Asia Pac J Clin Nutr 2017;26(2):317-325 317

Original Article

Household dietary diversity and child stunting in East

Java, Indonesia

Trias Mahmudiono SKM, MPH (Nutr.), GCAS, PhD1,2, Sri Sumarmi SKM, MSi, Dr1,
Richard R Rosenkranz PhD, FACSM2
1
Department of Nutrition, Faculty of Public Health, Universitas Airlangga, Indonesia
2
Department of Food, Nutrition, Dietetics & Health, College of Human Ecology, Kansas State University,
USA

Background and Objectives: More than one-quarter of under-five children in the developing world are stunted,
and those with poor nutrient intake are at risk of irreversible cognitive impairment. The purpose of this study was
to determine the relationship between dietary diversity and child stunting in an Indonesian context. Methods and
Study Design: Dietary diversity was assessed using a maternal-reported checklist of 12 food groups, summed as
a Household Dietary Diversity Score. Stunting was defined as ≤-2.0 height-for-age z-score by WHO-Anthro 2005.
Trained interviewers administered the household dietary diversity questionnaire to 768 households with children
aged <5 years in East Java, Indonesia. Logistic regression models were constructed to test the association be-
tween dietary diversity and child stunting. Results: The prevalence of child stunting was 39.4%, and the percent-
age of households consuming food groups high in protein and calcium, like dairy products (41%), and
meat/poultry, (65%) was lower compared with other food groups. The unadjusted model revealed that higher die-
tary diversity scores were associated with lower likelihood of child stunting (OR=0.89; 95% CI=0.80–0.98). This
relationship remained significant after adjustment for family size, maternal literacy, food expenditure, breastfeed-
ing, energy, and protein intake (OR=0.89; 95% CI=0.80–0.99). Conclusions: The dietary diversity score was
moderate, with consumption of dairy products and meat/poultry lowest among 12 food groups. Hence, population
interventions should focus on promoting food groups currently lacking in maternal and child diet, including those
rich in growth-promoting nutrients like dairy, meat/poultry. These results, from an Indonesian context, confirm
the widely observed protective relationship between dietary diversity and child stunting.

Key Words: dietary diversity, child stunting, malnutrition, food pattern, Indonesia

INTRODUCTION Among the three groups of priority policy responses in

Childhood stunting is a major nutritional problem in de- the SUN framework are nutrition-specific intervention.
veloping countries.1 Stunting is defined as height-for-age This intervention has nutritional improvement as the pri-
z-score below -2.0, using the current World Health Or- mary goal, and should be accessible to all individuals and
ganization (WHO) standards from the Multi Growth Ref- their households, especially from pregnancy to the first
erence Study.2 In 2011, around 26% of world’s children two years of life, and at times of illness or distress.5 To
under the age of five years were stunted.2 Based on the prevent a lifetime of lost potential due to stunting, em-
United Nations Children’s Fund (UNICEF) conceptual phasis should be placed on the importance of providing a
framework on malnutrition, nutritional status is influ- supportive nutritional environment through timely and
enced by three broad factors: food, health, and care. 3 appropriate complementary feeding during the first 1,000
Stunting is caused by long-term insufficient nutrient in- days of life,4-6 that involves a diverse diet.7-10
take,4 and frequent infections.3 There is an association One method of assessing dietary quality employs the
between childhood stunting and suboptimal brain devel- dietary diversity score.11 The dietary diversity score (DDS)
opment, which might impair children’s cognitive ability, measures the sum of diverse food categories, irrespective
school performance, and potential earning when they en- of the amount consumed individually for the last 24
ter the workforce.1-3 The WHO report of a colloquium on hours.12-13 The DDS, as measured by a quantitative num-
childhood stunting highlighted the importance of the
1,000-day period (from conception to 24 months of age) Corresponding Author: Dr Trias Mahmudiono, Department of
when the foundation is laid for an individual’s physical Nutrition, Faculty of Public Health, Universitas Airlangga, Jl.
size, as well as their physiological and intellectual capaci- Mulyorejo Kampus C, Surabaya 60115, Indonesia.
ties in later life.1 With detrimental effects of child stunt- Tel: +62-31-5964808, Fax: +62-31-5964809
Email: [email protected]
ing, many countries affiliated with The World Health
Manuscript received 06 September 2015. Initial review com-
Assembly have committed to achieve 40% reduction by
pleted 08 October 2015. Revision accepted 19 November 2015.
2025 through the Scaling up Nutrition (SUN) program.3 doi: 10.6133/apjcn.012016.01
318 T Mahmudiono, S Sumarmi and RR Rosenkranz

ber of food groups, has become a widely used method of in 2007 to 35.8% in 2010.36 To the best of our knowledge,
determining variety in the diet, and by proxy, nutrient there is no previous study conducted in East Java specifi-
adequacy.12 Several classification systems have evolved cally, or more generally within Indonesia, a country of
in determining DDS with the number of food groups var- 240 million people, to test the hypothesis that dietary di-
ying from 7,14 8,15 9,12-13 to 1213 groups. versity decreases the likelihood of child stunting. A pre-
One of the major nutritional problems in the diets of vious study involving a population-based sample of
developing countries is the lack of dietary diversity, 446,473 rural and 143,807 urban children in poor areas in
mainly comprising plant-based food sources, but with Indonesia showed that higher household expenditure on
limited intake of fruits and vegetables.16 Dietary quality animal-source and non-grain foods lowered the risk of
and diversity have changed; secondary data analysis in 6 stunting among children 0-59 months old.37 The study of
developing countries showed that the trend in dietary en- dietary diversity within an Indonesian context is specifi-
ergy availability steadily increased from 1970 to 2002.17 cally interesting because of the rapid nutrition transition
Increases in dietary energy availability were likely due to this country is undergoing that is believed to be the result
the increased consumption of edible oil, and in fats as per- of shifted food preferences toward modern Western
centage of dietary consumption.18-19 foods.38 Indonesia is an archipelago country with rich
Several studies have reported that low dietary diversity natural resources that gives vast food source availability
is associated with increased likelihood of child stunting,20 compared with previous studies related to dietary diversi-
and double burden of malnutrition.21 A low DDS also has ty and stunting in countries with limited natural resources,
been associated cardiovascular risk,22 dyslipidemia,23 and like those in sub-Saharan Africa.26,39 Studies in the
higher probability of metabolic syndrome.24 While food Southeast Asia region with relatively abundant natural
insecure-households have been linked with negative nu- resources of food, such as in Cambodia8,28 and Malay-
trition outcomes such as poor linear growth and child sia,27 only incorporated samples from rural areas, urban
stunting,25 the association between food insecurity and areas, or both with the data still producing equivocal re-
child stunting was independent with regard to dietary sults. To strengthen the evidence in Southeast Asian set-
diversity, especially in semi-arid areas.26 Previous studies tings, our study included participants from both rural and
in Southeast Asia have been mixed regarding the associa- urban settings. Even further, we incorporated different
tion between dietary diversity and child stunting. A study types of geographical locations from coastal to mountain-
in a Malaysian indigenous population showed that better ous areas. With the magnitude of child stunting in Indo-
dietary diversity was associated with less likelihood of nesia and absence of scientific evidence in an Indonesian
child stunting,27 while a similar study in rural Cambodia context, we aimed to determine the relationship between
showed no significant association.28 dietary diversity and child stunting in East Java, Indone-
A study in Ethiopia revealed that the dietary diversity sia.
among school-aged children was low, and even though
not statistically tested in relation to dietary diversity, the MATERIALS AND METHODS
article did highlight that the school-aged children were A cross-sectional study was conducted in 8 districts rep-
also lacking in animal-source food (ASF) intake.29 In resenting both urban and rural areas, as well as coastal
Ethiopia, intake of ASF was a protective factor against and mountainous regions in East Java Province, Indonesia.
stunting30 and zinc deficiency.31 One might argue that the With an initial sample of 768 households with under-five-
low dietary diversity score was driven mainly by the lack year-old children, 736 were included in the final analysis.
of ASF intake that was found to have a wide array of mi- The sample was determined using a multi-stage cluster
cronutrients essential for enriching type II nutrients (the random sampling following the census block from the
growth nutrients) such as: vitamin A, vitamin B-12, ribo- Indonesian Buerrau of Statistics. The sampling technique
flavin, calcium, iron and zinc.32 Specifically, zinc is an was done as follows: three sub-districts were randomly
essential trace element with special importance in the selected from each district/municipality, then two villages
immune system.33 A study of Peruvian infants showed an were randomly selected from each sub-district, and final-
association with poor growth when the body has a pro- ly 16 households were randomly selected in every village
longed short supply of these micronutrients.34 The im- block. Samples were 96 households in each dis-
mune system might also be depressed due to inability of trict/municipality. Consent was obtained from each par-
low fat mass to secrete leptin that helps stimulate immune ticipant in written form after information regarding the
system.35 Due to a weakened immune system, it is be- purpose of the study was explained. All identities of the
lieved that children are prone to repeated bouts of infec- participants were kept confidential.
tion, and thereby become stunted.3 Therefore, sufficient In this study, a single 24-hour maternal-reported food
intake of ASF increases fat stores and provides nutrients recall data was applied to the checklist of 12 food groups
that are essential for growth and micronutrients that sup- proposed by FAO.13 Presence or absence of the food
port the immune system.29 groups consumed in the last 24 hours then determined the
The prevalence of child stunting in Indonesia was household dietary diversity score (HDDS). The score is
36.8%, 35.6% and 37.2% in 2007, 2010 and 2013, respec- continuous, ranging from 0 to 12, based on whether
tively.36 Some have suggested that the East Java Province members of the household consumed any of the 12 food
is a microcosm of Indonesia in terms of child health out- groups in the last 24 hours prior to the interview.40 All the
come achievement in Indonesia. Based on the data of food items consumed by the participants were categorized
Indonesia’s Basic Health Research (Riskesdas), the into 12 food groups which were: staple food (rice, cere-
prevalence of stunting in East Java increased from 34.8% als), tubers/roots, vegetables, fruits, fish (including dried
 Dietary diversity and child stunting 319

fish and seafood), meat (including poultry) egg, nuts and ers of the children. This study utilized the Food Security
seeds, milk (including all dairy products), spices, oils and Survey data from the Food Security Board of East Java
fats (coconut products were included), and sweets. Province in Indonesia that was anonymous, and was ap-
Trained interviewers administered the HDDS to 768 proved by the Committee on Research Involving Human
household members with children aged <5 years in East Subjects/Institutional Review Board (IRB) for Kansas
Java, Indonesia. Children’s weight was measured using a State University, USA with proposal number 7955.
standardized electric scale (Camry EB6571, Guangdong
China) in at least 0.1 kg increments. Standing height was RESULTS
measured by using Vktech Stature Meter height measure Household and Environmental Characteristics
or microtoise in 0.1 cm increments. Stunting was defined The characteristics of the households were mostly con-
as ≤-2.0 height-for-age z-score (HAZ) by WHO-Anthro sisting of a nuclear family, with total number of house-
2005. hold members of ≤4 people (55.4%) and the main source
Maternal attributes and socio-demographic factors of drinking water was a well (80%). Thirty-seven percent
were measured using a structured face-to-face interview of households had 5 to 6 household members living under
performed by trained interviewers. Maternal attributes the same roof. Approximately 65% of households had
included were maternal literacy, and exclusive breastfeed- only one working member of the household. Almost all
ing status. Socio-demographic factors included total households had a literate father (92.5%), indicating suffi-
monthly expenditure, food expenditure, source of drink- cient capability of understanding written information. The
ing water, number of under-five-year-old children, and mean monthly family income was 92 USD, which was
family income. greater than the mean monthly expenditure (64 USD).
SAS version 9.4 was used for statistical analyses. Lo- Almost 47.8% of the family income was used for pur-
gistic regression models were constructed to test the asso- chasing food.
ciation between HDDS and child stunting. We developed The prevalence of child stunting was 39.4% and the
four models of logistic regression to test our hypothesis mean HDDS was 9.1, a score falling between moderate
that better dietary diversity score was a protective factor, and high. More than 90% of the households were con-
or related to decreased likelihood of child stunting in In- suming staple food (rice; 99.6%), vegetables (97.8%),
donesia. The first model included only HDDS, while the legumes (99.0%), egg (91.5%), and sweets (91.7%) and
second model also included family size and maternal lit- fat/oil (99.6%). Wider variation was observed in terms of
eracy as covariates. The third model added the food ex- fruit (64.8%), spices (78.2%), meat/poultry (65.0%) and
penditure variable as an estimate of food security. In the fish (88.0%) food group consumption. The least con-
fourth and final model, we incorporated all potential co- sumed food group was dairy products, with only 41.5% of
variates (family size, food expenditure, breastfeeding households reporting consumption in the last 24 hours.
status, total energy intake, and protein intake) to see
whether the observed association in the original model Dietary diversity and stunting
was robust. For all statistical analyses, the results were Looking at univariate correlations between “non-
deemed significant if the obtained p-values were less than nutritional” independent variables and child stunting, only
0.05. food expenditure had a significant association (p=0.04);
while family size (p=0.81), source of drinking water
Ethical clearance (p=0.67), geographical location (p=0.36) and maternal
Prior to the interview and anthropometric measurement, literacy (p=0.34) were not significant. For the nutritional-
the trained interviewers explained the purpose of the ly related independent variables, dietary diversity (p=0.03)
study and asked for oral informed consent from the moth- and protein intake (p=0.04) was significantly associated

Table 1. Mean score of individual food group consumed in household for the last 24 hours in association with child
stunting

95%CI
Food groups Mean±SD p-value Unadjusted OR
Lower Upper
Dairy 0.41±0.49 0.11 1.32 0.94 1.85
Egg 0.92±0.28 0.18 0.67 0.38 1.20
Fish* 0.88±0.32 0.03 1.83 1.07 3.15
Fruit 0.65±0.48 0.19 0.78 0.55 1.13
Legumes 0.99±0.09 0.79 1.25 0.24 6.60
Meat & poultry* 0.65±0.48 0.04 0.68 0.47 0.98
Oil & fat 0.99±0.07 0.64 0.56 0.05 6.45
Spices* 0.79±0.40 0.04 0.66 0.44 0.99
Staple (rice) 0.99±0.05 0.71 0.58 0.04 9.55
Sweet 0.91±0.28 0.90 1.04 0.59 1.83
Vegetables 0.98±0.15 0.37 0.62 0.22 1.75
*
Significant at α=0.05 for particular dietary diversity score of food group consumed in the last 24 hours in household with and without
stunted children. The unit of the mean and standard deviation on each food groups was based on binary score (0=not consuming; 1= con-
suming food groups in the last 24 hours).
320 T Mahmudiono, S Sumarmi and RR Rosenkranz

Table 2. Cross tabulation of fish group consumption and child stunting status

Child nutritional status, n (%)

Fish groups Chi-squared test
Not stunted Stunted
Salt-water fish Yes 151 (21.7) 105 (15.1) 0.232
No 279 (40.1) 160 (23.0)
Fresh-water fish Yes 14 (2.0) 10 (1.4) 0.717
No 416 (59.9) 255 (36.7)
Dried fish* Yes 30 (4.3) 33 (4.7) 0.015
No 400 (57.6) 232 (33.4)
Fish group in the calculation of dietary diversity score consist of salt-water fish, fresh-water fish and dried fish.
*
Consumption of dried fish was significantly different at α=0.05 between household with and without child stunting.

Figure 1. Predicted probability for stunting based on the final model (dietary diversity score adjusted for family size, food expenditure,
breastfeeding status, total energy intake, and protein intake).

with child stunting, while breastfeeding status (p=0.71) predictive variable, and binary response of stunting or not
and total energy intake was not significant (p=0.16). stunted. There was a significant protective association
Binary logistic regression involving all food groups as between DDS and child stunting with OR=0.89 (95%
independent variables and child stunting as dependent CI=0.80–0.98). The results of the final model showed that
variables showed only consumption of fish (p=0.03), the significant association between HDDS and child
spices (p=0.04) and meat/poultry (p=0.04) were statisti- stunting persisted across models. The final model showed
cally significant (Table 1). Fish consumption consisted of that HDDS was inversely associated with child stunting
fresh-water fish, salt-water fish, as well as dried fish. In with OR=0.89 (95% CI=0.80–0.99). This means that with
the last 24 hours, the mean consumption of salt-water fish a one-point increase in household dietary diversity score,
(27.9 g, SE=1.3) and dried fish (7.7 g, SE=0.6) was high- the likelihood of child stunting was decreased by more
er, compared with fresh-water fish (3.6 g, SE=0.4). As than 10% (Figure 1).
shown in Table 2, the chi-squared test within each type of Subsequent logistic regression was performed to strati-
fish consumption and child stunting showed a significant fy our final model with breastfeeding status as: exclusive-
association only for dried fish (p=0.02). The odds ratio of ly breastfed (until 4 months), partially breastfed, and not
overall fish group consumption was 1.83 (95% CI=1.07– breastfed (Table 3). The results showed only among non-
3.15) indicating fish consumption was a risk factor for breastfed children that the association between HDDS
child stunting. In contrast, the odds ratio of both spices and child stunting remained statically significant
and meat/poultry group showed inverse associations with (OR=0.88; 95% CI=0.078-0.99) adjusted for family size,
child stunting. The odds ratio of spices consumption was food expenditure, total energy intake and total protein
0.66 (95% CI=0.44–0.99), while meat or poultry con- intake.
sumption was 0.68 (95% CI=0.47–0.98).
All four models in the logistic regression consistently DISCUSSION
showed significant associations between HDDS and child The objective of the current study was to determine the
stunting. The first model used dietary diversity score as a relationship between dietary diversity and child stunting
 Dietary diversity and child stunting 321

Table 3. Logistic regression coefficients of predictive power of variables included in the model for child stunting by breastfeeding status

Exclusively breastfed (until 4 months) Partially breastfed Not breastfed

(n=62) (n=144) (n=530)
Variable
95% CI 95% CI 95% CI
p-value OR p-value OR p-value OR
Lower Upper Lower Upper Lower Upper
HDDS 0.91 1.02 0.67 1.55 0.43 0.90 0.70 1.17 0.04 0.88* 0.78 0.99
family size 0.46 1.62 0.45 5.89 0.57 1.26 0.57 2.76 0.36 0.84 0.59 1.22
food expenditure 0.85 1.15 0.29 4.47 0.25 1.61 0.71 3.61 0.48 1.15 0.79 1.67
Energy intake 0.28 1.00 0.99 1.01 0.29 1.00 1.00 1.00 0.74 1.00 0.99 1.00
Protein Intake 0.97 0.99 0.76 1.30 0.13 0.97 0.93 1.01 0.34 0.99 0.98 1.01
*
HDDS (Household Dietary Diversity Score) significantly associated with child stunting at α=0.05 only for non-breastfed children in a logistic regression model involving independent variables (family size, food
expenditure, total energy intake and total protein intake).
322 T Mahmudiono, S Sumarmi and RR Rosenkranz

in an Indonesian context. Our study confirmed the hy- Asian countries, significant associations between better
pothesis that dietary diversity, as measured using house- HDDS and decreased likelihood of child stunting also
hold dietary diversity score, was a protective factor for have been reported in Bangladesh,46 and Ethiopia.39
child stunting. We found a significant protective associa- The strength of association between HDDS and child
tion between household consumption of meat/poultry and stunting in our study was relatively small (OR=0.89; 95%
child stunting. The proportion of food groups that were CI=0.80–0.99) compared with the previous study in
rich in growth-promoting nutrients such as animal protein Bangladesh for children aged 24-59 months (OR=0.69;
was low, with only less than 65.0% of households con- 95% CI=0.66–0.73),46 but slightly greater than the study
suming meat/poultry in the last 24 hours. This association from Cambodia (OR=0.95; 95% CI=0.91–0.99).8 Unfor-
aligns with the results of study conducted in Cambodia tunately, the odds ratio from the study Malaysian popula-
that showed children who consumed animal source food tion was not available for comparison.27 Relative similari-
were less likely to be stunted.8 A previous study in Indo- ty in the strength of association between study in Cambo-
nesia highlighted the association between higher propor- dia,8 and our study might be attributable to the fact that
tions of total household expenditure on animal foods with both studies incorporated rural and urban settings, while
decreased likelihood of child stunting.41 The protective studies in Bangladesh,46 were based on rural population
effect of meat consumption was also confirmed in a study only. Rural settings had lower HDDS compared with ur-
focused on infants and toddlers in four diverse low- ban settings, as food availability in urban sites were more
income countries.42 abundant and diverse.29 Subsequent logistic regression
Another food group that showed a significant associa- with stratification of breastfeeding status showed that the
tion with child stunting was fish consumption. The effect protective effect of HDDS was only true for non-
size of fish group consumption was not aligned with pre- breastfed children (OR=0.88; 95% CI=0.08–0.99). This
vious studies where fish consumption was negatively as- finding might be related to the evidence that children who
sociated with child stunting.43 We hypothesize that due to were still breastfed were more likely to have limited di-
the high proportion of dried fish consumption in the pre- versity.46
sent study population, our results were masked by socio- The mean household dietary diversity score was 9.1,
economic status (SES) of the household. The cheap price which can be considered medium to high diversity. This
of dried fish, compared with freshwater fish, might ex- was almost twofold higher than a similar study using
plain this significant relationship. Our chi-squared analy- HDDS based on 12 food groups conducted in a rural
sis within each type of fish consumption also supported Cambodian population with mean HDDS score of only
our hypothesis where only dried fish was significantly 4.7.28 In the present study, six food groups were con-
associated with child stunting. In rural areas of Indonesia, sumed by more than 90% of the households: rice, vegeta-
consumption of small whole fish with bones, which are bles, legumes, egg, sweets and fat/oil. Rice is a staple
readily available sources of Fe, calcium, Zn and vitamin food of Indonesians that supplies around 70% of total
A, is often limited because of economic constraints.44 energy. Fat/oil was presumably used for cooking methods
Similarly, we posit that household spice consumption such as frying. Rice and oil are a poor source of protein,
showing a protective effect toward child stunting was which promotes child growth and prevents stunting. Leg-
actually an indication of SES. That is, higher SES house- umes were also widely consumed, and are rich in plant
holds were likely to include spices in their meal, com- protein, but may have lower bioavailability compared
pared with lower SES households. with animal protein. Hence, the ability of legumes to sup-
Our first logistic regression model showed a significant port child growth might not be as robust.
unadjusted association between HDDS and child stunting. Another food group that was highly consumed was
This relationship remained significant in the final model, vegetables. Vegetables are nutrient dense with a lot of
after adjustment for family size, food expenditure, breast- vitamins and micro-minerals essential for health. In Indo-
feeding status, total energy intake, and protein intake. A nesia, however, cooking methods such as boiling and
similar study using 12 food groups via HDDS in Cambo- steaming vegetables might diminish some of valuable
dia showed no significant association, despite a lower water-soluble vitamins when eaten. Fruits, as well as veg-
mean HDDS score, compared with our study.28 This dif- etables, are generally nutrient dense, particularly rich in
ference may be due to the study setting in rural Cambodia vitamins and minerals that help maintain body metabo-
that was more homogeneous than our study, which incor- lism. In our study, only 64.8% of households consumed
porated both urban and rural areas, as well as coastal and fruit in the last 24 hours, and this was not a significant
mountainous regions. Our study showed similar results to predictor of child stunting. The least commonly con-
another study in Cambodia where stunting was negatively sumed food group was dairy products, at 41.5% of the
associated with dietary diversity.8 That study, however, households. Dairy products are a prime source of calcium
used a measure of 6 food items from seven food groups, that is essential for bone development and growth. Be-
according to the WHO’s infant and young child feeding sides animal protein, dairy’s growth-stimulating effect is
(IYCF) model.45 The study by Darapheak et al used data believed to be insulin-like growth factor 1 (IGF-1) con-
from the Cambodia Demographic and Health Survey tained in these foods.47 A case-control study in Iran found
(CDHS) in 2005,8 which, in terms of population, is more significantly lower mean dairy intake of stunted children
representative than the McDonald et al study.28 Our re- than their normal counterparts.48 In our study, however,
sults were also in accord with a study in a Malaysian set- we did not find a significant relationship between dairy
ting that showed a negative association between HDDS products and stunting.
based on 15 food groups and HAZ.27 Outside Southeast The prevalence of child stunting in the current study
 Dietary diversity and child stunting 323

was 39.4% which was slightly higher than the national sign, should be implemented to confirm protective prop-
prevalence. Since the prevalence of child stunting was erties of dietary diversity and child stunting. Subsequent
way above 5%, it was less likely that the observed stunt- logistic regression stratified by breastfeeding status
ing was due to “healthy shortness” where child’s short- showed that the protective association between HDDS
ness was attributable solely due to inherited genes and not and child stunting was only significant among non-
because of inadequate nutrition or repeated bouts of in- breastfed children. It should be considered that stunting is
fection.49 Maternal short stature has been associated with a multifactorial disorder affected not only by dietary fac-
child stunting,50-52 and lends weight to the healthy short- tors, but also genetics, birth weight, and metabolic condi-
ness narrative. Unfortunately, we did not measure mater- tions.
nal height to test this hypothesis in our study.
In our study, food expenditure was significantly asso- Conclusion
ciated with child stunting, while other non-nutritionally In the scarcity of evidence in Indonesian setting, our
related factors such as family size, the source of drinking study added more weight to the hypothesis that better
water and geographical location were not significantly dietary diversity was associated with decreased likelihood
related. Our results conflicted with the study in rural Ar- of child stunting. These results confirm the widely ob-
menia that showed the association between non- served protective relationship between dietary diversity
nutritional factors particularly family size with child and child stunting. Hence, population interventions
stunting.53 Unsafe source of drinking water was a risk should focus on promoting food groups currently lacking
factor for child stunting in Tanzania,51 but not in our in maternal and child diets, including those rich in
study. We believe our results were affected by the fact growth-promoting nutrients such as dairy, poultry, and
that the majority of participants acquired drinking water meat.
from a well. Differences in ecological settings and geo-
graphical areas seem to be reflected in different dietary ACKNOWLEDGEMENTS
diversity, be it in rural28 or urban10,54 settings, as well as The first author (TM) received a scholarship to attend the course
in arid and semi-arid area.55 However, we failed to see an and workshop at the Harvard School of Public Health entitled
association between geographical locations and child ‘Analysis of Health and Nutrition Data from Low-Income
Countries’ and for data analysis and manuscript preparation
stunting in our study.
from the Higher Education Network Ring Initiative (HENRI)
The strength of our study includes the implementation Program, a partnership supported by a grant from the United
of the census block method to capture better representa- States Agency for International Development-Indonesia (Coop-
tion of diverse geographical areas in East Java Province erative Agreement AID-497-A-11-00002) to the Harvard School
including rural and urban setting. Further, we incorpo- of Public Health in partnership with, and with in-kind contribu-
rated different types of geographical locations from tions from, the SEAMEO Regional Centre for Food and Nutri-
coastal, agricultural low-lands, to mountainous areas. tion, University of Mataram, Andalas University, the Summit
Furthermore, the final logistic regression model was used Institute of Development, and Helen Keller International, with
to adjust for several covariates known to have confound- additional contributions from the Rajawali Foundation and Har-
ing effect on the association between dietary diversity and vard Kennedy School Indonesia Program.
child stunting, lending credence to the observed associa-
AUTHOR DISCLOSURES
tions. Last, this study was administered by trained-
None of the authors has a conflict of interest to report. This
interviewers in a manner that maternal literacy would not study was funded by the Food Security Board of the East Java
affect any of the outcomes. Province, Indonesia.
Some limitations should be noted for the interpretation
of our findings. This study was cross-sectional, so it does REFERENCES
not allow us to infer any causal relationship. The fact that 1. WHO. Childhood Stunting: Challenges and opportunities.
the HDDS calculation was based on a single 24-hours Report of a Promoting Healthy Growth and Preventing
dietary food recall is less robust compared with repeated Childhood Stunting colloquium. 2013/10/14 [cited 2015/9/6];
recalls, even for the common dietary measure of total Available from: http://www.cmamforum.org/Pool/Resour
energy.56 Some of food groups rich in growth promoting ces/WHO-childhood-stunting-challenges-and-opps-2014.pdf.
nutrients that have greater daily variation than energy 2. WHO. Reliability of anthropometric measurements in the
WHO Multicentre Growth Reference Study. Acta Paediatr
intake might be misreported in a single measure. These
Suppl. 2006;450:38-46. doi: 10.1080/08035320500494464.
include protein,57 zinc, iron, and calcium.58 Our HDDS 3. UNICEF. Improving Child Nutrition: The Achievable Im-
was measured based on maternal 24-hours dietary food perative for Global Progress. New York: UNICEF. 2013/4
recall, which may have lower agreement with child’s die- [cited 2015/9/6]; Available from: http://www.unicef.org/
tary diversity for food groups rich in micronutrients (e.g., gambia/Improving_Child_Nutrition_-_the_achievable_impe
dairy, meat/poultry, and fruits and vegetables).59 Although rative_for_global_progress.pdf.
we did not go to the extent of making a refined assess- 4. Save the Children. Nutrition in the first 1,000 days: State of
ment of HDDS to include “quantity” of at least 15 gr/day the World’s Mothers 2012. 1st ed. Westport, CT: Save the
for a food group to be considered “consumed”, we did Children. 2012/5 [cited 2015/9/6]; Available from:
base our assessment on a 24-hour dietary recall. Hence, http://www.savethechildren.org.nz/assets/599/Mothers%202
we were able to take into account the effect of potential 012%20Asia%20lr.pdf.
5. SUN. A Road Map for Scaling-Up Nutrition (SUN).
nutritional confounders such as total energy intake and
2010/9/17 [cited 2015/9/6]; Available from: http://scali
total protein intake. A more rigorous study, using ran-
domized controlled trial or prospective cohort study de-
324 T Mahmudiono, S Sumarmi and RR Rosenkranz

ngupnutrition.org/wp-content/uploads/pdf/SUN_Road_Map. 22. Truthmann J, Richter A, Thiele S, Drescher L, Roosen J,

pdf. Mensink GB. Associations of dietary indices with bi-
6. WHO. Essential nutrition action: Improving maternal, new- omarkers of dietary exposure and cardiovascular status
born, infant and young child health and nutrition. 1st ed. among adolescents in Germany. Nutr Metab (Lond). 2012;
Geneva, Switzerland: World Health Organization. 2013 [cit- 9:92. doi: 10.1186/1743-7075-9-92.
ed 2015/9/6]; Available from: http://apps.who.int/iris/bits 23. Li Y, Wedick N, Lai J, He Y, Hu X, Liu A et al. Lack of
tream/10665/84409/1/9789241505550_eng.pdf. dietary diversity and dyslipidaemia among stunted over-
7. Parlesak A, Geelhoed D, Robertson A. Toward the preven- weight children: the 2002 China National Nutrition and
tion of childhood under nutrition: diet diversity strategies us- Health Survey. Public Health Nutr. 2011;14:896-903. doi:
ing locally produced food can overcome gaps in nutrient 10.1017/S1368980010002971.
supply. Food Nutr Bull. 2014;35:191-9. 24. Azadbakht L, Mirmiran P, Esmaillzadeh A, Azizi F. Dietary
8. Darapheak C, Takano T, Kizuki M, Nakamura K, Seino K. diversity score and cardiovascular risk factors in Tehranian
Consumption of animal source foods and dietary diversity adults. Public Health Nutr. 2006;9:728-36. doi: 10.1079/
reduce stunting in children in Cambodia. Int Arch Med. PHN2005887.
2013;6:29. doi: 10.1186/1755-7682-6-29 25. Larrea C, Kawachi I. Does economic inequality affect child
9. Sunguya B, Poudel K, Mlunde L, Shakya P, Urassa D, Jim- malnutrition? The case of Ecuador. Soc Sci Med. 2005;60:
ba M et al. Effectiveness of nutrition training of health 165-78. doi: 10.1016/j.socscimed.2004.04.024.
workers toward improving caregivers’ feeding practices for 26. Bukania Z, Mwangi M, Karanja R, Mutisya R, Kombe Y,
children aged six months to two years: a systematic review. Kaduka L et al. Food insecurity and not dietary diversity is a
Nutr J. 2013;12:66. doi: 10.1186/1475-2891-12-66. predictor of nutrition status in children within semiarid agro-
10. Mallard S, Houghton L, Filteau S, Mullen A, Nieuwelink J, ecological zones in eastern Kenya. J Nutr Metab. 2014;
Chisenga M et al. Dietary diversity at 6 months of age is as- 907153. doi: 10.1155/2014/907153.
sociated with subsequent growth and mediates the effect of 27. Ey Chua EY, Zalilah MS, Ys Chin YS, Norhasmah S. Die-
maternal education on infant growth in urban Zambia. J Nutr. tary diversity is associated with nutritional status of Orang
2014;144:1818-25. doi: 10.3945/jn.114.199547. Asli children in Krau Wildlife Reserve, Pahang. Malays J
11. Drewnowski A, Henderson SA, Shore AB, Fischler C, Pre- Nutr. 2012;18:1-13.
ziosi P, Hercberg S. Diet quality and dietary diversity in 28. McDonald CM, McLean J, Kroeun H, Talukder A, Lynd LD,
France: implications for the French paradox. J Am Diet As- Green TJ. Household food insecurity and dietary diversity as
soc. 1996;96:663-9. correlates of maternal and child undernutrition in rural
12. Steyn NP, Nel JH, Nantel G, Kennedy G, Labadarios D. Cambodia. Eur J Clin Nutr. 2014;69:242-6. doi: 10.1038/
Food variety and dietary diversity scores in children: are ejcn.2014.161.
they good indicators of dietary adequacy? Public Health 29. Herrador Z, Perez-Formigo J, Sordo L, Gadisa E, Moreno J,
Nutr. 2006;9:644-50. doi: 10.1079/PHN2005912. Benito A et al. Low dietary diversity and intake of animal
13. FAO. Guidelines for measuring household and individual source foods among school aged children in Libo Kemkem
dietary diversity. 1st ed. Rome, Italy: Food and Agriculture and Fogera Districts, Ethiopia. PloS One. 2015;10:e0133435.
Organization of the United Nations. 2010 [cited 2015/9/6]; doi: 10.1371/journal.pone.0133435.
Available from: http://www.fao.org/fileadmin/user_upload/ 30. Herrador Z, Sordo L, Gadisa E, Moreno J, Nieto J, Benito A
wa_workshop/docs/FAO-guidelines-dietary-diversity2011.p et al. Cross-sectional study of malnutrition and associated
df. factors among school aged children in rural and urban set-
14. Wright M, Bentley M, Mendez M, Adair L. The interactive tings of Fogera and Libo Kemkem districts, Ethiopia. PloS
association of dietary diversity scores and breast-feeding sta- One. 2014;9:e105880. doi: 10.1371/journal.pone.0105880.
tus with weight and length in Filipino infants aged 6-24 31. Herrador Z, Sordo L, Gadisa E, Buño A, Gómez-Rioja R,
months. Public Health Nutr. 2015;18:1762-73. doi: 10.1017/ Iturzaeta JM et al. Micronutrient deficiencies and related
S136 8980015000427. factors in school-aged children in Ethiopia: a cross-sectional
15. Steyn N, Nel J, Labadarios D, Maunder EMW, Kruger HS. study in Libo Kemkem and Fogera districts, Amhara Re-
Which dietary diversity indicator is best to assess micronu- gional State. PloS One. 2014;9:e112858. doi: 10.1371/ jour-
trient adequacy in children 1 to 9 y? Nutrition. 2014;30:55- nal.pone.0112858.
60. doi: 10.1016/j.nut.2013.06.002. 32. Lanou AJ, Berkow SE, Barnard ND. Calcium, dairy prod-
16. Ochola S, Masibo P. Dietary intake of schoolchildren and ucts, and bone health in children and young adults: a reeval-
adolescents in developing countries. Ann Nutr Metab. 2014; uation of the evidence. Pediatrics. 2005;115:736-43. doi: 10.
64(Suppl 2):24-40S. doi: 10.1159/000365125. 1542/peds.2004-0548.
17. FAO. The double burden of malnutrition. Case studies from 33. Kloubert V, Rink L. Zinc as a micronutrient and its preven-
six developing countries. FAO Food Nutr Pap. 2006;84:1- tive role of oxidative damage in cells. Food Func. 2015;6:
334. 3195-204. doi: 10.1039/c5fo00630a.
18. Ng SW, Popkin BM. Impacts of China’s edible oil pricing 34. Iannotti LL, Zavaleta N, León Z, Shankar AH, Caulfield LE.
policy on nutrition. Soc Sci Med. 2008;66:414-26. doi: 10. Maternal zinc supplementation and growth in Peruvian in-
1016/j.socscimed.2007.09.015. fants. Am J Clin Nutr. 2008;88:154-60.
19. Popkin BM. The nutrition transition: an overview of world 35. Briend A, Khara T, Dolan C. Wasting and stunting--
patterns of change. Nutr Rev. 2004;62:140-3. doi: 10.1301/ similarities and differences: policy and programmatic impli-
nr.2004.jul.S140–S143. cations. Food Nutr Bull. 2015;36:S15-S23.
20. Labadarios D, McHiza ZJ, Steyn NP, Gericke G, Maunder 36. Balitbangkes. RISKESDAS 2013. 2013/12/1 [cited
EM, Davids YD et al. Food security in South Africa: a re- 2015/9/6]; Available from: http://www.academia.edu/
view of national surveys. Bull World Health Organ. 2011; 9134121/Laporan_Riskesdas_2013.
89:891-9. doi: 10.2471/BLT.11.089243. 37. Sari M, de Pee S, Bloem M, Sun K, Thorne Lyman A,
21. Deleuze Ntandou Bouzitou G, Fayomi B, Delisle H. Child Moench Pfanner R et al. Higher household expenditure on
malnutrition and maternal overweight in same households in animal-source and nongrain foods lowers the risk of stunting
poor urban areas of Benin. Santé. 2005;15:263-70. among children 0-59 months old in Indonesia: implications
 Dietary diversity and child stunting 325

of rising food prices. J Nutr. 2010;140:195-200S. doi: 10. among children in Tehran, Iran. J Health Popul Nutr. 2013;
3945/jn.109.110858. 31:202-10.
38. Lipoeto NI, Geok Lin K, Angeles-Agdeppa I. Food con- 49. Prendergast AJ, Humphrey JH. The stunting syndrome in
sumption patterns and nutrition transition in South-East Asia. developing countries. Paediatr Int Child Health. 2014;34:
Public Health Nutr. 2013;16:1637-1643. doi: 10.1017/S136 250-65. doi: 10.1179/2046905514Y.0000000158.
8980012004569. 50. Ozaltin E, Hill K, Subramanian SV. Association of maternal
39. Motbainor A, Worku A, Kumie A. Stunting is associated stature with offspring mortality, underweight, and stunting
with food diversity while wasting with food insecurity in low- to middle-income countries. JAMA. 2010;303:1507-
among underfive children in east and west Gojjam zones of 16. doi: 10.1001/jama.2010.450.
Amhara region, Ethiopia. PloS One. 2015;10:e0133542. doi: 51. Chirande L, Charwe D, Mbwana H, Victor R, Kimboka S,
10.1371/journal.pone.0133542. Issaka AI et al. Determinants of stunting and severe stunting
40. Swindale A, Bilinsky P. Household Dietary Diversity Score among under-fives in Tanzania: evidence from the 2010
(HDDS) for Measurement of Household Food Access: Indi- cross-sectional household survey. BMC Pediatr. 2015;15:
cator Guide VERSION 2. Washington, D.C.: Food and Nu- 165. doi: 10.1186/s12887-015-0482-9.
trition Technical Assistance Project (FANTA); 2006. 2006/9 52. Hambidge KM, Mazariegos M, Kindem M, Wright LL,
[cited 2015/9/6]; Available from: http://www.fantaproject. Cristobal-Perez C, Juárez-García L et al. Infant stunting is
org/sites/default/files/resources/HDDS_v2_Sep06_0.pdf. associated with short maternal stature. J Pediatr Gastroen-
41. Sari M, de Pee S, Bloem M, Sun K, Thorne Lyman A, terol Nutr. 2012;54:117-9. doi: 10.1097/MPG.0b013e3182
Moench Pfanner R et al. Higher household expenditure on 331748.
animal-source and nongrain foods lowers the risk of stunting 53. Demirchyan A, Petrosyan V, Sargsyan V, Hekimian K. Pre-
among children 0-59 months old in Indonesia: implications dictors of stunting among children aged 0-59 months in a ru-
of rising food prices. J Nutr. 2010;140:195-200S. doi: 10. ral region of Armenia: acase-control study. J Pediatr Gastro-
3945/jn.109.110858. enterol Nutr. 2016;62:150-6. doi: 10.1097/MPG000000000
42. Krebs N, Mazariegos M, Tshefu A, Bose C, Sami N, 0000901.
Chomba E et al. Meat consumption is associated with less 54. Savy M, Martin-Prével Y, Danel P, Traissac P, Dabiré H,
stunting among toddlers in four diverse low-income settings. Delpeuch F. Are dietary diversity scores related to the socio-
Food Nutr Bull. 2011;32:185-91. economic and anthropometric status of women living in an
43. Baye K, Guyot JP, Icard-Vernière C, Mouquet-Rivier C. urban area in Burkina Faso? Public Health Nutr. 2008;11:
Nutrient intakes from complementary foods consumed by 132-41. doi: 10.1017/S1368980007000043.
young children (aged 12-23 months) from North Wollo, 55. Segnon AC, Achigan-Dako E. Comparative analysis of di-
northern Ethiopia: the need for agro-ecologically adapted in- versity and utilization of edible plants in arid and semi-arid
terventions. Public Health Nutr. 2013;16:1741-50. doi: 10. areas in Benin. J Ethnobiol Ethnomed. 2014;10:80. doi: 10.
1017/S1368980012005277. 1186/1746-4269-10-80.
44. Wijaya-Erhardt M, Muslimatun S, Erhardt JG. Fermented 56. Ma Y, Olendzki BC, Pagoto SL, Hurley TG, Magner RP,
soyabean and vitamin C-rich fruit: a possibility to circum- Ockene IS et al. Number of 24-hour diet recalls needed to
vent the further decrease of iron status among iron-deficient estimate energy intake. Ann Epidemiol. 2009;19:553-9. doi:
pregnant women in Indonesia. Public Health Nutr. 2011; 10.1016/j.annepidem.2009.04.010.
14:2185-96. doi: 10.1017/S1368980011000954. 57. Ouellette CD, Yang M, Wang Y, Vance T, Fernandez ML,
45. UNICEF. Infant and Young Child Feeding: Programming Rodriguez N et al. Number of days required for assessing
Guide. New York: United Nations Children’s Fund usual nutrient and antioxidant intakes in a sample from a
(UNICEF). 2012/6 [cited 2015/9/6]; Available from: U.S. healthy college population. Nutrition. 2014;30:1355-9.
http://www.unicef.org/nutrition/files/Final_IYCF_programm doi: 10.1016/j.nut.2014.03.030.
ing_guide_June_2012.pdf. 58. Poslusna K, Ruprich J, de Vries J,H.M., Jakubikova M, van't
46. Rah JH, Akhter N, Semba RD, de Pee S, Bloem MW, Veer P. Misreporting of energy and micronutrient intake es-
Campbell AA et al. Low dietary diversity is a predictor of timated by food records and 24 hour recalls, control and ad-
child stunting in rural Bangladesh. Eur J Clin Nutr. 2010; justment methods in practice. Br J Nutr. 2009;101(Suppl 2):
64:1393-8. doi: 10.1038/ejcn.2010.171. S73-S85. doi: 10.1017/S0007114509990602.
47. Michaelsen K. Cow's milk in the prevention and treatment 59. Nguyen PH, Avula R, Ruel MT, Saha KK, Ali D, Tran LM et
of stunting and wasting. Food Nutr Bull. 2013;34:249-51. al. Maternal and child dietary diversity are associated in
48. Esfarjani F, Roustaee R, Mohammadi-Nasrabadi F, Esmaill- Bangladesh, Vietnam, and Ethiopia. J Nutr. 2013;143:1176-
zadeh A. Major dietary patterns in relation to stunting 83. doi: 10.3945/jn.112.172247.