Introduction

Indonesia still faces major challenges in alleviating malnutrition, especially among under-five children. Stunting is the common nutrition problems among this age group, which are closely related to inadequate nutrient intake from the diet. This situation has not improved significantly over the last decade, with stunting amongst under-five children being high i.e. 37.2% in 2013 [1] and 30.8% in 2018 [2]. Linear programming analysis of 2010 national dietary survey data amongst under-two children and two studies in sub-population groups of under-two children revealed that iron, zinc, calcium, niacin and folate are typical problem nutrients (PNs) in complementary feeding diets of young children in Indonesia [3].

The use of suitable locally-available foods has been emphasized by the World Health Organization (WHO) and UNICEF in the Global Strategy for Infant and Young Child Feeding [4] as well as the Food-based Dietary Guideline (FBDGs) in many countries. However, despite the potential for locally-available nutrient-dense foods to increase dietary adequacy, their usage in the nutrition-specific program must still be optimized.

WHO has provided “Optifood”, a software that uses Linear Programming (LP) to formulate food-based recommendations (FBRs) that is locally specific and affordable.These FBRs which take into account cultural diversity and differences in food availability are more likely to result in long-term improvements in dietary practices than general recommendations. On the other hand, food security can affect diet quality and therefore may influence the number and type of problem nutrients in the diet [5–7]. Dietary pattern also changes along the childhood period from infancy, complementary feeding period, and later period. Arecent study found that dietary quality declines with age and may begin as early as 1 year old [8], or even sooner [9].

Understanding whether number and type of PNs are affected by age and food security at the district level is critical for developiong a more effective approach for implementing nutrition-specific and nutrition-sensitive interventions to reduce stunting. This information is especially important in districts with a high prevalence and number of stunted children are high (the stunting priority districts) which have grown from 8 districts in 2017, to 100 districts in 2018, 160 districts in 2019, and eventually cover 514 districts in Indonesia in 2020–2024 [10]. The aim of this study is to identify PNs in the diet of under-five children in the 37 stunting priority districts in Indonesia and determine whether the number of PNs were associated with age of the children and the district’s food security status.

Methods

This study used secondary data from the Ministry of Health of Indonesia’s 2016 National Nutrition Consumption Monitoring Survey (Survei Pemantauan Konsumsi Gizi) for under-five children. The sample was randomly selected from households with children under the age offive, using Probability Proportion to Size (PPS). Dietary intake data was collected using a single 24-hour dietary recall.

Data of 11,040 children aged 6–59 from 37 selected districts in 33 provinces in Indonesia were checked for completeness. Children with missing data on dietary intake, age, or breastfeeding status were excluded from the analysis. Out of 37 selected districts, 32 districts were selected based on the following criteria: (1) the districts were included in the list of 160 stunting priority districts according to the Government of Indonesia stunting reduction program in 2019; (2) the districts had the highest stunting prevalence within each province; and (3) the local specific FBRs of the districts has not been developed. Pasrticularly for East Java Province, five districts were selected based on those three criteria and represented different geographic typologies i.e. highlands, lowlands, and coastline.

Linear programming analysis

The LP analysis was performed using linear optimization (Excel’s solver function) in Microsoft Excel software or WHO Optifood V4.0.14.0 (June 16^th, 2015). LP-Optifood analysis was done in several steps as follows: 1. preparing the list of foods consumed by subjects to be entered into Optifood software; 2. preparing the list of food groups and sub-groups, including the frequency of consumption; 3. performing the "check diet" for feasibility analysis of the FBR; 4. analysing the data to get 2-best diets (optimal-case-scenario) which was closest to existing food pattern i.e. median weekly frequency of food groups (Best Diet FP) and which may deviate from the existing food pattern (Best Diet No-FP), the worst-case (%RNI when each nutrient was minimized) dan the best-case (%RNI when each nutrient was maximized).Based on Best Diet No-FP, Best-case and Worst-case each nutrient was then categorized as either absolute problem nutrient (PN_a), partiap problem nutrient (PN_p), dietary inadequate (DI) or dietary adequate (Table 1); 6. identifying the proposed nutrient-dense foods of sub-groups to fill the nutrient gaps; 7. determining the most optimal FBRs by comparing several FBR alternatives. For the nutritional adequacy target, the Indonesian Recommended Nutrient Intake (Angka Kecukupan Gizi) 2019 [18] was used as reference (Table 2).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Criteria used in LP analysis for defining problem nutrients and dietary inadequacy, based on %RNI.

https://doi.org/10.1371/journal.pone.0314552.t001

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 2. Nutrient recommendation based on Indonesian RNI¹ used in the LP analyses.

https://doi.org/10.1371/journal.pone.0314552.t002

Results

In total, dietary intake data from 9,847 under-five children were included in LP analysis. Number of dietary data varied with a range of 92 to 308 under-five children per district (average 266 children). The number of food items included in LP analyses varied greatly based on the dietary intake data. The minimum number of food items was 16 amongst 6-11mo infants in Nunukan (North Kalimantan) and the maximum number was 130 amongst 36-59mo children in Bantul (DI Yogyakarta), Table 3.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 3. Number of food items identified measure by a single 24 hour dietary recall for LP analyses, by district and age group.

https://doi.org/10.1371/journal.pone.0314552.t003

Majority of the 37 districts had partial or absolute PNs in their food pattern. Infants (6-11mo) had the largest percentage of districts with at least one PN (35/37 = 95%). While the proportions of district with PNs decreased in the second year of life (25/37 = 68%), the proportion increased toward the end of the under-five year period (28/37 = 76% in 24-35mo and 33/37 = 89% in 36-59mo children). The top-3 PNs in each age group were: iron, folate, vitamin B12 (6-11mo); folate, calcium, zinc (12-23mo); folate, calcium, vitamin C (24-35mo); and calcium, folate, vitamin C (36-59mo), Fig 1. Besides the PNs, some nutrients were not adequate (dietary inadequate) as its minimized RNI is still below the EAR (<65% RNI). On average, most districts had 1–2 PNs and additional 1–3 dietary inadequacy (Tables 4 and 5).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. Top 3 problem nutrients in the 37 stunting priority districts, by age group.

https://doi.org/10.1371/journal.pone.0314552.g001

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 4. Identified problem nutrients and dietary inadequacy in 6-23mo children, by district.

https://doi.org/10.1371/journal.pone.0314552.t004

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 5. Identified problem nutrients and dietary inadequacy in 24-59mo children, by district.

https://doi.org/10.1371/journal.pone.0314552.t005

Out of the 37 districts, 31 districts (84%) were categorized as food secure and 6 districts (16%) as food insecure based on FSVA priorities. Nineteen districts (51%) had at least 6 PNs in all the four age groups based on LP analyses. Based on combined categories of food security and number of PNs, almost half of districts (43%) were food secure but had at least 6 PNs (Fig 2). The independent t-test between number of PNs and food security status was non-significant (p = 0.501). The number of nutrients which were adequate varied by age group, with the least number in the 35-59mo age group (0–1 nutrients only) followed by 24-35mo (1 nutrient), 12-23mo (2 nutrients), and 6-11mo (3 nutrients).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 2. Distribution of 37 stunting priority districts by food security status and number of problem nutrients in under-five children.

• Categorization of food security status based on Food Security Vulnerability Atlas (FSVA) 2018: 1 = severe food insecure; 2 = moderate food insecure; 3 = mild food insecure; 4 = low food secure; 5 = middle food secure; 6 = high food secure, • Categorization of problem nutrient, based in total number of PN in all four age groups with median [6] as the cut-off, low PN< 6 problem nutrients; high PN ≥6 problem nutrients.

https://doi.org/10.1371/journal.pone.0314552.g002

Nutrient-dense foods with potential to improve dietary adequacy of the problem nutrients were identified from LP analyses and were included in the final FBRs (S1–S5 Tables) After incorporating these nutrient-dense foods in the FBRs, the number of districts with dietary inadequacy decreased for the top-3 problem nutrients in each age group, with the exception of iron for infants (Fig 3).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 3. Number of districts with dietary inadequacy before and after nutrient-dense foods (NDFs) were included in the diet, based on LP analysis in Module 3 (worst-case scenario).

https://doi.org/10.1371/journal.pone.0314552.g003

Discussion

This study is the first study to identify problem nutrients of under 5-years-old children using national data from 37 stunting priority districts in Indonesia. Among the top 3 PNs, folate was the PNs in all age groups, iron was considered problem in almost all districts at earlier age (6-11mo), while calcium was considered problem at later age (11mo and older). However, the number of PNs was not associated with the district’s food security status.

Number of districts with PNs was highest among youngest infants aged 6-11mo. This finding was consistent with our prior analysis using the 2013 Indonesian Basic Health Survey data, which revealed that children 6-11mo had four PNs (iron, zinc, calcium, niacin) as compared to no PN in the older children aged 12-23mo [3]. However, in these 37 stunting priority districts, we also found that older children above 24mo had more PNs than their 12-23mo counterparts, despite more variety of foods they consumed [20].

The fact that all districts had PN(s) and/or dietary inadequacy in their food patterns suggests that feeding practices needs to be improved, particularly in terms of intakes of animal source foods, fruits, and vegetables which are rich in iron, calcium, zinc, B12, folate, and vitamin C. Recent review of 23 studies employing diet modelling on children aged 6-23mo found that the diets are inadequate in iron, zinc, and calcium. The review also found that requirements for vitamin A, thiamine, riboflavin, niacin, folate, and vitamin C were not always met [21]. Previous study reported that consumption of plant based foods (e.g. rice, maize, legumes), fibre, and polyunsaturated fat, which was common among children aged 6–24 months old, was inversely associated with nutrient density of the complementary diet for vitamin C, calcium, and iron [22].

Unexpectedly, folate was a PN in all age groups, a finding which was not reported in previous study that used a nationally representative data from Indonesia [3]. Folate is naturally present in a wide variety of foods, including vegetables (especially dark green leafy vegetables), fruits and fruit juices, nuts, beans, eggs, and rice [13, 23], which are commonly found in almost all areas in Indonesia. The finding suggests that consumption of fruits and vegetables is not yet optimal and was in line with national figure from Indonesian Basic Health Survey Report 2018 that over 90% of Indonesian population did not consume fruits and vegetables adequately in terms of frequency and portion [2].

Iron is the most common PN among infants aged 6–11 months, which confirmed findings in previous studies [3, 21]. It is known that iron concentrations in breastmilk decrease as infants grow older [24], therefore more iron from complementary feeding is required to meet the RNI. The challenge for 6–11 mo children to meet all of their nutrient requirements from locally available and typically unfortified foods has been primarily caused by predominantly cereal-based and monotonous diet, resulting in low dietary diversity, low nutrient density, and poor micronutrient bioavailability [25]. Although infants who consumed fortified infant products and fortified formula milk had significantly higher intakes of iron compared to infants who did not consume these products, intakes were still inadequate for iron, due to inadequate nutrient densities of fortified infant foods, small quantities consumed, and/or infrequent consumption of these products [9, 26].

Across all age groups, folate and calcium were the most problematic nutrients. Inadequate intake of folate during infancy and early childhood has negative consequence on brain development [27]. The absence of calcium as PN at 6-11mo is most likely due to breastmilk, which we included in the LP analyses in accordance with the public health message of breastfeeding on demand. According to the Indonesian Basic Health Survey 2018, the national rate of exclusive breastfeeding amongst infants under the age of 6 months was only 37.3% [2]. This suggests that dietary adequacy of calcium may have been over-estimated in our analyses and that calcium may still be PN given the actual rate of exclusive breastfeeding.

The LP analysis of these 37 districts revealed that zinc, a type-II nutrient associated with stunting, was not identified as PNs in most districts, a finding which was in contrast with previous study [21]. In our analysis, zinc was found as the PNs only in some districts, mainly amongst the 36-59mo children. This suggests that food basket and food pattern of these under-five children are generally sufficient to ensure adequacy of zinc intake. Fish was zinc-rich food which was commonly consumed in many districts in Indonesia and fish consumption has increased significantly due to promotion to consume fish conducted by Ministry of Fishery (the Gemarikan which means “Let’s Eat Fish” movement). In Indonesia, fish consumption rose from 38.14 kg/capita.year in 2014 to 55.95 kg/capita.year in 2019 [28].

Our finding was consistent with previous study that dietary quality starts to decline as early as one year old [3]. In younger age, children experienced variety of food for the first time. As children get older, they gain greater control over what they eat, which can have favorable or negative impact on dietary quality [29, 30]. This emphasizes the need of promoting healthy and balanced diet beyond the complementary feeding period. Nutrition education in early childhood education centres can be the potential channel to establish healthy eating habit amongst toddlers who begin to make dietary choices. One of the approaches can be to teach children to identify and categorize food which can be consumed sometimes (i.e., unhealthy) or anytime (i.e., healthy) as previous study found that pre-schoolers who were better at identifying food as healthy or unhealthy were more likely to select healthier option [22].

The finding that food insecurity had no significant influence on the number and type of PNs suggest that the problem was more with food consumption (i.e. less favourable feeding and dietary practices) rather than food availability. In addition, parents may compromise their food intake to protect their younger children from food shortage in the food insecure household and earlier study found that food security has linear relationship with diet quality in adults while pattern was inconsistent in children [20].

Most of the PNs found in these 37 districts were absolute PNs, which suggest limitation in the current food basket of the children. However, this does not necessarily mean that the community food basket is inadequate; rather, the foods with potential to increase of the PNs may be available locally but are not commonly fed to children. In our earlier study in East Lombok we promoted anchovy which was not traditionally included in children’s diet but was locally available and consumed by older family members. The result showed that introduction of anchovy powder can increase calcium intake which was the PN amongst the under-two-year-old children, suggesting that it is possible to introduce nutrient-dense food from the family food basket into the child’s food basket [11]. Furthermore, our current LP analysis has shown that the introduction of nutrient-dense foods using traditional recipes of composite dishes (dishes containing three or more food groups) have also been shown to meet the requirement of the PNs iron, folate, vitamin C in several districts across different regions/islands.

The LP analyses conducted in this study were the first to identify the PNs in 33 out of 34 provinces in Indonesia. LP analyses were also conducted by trained nutritionists who were familiar with the local food patterns and therefore could put the FBRs into a local context. The limitations of the study were the small sample size (<50) in some district-age groups, especially for the 6-11mo infants, which may limit the number of identified food items consumed. Nonetheless, we used the most recent national representative dataset, which can still be used to develop local specific FBRs.

Since the aim of LP analyses is optimization, not all FBRs can ensure dietary adequacy of all nutrients. Dietary adequacy was particularly difficult to meet for iron in 6-11mo infants; for folate in 12-23mo children and 24-35mo children; and for calcium and folate in 36-59mo children. LP analyses indicated that the gap for these nutrients in the worst-case scenario (minimized scenario) ranged between 30–40%RNI; or equivalent to nutrient gap of 25–35%RNI when compared with 65%RNI (≈EAR level). In our earlier study, we used shredded chicken liver, shredded fish, and anchovy powder to increase intakes of the PNs iron, zinc, and calcium [31]. We also found that mungbean was rich in folate. Using these nutrient-dense foods, the ‘nutrient gap’ identified in our analyses of these 37 districts can be achieved by providing shredded liver (abon hati) of 16–19 grams/day (for iron), dried anchovy powder of 14–17 grams/day (for calcium), or mung beans of 22–27 grams/day (for folate), all ofwhich are feasible portions for children. Several nutrient-dense foods identified from the LP analyses in the current study can be promoted further to increase intakes of the typical PNs amongst the under-five children. These include fish, seaweed, sugar-palm fruit, and mung beans; as well as local recipes in form of composite dishes which included starchy staples, animal source protein, and vegetables.

Our findings showed that problem nutrients and dietary inadequacy were typical in diet of under-five children, even in food-secure areas. Locally available nutrient-dense foods have the potential to improve dietary adequacy in most districts for most problem nutrients, and should thus be promoted through FBRs (the nutrition specific intervention) as well as by ensuring local availability and access such as through agriculture, fishery, and market intervention (the nutrition sensitive intervention). Additional intervention(s), particularly for iron in infants, are required to ensure dietary adequacy.

Supporting information

Acknowledgments

We are grateful to the Directorate of Nutrition, Ministry of Health Indonesia for providing access to dietary data for LP analysis and to the academic partners who developed the draft of FBRs in LP Optifood analyses: Health Polytechnic Ministry of Health of Indonesia (Poltekkes Kemenkes RI), including Poltekkes Kemenkes Jakarta II (Dara Humaira, Ratih Puspitaningsih), Kendari (Sri Rizki), Makassar (Manjilala), Malang (Sugeng Iwan), Mataram (Irianto), Medan (Tetty H. Doloksaribu), Padang (Edmon), Pangkal Pinang (Ori P. Enardi), Pontianak (Martinus Ginting), Riau (Muharni), Semarang (Heni Hendriyani), Surabaya (Annas Buanasita), Sorong (Anjar Briliannita), Tasikmalaya (Priyo Sulistiyono), and Yogyakarta (Primiaji), as well as Universitas Airlangga (Dini R. Andrias), Universitas Brawijaya (Nia Wirawan, Ilmia Fahmi), Universitas Diponegoro (Sulistyawati, Naintina Lisnawati), and Universitas Udayana (Kadek T. Adhi).