-nine stunted children [height-for-age < Intelligence Scale for Children and the WRAT (27).
-2 SD of the National Center for Health Statistics (NCHS) 3) The Peabody Picture Vocabulary Test (PPVT; 28) was
reference values (24)] aged 9-24 mo were identified from a used to measure the children’s language comprehen-
house-to-house survey of poor Kingston neighborhoods. They sion. The PPVT has been used frequently in Jamaica
were randomly assigned to four groups: control, supplemented, and is reliable over time and correlates with the Stanford
stimulated, or both treatments. A fifth group of 32 nonstunted Binet and WRAT scores (27).
4) Raven’s Progressive Matrices (29) were used to mea-
children (heights > - I SD of the NCHS reference values)
sure the children’s visual reasoning ability.
matched for age ( ± 3 mo), sex, and neighborhood was also
5) Categorical fluency refers to an individual’s capacity to
studied. Other selection criteria were type of housing, maternal
generate items from a specified semantic category (30).
education below a certain standard, singleton birth, birth
The children were asked to name as many items as
weight > 1 .8 kg, and no reported physical or mental handicaps.
The supplement was a 1-kg milk-based formula that was de-
livered to the homes weekly. The stimulation was similar to TABLE 1
programs we used previously (25) and comprised weekly home Test-retest correlations of intelligence quotient, cognitive function, and
visits during which the visitor demonstrated play with home- school achievement tests over 2 wk

made toys. All groups were visited weekly. The control sub-
Test r
jects were visited in an attempt to control for any benefits that
supplemented children may have received from extra attention. Stanford Binet (n = 19) 0.96
One hundred twenty-two of the 127 stunted children who Peabody Picture Vocabulary Test (n = 14) 0.83

had participated in the study until the end of the intervention Ravens Progressive Matrices (n 17) 0.70
Free recall (n = 13) 0.80
were located. Of the remaining five (two from the group
Fluency (n 19) 0.87
receiving both treatments and one each from the other three
French learning test (n = 14) 0.82
groups), three had migrated and two could not be located. All
Verbal analogies (n 19) 0.75
of the 32 nonstunted children were located. There was no Digit span (n = 19) 0.70
significant difference between the original samples and the Corsi blocks (n 19) 0.83
samples followed up in any measure examined, including mi- Pegboard
tial age, height-for-age, and developmental quotient. To in- Dominant hand (F’ 17) 0.90
crease the number of nonstunted children to facilitate compar- Nondominant hand (n = 17) 0.84
isons between those stunted and nonstunted, an additional 52 Wide range achievement test

children were added. They had been identified in the original Reading (n 18) 0.99
Arithmetic (n = 18) 0.93
survey when they were between 9 and 24 mo of age, lived in
Spelling (n = 18) 0.95
the same neighborhoods as the study children, and had
 SUPPLEMENTATION IN STUNTED CHILDREN 249

possible in 1 mm from three sets: things to eat, animals, Anthropometry

and things to wear.
Weight, height, head and midupper arm circumferences, and
6) Verbal analogies consisted of a set of 1 1 analogicai
triceps and subscapular skinfold thicknesses were measured by
reasoning problems that were read to the child for him
using standard techniques. These will be reported in another
or her to complete.
paper.
7) Free recall measures long-term semantic memory and
comprises a list of 15 words arranged in random order.
Statistical analysis
The list was read to the child and he or she had to recall
as many words as possible within 1 mm. To reduce the number of statistical tests performed and to
8) The French learning test is a paired-associate learning identify the underlying constructs, a factor analysis of the
task in which the child is presented with a sheet con- developmental measures was conducted. Principal-components
taming drawings of six objects whose French names analysis was used with vanimax rotation. A similar factor
were to be learned (3 1). The tester named each object in analysis was also conducted with the variables describing the
French while indicating the relevant picture. The child children’s stimulation at home.

Downloaded from https://academic.oup.com/ajcn/article/66/2/247/4655658 by guest on 01 September 2020

was then asked to identify the pictures when they were Multiple regression analyses were used to determine the
named. effects of the interventions. In each analysis, age and sex were
9) The digit span measures auditory working memory. The entered and then the social background variables that were
child was asked to repeat strings of numbers of increas- significantly associated with one or more test scores in prelim-
ing length.
mary analyses were offered stepwise (mother’s education and
10) Corsi blocks, a set of nine black blocks arranged in a PPVT score, the stimulation factors, the number of free school
quasi-random array, measure visual-spatial working
books, possessions, time in school, and lack of money for
memory (32). The child had to touch the blocks in the
school attendance). After this the interventions were entered
same order as the tester touched them. An increasing
(supplementation coded O/l, stimulation coded 0/1). In a final
number of blocks were indicated.
step, interactions between the two interventions and between
1 1) The Lafayette Grooved Pegboard Test was used to
the interventions and the children’s socioeconomic status and
measure fine motor coordination (33).
their caretaker’s PPVT score were entered stepwise. We had
originally hypothesized that an interaction existed between the
Questionnaire two treatments and the interaction terms between social back-
Mothers were interviewed at home to obtain details of ma- ground variables and the interventions were entered because
ternal characteristics, standard of housing, and household pos- significant ones were reported in a supplementation study in
sessions. In addition, 13 questions were asked about the Guatemala (20).
amount of stimulation in the home. Factor analysis was con- To compare the stunted children with the nonstunted chit-
ducted on the responses from all groups combined by using a dren, a similar model was used. Multiple regression of each
principal components analysis with varimax rotation and four cognitive factor was carried out by using data from all the
factors were extracted, which are shown in Table 2. These stunted and nonstunted groups, entering dummy variables for
were factor 1 , books and paper; factor 2, games and trips; factor each of the four stunted groups instead of the stimulation and
3, verbal stimulation; and factor 4, writing materials. supplementation variables. The program SPSSx was used for
statistical analyses (SPSS Inc. Chicago).

RESULTS
TABLE 2
Results of factor analysis of the home stimulation items The characteristics of the children, caretakers, and homes are
shown in Table 3. There were no significant differences among
Factor and item Loading igen the four stunted groups in any of the variables. When compared
value
with the stunted groups combined, the nonstunted group was
Factor 1 (books and paper) significantly taller, had been enrolled in school longer, and had
Number of school books purchased 0.68 been absent from school less often because of insufficient
Number of exercise books 0.66 2.47 money. Their homes were also of a better standard and had
Paper available at home 0.65 more household possessions. The nonstunted group also had
Number of books at home (not school books) 0.49
significantly higher scores (P < 0.001) for home stimulation
Factor 2 (games and trips)
factor 1 (books and paper) than the stunted groups combined,
Adult plays outdoor games with child 0.71
which were not different among themselves. There were no
Number of trips in past year 0.65 1.73
Adult plays indoor games with child 0.53
group differences in the other home stimulation factors.
Number of toys and games child has 0.41 The mean scores of the intelligence quotient, cognitive func-
Factor 3 (verbal stimulation) tion, pegboard, and school achievement tests are shown in
Adult reads to child 0.77 1.16 Table 4. In all tests the higher score is the better, except for the
Adult teaches child school work 0.68 pegboard, which is scored by time, so a lower score is better.
Adult tells stones to child 0.44 Factor analysis of all the test scores from all the groups pooled
Factor 4 (writing material) yielded three factors (Table 5), with most of the cognitive and
Number of crayons for use at home 0.74
the educational tests loading on factor 1 (general cognitive
Number of pencils or pens for use at home 0.7 1 1.05
factor). The pegboard and the Corsi blocks test scores toaded
250 GRANTHAM-MCGREGOR ET AL

TABLE 3
Characteristics of the child, caretaker, and home at follow-up

Control Stimulated Supplemented Both Nonstunted

(n l9M,l3F) (n 16M,13F) (n 17M,13F) (n l8M,l3F) (n45M,39F)

Child
Age (y) 7.7 ± 0.4’ 7.7 ± 0.4 7.7 ± 0.3 7.7 ± 0.3 7.9 ± 0.42
Height-for-age (Z score) - 1 . 1 ± 0.8 - 1.0 ± 1.0 -0.9 ± 0.9 - I .2 ± 0.8 0.8 ± 0.8
Weight-for-height (Z score) -0.7 ± 0.7 -0.6 ± 0.7 -0.7 ± 0.8 - I .0 ± 0.8 -0.5 ± 0.7
Terms at school 3.8 ± 2.0 4.6 ± 1.5 3.7 ± 1.9 3.8 ± 1.8 5.0 ± l.8
Free school books 2.1 ± 2.0 2.7 ± 2.5 2.0 ± 2.2 2.1 ± 1.8 2.5 ± 2.0
Missed school because of no money (%)
Never 43.7 67.9 32.3 58.6 68.3
1 wk 34.4 25.0 38.7 27.6 25.6

Downloaded from https://academic.oup.com/ajcn/article/66/2/247/4655658 by guest on 01 September 2020

> I wk 21.9 7.1 29.0 13.8 6.1
Home
Housing rating 8.6 ± 1.9 8.1 ± 1.7 8.2 ± 1.8 8.4 ± 1.8 9.0 ± l.7
Possessions 2.5 ± 1.4 2.4 ± 1.3 2.3 ± 1.3 2.5 ± 1.3 3.1 ± 1.l
Mother
Age (y) 29.9 ± 8.2 32.0 ± 8.1 34.5 ± 9.2 32.5 ± 7.8 33.5 ± 7.5
Peabody Picture Vocabulary Test 83.4 ± 19.5 86.7 ± 23.0 87.6 ± 19.0 88.8 ± 22.2 92.6 ± 25.4
Took no secondary school exams (%) 7 1.9 57. 1 80.6 75.9 68.3
Unemployed (%) 68.7 42.9 54.8 55.2 36.66

‘1 ± SD.
2.3.5 Significantly different from the pooled stunted group (Student’s a’ test): 2 p < 0.01, p < 0.001, p < 0.05.
4,6 Significantly different from the pooled stunted group (chi-square test): P < 0.05, 6 p < 0.01.

on factor 2 (perceptual motor), the fluency and free recall tests There was no significant main effect of supplementation in
loaded on factor 3 (long-term semantic memory). any regression. The children who had received stimulation had
significantly higher scores in factor 2 (perceptual-motor), in
Intervention effects addition, the interaction term between mother’s PPVT score

To determine whether there were any remaining benefits and supplementation was significant in this regression (Table
from the interventions, a series of multiple-regression analyses 6). The children whose mothers had higher PPVT scores ben-
of each of the three cognitive factors were conducted. Age and efited the most.
sex were entered first, then the social background variables Although there was a significant effect of intervention on
were offered stepwise; after this the two interventions were only one factor, Table 4 shows that the control group’s scores
entered. Finally, interaction terms between the treatments, on the tests were lower than the intervention group’s on nearly
stimulation and supplementation, and between each interven- all tests. We therefore conducted sign tests comparing each of
tion and possessions, and the mother’s PPVT score were the intervention groups with the control. This takes into ac-
entered. count the direction of the difference between the groups (higher

TABLE 4
Mean scores of the cognitive fine motor and educational tests in the four stunted groups and nonstunted group’

Control Stimulated Supplemented Both Nonstunted

Test (n 32) (n 29) (n = 31) (n = 30) (n 84)

Stanford Binet 73.3 ± 8.2 76.9 ± 9.2 76.4 ± 9.8 76.4 ± 8.9 81.9 ± 9.2
Analogies 5.7 ± 1.5 6.0 ± 1.2 5.8 ± 1.4 6.1 ± 1.3 7.1 ± 1.4
Pegboard
Dominant hand I 12.1 ± 28.4 97.1 ± 20.4 109.2 ± 41.8 100.6 ± 26.9 88.7 ± 13.3
Nondominant hand 159.1 ± 68.4 123.8 ± 35.3 152.4 ± 87.9 142.2 ± 49.0 120.1 ± 47.8
Peabody Picture Vocabulary Test 34.7 ± I 1.0 42.2 ± 13.0 40.0 ± I 1.5 38.5 ± 13.8 50.5 ± 13.4
Ravens Progressive Matrices 1 1.8 ± 2.5 12.6 ± 2.9 12.1 ± 3.4 13.0 ± 3.0 13.6 ± 3.3
fluency 27.8 ± 5.5 27.9 ± 6.5 27.6 ± 6.3 28.1 ± 8.0 30.5 ± 7.6
French learning 23.3 ± 5.7 22.9 ± 4.7 25.1 ± 7.0 23.6 ± 5.3 26.0 ± 6.6
Digit span 6.0 ± 1.3 6.3 ± 1.6 6.1 ± 1.6 6.2 ± 1.8 6.6 ± 1.5
Corsi blocks 5.7 ± 2.4 5.9 ± 2.2 5.9 ± 2.1 6.2 ± 2.2 7.2 ± 2.1
Free recall 27.3 ± 5.8 26.9 ± 5.4 28.7 ± 7.4 27.0 ± 4.8 28.4 ± 5.3
Reading 17.2 ± 6.9 18.8 ± 9.4 18.8 ± 9.5 20.0 ± 5.1 27.4 ± 13.5
Spelling 10.2 ± 3.4 1 1.3 ± 5.6 10.7 ± 5.3 1 1.0 ± 3.1 16.1 ± 8.4
Arithmetic 14.5 ± 5.0 15.5 ± 5.7 14.8 ± 5.6 15.7 ± 5.2 19.1 ± 4.9

‘ I ± SD.
 SUPPLEMENTATION IN STUNTED CHILDREN 251

TABLE 5 TABLE 7

Factor analyses of all the cognitive function. intelligence quotient. and Regression coefficients and SEs from the multiple regressions of the
school achievement tests for all groups pooled general cognitive and the perceptual-motor factors. comparing the 6tir
stunted groups with the nonstunted group
Factor and item Loading Eigen value
Dependent variables
Factor I : general cognitive function
Reading 0.78 6.2 Gen eral cognitive Perceptual-motor
Spelling 0.78 factor’ facur
Stanford Binet 0.74
Independent variables B SE B SE
Verbal analogies 0.69
Arithmetic 0.67 Age 0.39 0.162 St) 0.212
Peabody Picture Vocabulary Test 0.64 Sex 0.15 0.12 -0.42 0.13’
Digit span 0.61 Mothers PPVT 0.007 0.(X)32
French learning 0.58 Stiiulation
Stiniulation
factor I
2
0.36
0.15
0.07 -

-
-

Downloaded from https://academic.oup.com/ajcn/article/66/2/247/4655658 by guest on 01 September 2020

Ravens Progressive Matrices 0.56 factor 0.()6
Factor 2: perceptual-motor factor School books 0.06 0.032

Pegboard
Dominant
Nondominant
hand
hand
-0.92
-0.89
.4
Control
Stinulated
Both
-0.38
-

-0.32
0.32
0.182
0. I 9
0. I9
-
0.66

-
0.05
0.2 1
O.l9
0.20
0.20
Corsi blocks 0.42 Supplemented -0.28 0.19 0.41 0.20

Factor 3: long-term semantic memory R 0.30 0.19

Free recall 0.86 1.1
I Other significant covariates: mother’s PPVT. stimulation factors I and
Fluency 0.58
2. and free school books.
2 p < 0.05.
‘P < 0.01.
or lower) on each test but does not address the magnitude of the 4P < 0.0()l.
difference. The supplemented group and the group that re-
ceived both treatments had better scores than the control sub-
group receiving both treatments. There were no group differ-
jects on significantly more tests than would be expected by
ences in the tong-term memory factor.
chance ( I4 of I 5, P < 0.01 in both cases). The stimulated
group did better than the control subjects in I 3 of I 5 tests (P <
Early measures and later intelligence
0.05).
Finally. we examined the relations between anthropometric
Comparison with nonstunted children measures taken at enrollment, at the end of intervention, and at
follow-up and the Stanford Binet intelligence quotient scores in
To compare the four stunted groups and the nonstunted
the stunted children only. Partial correlations coefficients were
group the multiple regressions of the cognitive factors were
calculated between intelligence quotient and each height.
repeated by using the five groups coded as four dummy van-
weight, and head circumference measure while controlling for
ables, instead of the stimulation and supplementation variables.
age at enrollment and at the time of follow-up, sex, and the
All other variables remained the same as in previous regres-
interventions. Enrollment height was significantly associated
sions. The nonstunted group had significantly higher scores
with the children’s intelligence quotient scores (r = 0.2, P <
than the control group on the general cognitive factor, but there
0.05) whereas height after intervention and at follow-up were
were no other group differences in this factor (Table 7). The
not. There were significant correlations between head circum-
nonstunted group also had significantly higher scores on the
ference and intelligence quotient on all three occasions (enroll-
perceptual motor factor than both the control and supplemented
ment: r = 0.28, P < 0.005; end of intervention: r = 0.25. P <
groups, but was not different from the stimulation group or the
0.01; currently: r = 0.2, P < 0.05). There was flO relation
between weight-for-height and Stanford Binet scores.
TABLE 6 To determine the relative importance of the early and COfl-

Regression coefficients and SEs from the multiple regression of the current anthropometric measures, a multiple regression was
perceptual-motor factor in ie stunted groups’ calculated predicting the Stanford Binet intelligence quotient.
Enrollment and final age, sex. supplementation. and stimula-
Perceptual-i notor factor
tion were entered in the first step. all the socioecononhic status
Independent variables B SE variables used previously were then offered stepwise in the
second step. then the anthropometnic measurements height.
Sex -0.46 0.182
Age 1.13 0.26 head circumference, and weight-for-height. on enrollment, at
Mothers PPVT 0.01 0.0042 the end of intervention, and concurrently were offered step-
Supplementation -0.02 0.18 wise. Enrollment head circumference made an independent
Stimulation 0.41 0.182 contribution to the variance. Once this had entered the equation
PPVT x supplementation 0.02 0.0092 no other early or concurrent anthropornetric measurement was
R2 0.22 significant. To determine whether head circumference at en-
‘ PPVT. Peabody Picture Vocabulary Test.
rollment was a proxy for size at birth, we repeated the ‘regres-
2 p < 0.05. sion entering birth weight as an extra variable in the llrst step.
3P < 0.01. Birth weight did not make a significant contribution to the
252 GRANTHAM-MCGREGOR ET AL

variance, and head circumference at enrollment remained sig- indicate that benefits resulting from early childhood interven-
niticant (Table 8). tions vary with age. It is therefore possible that the children in
the present study will follow the same pattern and greater
benefits from the interventions may become apparent during
DISCUSSION
adolescence. However. it is probable that when children are
Intervention effects living in extremely poor circumstances early childhood inter-
ventions should be continued at least through to school age, if
Four years after the end of intervention, the children who
not longer, if early substantial gains are to be maintained.
received supplementation, stimulation, or both showed a sig-
nificant benefit only in the perceptual-motor factor. Stimula- The effect of stunting
tion had a significant main effect on this factor whereas sup-
plementation only benefited children whose mothers had The nonstunted children came from the same neighborhoods
as the stunted children; however, small but significant differ-
higher PPVT scores. However. there was an extremely small
ences were found in the quality of their home backgrounds.
but wide-ranging effect, as shown by the fact that children who

Downloaded from https://academic.oup.com/ajcn/article/66/2/247/4655658 by guest on 01 September 2020

After taking these differences into account, the nonstunted
received any of the interventions had higher scores than the
children had higher scores than the control stunted children in
control subjects on almost all tests. This is probably the first
both the general cognitive and the perceptual-motor factors.
demonstration of benefits from supplementing stunted children
These factors included all the educational and intelligence tests
in early childhood that lasted for 4 y. We showed long-term
benefits from a stimulation program in severely malnourished and most of the cognitive tests as well as a measure of fine
motor coordination. Early height measures were more pnedic-
children previously (25). The children in the study in Cali,
tive of later intelligence quotient than current height. Stunting
Colombia ( I 8). showed benefits 3 y after intervention but the
in early childhood was therefore associated with an extensive
effects of supplementation could not be separated from those of
stimulation because there was no group that received stimuta- range of deficits in cognitive function and educational attain-
ment. This indicates that early childhood stunting is a serious
tion alone. In the present study, the interaction between sup-
risk factor for poor development in later childhood.
plementation and the mother’s PPVT score may be analogous
A considerable amount of research has been conducted into
to the interaction found in the follow-up of the Guatemalan
study (20). In that study, children who reached higher grades in the possibility of long-term programming of cardiovascular
function and glucose and lipid metabolism originating from
school benefited from early childhood supplementation
whereas those from lower grades did not. poor nutrition in utero (35). In the present study. birth weight
was not related to cognitive development. However, the pre-
At the end of the intervention there were marked benefits
dictive power of height and head circumference at enrollment
from both supplementation and stimulation and their effects
was impressive. It may be that undernutrition in early child-
were additive. The size of the benefits remaining 4 y later was
hood changes children’s developmental trajectory unless they
therefore less than anticipated and no suggestion remained of
an additive effect, which was disappointing.
have substantial long-term rehabilitation. U
There is evidence from preschool-intervention programs in We thank J Burrowes for assistance in testing the children.
the United States that benefits from early childhood stimulation
programs may decline in the first few years after the program
ends but reappear at adolescence (34). Further, in the Guate- REFERENCES
malan supplementation study (20), no benefits were apparent
I . Keller W. The epidemiology of stunting. In: Waterlow JC. ed. Linear
when the children were 6 y old, although at follow-up in
growth retardation in less developed countries. Nestle Nutrition Work-
adolescence wide-ranging benefits were found. These studies
shop 4. New York: Raven Press, 1988:17-40.
2. UNICEF. The state of the world’s children 1994. New York: Oxford
TABLE 8 University Press. 1994.
Regression coefficients and SEs from the multiple regression of the 3. Lasky RE. Klein RE. Yarbrough C. Engle PL. Lechtig A. Martorell R.
Stanford Binet intelligence quotient on early anthropometric measures in The relationship between physical growth and infant behavioural de-
the stunted children’ velopment in rural Guatemala. Child Dev I 981:52:220-6.
4. Powell CA. Grantham-McGregor SM. The ecology of nutritional
Independent variables B SE status and development in young children in Kingston. Jamaica. Am J
Clin Nuts 1985:41:1322-31.
Age -9.77 7.86
5. Christiansen N, Vuon L. Clement J, Herrera MG. Malnutrition, social
Initial age 0.33 0.68
environment and cognitive development of Colombia infants and
Supplementation 0.66 1.52
preschoolers. Nutr Rep Int 1977:16:93-102.
Stimulation 1.5 1.53
6. Jamison DI. Child malnutrition and school performance in China. J
Sex 1.91 1.74
042
Dcv Econ l977:20:299-3()9.
Mothers PPVT 0.14
7. Agarwal DK. Upadhyay 5K. Tripathi AM. Agarwal KN. Nutritional
Head circumference I .85 0.63’
status, physical work capacity and mental function in school children.
Stimulation factor 1 1.97 0.93
New Delhi: Nutritional Foundation of India. 1987. (Scientific report
Birth weight -0.44 1.74
no. 6.)
R2 26
8. Clarke NM. Grantham-McGregor SM. Powell C. Health and nutrition
I PPYT. Peabody Picture Vabulary Test.
predictions of school failure in Kingston, Jamaica. Ecol Food Nutr
2 p < 0.001.
1991:26:47-57.
.‘ P < 0.01. 9. Martorell R. Mendoza F. Castillo R. Poverty and stature in children.
4 P < 0.05. In: Waterlow JC. ed. Linear growth retardation in less developed
 SUPPLEMENTATION IN STUNTED CHILDREN 253

countries. Nestle Nutrition Workshop 14. New York: Raven Press. 2 1 . Hsueh AM. Meyer B. Maternal dietary supplementation and 5 year old
I 988:57-74. Stanford Binet IQ test on the offspring in Taiwan. Fed Proc 981:40:
10. Joos S. Pollitt E, Mueller W. Albright D. The Bacon Chow Study: 897 (abstr).
maternal nutritional supplementation and infant behavioral develop- 22. Grantham-McGregor SM, Powell CA. Walker SP. Himes JH. Nutri-
ment. Child Dev 1983;54:669-76. tional supplementation. psychosocial stimulation. and mental develop-
I 1 . Waber DP. Vuori-Christiansen L. Ortiz N. et al. Nutritional supple- ment of stunted children: the Jamaican study. Lancet I 99 1 :338: 1-5.
mentation, maternal education, and cognitive development of infants at 23. Walker SP, Powell CA, Grantham-McGregor SM. Himes JH. Chang
risk of malnutrition. Am J Clin Nutr 198 l;34:807-l3. SM. Nutritional supplementation. psychosocial stimulation. and the
l2. Freeman HE. Klein RE. Townsend JW, Leghtig A. Nutrition and growth of stunted children: the Jamaican study. Am J Clin Nutr
cognitive development among rural Guatemalan children. Am J Public
I 991:54:642-8.
Health I 980:70:1277-88. 24. Hamill PVV. Drizd TA, Johnson CL. Reed RB, Roche AF. Growth
13. Chavez A. Martinez C. Growing up in a developing community.
curves for children, birth-18 years. Vital Health Stat II 1978:165.
Guatemala City: INCAP. 1982.
(DHEW publication no. 78-1650.)
14. Grantham-McGregor SM. Field studies in early nutrition and later
25. Powell CA. Grantham-McGregor SM. Home visiting of varying fre-
achievement. In: Dobbing J. ed. Early nutrition and later achievement.

Downloaded from https://academic.oup.com/ajcn/article/66/2/247/4655658 by guest on 01 September 2020

quency and child development. Pediatrics l989;84: 157-65.
London: Academic Press. 1987:128-74.
26. Jastak J, Bijou S. Wide range achievement test. Wilmington. DE: CL
IS. Mora JO, Clement J, Chnstiansen N, Ortiz N, Vuori L. Wagner M.
Story Co. 1946.
Nutritional supplementation, early stimulation. and child development.
27. Grantham-McGregor SM. Powell C. Walker S. Chang S. Fletcher
ICBF-Harvard-Giessen Research Project on Malnutrition and Mental
P. The long-term follow-up of severely malnourished children who
Development. Bogota, Colombia. In: J Brozek J, ed. Behavioral effects
participated in an intervention programme. Child Dcv
of energy and protein deficits. Washington. DC: National Institutes of
I 994:65:428-39.
Health. 1979:255-69. (NIH publication no. 79-1906.)
28. Dunn LM. Peabody picture vocabulary test. Nashville: American
16. Husaini MA, Karaydi L, Husaini YK, Sandjaja, Karyadi D. Pollitt E.
Guidance Service. 1965.
Developmental effects of short-term supplementary feeding in nutri-
29. Raven JC. Court JH. Raven J. Ravens progressive matrices. London:
tionally-at-risk Indonesian infants. Am J Clin Nutr 1991:54:799-804.
17. McKay H. Sinesterra L, McKay A. Gomez H. Lloreda P. Improving
HK Lewis and Co Ltd. 1986.

cognitive ability in chronically deprived children. Science 30. Semel E. Wiig E. Clinical evaluation of language functions. Coluni-

I 978:2()0:270-8. bus, OH: Charles Mervill, 1980.

18. McKay A, McKay H. Preventing school failure: the relationship be- 3 1 . Baddeley A. Meeks Gardner J. Grantham-McGregor SM. Cross-cul-
tween preschool and primary education. Ottawa: International Devel- tural cognition: developing tests for developing countries. AppI Cognit
opment Research Center, 1983. (No. 172e 36-41.) Psychol I 995:9:173-95.
19. Chavez A. Martinez C. Soberanes B, Dominguez L, Avila H. Nutrition 32. Milner B. Interhemispheric differences in the localisation of psycho-
of Adolescent Girls Research Program research report. Series 4. Early logical processes in man. Br Med Bull 197 1:27:272-7.
nutrition and physical and mental development in Mexican rural ad- 33. Lafayette Instrument Company. Grooved pegboard test. LaLtyette. IN:
olescent females. Washington, DC: International Center for Research Lafayette Instrument Company. I 989.
on Women, 1994. 34. Myers R. The twelve who survive. London: Routledge. 1992.
20. Pollitt E, Gorman K. Engle PL, Martorell R, Rivera J. Early supplemen- 35. Barker DJB. ed. Fetal and infant origins of adult disease. London:
tary feeding and cognition. Monogr Soc Res Child Dcv 1993:58:235. British Medical Journal. 1992.