5/21/2018 J. Nutr.-2005-Watanabe-1918-25

1/8

Community and International Nutrition

Early Childhood Development Interventions and Cognitive Development ofYoung Children in Rural Vietnam1

Koichiro Watanabe,2 Rafael Flores, Junko Fujiwara,* and Lien Thi Huong Tran*

Rollins School of Public Health at Emory University, Atlanta, GA 30322 and *Save the ChildrenJapanVietnam Office, Hanoi, Vietnam

ABSTRACT Little is known about the long-term benefits of interventions that aim to promote early childhooddevelopment programs. The goal of this research was to determine whether an early childhood developmentintervention added to a nutrition intervention during preschool ages had lasting effects on the cognitive develop-ment of school-age children in communes of Thanh Hoa province in rural Vietnam. The study focused on a totalof 313 children aged 6.58.5 y (grades 1 and 2 in primary school) in 2 communes that were exposed to nutritionintervention or nutrition and early childhood development (ECD) intervention from 1999 to 2003. Measurements ofheight and cognitive test scores (Ravens Progressive Matrices Test) were collected from the children; householdcharacteristics were determined by interviews with mothers. Longitudinal analysis was performed by integratingthe data with that collected from the same children in past surveys. Significant effects of the ECD interventioncompared with the nutrition intervention were detected. The beneficial effect of ECD intervention on the cognitivetest scores was large for the most nutritionally challenged children whose height-for-age Z-scores declined orremained in the stunted range. The findings help provide useful insights into the development of an effectiveintegrated model of ECD and nutrition intervention for children in rural Vietnam. J. Nutr. 135: 1918 1925, 2005.

KEY WORDS: child development growth nutrition Vietnam

In 2004 it was estimated that 162 million preschool chil-dren, or 33% of children 5 y old in developing countries,were stunted [height-for-age Z-scores (HAZ)3 2] (1). Thenegative effect of malnutrition on child survival has been welldocumented. For example, children who are underweight have

an 2- to 8-fold higher risk of death than those who are betternourished (2). Malnutrition in early childhood impairs func-tional performance in adulthood; this impairment may bephysical as well as cognitive. Economic losses due to thephysical and cognitive losses are substantial. Malnutrition inchildhood decreases intellectual potential and productivity inadulthood. Furthermore, physical losses in adulthood also haveintergenerational effects on child health and nutrition (3).Improving the physical and cognitive development of childrenin lower-income communities will have far-reaching effectsthroughout the less-developed world. Effective program de-signs for nutrition and early childhood development need aclearer understanding of their effects on physical and cognitivedevelopment.

The negative effect of malnutrition on cognitive develop-

ment has been demonstrated in multiple locations around thworld. Some cross-sectional studies found associations between chronic malnutrition, marked by stunting, and poocognitive function among school-age children (46). Thesstudies include one in Vietnam, although it did not control fo

poverty. Other longitudinal studies showed that height-for-agduring infancy was strongly associated with cognitive performance in late childhood (7).

The positive effect of nutrition interventions on cognitivdevelopment has also been shown. Several supplementationstudies suggested the potential for greater cognitive development through improved nutrition at early ages in populationwith endemic undernutrition (814). Although not conclusive, it is generally agreed the first 3 y of life constitute themost vulnerable period, one in which nutritional status is oparticular importance (15). Several studies reported evidencof long-lasting benefits from early supplementation on cognition when the children reached school age (11,14). The benefits of supplementation during early childhood on cognitivedevelopment may not be apparent at school age, but they arpresent by adolescence through interaction between schoolinand nutritional improvement (11,16).

What is much less documented, less well understood, andthus a subject of considerable, sometimes acrimonious debateis whether adding an early childhood development (ECDintervention, defined as preschool-based activities and parenting education, to a nutrition intervention in communities inwhich growth stunting is endemic can improve cognitivperformance. Because a majority of the children served b

1 Supported by Joint Japan World Bank Graduate Scholarship Program andSave the Children Japan. Any opinion, findings, conclusion, or recommendationsexpressed in this publication are those of the author(s) and do not necessarilyreflect the views of the supporting agencies.

2 To whom correspondence should be addressed.E-mail: watanabe@savechildren.or.jp.

3Abbreviations used: ECD, early childhood development; GEE, generalizedestimating equation; GLM, generalized linear model; HAZ, height-for-age Z-score; NERP, nutrition education rehabilitation program; PDI, positive devianceinquiry; WAZ, weight-for-age Z-score.

0022-3166/05 $8.00 2005 American Society for Nutritional Sciences.Manuscript received 7 December 2004. Initial review completed 26 January 2005. Revision accepted 24 May 2005.

1918

5/21/2018 J. Nutr.-2005-Watanabe-1918-25

2/8

existing ECD programs in many of the developing countriesbelong to the age group 12 y away from entry into primaryschool, the effects of ECD interventions for this age groupshould be investigated. Hence, the objectives of the presentstudy were to assess the existence of and potential effects of anECD project during the preschool years (ages 45 y) on thecognitive development of school-age children in rural com-munes in Vietnam where a nutrition project was implemented

when the children were 03 y old.We hypothesized that the cognitive development of chil-dren participating in the ECD project would improve morethan that of nonparticipating children. Nutritional status andcognitive development of the children are often confoundedby the household environment, such as economic conditions,age, and education level of the mothers; these were taken intoaccount in our assessments.

SUBJECTS AND METHODS

Population and setting. Data from 2000, the most recent dataavailable, showed that the prevalence of stunting (HAZ 2)among children 5 y old in Vietnam was 36% (17). Thanh Hoa

province is located in the north central region, where the prevalenceof stunting among children 5 y old in 2000 was 40%, among thehighest in the country (18).

Save the Children Japan implemented a nutrition interventionusing Positive Deviance Inquiry (PDI) for 2 y from 1999 to 2000 at5 communes in Vinh Loc district, Thanh Hoa province, targeting allthe children aged 036 mo in the communes. The PDI approachidentifies successful child-caring practices of poor families who havewell-nourished children (19). The intervention communes were se-lected on the basis of 4 criteria set by the implementing organization:high prevalence of child malnutrition, poor socioeconomic condi-tions, absence of prior or current participation in the national nutri-tion program, and the existence of leaders interested in the project.The main components of the nutrition intervention included bi-monthly growth monitoring for all children and 9 sessions of a 12-dnutrition education rehabilitation program (NERP) conducted everymonth by local health volunteers in their hamlets, targeted forseverely malnourished children [weight-for-age Z-score (WAZ) 3]. Approximately half of the families participated in at least 1

NERP session, where they learned nutrition- and health-seekingbehaviors and fed the children locally available nutritious foods. Inaddition, antenatal care services, home gardening, and a savings andcredit program were promoted to help strengthen and sustain thenutritional benefits.

Of the 5 communes, 2 were followed up with an early childhooddevelopment (ECD) project for 2 y during 20022003, targetingchildren of preschool age (45 y old). The 2 communes were selectedbecause they had less access to quality preschool services. Althoughthe majority of the children aged 45 y in the 2 communes attendedpreschool at baseline, the attendance and the quality of preschoolteachers in the 2 communes were lower than those in the other 3

communes. The ECD intervention strengthened existing center-based preschooling through material support and teacher training onchild-centered teaching methods. It also supported parental behaviorsthrough 1-d training sessions for fathers and mothers separately everymonth on 10 different topics concerning child care and development.In addition, the interventions included the establishment of a smalllocal library for parents and promoted play corners in the homes ofparticipating children.

Anthropometric measurements of height and weight were col-lected for all of the participating children in 3 surveys conducted in1999, 2000, and 2001; at that time, mothers were interviewed abouttheir behavior and knowledge of child care. The present study focusedon primary school children living in the project communes who wereaged 6.58.5 y in 2004 and were exposed directly to 1 or both of the2 different interventions from 1999 to 2003.

Sampling. Two communes were selected as samples for thisresearch and represent the 2 study populations. The first communerepresents those populations exposed only to the nutrition interven-

tion; the second commune represents those exposed to both thnutrition intervention and early childhood development. The maiselection criteria for the 2 communes were similar socioeconomiconditions and ecologic characteristics. Baseline data on wealth ranking, ethnic composition, and ecologic characteristics collected fromthe district in 1998 were used to identify the sample communes. Wstudied nearly all of the children (431 of 474) listed in the samplingframe of children aged 6.58.5 y in the 2 communes. The samplinframe was established on the basis of the data collected in pas

surveys, which were updated by the most recent census recordavailable at the Commune Peoples Committees. The criteria foeligibility for the sampling frame were that the children had beenliving in the commune since 1998, were currently registered in thsame commune, and were between the ages of 6.5 to 8.5 y.

Data collection. Standing height was measured to the nearest mm with a standard technique that used the infant/child/adult heighmeasuring board (Shorr Production) (20,21). The survey team watrained in anthropometry standardization exercises conducted by senior program officer of Save the Children Japan, to avoid interobserver measurement bias.

Ravens Colored Progressive Matrices test was used to measure thcognitive performance of the children. The test measures the abilitto develop new insights and information from what is already perceived or known (22). This test was used to assess the cognitiv

ability of children in several international studies in developincountries such as Guatemala (11), Kenya, Egypt, and Mexico (5)The test generally has had high internal consistency and retestreliability (0.8); factor analytical studies showed that the test is good indicator for Spearmans g-factor (23). The standard version oRavens Progressive Matrices consists of 5 scales (AE), with 12 itemin each scale. Each item contains a figure with a missing piece, belowwhich alternative pieces are placed to complete the figure. Each seinvolves a different principle for obtaining the missing piece; withia set, the items are arranged in increasing order of difficulty. According to the instructions given by the trainer, Ravens Colored Matriceversion was administered. The Colored Matrices version, consistinof 3 scales, Test-A (12 items), Test-B (12 items), and Test-AB (1items), is designed to assess with greater precision the intellectuaprocesses of young children (22). The same approach was taken in

similar research conducted in Guatemala (11).Because the test has not been standardized locally, interpretationof the scores were made only through comparisons of the group meanscores within the study population. Field workers administered thtest one-on-one for each individual child, taking 15 min per test pechild. A trainer, affiliated with the Research and Training Center foCommunity Development in Hanoi, conducted training on the tesmethods and use of materials made available by the Center for thiresearch (24). To minimize examiners biases, children were assignerandomly to field workers, and no information was provided to thfield workers regarding interventions, nutritional status, and socioeconomic level of children. The workers were trained repeatedly oskills and attitudes against biases throughout the test administrationThe training included trial test administration, which confirmeagreement between the test results obtained by the workers and th

trainer.In addition, selected information was gathered on maternal characteristics (age, education, number of deliveries, occupation), household characteristics (number of children and family members, presence of grandparents in the household), and childcare (preschoolinhistory) through interviews with mothers, conducted by field workerin Vietnamese. Data on household monthly income were collectedfrom commune-level secondary data sources. A questionnaire wadeveloped in English, translated into Vietnamese, and pretested inthe field before revisions and completion.

The research team consisted of 16 field surveyors; 4 were assignedexclusively for the Ravens tests and 2 for anthropometric measurements. They worked in 2 groups under the supervision of a teamleader. Half of the field workers were recruited for short-term work amain surveyors from universities or nongovernment organizations

whereas the remaining workers were recruited from local partners aassistants to the main surveyors. They had previous experience withdata collection in rural Vietnam and were trained in data collection

COGNITIVE DEVELOPMENT OF VIETNAMESE CHILDREN 191

5/21/2018 J. Nutr.-2005-Watanabe-1918-25

3/8

techniques before doing field work. The field workers were assignedspecific data collection tasks that were fixed throughout the durationof the survey to minimize interobserver errors.

Data management and analysis. Names and dates of birth ofmothers and children were obtained from the demographic recordbook held in the hamlet and surveys conducted previously in thecommunes by SC Japan. Then they were cross-checked at registrationpoints in the survey sites, double-checked by field workers beforemeasurements were taken, and finally checked by the team leaderafter data collection. All data collection forms were reviewed by theteam leader each evening and any discrepancies were discussed. Datacollected in the field survey were entered into EPI Info 2000 (CDC)immediately after collection.

The 2 main outcome variables were HAZ and Ravens test scores.Heights were compared with the international National Center forHealth Statistics/WHO/CDC reference standards (25) and convertedto HAZ using Epi Info 2000. Ravens test scores were calculated bysumming total correct answers across the 3 scales. All data wereentered into Epi Info 2000 and analyses were performed using SAS8.0 (SAS Institute) (26).

For data analysis, the present datasets were merged with the dataof previous surveys, which included weight/height of children atearlier ages, household characteristics, and child feeding/caring prac-

tices of mothers. Bivariate analyses were performed using t tests forcontinuous variables and 2 tests for categorical variables. The gen-eralized linear model (GLM) was used to assess effects of interven-tions on HAZ and Ravens tests scores. Similarly, generalized esti-mating equations (GEEs) were used to assess the effects onproportions. The GLM and GEE models control for potential con-founders to compute adjusted mean scores and proportions of eacheffect group for comparison. P-values 0.05 were considered signif-icant.

Human subjects and ethical considerations. The project wasapproved by the Emory University Institutional Review Board beforethe present research was launched. Written consent was receivedfrom parents of the children who voluntarily participated in theresearch. Field workers were trained in ethical considerations andwere under strict orders not to discuss information pertaining to

subjects with members of the community or with other non-project-related persons. All of the data collected in the researchwere kept confidential and used only for aggregated data analysis withanonymity.

RESULTS

Characteristics of subjects. Of the 431children studied313 or 73% were matched with baseline data obtained in 1998to construct a longitudinal data set. The remaining 118 children were excluded from the data analysis. Main characteristics of the studied children did not differ significantly fromthose excluded. Characteristics of the 313 study participant

are summarized inTable 1; each intervention had similar agand sex structures. Children ranged from 77 to 102 mo of agewith group means of 9193. The mean number of children pefamily was small (1.5) and the majority of mothers had children (95%). Mothers were young (mean age 33 y) antheir education level was low; more than one-third had completed only primary school. Farming was the primary occupation of mothers in the communes (9798%); 23% of childrenlived with their grandparents. Most of the household variablewere similar for the 2 interventions. They differed from eachother, however, in maternal education (P 0.05), child ag(P 0.01), parity (P 0.02), and number of children (P 0.01). The ECD-and-nutrition intervention commune hadfewer children per household, they were younger, and thei

mothers had less education. Those variables were controlledfor in multiple variable analyses.At baseline, children in the ECD-and-nutrition interven

tion commune did not differ nutritionally from those in thnutrition-only intervention commune on anthropometric indicators measured (Table 2). Some of the key feeding practiceas well as prenatal care reported by mothers were poorer for thECD-and-nutrition commune. Compared with nutrition-onlmothers, ECD-and-nutrition mothers were less likely to havhad 3 prenatal checks (26 vs. 53%, P 0.01) and startedgiving complementary food nearly 1 mo earlier (3.9 vs. 4.6 moP 0.02).

Nutritional status in terms of linear growth at baseline anat the time of the 2004 surveys was compared for all 313

children by GLM and GEE, which detected a significandecrease in the proportion of stunting over the 5 y frombaseline to 2004 for both of the interventions; the adjustedproportions of stunted children (2 HAZ) was reduced b

TABLE 1

Characteristics of subjects and their households by study group1,2

Variable

Intervention

P-valueECD and nutrition

(n 141)Nutrition only

(n 172)

Age of child,mo 93.4 6.6 90.9 7.3 0.01Age of mother,y 32.7 5.6 33.3 5.5 0.35Household income,3 1000 dong 91 19 89 18 0.55Household expenditures,3 1000 dong 447 186 422 198 0.29Size of family, n 4.7 1.2 4.8 1.1 0.44Parity,n 2.3 0.8 2.6 1.0 0.02Male sex, n (%) 74 (52.5) 83 (48.3) 0.462 children, n (%) 42 (29.8) 76 (44.2) 0.01Has a younger sibling, n (%) 32 (23.0) 50 (30.3) 0.15In 1st grade, n (%) 44 (31.2) 79 (45.9) 0.01Mothers education primary school, n (%) 67 (48.2) 61 (37.0) 0.05Mothers occupation, farmer, n (%) 130 (97.0) 160 (97.6) 0.77Living with grandparents, n (%) 31 (23.0) 38 (23.3) 0.94Had preschool experience, n (%) 135 (100.0) 160 (99.4) 0.36

1

Values are means

SD or n (%).2 P-values for continuous variables were calculated by Students 2-sided ttest.P-values for categorical variables were calculated using Pearsons 23 1 US$ 16,000 Vietnamese dong.

WATANABE ET AL.1920

5/21/2018 J. Nutr.-2005-Watanabe-1918-25

4/8

16% (P 0.01) and 13% (P 0.01) for the ECD-and-nutrition and nutrition-only interventions, respectively (Ta-ble 3). Severe stunting was decreased only for the ECD-and-nutrition children, with a reduction of adjusted proportions 3 HAZ of 7.8% (P 0.01). Overall, children in theECD-and-nutrition intervention had a slightly larger reduc-tion in the proportion of severe stunting. However, there wasno difference in the adjusted mean HAZ scores between thenutrition-only and ECD-and-nutrition intervention com-munes at baseline as well as in the 2004 surveys, indicating no

additional effects of ECD interventions on the nutritionalstatus of children.

Second, scores on the Ravens tests were modeled withGLM to assess the main effects (Table 4). The results showed

that test scores were significantly higher for the ECD-andnutrition than for the nutrition-only intervention group (adjusted mean scores 16.5 vs. 14.9, P 0.05). The difference inthe scores was particularly greater among stunted children(adjusted mean scores 16.8 vs. 12.8, P 0.01). Test scores ostunted and nonstunted children differed for the nutritiononly intervention (adjusted mean scores 17.0 vs. 12.8, P 0.01), whereas almost no difference was detected for children exposed to the nutrition-and-ECD intervention.

Additional models were also applied to assess whethe

initial height, change in height, and their interactions modified the effects (Table 4). Adjusting for initial height did nochange the effect, showing that the effects were not biased bythe different distributions of stunting at baseline in the groups. Further adjustment for change in height and the interactions between the change and initial height slightly reduced the difference in scores between the interventions (P 0.06). Even after this adjustment, however, the effect of thECD intervention on cognitive test scores among stuntechildren remained significant (P 0.01).

Finally, scores on the Ravens tests were modeled withGLM to compare how changes in the nutritional status ochildren from baseline to 2004 affected cognitive test scorebetween the 2 interventions (Table 5). Nutritional status a

baseline and in 2004 were cross-tabulated to form 4 groupsdrop (HAZ dropped from 2 to 2), remainestunted (HAZ stayed 2), never stunted (HAZ stayed2), and catch up (HAZ improved from 2 to 2)Adjusted differences in cognitive test scores between thgroups were detected only for the nutrition-only intervention(P 0.0010.06). Mean scores from the 4 groups (Fig. 1illustrate the dose-response relation between the change innutritional status over time and cognitive test scores for thenutrition-only intervention commune, which was not presenfor the ECD-and-nutrition intervention.

DISCUSSION

This study examined the effects of an early childhoodevelopment project for children aged 45 y on the cognitiv

TABLE 2

Nutritional status and key maternal behaviors at baseline and 2000 surveys in school-aged children

previously exposed to ECD-and-nutrition or nutrition-only intervention1,2

Variable

Intervention

P-valueECD and nutrition

(n 141)Nutrition only

(n 172)

WAZ 1.93 0.84 1.89 0.77 0.68HAZ 1.59 1.22 1.54 0.98 0.70WHZ3 1.19 0.75 1.20 0.69 0.88Underweight (WAZ 2), n (%) 70 (49.7) 81 (47.1) 0.65Stunting (HAZ 2), n (%) 53 (37.6) 54 (31.4) 0.25Wasting (WHZ 2), n (%) 16 (11.4) 19 (11.1) 0.93Premature birth, n (%) 11 (7.9) 15 (9.1) 0.71Smaller at birth, n (%) 62 (44.6) 60 (36.4) 0.143 antenatal care visits, n (%) 36 (25.5) 91 (52.9) 0.01Stop breast-feeding,mo 17.7 4.8 18.6 6.7 0.16Start complementary feeding, mo 3.9 1.9 4.6 3.1 0.02Participation in NREP, n (%) 62 (44.6) 91 (55.2) 0.07Participation in growth monitoring, within 3 mo, n (%) 97 (71.3) 87 (54.7) 0.01

1Values are means SD or n (%).2 P-values for continuous variables were calculated by Students 2-sided ttest.P-values for categorical variables were calculated using Pearsons 23 WHZ, weight-for-height Z-score.

TABLE 3

Anthropometric status of children exposed to

ECD-and-nutrition or nutrition-only intervention,

at the 2004 survey and change over 5 y1

Variable

InterventionDifference

2

ECD and nutrition(n 141) Nutrition only(n 172) P-value

2004Mean HAZ 1.42 0.74 1.36 0.83 0.12 0.18% 2 HAZ 21.3 18.0 4.8 0.30% 3 HAZ 0.7 4.1 3.1 0.19

Change3

Mean HAZ 0.17 1.00 0.18 0.84(% 2 HAZ) 16.3* 13.4*(% 3 HAZ) 7.8* 1.2

1Values are means SD or %.2 Differences between interventions were adjusted for initial height

at baseline, household income, and age of the child. P-values werecalculated using the GLM for continuous variables and GEE for pro-

portions.3 Change between baseline and 2004. P-values were calculatedusing McNemars tests. * Different from baseline, P 0.05.

COGNITIVE DEVELOPMENT OF VIETNAMESE CHILDREN 192

5/21/2018 J. Nutr.-2005-Watanabe-1918-25

5/8

development of the children who were previously exposed toa nutrition intervention at the age of 03 y. We collectedmeasurements of heights and cognitive test scores of children

when they reached school age in the 2 intervention communesin rural Vietnam. Overall, children exposed to Save the Chil-drens ECD interventions had significantly better scores onRavens progressive matrix tests than their counterparts whoexperienced only nutrition intervention. Effects were particu-larly large for currently malnourished children.

Overall, the current level of stunting among school-agchildren who were exposed to the Save the Childrens nutrition program was 20% in terms of the proportion of children

2 HAZ. Age-specific nutrition status by height-for-age inot available for this age group in Vietnam. A recent nationasurvey reported the prevalence of undernutrition among children 610 y old as 27% nationally and 28% for rural region(27). Although these data provide only an overview, it couldbe argued that nutritional improvement has occurred in thes

TABLE 4

Mean Ravens test scores of children exposed to ECD-and-nutrition or nutrition-only intervention, by current nutritional status1

Variable

Intervention

Difference

ECD and nutrition Nutrition only

n Score n Score

UnadjustedAll 141 16.6 5.4 170 16.1 6.1Normal2 111 16.7 4.9 140 17.0 5.5Stunted3 30 16.1 7.2 30 11.8 6.9

Adjusted for confounders4

All 128 16.46 0.58 152 14.91 0.56 1.55*Normal 101 16.10 0.53 124 17.03 0.48a 0.93Stunted 27 16.82 1.03 28 12.80 1.02b 4.03*

Adjusted for initial height5

All 128 16.62 0.59 152 15.08 0.57 1.54*Normal 101 15.96 0.54 124 16.96 0.48a 1.01Stunted 27 17.28 1.07 28 13.20 1.05b 4.08*

Adjusted for initial height andchange in height6

All 128 16.68 0.60 152 15.17 0.61 1.51

Normal 101 15.91 0.55 124 16.91 0.50a

1.00Stunted 27 17.44 1.14 28 13.43 1.19b 4.01*

1Values are means SD for unadjusted data and SEM for adjusted data. * Different between the interventions, P 0.05.2 HAZ 2.3 HAZ 2.4 PROC GLM was applied to compute least-squares means for different interventions, adjusted for household income, maternal education, child

age, and the presence of grandparents. Means in a column without a common letter differ, P 0.05.5Adjusting for initial height in addition to 4.6Adjusting for change in height from baseline to 2004 and its interaction with initial height, in addition to 5.

TABLE 5

Mean difference in Ravens test scores of groups by nutritional status change in school-aged children previously exposed to

ECD-and-nutrition or nutrition-only intervention

Variable

Intervention

ECD and nutrition Nutrition only

n Coefficient1 SE 95% CI n Coefficient1 SE 95% CI

Unadjusted (reference group catch up)Drop 6 0.35 2.53 ( 4.63, 5.33) 8 7.78* 2.23 ( 12.12, 3.39Remained stunted 24 0.73 1.56 ( 2.34, 3.79) 22 5.88* 1.56 ( 8.96, 2.81)Never stunted 82 1.63 1.22 ( 0.77, 4.03) 108 1.54 1.14 ( 3.77, 0.70)Catch up 29 32

Adjusted2 (reference group catch up)Drop 4 1.03 2.85 ( 6.65, 4.59) 8 7.16* 1.56 ( 11.43, 2.90Remained stunted 23 1.53 1.52 ( 1.46, 4.52) 20 5.40* 1.15 ( 8.46, 2.34)Never stunted 75 0.62 1.22 ( 1.78, 3.01) 96 2.15 2.85 ( 4.42, 0.11)Catch up 26 28

1

Coefficients from linear regression models in which the Ravens test score is the dependent variable; coefficients represent the mean deficit ithe Ravens test score relative to the reference group. * Different from the reference, P 0.05.2 PROC GLM was applied, adjusted for household income, maternal education, childs age, and the presence of grandparents.

WATANABE ET AL.1922

5/21/2018 J. Nutr.-2005-Watanabe-1918-25

6/8

communes. The difference was not significant, however, be-tween the 2 communes in terms of the amount of change overtime in mean HAZ scores. Those results indicate that therewere no additional effects of ECD interventions on the nutri-tional status of children.

With the nutritional improvement program, the nutritionalchange among children exposed to nutrition-only interven-tions had a linear association with cognitive development.

This finding, consistent with other studies that detected sig-nificant associations between cognitive performance amongschool-age children and their current or early nutritional sta-tus (57,11), suggests that continuing malnutrition is detri-mental to the cognitive development of children.

Children exposed to Save the Childrens ECD interven-tions had significantly better test scores than their counter-parts who experienced only the nutrition intervention. Differ-ences in test scores between stunted and nonstunted childrenwere minimal among children who had ECD interventions,whereas they were significantly larger in the nutrition-onlyintervention commune in which better-nourished childrenhad higher scores than malnourished children. Significanteffects of the ECD intervention were thus most noticeable forcurrently malnourished children whose test scores were signif-icantly higher than similarly malnourished children who wereexposed only to the nutrition intervention.

Furthermore, additional analyses controlling for initialheight, change in height over time, and their interactions didnot alter the findings without the adjustments, suggesting thatthe effect of the ECD intervention was separate from theeffects on physical growth. All of this evidence supports theprotective effect of ECD intervention against the negativecognitive effects of growth failure; it is possible that the effectsof Save the Childrens ECD intervention compensated forthe loss caused by poor physical growth in the earlier years.

Reviews of early childhood development programs for chil-dren in low-income families in the United States reportedevidence that the programs produced large effects on IQ dur-ing the early childhood years and sizable persistent effects on

school achievement (28). Those included longitudinal studieswhich found significant positive effects on IQ scores (29,30)achievement test scores (2932), school repetition (29,32)attendance (33), and special education (2931). For populations in developing countries with endemic undernutritionmany studies assessed the effects of food supplementation onthe cognitive development of children. In contrast, very fewexamined the additional effects of psychological stimulation

A longitudinal study in Cali, Colombia assessed the effectof integrated interventions of varying duration, consisting onutrition supplementation, health care, and an early educationprogram, on the cognitive development of preschool childrenaged 42 84 mo who were chronically malnourished. Thresults showed that nutrition supplementation alone beforinitiation of the preschool program had no effect on psychological development measured by various intellectual abilittests at several ages from 4 to 7 y, whereas the combinedinterventions of nutrition and early education had larger effects if they were started earlier and lasted longer (34). Thstudy design, however, did not allow us to compare the effectof nutrition supplementation alone with the combined interventions of supplementation and stimulation. Another study

in Bogota, Colombia compared the effects of supplementationonly, stimulation only, supplementation and stimulation, andno intervention on cognitive development of children aged036 mo. It detected clear effects of nutrition supplementation on the cognitive performance of children as measured bythe Griffith test. However, no added benefits of psychologicastimulation were found (12).

Evidence concurrent with this studys findings was providedby a study of stunted Jamaican children, aged 924 mo, whowere randomly assigned to nutrition-only, stimulation-onlynutrition-and-stimulation, and control groups (10) After th2-y intervention, the benefits from a combination of supplementation and stimulation were additive, and the childrenreceiving both treatments caught up to the nonstunted contro

group in developmental levels. A follow-up study of thoschildren at 7 y of age showed small, but significant, globabenefits of the interventions (9). That study indicated a plausible mechanism that the lower activity levels of undernourished children may not affect their development when they arin a stimulating environment.

The greater effect of the combined interventions in improving both physical growth and psychological developmenthan the effect of either one alone may be explained by theiinteractions at 3 different critical points: at the child levelbetween the child and his or her family or caregivers, and inthe design and delivery of programs (35). At the level of thchild, the negative cognitive effects of growth failure werevident, as shown in the nutrition-only commune in thistudy, which was protected by exposure to the ECD intervention. Psychosocial stimulation programs, although improvincognition, would also have positively affected physical growth(36), which was not detected in this study. This was possiblydue to the late timing of the ECD intervention (ages of 45 y)Effective timing and duration of ECD interventions warranfurther investigation. As illustrated by similar studies in Colombia (32), the effects of psychological stimulations werlarger if the interventions were applied earlier and for a longeperiod. A follow-up study on younger children aged 03 y athe time of this study, who were exposed to the ECD intervention at younger ages and for a longer period, may providuseful insights.

At the level of interaction between child and caregiversbetter-nourished children tend to be more active and exploratory and more able to elicit parental interaction. ECD inter

FIGURE 1 Mean test scores by change in nutritional status be-tween baseline and 2004 in school-aged children previously exposed tonutrition-only intervention or nutrition-and-ECD intervention. Values are

means and SEM,n 313. The 4 groups are defined as follows:1) Drop HAZ 2 at baseline and 2 in 2004, 2) Stunt HAZ 2 atbaseline and in 2004, 3) Norm HAZ 2 at baseline and in 2004; 4)Up HAZ 2 at baseline and 2 in 2004.

COGNITIVE DEVELOPMENT OF VIETNAMESE CHILDREN 192

5/21/2018 J. Nutr.-2005-Watanabe-1918-25

7/8

vention may have stimulated parents to acquire more respon-sive behaviors toward malnourished children than parentswith nutrition-only intervention. This study, however, cannotprovide information to further examine the critical points dueto lack of data on parental behaviors and child-parent inter-actions. It is of interest to determine which components of theECD interventions worked most effectively on child-parentinteractions. The ECD interventions combined the center-

based approach with the parental support component, whichcannot be separated in this study design. The preschool-basedactivities could have stimulated active learning and socialinteractions between the child and his or her environment,and the parental education component may have enhancedparents knowledge, attitude, and practices regarding childcare at home.

The nonverbal cognitive ability measured by Ravens Col-ored Progressive Matrices test is defined as eductive, whichrefers to the ability to make meaning out of confusion. It is notwithin the scope of this study to detail the interpretationbeyond the one summarized by Raven that eductive ability,compared with other cognitive abilities, is promoted whenparents involve their children in their own attempts to make

sense of difficult situations (37). This aspect of ability hasrelevance in assessing effects of ECD interventions. However,we can by no means assume that Ravens progressive matrix isthe single best tool with which to assess cognitive develop-ment in rural Vietnam. It would have been preferable toperform several other tests that have been used for differentpurposes in studies measuring childrens cognitive perfor-mance. School test scores and teacher ratings were of partic-ular interest because the effect is modified by access andquality of education at preschool and primary schools. Never-theless, the study provided a unique example of the utility ofRavens tests in assessing the cognitive development of chil-dren in developing countries.

We acknowledge several other limitations to this study in

addition to the points raised above. Unlike intervention trialsusing direct high-energy food supplementation or intensiveprovision of psychological stimulation, this study involvedinterventions designed for a real-life project with a primaryfocus on behavioral changes in feeding and child-care prac-tices by parents as well as the community. The absence of dataon the key behavioral factors did not allow us to investigateintermediate variables. If the Ravens test had been conductedat baseline, it would have provided stronger evidence that thedifferences could be attributed to the ECD intervention.Larger samples randomly selected from the entire project com-munes would have provided a statistically more valid estima-tion of measurements. Measurements of potential confounders,particularly regarding educational opportunities available forchildren and mothers, would have helped to better explainstudy outcomes and interpretations. The optimal study designwas beyond the resources of this study.

In summary, this study documented evidence that during aperiod of nutritional change, children who failed to catch upin linear growth are likely to lose the opportunity for cognitivebetterment in the early primary school years if appropriateinterventions are not given. The findings suggest that earlychildhood program interventions for preschool-age children(45 y) increase the potential for cognitive development,particularly for malnourished children, whose growth failureprevents proper cognitive development. All of these findingssupport the importance of an integrated program of nutritionand ECD to optimize the potential for holistic development ofchildren in an equitable manner.

The Government of Vietnam, in collaboration with inter-

national donors, has initiated efforts to develop an effectiveintegrated model of ECD and nutrition for preschool childrenSave the Children is expanding its ECD program to have awider scope with particular focus on the northern mountainous regions where people belonging to an ethnic minority liveThese regions have fallen behind during the current trend owidening regional disparity, and children there suffer fromchronic poverty and inferior access to health and education

services, as well as food insecurity. Such a context allows thesedisadvantaged children to lose their chance for physical andintellectual development. It is here that the contributionfrom nutrition and ECD programs are expected to be mosimportant. It is recommended, therefore, that the effects of theintegrated nutrition and ECD interventions be examined further through community intervention trials.

ACKNOWLEDGMENTS

The authors thank the following people who contributed to thsuccess of this study: field survey team members of Save the Children

JapanVietnam Office, who participated in the data collection; TuanTran, Thach Duc Tran, and Van Ha of Research and TraininCenter for Community Development, who provided training antools for the Ravens tests and other technical assistance.

LITERATURE CITED

1. The United Nations System Standing Committee on Nutrition (2004The 5th Report on the World Nutrition Situation. World Health OrganizatioGeneva, Switzerland.

2. Black, R. E., Morris, S. S. & Bryce, J. (2003) Where and why are 1million children dying every year? The Lancet 361: 22262234.

3. The United Nations System Standing Committee on Nutrition (1992The 2nd Report on the World Nutrition Situation, Vol. I. World Health OrganizationGeneva, Switzerland.

4. Hall, A., Khanh, L. N., Son, T. H., Dung, N. Q., Lansdown, R. G., Dar, D. THanh, N. T., Moestue, H., Kho, H. H. & Bundy, D. A. (2001) An associatiobetween chronic undernutrition and educational test scores in Vietnamese chidren. Eur. J. Clin. Nutr. 55: 801804.

5. Sigman, M., Neumann, C., Jansen, A. A. & Bwibbo, N. (1989) Cogntive abilities of Kenyan children in relation to nutrition, family characteristics, aneducation. Child Dev. 60: 14631474.

6. Moock, P. R. & Leslie, J. (1986) Childhood malnutrition and schoolinin the Terai region of Nepal. J. Dev. Econ. 20: 3352.

7. Mendez, M. A. & Adair, L. S. (1999) Severity and timing of stunting the first two years of life affect performance on cognitive tests in late childhoodJ. Nutr. 129: 15551562.

8. Freeman, H. E., Klein, R. E., Townsend, J. W. & Lechtig, A. (1980Nutrition and cognitive development among rural Guatemalan children. Am. JPublic Health 70: 12771285.

9. Grantham-McGregor, S. M., Walker, S. P., Chang, S. M. & Powell, C. A(1997) Effects of early childhood supplementation with and without stimulatioon later development in stunted Jamaican children. Am. J. Clin. Nutr. 66: 247253.

10. Grantham McGregor, S. M., Powell, C. A., Walker, S. P. & Himes, J. H(1991) Nutritional supplementation, psychosocial stimulation, and mental deveopment of stunted children: the Jamaican Study. Lancet 338: 15.

11. Pollitt, E., Gorman, K. S., Engle, P. L., Martorell, R. & Rivera, J. (1993

Early supplementary feeding and cognition. Monogr. Soc. Child Dev. 58: 110212. Waber, D. P., Vuori-Christiansen, L., Ortiz, N., Clement, J. R., Chris

tiansen, N. E., Mora, J. O., Reed, R. B. & Hererra, M. G. (1981) Nutritionasupplementation, maternal education and cognitive development of infant at risof malnutrition. Am. J. Clin. Nutr. 34: 807813.

13. Chavez, A., Martinez, C., Soberanes, B., Dominguez, L. & Avila, A(1994) Early Nutrition and Physical and Mental Development in Mexican Rur

Adolescent Families. International Center for Research on Women, WashingtonDC.

14. Pollitt, E., Watkins, W. E. & Husaini, M. A. (1997) Three-month nutrtional supplementation in Indonesian infants and toddlers benefits memory function 8 y later. Am. J. Clin. Nutr. 66: 13571363.

15. Grantham-Macgregor, S. M., Fernald, L. C. & Sethuranman, K. (1999Effect of health and nutrition on cognitive development in children in the first yearsof life. Food Nutr. Bull. 20: 5375.

16. Li, H., Barnhart, H. X., Stein, A. D. & Martorell, R. (2003) Effects of earchildhood supplementation on the educational achievement of women. Pediatric112: 11561152.

17. General Statistic Office and Vietnam Committee for Protection and Carfor Children (2000) Analysis of Result of Multiple Indicator Cluster Survey IGeneral Statistic Office of Vietnam, Hanoi, Vietnam.

WATANABE ET AL.1924

5/21/2018 J. Nutr.-2005-Watanabe-1918-25

8/8

18. National Institute of Nutrition & UNICEF (2003) Vietnam 20002002: AReview of the Nutrition Situation. National Institute of Nutrition and UNICEF,Hanoi, Vietnam.

19. Sternin, M., Sternin, J. & Marsh, D. (1998) Designing a Community-Based Nutrition Program Using the Hearth Model and the Positive Deviance

ApproachA Field Guide. Save the Children, Westport, CT.20. United Nations (1986) How To Weigh and Measure Children: Assess-

ing the Nutritional Status of Young Children in Household Surveys. Department ofTechnical Cooperation for Development and Statistical Office, New York, NY.

21. Infant/child/adult height measuring board [Online]. Shorr Productions.http://www.shorrproductions.com [accessed December 4, 2004].

22. Ravens Colored Progressive Matrices [Online]. Campbell Thomson &McLaughlin Ltd. http://www.jcravenltd.com/applications.htm [accessed Decem-ber 4, 2004].

23. Court, J. H. & Raven, J. (1995) Manual for Ravens Progressive Ma-trices and Vocabulary Scales. Oxford Psychologists Press, San Antonio, TX.

24. Attawell, K. (2003) International Longitudinal Research on ChildhoodPoverty: Practical Guidelines and Lessons Learnt from Young Lives. Young Lives,London, UK.

25. World Health Organization (1995) Physical Status: The Use and Inter-pretation of Anthropometry: Report of a WHO Expert Committee. WHO, Geneva,Switzerland.

26. SAS 8.0 [Online]. SAS Institute Inc. http://www.sas.com [accessed De-cember 4, 2004].

27. National Institute of Nutrition (2003) 2000 General Nutrition Survey.Medical Publishing House, Hanoi, Vietnam.

28. Barnett, W. S. (1995) Long-term effects of early childhood programsof cognitive and school outcomes. The Future of Children 5: 2550.

29. Campbell, F. A. & Ramey, C. T. (1994) Effects of early intervention ointellectual and academic achievement: a follow-up study of children from lowincome families. Child Dev. 65: 684698.

30. Levenstein, P., OHara, J. & Madden, J. The Mother-Child Home Programof the Verbal Interaction Project. In: As the Twig Is Bent: Lasting Effects oPreschool Programs (Consortium for Longitudinal Studies), pp. 237263. Erbaum, Hillsdale, NJ.

31. Jester, R. E. & Guinagh, B. J. (1983) The Gordon Parent EducatioInfant and Toddler Program. In: As the Twig Is Bent: Lasting Effects of PreschooPrograms (Consortium for Longitudinal Studies), pp. 103132. Erlbaum, HillsdaleNJ.

32. Palmer, F. (1983) The Harlem Study: effects by type of training, age otraining, and social class. In: As the Twig Is Bent: Lasting Effects of PreschooPrograms (Consortium for Longitudinal Studies), pp. 201236. Erlbaum, HillsdaleNJ.

33. Seitz, V., Rosenbaum, L. K. & Apfel, N. H. (1985) Effects of famisupport intervention: a ten-year follow-up. Child Dev. 56: 376 391.

34. Mckay, H., Sinisterra, L., Mckay, A., Gomez, H. & Lloreda, P. (1978Improving cognitive ability in chronically deprived children. Science (WashingtonDC) 200: 270278.

35. World Health Organization (1999) A Critical Link: Interventions foPhysical Growth and Psychological Development. Department of Child and Adolescent Health and Development. WHO, Geneva, Switzerland.

36. Super, C. M., Herrera, M. A. & Mora, J. O. (1990) Long term effects ofood supplementation and psychosocial intervention on the physical growth oColombian infants at risk of malnutrition. Child Dev. 61: 2949.

37. Raven, J. (2000) The Ravens Progressive Matrices: change and stability over culture and time. Cognit. Psychol. 41: 148.

COGNITIVE DEVELOPMENT OF VIETNAMESE CHILDREN 192