Abstract
Objective
Stunting impairs child development and long-term economic productivity. In Indonesia, prevalence has declined substantially but progress has slowed. We aim to understand the contextual factors of childhood stunting in two rural communities in the Sumedang district, West Java, which is prioritised by health authorities due to its high stunting prevalence.
Methods
We used an explanatory sequential mixed-methods design. A cross-sectional household survey (172 children under 5 years of age) informed the adaptation of interview guides for a qualitative phase consisting of 34 in-depth interviews with caregivers and community key informants, which were analysed inductively and integrated with the survey findings in a joint display.
Results
Measured stunting prevalence was similar in the two communities (20.5% and 29.8%) and higher than the routine district estimates. Exploratory models showed common associations such as with household wealth, consistent with the wider literature. The interviews uncovered important resources and risks. These included informal nutrition education networks, delegated caregiver arrangements, coping with food insecurity through borrowing, credit, subsistence farming and fishing and health-worker accompaniment and village-funded social assistance programmes as buffers.
Conclusions
Stunting in these communities was high and socially patterned. Sustained reductions will require ongoing intergenerational policies that address structural socioeconomic disadvantage and strengthen caregiving environments, rather than stand-alone nutrition or health interventions.
Keywords: Child Health, Growth, Qualitative research, Cross-Sectional Studies
WHAT IS ALREADY KNOWN ON THIS TOPIC
Stunting in Indonesia has decreased in recent years but remains a serious public health concern. Previous research has shown that poverty, caregiving practices and child illness play important roles.
WHAT THIS STUDY ADDS
Using an explanatory sequential design, this study pairs measurement with in-depth community accounts to show why stunting persists in two rural communities. Where the two communities were expected to differ, they carried a similar burden, and the qualitative work shows how distinct local configurations of poverty and health-system access converge on the same outcome—evidence that the drivers are structural and contextual rather than narrowly nutritional.
HOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE OR POLICY
Because stunting risk arises from locally specific combinations of poverty and health-system access, responses are likely to be more effective when tailored to community context and developed with local stakeholders, rather than applied uniformly.
Introduction
Stunting in children is associated with impaired child development and reduced economic capacity in adulthood.1 Globally, in 2022, stunting prevalence was 22.3% among children under 5 years of age.2 In the same year, the prevalence of stunting in Indonesia was 21.6%.3 Indonesia has made substantial gains in stunting since 2007, when the prevalence was 37%.3 However, recent progress has slowed, and a nuanced understanding of the ongoing social factors affecting stunting in Indonesia is needed.
To explore ongoing factors associated with childhood stunting in West Java, Indonesia, we conducted a mixed-methods study in Sumedang, a district with one of the country’s highest stunting prevalence rates (27.6%) according to the Ministry of Health Data.3 Using separate district government data, we selected two communities with contrasting reported stunting prevalence (3% and 15%),4 5 intending to use this approach to examine distinguishing features. However, as reported below, our measured prevalence in both communities was similar and higher than official estimates; because the anticipated contrasts were not borne out, we drew heavily on the qualitative phase of the study to understand shared contextual drivers of stunting across both settings.
Methods
Overall study design and setting
We conducted an explanatory sequential mixed-methods design to understand the risk and protective factors of childhood stunting in two communities in Sumedang, West Java, selected on the basis of lower (3%, Community 1) and higher (15%, Community 2) stunting prevalence according to the official statistics provided by the Health District. The quantitative phase estimated community stunting prevalence and explored associated factors. Findings from the quantitative phase then informed an explanatory qualitative phase examining contextual drivers of stunting. The overall data collection scheme is presented in figure 1. The project was developed with the support of the District Government of Sumedang Health Office. Community 1 has approximately 40 000 inhabitants, while Community 2 has approximately 25 000 residents. According to regional statistics, the density of healthcare facilities in Community 1 is lower than that in Community 2 (3 vs 17 facilities).6 7 Community 1 is farther from the district hospital (27.2 vs 2.8 km).8 9
Figure 1. Explanatory sequential mixed-methods data collection.
Quantitative methods and data analysis
Sampling and power calculation
Figure 1 gives an overview of the subject recruitment process. To estimate the prevalence of stunting among children under 5 years of age in each community, we conducted a cross-sectional survey using simple random sampling (online supplemental table S1). The study period, from recruitment to data collection, spanned from May to August 2025. Sample size calculations were based on assumed stunting prevalences of 3% in Community 1 (2069 children under 5 years of age) and 15% in Community 2 (1381 children under 5 years of age), derived from district figures.4 5 Allowing for the specified precision (5% in Community 1 and 10% in Community 2) and an anticipated refusal rate of up to 50%, the required sample sizes were 84 and 98 children, respectively.10 The quantitative phase was thus powered to estimate prevalence with prespecified precision, while association analyses were exploratory.
We obtained the complete list of eligible children from the Sumedang District Health Office5 database for each community. We then used simple random sampling to select the required number of children at each site and invited families to participate. Recruitment was conducted by community health workers/cadres and village midwives, alongside the first author and trained research assistants.
Outcome measures
Child anthropometric measures were obtained by trained research assistants using standard anthropometric equipment during home visits. Length/height-for-age Z-scores (HAZ) were calculated in StataBE 18 using the zscore06 package, based on the WHO Child Growth Standards. The underlying dataset, data dictionary and reproducible code scripts are available in Harvard Dataverse.11 Stunting was defined as a length/HAZ of ≤−2 SD.12
Predictor variables
Sociodemographic data on mother-child dyads were collected via surveys (table 1) using Research Electronic Data Capture (REDCap)13 14 for data collection and management. Variables included child characteristics (age, sex and illness or diarrhoea within the past 3 months), caregiver characteristics (age, education, occupation) and household characteristics (wealth status and food security). Household wealth status was assessed using the EquityTool, which provides a continuous wealth index score15 and categorical wealth quintiles (quintile 1, least wealthy).16 Continuous scores can be negative (range observed in our study: from −0.54 to 0.98), with lower scores reflecting poorer households. Food insecurity was assessed using the Food Insecurity Experience Survey (FIES) global standard scale, which comprises eight questions to categorise households based on item severity values (online supplemental table S2).17 Questions regarding child health and healthcare access were drawn from the Indonesian Nutritional Status Survey (Survei Status Gizi Indonesia).18 All instruments used were validated and widely accepted standard tools, supplemented by straightforward, study-specific demographic questions. The survey instrument was piloted with three local research assistants who had extensive experience working within the participating communities.
Table 1. Characteristics of surveyed children, caregivers and households in communities 1 and 2.
| Characteristic | Community 1 (n=78) |
Community 2 (n=94) |
|---|---|---|
| Child characteristics | ||
| Child age in months, median (IQR) | 33 (19 to 45) | 30.5 (18 to 45) |
| Child sex, N (%) | ||
| Female | 36 (46.2) | 46 (48.9) |
| Length/HAZ, median (IQR) | −1.2 (−1.9 to −0.3) | −1.1 (−2 to −0.3) |
| Stunted, N (%) | 16 (20.5) | 28 (29.8) |
| Diarrhoea as the cause of illness in the past 3 months, N (%) | 8 (10.3) | 7 (7.5) |
| Caregiver characteristics | ||
| Primary caregiver age in years, median (IQR) | 30 (27 to 34) | 32.5 (26 to 38) |
| Primary caregiver, N (%) | ||
| Mother | 72 (92.3) | 82 (87.2) |
| Grandmother | 5 (6.4) | 8 (8.5) |
| Aunt | 0 (0) | 3 (3.2) |
| Legal guardian | 1 (1.3) | 1 (1.1) |
| Primary caregiver education, N (%) | ||
| Elementary school | 8 (10.3) | 25 (26.6) |
| Junior high school | 28 (35.9) | 29 (30.9) |
| Senior high school | 24 (30.8) | 22 (23.4) |
| Technical high school | 8 (10.3) | 11 (11.7) |
| College or university | 10 (12.8) | 7 (7.5) |
| Caregiver primary occupation, N (%) | ||
| Housewife | 60 (76.9) | 70 (74.5) |
| Market | 9 (11.5) | 14 (14.9) |
| Professional | 6 (7.7) | 3 (3.2) |
| Daily wage worker | 2 (2.6) | 1 (1.1) |
| Farming or agriculture | 0 (0) | 2 (2.1) |
| Household help | 0 (0) | 4 (4.3) |
| Others* | 1 (1.3) | 0 (0) |
| Household characteristics | ||
| Individuals in a household, people (mean±SD) | 4.6±1.1 | 4.5±1.3 |
| Wealth quintile distribution, N (%) | ||
| 1 | 0 (0) | 1 (1.1) |
| 2 | 2 (2.6) | 17 (18.1) |
| 3 | 14 (17.9) | 22 (23.4) |
| 4 | 31 (39.7) | 20 (21.3) |
| 5 | 31 (39.7) | 34 (36.2) |
| Wealth index score, median (IQR)† | 0.5 (0.3 to 0.7) | 0.3 (−0 to 0.7) |
| FIES score‡ | 0.04 (−0.8–0.6) | 0.04 (−0.8–0.6) |
Other occupations: daily wage workers, farming or agriculture, housekeeping, market or street vendors and no occupation.
Wealth index score and quintiles: calculated using the EquityTool Indonesia instrument.
Calculated using the FIES Global Standard Scale. Food insecurity was assessed using the 8-item scale, and the score (FIES θ score) was estimated using a Rasch one-parameter logistic model and entered into the regression as a continuous variable.
FIES, Food Insecurity Experience Survey; HAZ, height-for-age Z-score.
Quantitative data analysis
Descriptive statistics and comparisons between the two study communities were assessed using the Wilcoxon rank-sum test, two-sample t-test, χ2 test or Fisher’s exact test, as appropriate. Associations between stunting (HAZ≤−2) and predictor variables were assessed using bivariate and multivariable logistic regression. Associations between continuous HAZ and the predictors were assessed using both simple and multiple linear regression. For logistic regression, model fit was assessed using tests for model specification, multicollinearity and the Hosmer-Lemeshow goodness-of-fit test. For linear regression, we performed tests of model specification, multicollinearity and heteroscedasticity. For all models, variables were selected using a theory-driven directed acyclic graph19 constructed from expert knowledge and literature review (online supplemental figure S1).
Qualitative methods and data analysis
Qualitative instrument development
We designed semistructured, in-depth interview guides to explore stakeholders’ perceptions of healthcare infrastructure and factors associated with stunting (online supplemental appendix S1). Given our sequential quantitative-qualitative design, we adapted these guides to include factors uncovered in the quantitative phase. Key modifications targeted important themes from the quantitative survey, such as local financial aid programmes, the impact of a recent dam construction project and community care management.
Qualitative population and sampling
We recruited community key informants (doctors, nutritionists, community health workers/cadres, village midwives and local government officers) (n=7 per community; n=14 total) through purposive sampling to capture a range of experiences and work roles. In addition, we purposively sampled five caregivers of children without stunting and five caregivers of children with stunting per community (n=20 total).
Qualitative data collection
The first author is fluent in Bahasa and conducted the interviews in this language, except for two instances where participants preferred to speak Sundanese. A research assistant translated these two interviews into Bahasa after they were transcribed. Interviews were conducted either at a community health centre or at the participant’s home, according to the participant’s preference. Interviews lasted 17–77 min. All interviews were audio-recorded with consent and subsequently transcribed. Participants were approached in person through the community health workers/cadres and village midwives who had assisted with recruitment. None of the participants approached for an interview declined to participate. Interviews were conducted with the participant and interviewer present; however, for caregivers who were interviewed, health workers/cadres/village midwives were also present without interfering with the interview. No repeat interviews were conducted. Field notes were taken during and after each interview. Interviews continued until the study team judged that no new themes were emerging. Transcripts were not returned to participants for comment.
Research team and reflexivity
All interviews were conducted by the first author (DNE), an Indonesian physician with graduate training in quantitative, qualitative, and mixed-methods research. DNE had no prior relationship with study participants. At enrolment, participants were informed of the study’s purpose, the interviewer’s background as a researcher and the reasons for undertaking the research.
Qualitative data analysis
The first author first open-coded 34 interviews and used thematic analysis to develop a codebook. The codebook was cross-checked by PR and AVS, and disagreements were resolved through consensus. Subsequently, all interviews were coded in Dedoose20 by the first author. Coding was cross-checked by the remaining authors. The final thematic meta-categories were developed inductively. The codebook with code frequencies is provided in online supplemental tables S6 and S7. The analysis followed an inductive thematic-analysis approach.21 Participants did not check transcripts or findings. Instead, findings were triangulated across participant groups and against quantitative results using a joint display.
Joint display
To represent the integration of qualitative and quantitative findings, we created a joint display highlighting pertinent findings from both studies and exploring divergences and convergences across the results.
This mixed-methods study is reported in accordance with the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE)22 23 guideline for the quantitative component (online supplemental B), the Standards for Reporting Qualitative Research (SRQR)24 (online supplemental C) and the Consolidated Criteria for Reporting Qualitative Research (COREQ)25 for the qualitative component (online supplemental D). The mixed-methods integration is additionally reported against the Good Reporting of a Mixed Methods Study (GRAMMS) criteria.26 Completed COREQ and GRAMMS checklists are provided in online supplemental D.
Results
Eligible participants were screened, with 84 identified in Community 1 and 98 in Community 2. Of these, 10 participants declined to participate (Community 1, n=6; Community 2, n=4). 78 participants from Community 1 and 94 participants from Community 2 were recruited into the study (figure 1). Primary participant characteristics and intercommunity comparisons are presented in table 1. There were few differences in the sociodemographic profile of the two communities, although the community wealth quintile was significantly lower in Community 2. The two communities were similar in child age, child sex and household size. Community 2 carried a heavier burden of poverty: 18 households (19.1%) fell in the two lowest wealth quintiles compared with 2 (2.6%) in Community 1, and caregivers more often had only elementary schooling (26.6% vs 10.3%). Importantly, measured stunting prevalence was higher than expected and similar in the two communities (Community 1, 20.5%; Community 2, 29.8%), in contrast to prior district estimates.
Exploratory association analyses
Full unadjusted and adjusted models are provided in online supplemental tables S3 and S4. No unexpected associations emerged: the directions of the age and household wealth were consistent with the extensive stunting literature, and most CIs were wide and crossed the null, as expected for a survey powered to estimate prevalence rather than to test associations. Descriptively, however, the quantitative phase revealed several context-specific signals that we used to design our qualitative questions. These included economic stress in a subset of households (visible in the wealth quintile distribution); reports of meal-skipping, food borrowing and related coping strategies in the food insecurity questionnaire (online supplemental repository tables R1–R5) and field observations during the survey on the impact of a recent dam-construction project and families’ reliance on shared or delegated child care.
Qualitative findings
We conducted semistructured interviews to explore factors associated with stunting with caregivers (n=20) and key informants (nurses, physicians, midwives, community health workers/cadres and local government, n=14). Interviews were distributed equally between the two communities. Demographic characteristics of interview participants are summarised in online supplemental table S5. Five major categories emerged from our thematic analysis, described in detail below. Code frequencies disaggregated by community and child stunting status are presented in online supplemental tables S6 and S7; unless otherwise noted, no significant differences in code frequencies across these categories were observed.
Drivers and impact of education on nutrition
Some stakeholders in both communities emphasised the importance of caregiver nutrition-related knowledge and education contributing to adequate breastfeeding, child feeding practices and nutritional status (5 of 14 stakeholders’ interviews):
Based on my observations, some already have good breastfeeding practices because their education is sufficient; most have completed at least through Senior High School or college.
- Community key informant #9, 3 years of experience, Community 1
Some caregivers agreed, with 4 of 20 mentioning the importance of education despite financial constraints, which were more frequently mentioned by caregivers with stunted children:
Higher education is important, even though household income is low, and we face hardships.
- Caregiver #8 of a stunted child, 33 years old, Community 2
At the same time, key informants frequently described how financial constraints and gender norms constrained access to education. For example:
There is a prevailing mindset within the community. Some parents think, “Why send girls to high school when they’ll just end up in the kitchen anyway?”
- Community key informant #4, 8 years of experience, Community 2
To address the need for maternal nutrition education, public health centres offer classes. This theme was noted by almost all community key informants in both communities (11 of 14).
We have a nutrition class where we teach parents how to prepare complementary foods based on the child’s age.
- Community key informant #5, 9 years of experience, Community 2
7 out of 20 caregivers reported taking informal classes arranged by the community health centres, including classes on pregnancy and nutrition.
Yes, I joined them [pregnancy class]. Additionally, we will learn how to breastfeed and prepare complementary foods for our children.
- Caregiver #10 of a stunted child, 28 years old, Community 2
Prenatal factors affecting growth
About a third of community key informants mentioned antenatal factors that contribute to prematurity and increase the risk of stunting.
Some mothers had preeclampsia that may lead to premature birth and low birth weight, and eventually they can also be stunted.
- Community key informant #13, 11 years of experience, Community 1
However, there were also contrary experiences:
My three children were all born premature, and their birth weight was low, but they were never diagnosed as stunted.
- Caregiver #11 of a non-stunted child, 25 years old, Community 1
Additionally, 8 of 14 key informants mentioned that maternal age may contribute to the risk of stunting, especially if mothers are too young to have a child.
It [the mother] is young, which could be related to the issue of stunting. If a pregnant woman is only 14 years old, this likely affects the birth of the baby, correct?
- Community key informant #10, 1 year of experience, Community 1
Maternal supplementation with iron and folic acid tablets was also commonly mentioned. However, 11 of 20 caregivers noted barriers to adherence, including side effects and forgetfulness. For example:
I was afraid [of drinking IFA] because I was afraid my baby would be too big. When I was 7 months pregnant, I stopped taking it. I was scared, but the baby was eventually born small, only 2.7 kg.
- Caregiver #18 of a stunted child, 25 years old, Community 1
Feeding practice and food experience
Some key informants (6 of 14) mentioned the absence of the primary caregiver and relatives’ alternative caregiving arrangements as important influences on child feeding practices. For example:
That’s the difference when a child is cared for by someone else. The focus is often solely on ensuring the child eats, rather than on the quality of the food. The priority is simply that the child is full.
- Community key informant #6, 12 years of experience, Community 2
However, this theme was rarely mentioned by caregivers, although all 20 were homemakers.
2 of 14 community key informants noted that the most crucial period to improve children’s nutrition is within the first 2 years of life:
Stunting can still be prevented or addressed, but this must be done before the child turns 2 years old… After 2 years, correcting nutritional deficiencies is already quite challenging, but if intervention occurs before then, it is still possible to improve outcomes with more complete and higher-quality nutrition.
- Community key informant #6, 12 years of experience, Community 2
Poverty and health systems
Caregivers and key informants frequently mentioned financial difficulties, as well as common coping strategies such as borrowing from relatives and neighbours:
If my child wants something, such as milk, I feel bad because my financial situation is unstable. Sometimes I go to the stalls, get what I need, and pay when I have the money.
(Caregiver #20 of a stunted child, 40 years old, Community 1)
Another common coping strategy was subsistence farming and fishing. This was more common in Community 1 (6/17 vs 2/17 interviews in Community 2), likely related to the presence of fishing opportunities on the local reservoir:
R: There’s plenty of fish in [the dam reservoir]. Her father loves fishing.
I: What kinds of fish are there?
R: Tilapia and common carp fish.
I: Does she like eating the fish from there?
R: Yes, she eats fish often.
- Caregiver of a non-stunted child, 34 years old, Community 1
Subsistence strategies were more commonly mentioned by caregivers of non-stunted children (4/10 vs 1/10 interviews).
Issues with healthcare accessibility and affordability were mentioned more frequently in Community 1 (24 total mentions across 17 interviews, compared with 15 across 17 interviews in Community 2), likely due to the sparser density of healthcare facilities and the need to travel through mountainous terrain.
Yes, I had to cover the fare myself from here [house]. The hospital visit was free under universal health coverage.
- Caregiver #13 of a non-stunted child, 30 years old, Community 1
One key informant in Community 1 mentioned that accompaniment by health workers was an essential strategy for overcoming these barriers:
Yesterday, we accompanied them [stunted children] to the pediatrician at the hospital, and next Friday we will do it again.
- Community key informant #8, more than 5 years of experience, Community 1
Village funds are also important, as they are used for various health needs, including supplementary foods beyond those provided by the government, ambulance services and salaries for the cadres.
The village cooks it daily and funds it. Thus, children with stunting receive additional supplementary food sent to their house.
- Community key informant #2, 4 years of experience, Community 2
However, some villages lack ambulances to transport patients.
In our village, we do not have an ambulance, but we have a transport car. We can request that the village fund the ambulance. I have applied for it, but I have not yet received it.
- Community key informant #14, 2 years of experience, Community 1
Conditions complicating management
As stunting is often stigmatised, the shame of being diagnosed leads to hesitation among caregivers to bring their children to health posts. This theme was mentioned at a higher frequency among caregivers of stunted children (4/10) than among caregivers of non-stunted children (2/10).
I feel ashamed when I come to the village health post, because of the stunting diagnosis.
- Caregiver #6 of a stunted child, 42 years old, Community 1
This theme was validated by the community key informants:
Furthermore, they don’t want to go to the Integrated Village Health Post because their child is diagnosed with stunting; they feel embarrassed.
- Community key informant #1, 5 months of experience, Community 2
Some key informants in both communities (6/14 total) emphasised the role of other diseases, especially tuberculosis and diarrhoea, in complicating the management of stunting:
Sometimes, stunting can be linked to the possibility of having TB. For instance, if a child hasn’t gained weight for a few months, I’ll conduct a Mantoux test here. Getting tested is now very easy at the Puskesmas.
- Community key informant #5, 9 years of experience, Community 2
This theme was also mentioned by 3/10 caregivers with stunted children but not mentioned by caregivers of non-stunted children:
R: She’s been sick often in the last three months.
I: What are the symptoms?
R: Cough and cold, then fever. Diarrhea as well.
I: How long does it take for the diarrhea to stop?
R: Three days. But it happens eight to ten times a day. Then, she recovers. After three days or a week, she’s sick again.
- Caregiver #17 of a stunted child, 25 years old, Community 1
Integration of the results
Table 2 presents a joint display summarising key quantitative and qualitative findings, highlighting areas of concordance and discordance.
Table 2. Joint display of the mixed-methods results.
| Meta-category | Quantitative signal | Questions carried into the qualitative guides | Qualitative finding | Integrated interpretation |
|---|---|---|---|---|
| Drivers and impact of education on nutrition | Caregiver education not associated with stunting or HAZ (online supplemental tables S3 and S4). | Core domain: sources of nutrition knowledge, formal and informal. | Nutrition knowledge travels through health-centre classes and family networks rather than formal schooling; cost and gender norms limit schooling. | Partly concordant. Informal nutrition education plausibly offsets lower formal education. |
| Prenatal factors affecting growth | Older caregiver age was associated with lower odds of stunting and higher HAZ (online supplemental tables S3 and S4), consistent with the prior literature. | Core domain: antenatal care, maternal age and supplementation. | Adolescent pregnancy is viewed as a key risk; side-effect fears limit IFA adherence. | Partly concordant. Caregiving experience accrued with age may protect; a secular (cohort) contribution cannot be excluded. |
| Feeding practice and food experience | Older child age was associated with lower HAZ (online supplemental table S4) — the expected pattern of age-related growth faltering. | Added after the quantitative phase: community care management and alternative caregiving arrangements. | Feeding becomes harder after 2 years of age; delegated care prioritises fullness over food quality; low-quality foods are readily available. | Concordant. Age-related HAZ decline is explained by evolving feeding challenges, intercurrent illness and caregiving arrangements. |
| Poverty and health systems | Higher household wealth was associated with higher HAZ (online supplemental table S4); measured stunting prevalence exceeded routine district estimates. | Added after the quantitative phase: household economics, local financial-aid programmes and the impact of the recent dam construction. | Financial strain and coping (borrowing and food credit) are ubiquitous; subsistence farming and reservoir fishing buffer food access; health workers and village funds bridge access barriers. | Partly concordant. Economic and access-related risks offset one another, so a similar burden can arise through different local configurations of risk. |
| Conditions complicating management | Recent diarrhoea not associated with stunting or HAZ (online supplemental tables S3 and S4). | Core domain: intercurrent illness, care-seeking and barriers to attendance. | Caregivers of stunted children describe recurrent diarrhoea and stigma-related reluctance to attend health posts; informants describe tuberculosis work-ups for faltering children. | Discordant. Survey recall likely under-captures recurrent illness; reported morbidity may partly follow rather than cause stunting, and stigma may delay care. |
Quantitative associations are exploratory (the survey was powered for prevalence estimation; see Methods).
‘Added after the quantitative phase’ marks interview-guide adaptations documented in Qualitative instrument development; core domains were in the original guides.
HAZ, length/height-for-age Z-score; IFA, iron and folic acid.
Discussion
Stunting is a global paediatric concern, and it is of particular priority in Indonesia, where previous rapid improvements in child growth outcomes have stalled over the last decade.27 28 In this study, we used an explanatory sequential mixed-methods design to understand the contextual factors of stunting in two communities within a high-burden district.4 Integrating the survey and interview findings in a joint display (table 2), we focused on contextual conditions under which stunting persists.
Stunting prevalence
The prevalence of stunting we found in a systematic random sample is higher than the officially reported rate for the study area.4 This is consistent with the well-recognised challenges of routine, community-level nutrition data collection. Nutrition monitoring in the region relies entirely on community health workers, who face intense time pressure and multiple competing priorities.29 As a result, there may be limited time to visit families with sicker children and more barriers to accessing health centres, leading to the censoring of these cases from reported statistics. Furthermore, workers are required to report health statistics electronically, and several participants reported persistent connectivity issues during our interviews, which likely led to errors in data entry.30 31
Drivers and impact of education
Prior studies have highlighted that caregivers’ higher education lowers the odds of childhood stunting.32–34 However, our results show no correlation between maternal education and stunting (table 2). A plausible explanation, supported by our qualitative data, is the abundance of informal education, including education offered by health workers, and learning through family and relational networks. These informal offers may help offset any disadvantage in infant nutrition occasioned by a lack of access to formal education.
Prenatal factors affecting growth
In exploratory models, older caregiver age was associated with higher HAZs. Similarly, informants highlighted the risk to child health occasioned by young maternal age, especially adolescent pregnancy (table 2). Prior research in Indonesia has inconsistently demonstrated an association between stunting and maternal age, although the association is stronger for adolescent pregnancies.35 Accrued caregiving experience may also be beneficial to a child’s nutritional well-being.36 These age associations nevertheless warrant cautious interpretation: because stunting prevalence in Indonesia has declined steadily, the better height-for-age of more recently born children may reflect mostly secular improvements in early-life conditions.
Feeding practice and food experience
In exploratory models, older child age was associated with lower height-for-age, which aligns with another study in West Java.37 It is well known that the HAZ tends to decrease in at-risk children during the first few years of life.38 Qualitatively, in our sample, these phenomena can be partly explained by the challenges of feeding more and diverse foods to growing children; intercurrent episodes of acute illness (diarrhoea) challenging growth; and competing priorities, such as the need to work and implement alternative childcare arrangements. Food insecurity was not associated with stunting in our exploratory models; one possible explanation, raised by participants, is the displacement of high-quality foods by cheap processed foods.
Poverty and health systems
In exploratory models, greater household wealth was associated with higher height-for-age. This aligns with findings from a study in North Maluku.39 Additionally, another study conducted in West Java found that building materials, as a sanitation-related parameter, were associated with stunting.37 Notably, economic and access-related risks appeared to offset one another across settings (table 2), such that a similar burden can arise through different local configurations of risk.
Conditions complicating management
Diarrhoeal disease is well known to be associated with stunting.1 40 However, our quantitative data showed no association with stunting, although caregivers reported recurrent, chronic diarrhoea (table 2). This highlights the fact that, although stunting can, in many cases, be exacerbated by intercurrent illness, it also results in significant perceived morbidity in affected children, with caregivers reporting diarrhoea and decreased appetite.
Strengths and limitations
Our study had important strengths and limitations. First, we employed an explanatory sequential mixed-methods design, using a joint display to triangulate results. Second, we mitigated potential selection bias by using a simple random sampling method to include households beyond those directly referred by public health centres. We sampled diverse voices in the qualitative component. At the same time, our quantitative sample was small, powered primarily to estimate the point prevalence of stunting; as a result, the lack of association between stunting and several examined predictors, such as education, may reflect type II error; the cross-sectional design also precludes causal inference. In addition, our mixed-methods approach revealed highly contextual features of the two communities studied, which may not be generalisable to other regions of Indonesia. However, identifying contextual balancing factors (eg, the interplay between healthcare access and poverty across the two communities) underscores the need for highly contextualised assessments in any community or region where stunting is a significant problem.
Conclusions
In a mixed-methods study conducted in two rural communities in West Java, we found a higher-than-expected prevalence of stunting, underscoring the value of periodic household-based assessment to complement routine monitoring. We identified a complex interplay of factors, including poverty, healthcare access and community-embedded supports, such as village funds and health-worker accompaniment, that shaped child health in highly contextual ways, underscoring the need for contextually tailored, community-based approaches.
The quantitative associations we observed were small and not novel. Although this may partly reflect the limited power of our exploratory analyses, it is also consistent with Indonesia’s rapid and sustained progress against stunting (from an overall prevalence of 37% in 2007 to 21.6% in 2022).3 The relatively small associations with known modifiable factors observed in our study underscore this reality, as significant improvements in poverty41 and education42 are evident nationally. It is important, more than ever, to sustain the gains in the social safety net that have been made so that their benefits continue to accrue to future generations.
Supplementary material
Acknowledgements
We are also grateful to all study participants, including health providers, caregivers and children, who shared their experiences to help us better understand childhood stunting. We would like to sincerely thank the District Government of Sumedang Health Office staff and research assistants involved in the study for their invaluable support. This work was conducted with support from the Harvard Medical School Master of Medical Sciences in Global Health Delivery programme, Harvard University and the Ronda Stryker and William Johnston MMSc Fellowship.
Footnotes
Funding: This work was conducted with support from the Harvard Medical School Master of Medical Sciences in Global Health Delivery programme, Harvard University and the Ronda Stryker and William Johnston MMSc Fellowship.
Provenance and peer review: Not commissioned; externally peer reviewed.
Patient consent for publication: Not applicable.
Ethics approval: This study involved human participants and was approved by the relevant ethical committee in Indonesia (District Government of Sumedang Health Office Research Permit; number: KS.01.02.03/388/X/2023). The study was also approved by the Institutional Review Board (IRB) of Harvard University (approval number IRB24-0269). Participants provided informed consent to participate in the study before taking part.
Data availability free text: De-identified raw datasets, statistical code and qualitative codebook are available on Harvard Dataverse (https://doi.org/10.7910/DVN/G5CTKR).
Patient and public involvement: Patients and/or the public were not involved in the design, conduct, reporting or dissemination plans of this research.
Data availability statement
Data are available in a public open access repository.
References
- 1.Millward DJ. Nutrition, infection and stunting: the roles of deficiencies of individual nutrients and foods, and of inflammation, as determinants of reduced linear growth of children. Nutr Res Rev. 2017;30:50–72. doi: 10.1017/S0954422416000238. [DOI] [PubMed] [Google Scholar]
- 2.World Health Organization Prevalence of stunting in children under 5 (%) 2024. [10-Feb-2024]. https://data.who.int/indicators/i/A5A7413/5F8A486 Available. Accessed.
- 3.Ministry of Health of Indonesia . Hasil Survei Status Gizi Indonesia (SSGI) 2022. 2023. [Google Scholar]
- 4.Sumedang District Health Office 2024. Compilation of stunting and malnutrition toddler 0–59 months old: results of 2023 toddler weighing month (BPB) activities [unpublished database]
- 5.Sumedang District Health Office 2024. Database of children located in community health centers 1 and 2 [unpublished database]
- 6.BPS-Statistics Sumedang Regency . Kecamatan Darmaraja Dalam Angka 2024. XXIV. 2024. [Google Scholar]
- 7.BPS-Statistics Sumedang Regency . Kecamatan Sumedang Utara Dalam Angka 2024. XXIV. 2024. [Google Scholar]
- 8.Google Maps . Distance from the Community 1 Health Center to the District Hospital. 2024. [Google Scholar]
- 9.Google Maps . Distance from the Community 2 Health Center to the District Hospital. 2024. [Google Scholar]
- 10.Select Statistical Service Population proportion – sample size. 2022. [1-Dec-2023]. https://select-statistics.co.uk/calculators/sample-size-calculator-population-proportion/ Available. Accessed.
- 11.Esterini DN, Venzor Strader A, Rohloff P. Identifying factors associated with childhood stunting in West Java, Indonesia: an explanatory sequential mixed-methods study. Harvard Dataverse; 2026. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.World Health Organization . Guideline: assessing and managing children at primary health-care facilities to prevent overweight and obesity in the context of the double burden of malnutrition. 2017. [29-Mar-2025]. https://iris.who.int/handle/10665/259133 Available. accessed. [PubMed] [Google Scholar]
- 13.Harris PA, Taylor R, Thielke R, et al. Research electronic data capture (REDCap)--a metadata-driven methodology and workflow process for providing translational research informatics support. J Biomed Inform. 2009;42:377–81. doi: 10.1016/j.jbi.2008.08.010. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Harris PA, Taylor R, Minor BL, et al. The REDCap consortium: building an international community of software platform partners. J Biomed Inform. 2019;95:103208. doi: 10.1016/j.jbi.2019.103208. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Metrics for Management Methods: EquityTool. [14-Jul-2025]. https://equitytool.org/methods/ Available. Accessed.
- 16.Metrics for Management EquityTool: Indonesia. 2020. [1-Jan-2024]. https://www.equitytool.org/indonesia/ Available. Accessed.
- 17.FAO SDG indicator 2.1.2—using the Food Insecurity Experience Scale (FIES) 2018. [1-Jun-2024]. https://elearning.fao.org/pluginfile.php/491591/mod_scorm/content/5/story_content/external_files/SDG2.1.2_lesson4.pdf Available. Accessed.
- 18.Badan Penelitian dan Pengembangan Kesehatan Kuesioner SSGI 2022 individu. 2022. [1-Jul-2026]. http://repository.badankebijakan.kemkes.go.id/id/eprint/4747/ Available. Accessed.
- 19.Textor J, van der Zander B, Gilthorpe MS, et al. Robust causal inference using directed acyclic graphs: the R package ‘dagitty’. Int J Epidemiol. 2017:dyw341. doi: 10.1093/ije/dyw341. [DOI] [PubMed] [Google Scholar]
- 20.Dedoose . Dedoose version 10.0.59: cloud application for managing, analyzing, and presenting qualitative and mixed method research data. SocioCultural Research Consultants, LLC; 2026. www.dedoose.com Available. [Google Scholar]
- 21.Braun V, Clarke V. Using thematic analysis in psychology. Qual Res Psychol. 2006;3:77–101. doi: 10.1191/1478088706qp063oa. [DOI] [Google Scholar]
- 22.von EE, Altman DG, Egger M, et al. The STROBE reporting checklist. https://www.resources.equator-network.org/reporting-guidelines/strobe/strobe-checklist.docx n.d. Available.
- 23.von EE, Altman DG, Egger M, et al. In: The EQUATOR Network Reporting Guideline Platform [Internet] Harwood J, Albury C, de BJ, et al., editors. The UK EQUATOR Centre; 2025. The strobe reporting checklist.www.resources.equator-network.org/reporting-guidelines/strobe/strobe-checklist.docx Available. [Google Scholar]
- 24.O’Brien BC, Harris IB, Beckman TJ, et al. Standards for reporting qualitative research: a synthesis of recommendations. Acad Med J Assoc Am Med Coll. 2014;89:1245–51. doi: 10.1097/ACM.0000000000000388. [DOI] [PubMed] [Google Scholar]
- 25.Tong A, Sainsbury P, Craig J. Consolidated criteria for reporting qualitative research (COREQ): a 32-item checklist for interviews and focus groups. Int J Qual Health Care. 2007;19:349–57. doi: 10.1093/intqhc/mzm042. [DOI] [PubMed] [Google Scholar]
- 26.O’Cathain A, Murphy E, Nicholl J. The quality of mixed methods studies in health services research. J Health Serv Res Policy . 2008;13:92–8. doi: 10.1258/jhsrp.2007.007074. [DOI] [PubMed] [Google Scholar]
- 27.Rueda-Guevara P, Botero Tovar N, Trujillo KM, et al. Worldwide evidence about infant stunting from a public health perspective: a systematic review. Biomedica. 2021;41:541–54. doi: 10.7705/biomedica.6017. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28.Laksono AD, Wulandari RD, Amaliah N, et al. Stunting among children under two years in Indonesia: does maternal education matter? PLoS One. 2022;17:e0271509. doi: 10.1371/journal.pone.0271509. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29.Rotella R, Morales-Suarez-Varela M, Llopis-Gonzalez A, et al. Assessment of stunting and its effect on wasting in children under two in rural Madagascar. Children (Basel) 2025;12:686. doi: 10.3390/children12060686. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30.UNICEF, Pulse Lab Jakarta Final report: assessing the implementation of Indonesia’s national nutrition information system (e-PPGBM) [1-Nov-2025]. https://www.unglobalpulse.org/ Available. Accessed.
- 31.UNICEF, Ministry of National Development Planning/National Development Planning Agency, Indonesia Final report: formative evaluation of the national strategy to accelerate stunting prevention. [1-Nov-2025]. https://www.unicef.org/indonesia/ Available. Accessed.
- 32.Widyaningsih, Mulyaningsih, Rahmawati . Determinants of socioeconomic and rural-urban disparities in stunting: evidence from Indonesia. Rural Remote Health; 2022. [DOI] [PubMed] [Google Scholar]
- 33.Rannan-Eliya RP, Hossain SMM, Anuranga C, et al. Trends and determinants of childhood stunting and underweight in Sri Lanka. Ceylon Med J. 2013;58:10–8. doi: 10.4038/cmj.v58i1.5357. [DOI] [PubMed] [Google Scholar]
- 34.Mzumara B, Bwembya P, Halwiindi H, et al. Factors associated with stunting among children below five years of age in Zambia: evidence from the 2014 Zambia demographic and health survey. BMC Nutr. 2018;4:51. doi: 10.1186/s40795-018-0260-9. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 35.Laksono AD, Sukoco NEW, Rachmawati T, et al. Factors related to stunting incidence in toddlers with working mothers in Indonesia. Int J Environ Res Public Health. 2022;19:10654. doi: 10.3390/ijerph191710654. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 36.Fall CHD, Sachdev HS, Osmond C, et al. Association between maternal age at childbirth and child and adult outcomes in the offspring: a prospective study in five low-income and middle-income countries (COHORTS collaboration) Lancet Glob Health. 2015;3:e366–77. doi: 10.1016/S2214-109X(15)00038-8. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 37.Permatasari TAE, Chadirin Y, et al. The association of sociodemographic, nutrition, and sanitation on stunting in children under five in rural area of West Java Province in Indonesia. J Public Health Res. 2023;12:22799036231197169. doi: 10.1177/22799036231197169. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 38.Katoch OR. Determinants of malnutrition among children: a systematic review. Nutrition. 2022;96:111565. doi: 10.1016/j.nut.2021.111565. [DOI] [PubMed] [Google Scholar]
- 39.Agho KE, Inder KJ, et al. Prevalence and risk factors for stunting and severe stunting among under-fives in North Maluku province of Indonesia. BMC Pediatr. 2009;9:64. doi: 10.1186/1471-2431-9-64. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 40.Beal T, Tumilowicz A, Sutrisna A, et al. A review of child stunting determinants in Indonesia. Matern Child Nutr. 2018;14:e12617. doi: 10.1111/mcn.12617. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 41.Badan Pusat Statistik . Profil kemiskinan di Indonesia Maret 2025. 2025. [Google Scholar]
- 42.Badan Pusat Statistik . Statistik pendidikan Indonesia, volume 13, 2024. 2024. [Google Scholar]