Exploring the Role of Food Security in Stunting Prevention Efforts in the Bondowoso Community, Indonesia

Abstract

Stunting—defined by the World Health Organization as a height-for-age z-score < −2 SD—signals chronic undernutrition that impairs both physical and cognitive development. This study investigates how the three pillars of food security (availability, access, utilization) influence stunting prevention efforts in the Bondowoso Regency, East Java, Indonesia. A cross-sectional survey of 113 mothers of stunted children (0–59 months) was analysed with Structural Equation Modelling using Partial Least Squares (PLS-SEM). The model reveals significant positive paths from food security pillars to composite stunting prevention behaviours (β = 0.18–0.86, p < 0.05), with availability emerging as the strongest predictor. These findings highlight food security as a lever for reducing the local stunting prevalence (local 32% vs. national 24.4%) and provide evidence for community-based nutrition programmes in similar agrarian districts. Strengthening food security is therefore essential to safeguarding child well-being in vulnerable Indonesian communities.

1. Introduction

Stunting is a form of linear growth failure, scientifically defined as a height-for-age z-score (HAZ) below −2 standard deviations from the WHO Child Growth Standards. It results from chronic undernutrition, recurrent infections, and poor-quality caregiving environments [1,2,3,4]. Beyond restricted stature, stunting impairs immune function, cognitive development, academic performance, and long-term earning potential [5].

The stunting process begins when a child’s body consistently lacks sufficient macronutrients (e.g., protein and energy) and micronutrients (e.g., iron, zinc, and vitamin A) or when nutrient absorption is hindered by repeated infections, such as diarrhoea or respiratory illnesses. These challenges, combined with unsupportive environmental and socioeconomic conditions, gradually hinder a child’s physical growth. Maternal nutritional status and health during pregnancy also play a vital role in determining the initial risk of stunting [6,7,8].

Stunting has long-term consequences, including impaired physical and cognitive development, lower educational achievement, reduced economic productivity, and heightened risk of chronic disease—factors that perpetuate intergenerational poverty. The prevention of stunting therefore requires a holistic approach, including improved maternal and child nutrition, exclusive breastfeeding, access to clean water and sanitation, basic health services, and robust food security and diversification programs. Interventions are most effective during the first 1000 days of life [7,8].

Food security—defined by the sustained availability, accessibility, and utilization of safe, nutritious food—has consistently been linked to improved child linear growth outcomes [9,10,11]. When one or more of these pillars is compromised, nutritional deficits, infection risk, and inadequate caregiving often intersect to hinder optimal growth during early childhood.

Tjenemundan et al. [1] highlighted the benefits of enhancing food security through community participation. Interventions such as nutrition education, homestead gardens, and food parcel distribution significantly increase awareness and strengthen community-driven nutrition efforts. Pardede and Saraan [12] further emphasized the importance of integrated food security strategies, noting that increased food production, dietary diversification, and nutrition outreach have reduced stunting in targeted groups. Additional feeding programs for toddlers and pregnant or breastfeeding mothers also play a key role in improving nutritional intake and preventing stunting.

Community-based initiatives—particularly nutrition education, homestead gardening, and targeted feeding—have shown promising results in several Indonesian districts [1,12]. However, no prior study has systematically examined how the three core pillars of food security jointly influence multidomain stunting prevention behaviours at the village level in East Java. This evidence gap provides the rationale for the present study.

Based on the 2022 Indonesian Nutrition Status Survey (SSGI) report, the prevalence of stunting in Indonesia is still at 24.4%, indicating that this problem is still a serious national challenge and requires comprehensive and sustainable intervention [13]. At the provincial level, East Java recorded a stunting prevalence of 19.2%, which, although lower than the national average, still places it as one of the provinces with a high burden of stunting.

Bondowoso Regency—specifically Wonosari District—reported one of the highest stunting rates in East Java at 32.1% in 2023 (internal data from the Bondowoso District Health Service, 2023). Although the region maintains year-round agricultural production, high rural poverty (17%) and low dietary diversity have contributed to the stunting burden. Wonosari District, composed of 12 villages, was selected for this study due to its high stunting prevalence, patterns of food insecurity, and representative rural characteristics. Seven villages with the highest stunting rates were sampled, with 161 stunted children identified, of whom 113 were selected as respondents.

Food security is a condition wherein a community or nation can ensure equitable access to adequate, safe, and nutritious food for all individuals. This includes both the physical availability of food and fair access to it [11,14]. Robust food security helps prevent stunting and other nutrition-related conditions by ensuring that children receive the nutrients necessary for growth and development [9,15]. Food diversification is also an important component in achieving good food security and preventing stunting. By consuming a variety of foods, children have access to a variety of different nutrients necessary for optimal growth. Food diversification helps avoid deficiencies in certain nutrients and achieve a good nutritional balance [16,17].

Furthermore, food quality and safety—through proper handling and chemical-free production—are essential to sustaining a nutritious diet [18]. Against this backdrop, the present study applies a cross-sectional SEM-PLS design to investigate the direct effects of food availability, access, and utilization on four composite stunting prevention behaviours among 113 mothers of stunted children in Wonosari Subdistrict. This research aims to clarify the relationship between food security dimensions and stunting prevention and to underscore the importance of food security as a cornerstone of child health in rural Indonesia.

2. Literature Review

2.1. Stunting Prevention Efforts

Effective stunting prevention hinges on an integrated “1000-day” package that starts before conception and continues until a child’s second birthday. Core elements include exclusive breastfeeding for at least six months, timely and diverse complementary feeding (MP-ASI), maternal nutrition counselling, and full immunization—each of which has been shown to reduce height-for-age deficits in Indonesian cohorts [11,15].

Several interventions have been implemented to prevent stunting, including supplementary-feeding programmes that supply additional nutrition to malnourished children and improve their growth [13]. The effectiveness of such programmes depends on feed quality and consistency. Community-empowerment initiatives—nutrition education, home-garden cultivation, and sanitation improvements—have also proved effective [1].

Randomized and quasi-experimental studies in Java and Sulawesi confirm that community-managed supplementary-feeding schemes combined with home-garden promotion and hygiene classes reduce stunting by 3–6 percentage points within twelve months [1,12,19]. Caregiver feeding style is a strong mediator between food availability and actual child intake; restrictive or coercive practices negate diet-diversity gains [5]. Indonesian evidence shows that responsive-feeding training delivered through Integrated Health Service Post (Posyandu) cadres increased Minimum Dietary Diversity compliance from 34% to 61% in six months [20].

Toddlers’ eating behaviour strongly influences nutritional intake. Parental practices—such as coercing a child to eat or restricting food variety—can impair nutrition and growth [4,5,21]. Children’s individual food preferences must also be considered; parents should offer a range of nutritious foods to ensure a balanced diet. A supportive family environment that models healthy eating can positively shape children’s food choices.

Beyond diet, deficits in water, sanitation, and hygiene (WASH) trigger environmental enteric dysfunction that limits nutrient absorption [22]. A meta-analysis across 13 Indonesian districts found that improved household latrines and ≥2 daily handwashing events were associated with an adjusted 18% reduction in stunting risk [23]. Poor sanitation and hygiene can lead to recurrent infections, which interfere with nutrient absorption and growth. Poor sanitation can cause recurrent infections that impede nutrient uptake and growth. Essential WASH practices include access to safe drinking water, proper handwashing before meals and after defecation, and sound waste management to prevent contamination [23,24,25].

Collectively, these studies demonstrate effective interventions; however, few have modelled how the pillars of food security interact with these behavioural domains. The present study therefore tests a structural model that links availability, access, and utilization to four prevention-behaviour clusters within a single agrarian district.

2.2. Food Security

Food security—operationalized as availability, access, and utilization—accounts for substantial variation in child growth outcomes across low- and middle-income countries. In Indonesian agricultural zones, households in the highest food security tertile exhibit HAZ scores that are, on average, 0.46 SD higher than those of food-insecure peers [13]. Seasonal availability shocks during the pre-harvest “lean” period have been linked to HAZ declines of 0.2 SD in East Java.

Adequate food security guarantees the availability of, access to, and utilization of sufficient, nutritious food for all community members—particularly pregnant women, breastfeeding mothers, and children [20]. A diverse and reliable food supply is vital for meeting nutritional needs. Sustainable local food production underpins availability, while efficient distribution systems ensure food is accessible to all population groups [26].

Limited physical and economic access amplifies stunting risk even where aggregate food supplies are adequate; a 2023 SEM-GSCA study across 112 Indonesian districts found that access elasticity (β = 0.37) exceeded that of availability (β = 0.29) [27]. Food access includes the economic and physical ability to obtain sufficient and nutritious food. Poverty and economic inequality often limit people’s access to nutritious food [26,28].

Utilization—reflected in dietary diversity, safe preparation, and child-care practices—mediates the access. Putra et al. [9] reported that each additional food group consumed raised the HAZ by 0.11 SD after adjusting for infection burden. Food utilization includes the ability to use available food to meet nutritional needs. Proper utilization entails using available food to meet nutritional needs, applying safe processing methods to minimize nutrient loss, and maintaining diverse diets to secure the full range of nutrients required for optimal growth [18,27,29].

Existing Indonesian studies typically isolate single pillars or behaviours. No research has yet quantified the combined effects of all three pillars on a multidimensional prevention index, creating a knowledge gap that this study addresses for Bondowoso’s high-burden villages.

3. Materials and Methods

3.1. Research Model Development

The research model (Figure 1) posits that the three pillars of food security—availability, access, and utilization—directly influence four clusters of stunting prevention behaviour. This specification is grounded in evidence that local food diversity, household income, and caregiver knowledge jointly shape linear growth trajectories among Indonesian toddlers [13,20,30,31]. In the model, each pillar is specified as a first-order reflective construct, while the prevention package is modelled as a second-order formative construct.

Numerous studies highlight the need to address food security at both community and household levels to prevent stunting effectively. In agricultural settings, maintaining local food security is crucial for lowering stunting rates. A diverse and adequate food supply—particularly during the pre-harvest lean season—is essential. Economic factors, such as household income, strongly influence food security and thereby affect stunting prevalence [13,30]. At the family level, access to diverse, nutritious, and affordable foods is vital. Training programmes for parents and Posyandu cadres on complementary food preparation have proved effective in boosting nutritional intake and reducing stunting [20,31].

Education and public awareness are additional pillars for enhancing food security and preventing stunting. Community-wide nutrition education ensures that children gain access to nutrient-rich foods, thereby lowering stunting rates [9]. Multiple studies report a strong association between household food security and stunting incidence. Strengthening household food security remains the primary strategy for combating stunting, even when other factors—such as macronutrient intake—show no significant correlation [15].

Accordingly, it is hypothesized that each food security pillar exerts a positive and direct effect on the composite prevention index (H1–H3). This hypothesis rests on the premise that guaranteeing access to diverse, nutritious foods, improving food quality and availability, and raising nutrition awareness are critical for reducing stunting.

3.2. Data Collection

This study was conducted in January 2025 in Wonosari District, Bondowoso Regency, where the 2023 district health census recorded a stunting prevalence of 32.1%—substantially higher than the provincial average of 21.3%. The seven villages with the highest case counts were purposively selected (Section 3.5). Stunting status had been determined by village midwives and health cadres using routine Posyandu growth-monitoring records. Households qualified as respondents if at least one child had been identified as stunted in these monthly records; height-for-age z-scores were calculated with WHO Anthro 3.2 software. A structured questionnaire was administered face to face by trained enumerators to collect information on (i) preventive practices, (ii) household food security, and (iii) sociodemographic characteristics.

Prior to data collection, a preliminary survey was carried out in collaboration with local health workers and village authorities to verify that the research tools were context-appropriate and would foster respondent engagement. Community feedback guided the design of a 27-item, 5-point Likert questionnaire (1 = strongly disagree; 5 = strongly agree) aimed at capturing stunting-related issues in depth. Content validity was reviewed by three public health experts, and a pilot test with 30 mothers (excluded from the final sample) produced Cronbach’s α values ≥ 0.80 for all scales, after which minor wording adjustments were made.

Data were collected face to face by trained enumerators, supported by community health-centre staff and village officials. Feedback from respondents highlighted the need to clarify the wording of several items and simplify the language; the question order was also adjusted to enhance interview flow. This iterative process refined the final instrument and ensured that the questionnaire was culturally appropriate and easily understood by the participants.

3.3. Subjects

This study was carried out in Bondowoso Regency, which comprises 23 subdistricts. Wonosari District—selected for its elevated stunting prevalence—contains 12 villages. Of these, the seven villages with the highest stunting counts were included in the sample. Posyandu records identified 161 children (0–59 months) classified as stunted in Wonosari; these children constituted the study population.

From the register of 161 stunted children (0–59 months), primary caregivers—specifically mothers—were selected as respondents, given that stunting prevention occurs mainly within the first 1000 days and mothers are the principal decision-makers for child feeding and hygiene.

The 0–59-month age range was selected because toddlerhood represents the period most vulnerable to growth and developmental disorders; the first 1000 days are widely recognized as the “golden window” for effective stunting interventions.

In addition, signs of stunting can be observed and measured accurately in this age group through routine growth monitoring. Including children up to 59 months old also captures those who have moved beyond the initial intervention period yet still exhibit the long-term effects of chronic malnutrition.

3.4. Variables and Indicators

For structural model analysis, each latent variable is represented by a set of indicators denoted by abbreviations for ease of reference.

Table 1. Research variables.

Variable	Sub Variable	Note	Indicator	Note
Stunting prevention efforts (STGs)	Health Prevention Practices	HPPs	Health insurance coverage	STG1
			Impact of social guidance	STG2
			Effect of the household environment	STG3
	Interventions to Prevent Stunting	IPSs	Proactive responses to stunting information	STG4
			Consistency in stunting prevention efforts	STG5
	Eating Behaviour in Toddlers	EBT	Decrease in instant food consumption	STG6
			Sufficiency of daily energy intake	STG7
			Provision of exclusive breastfeeding	STG8
			Adherence to the immunization schedule	STG9
			Timely introduction of complementary foods	STG10
			Incidence of infectious diseases	STG11
	Sanitation and Hygiene Practices	SHPs	Waste management discipline	STG12
			Accessibility of clean water	STG13
			Instilling handwashing habits in children	STG14
Food security (FC)	Food availability	AV	Fulfilment of nutritional food requirements	AV1
			Days lacking nutritious food	AV2
			Challenges in providing complementary foods	AV3
			Storage of food supplies	AV4
	Food accessibility	AC	Time invested in acquiring nutritious food	AC1
			Transportation expenses to reach the market	AC2
			Interruptions in transport or supply chains	AC3
	Food utilization	UT	Frequency of health facility visits	UT1
			Frequency of serving nutritious meals	UT2
			Receiving food assistance	UT3

lists the following codes: STGs (stunting prevention indicators), AV (availability), AC (access), and UT (utilization). The table systematically pairs each latent variable with its indicator codes, forming the basis for measurement in the analysis model.

3.5. Sampling Strategy

Using the Krejcie–Morgan formula (

Table 2. Krejcie and Morgan table [32].

N	S	N	S	N	S
10	10	220	140	1200	291
15	14	230	144	1300	297
20	19	240	148	1400	302
25	24	250	152	1500	306
30	28	260	155	1600	310
35	32	270	159	1700	313
40	36	280	162	1800	317
45	40	290	165	1900	320
50	44	300	169	2000	322
55	48	320	175	2200	327
60	52	340	181	2400	331
65	56	360	186	2600	335
70	59	380	181	2800	338
75	63	400	196	3000	341
80	66	420	201	3500	346
85	70	440	205	4000	351
90	73	460	210	4500	354
95	76	480	214	5000	357
100	80	500	217	6000	361
110	86	550	226	7000	364
120	92	600	234	8000	367
130	97	650	242	9000	368
140	103	700	248	10,000	370
150	108	750	254	15,000	375
160	113	800	260	20,000	377
170	118	850	265	30,000	379
180	123	900	269	40,000	380
190	127	950	274	50,000	381
200	132	1000	278	75,000	382
210	136	1100	285	100,000	384

) [32] for a population of 161 stunted children, a minimum sample of 113 respondents was required. Proportional-quota sampling—adjusted with a cluster-robust design effect of 1.2—allocated quotas to the seven Wonosari villages with the highest stunting counts. All eligible mothers of stunted toddlers were approached until the quotas were met (non-response = 8%), and design-based χ² tests confirmed the absence of a significant cluster bias (p > 0.10).

Posyandu registers identified the 161 stunted children (0–59 months) across Wonosari’s 12 villages. The seven highest-burden villages were therefore selected, and sample quotas were distributed in proportion to each village’s caseload;

Table 3. Sample.

No	Village	Number of Stunted Children	Sample
1	Lombok Kulon	33	28
2	Tumpeng	13	11
3	Jempong	10	9
4	Tangsil Wetan	29	25
5	Lorok	16	14
6	Bendo	14	12
7	Pasarejo	17	15
Sample			113

presents the final allocation. This purposive, proportionally weighted design ensured representative coverage of the district’s stunting hotspots while maintaining statistical efficiency.

3.6. Data Analysis

Data analysis in this research was carried out using the Partial Least Squares (PLS) technique from Structural Equation Modelling (SEM) with Smart-PLS 3.0. PLS was chosen because it suited the exploratory nature of this research. PLS path modelling provides a powerful solution, especially when the primary objective is prediction, the model is relatively complex, the sample size is small, and there are few dependent variables. Model assessment proceeded in two stages: (i) measurement model—factor loadings, composite reliability, AVE, and Heterotrait–Monotrait ratio; (ii) structural model—β-coefficients and t-values from 5 000-sample bootstrapping. The Standardized Root-Mean-Square Residual (SRMR) is reported to provide an overall fit index (acceptable if <0.08) [20].

4. Results

Based on the existing research objectives and questions, as well as to evaluate the proposed hypothesis, analysis was carried out using Structural Equation Modelling with the following research model.

4.1. PLS-SEM Analysis

Partial Least Squares Structural Equation Modelling (SmartPLS 3.0) was chosen for its robustness with moderate samples and non-normal data. A post hoc power calculation (effect size = 0.15, α = 0.05, three predictors) returned 0.87, exceeding the 0.80 benchmark. The overall model fit was adequate (SRMR = 0.064 < 0.08).

The model developed based on Research Model Development in Smart-PLS 3.0 was evaluated in two stages; namely, the measurement model was analysed to assess the reliability and validity of the construct, and the structural model was analysed to check the significance of the path coefficients in the research model.

4.1.1. Measurement Model

PLS-SEM was employed to validate the measurement model. Reliability and convergent validity were confirmed with composite reliability (CR) and average variance extracted (AVE), while discriminant validity was assessed via HTMT ratios and the Fornell–Larcker criterion. SmartPLS provided factor loadings, CR, and AVE estimates; all VIF values for availability, access, and utilization ranged from 1.42 to 2.11, well below the multicollinearity threshold of 3.0.

All factor loadings exceeded 0.50 (

Table 4. Measurement model.

Variables	Indicators	Note	Loading	CR	AVE
Health Prevention Practices (HPPs)	Health insurance coverage	STG1	0.957	0.956	0.878
	Impact of social guidance	STG2	0.919
	Effect of the household environment	STG3	0.935
Interventions to Prevent Stunting (IPSs)	Proactive responses to stunting information	STG4	0.956	0.961	0.924
	Consistency in stunting prevention efforts	STG5	0.967
Eating Behaviour in Toddlers (EBT)	Decrease in instant food consumption	STG6	0.860	0.947	0.751
	Sufficiency of daily energy intake	STG7	0.712
	Provision of exclusive breastfeeding	STG8	0.943
	Adherence to the immunization schedule	STG9	0.901
	Timely introduction of complementary foods	STG10	0.949
	Incidence of infectious diseases	STG11	0.811
Sanitation and Hygiene Practices (SHPs)	Waste management discipline	STG12	0.838	0.882	0.714
	Accessibility of clean water	STG13	0.774
	Instilling handwashing habits in children	STG14	0.917
Food availability (AV)	Fulfilment of nutritional food requirements	AV1	0.889	0.953	0.835
	Days lacking nutritious food	AV2	0.882
	Challenges in providing complementary foods	AV3	0.954
	Storage of food supplies	AV4	0.928
Food accessibility (AC)	Time invested in acquiring nutritious food	AC1	0.957	0.954	0.873
	Transportation expenses to reach the market	AC2	0.961
	Interruptions in transport or supply chains	AC3	0.883
Food utilization (UT)	Frequency of health facility visits	UT1	0.838	0.925	0.804
	Frequency of serving nutritious meals	UT2	0.897
	Receiving food assistance	UT3	0.952

, Figure 2). The AVE values ranged from 0.50 to 0.90, and the CR values ranged from 0.80 to 0.96, surpassing the recommended thresholds (AVE ≥ 0.50; CR ≥ 0.70). The HTMT ratios were ≤0.83, indicating satisfactory discriminant validity. Fornell–Larcker analysis (

Table 5. Discriminant validity.

Construct	1	2	3	4	5	6	7
AC	0.934
OF	0.431	0.914
EBT	0.589	0.798	0.867
HPPs	0.576	0.857	0.847	0.937
IPSs	0.339	0.785	0.771	0.711	0.961
SHPs	0.407	0.787	0.799	0.803	0.696	0.845
OUT	0.621	0.804	0.758	0.818	0.603	0.702	0.897

) showed that the square root of each construct’s AVE exceeded its inter-construct correlations, further confirming discriminant validity. Collectively, these results indicate that the measurement model meets accepted reliability and validity criteria, providing a sound basis for subsequent structural analysis.

4.1.2. Structural Model

Hypothesis testing in this research was carried out using the Partial Least Squares Structural Equation Modelling (PLS-SEM) approach. The structural model was evaluated to assess model quality and test research hypotheses through a bootstrapping process, using a two-tailed t-test at a significance level of 5%. The path coefficient was considered significant if the t-value was greater than 1.96.

Table 6. Path coefficient testing: relationships between constructs.

Path Coefficients	Original Sample	Standard Deviation	T Statistics	p Values	Decision
AC → EBT	0.258	0.048	5.397	0.000	Supported
AC → HPPs	0.178	0.043	4.173	0.000	Supported
AC → IPSs	0.038	0.055	0.700	0.484	Not supported
AC → SHPs	0.024	0.086	0.273	0.785	Not supported
OFF → EBT	0.584	0.086	6.809	0.000	Supported
OFF → HPPs	0.597	0.071	8.358	0.000	Supported
OFF → IPSs	0.856	0.088	9.746	0.000	Supported
OFF → SHPs	0.633	0.103	6.174	0.000	Supported
OUT → EBT	0.127	0.108	1.183	0.238	Not supported
OUT → HPPs	0.228	0.080	2.862	0.004	Supported
OUT → IPSs	−0.109	0.108	1.005	0.315	Not supported
UT → SHPs	0.178	0.112	1.598	0.111	Not supported

and Figure 3 present the structural model results. Most paths achieved statistical significance, thereby supporting the study hypotheses. For instance, access (AC) exerted a positive effect on Toddler Eating Behaviour (EBT) (β = 0.258; t = 5.397; p < 0.001), explaining 46% of the variance in EBT (R² = 0.46; f² = 0.08). Enhanced access to food resources therefore promotes healthier eating patterns among toddlers, a critical factor in stunting prevention.

However, not all path coefficients support the hypothesis. The relationship between AC and Interventions to Prevent Stunting (IPSs) is not significant (original sample = 0.038, t-value = 0.700; p = 0.484), indicating that access alone is not strong enough to influence the effectiveness of stunting interventions. This implies that other factors may play a greater role in the success of this intervention. Likewise, the relationship between AC and Sanitation and Hygiene Practices (SHPs) was not significant (original sample = 0.024, t-value = 0.273; p = 0.785), indicating that access does not have a substantial impact on sanitation practices. This pattern mirrors findings from Dewi et al. [27] where hygiene behaviour was mediated by maternal knowledge rather than physical access.

In contrast, resource availability (AV) shows a strong influence on several variables. The relationship between AV and EBT is significant, with an original sample of 0.584 and a t-value of 6.809 (p = 0.000), supporting the hypothesis that availability directly influences eating behaviour. This highlights the important role of ensuring a steady supply of nutritious food in establishing healthy eating patterns. The relationships between AV and Health Preventive Practices (HPPs), IPSs, and SHPs were also significant, with original sample values of 0.597, 0.856, and 0.633, respectively, all with p values of 0.000. These findings emphasize the importance of resource availability in improving Health Prevention Practices, stunting interventions, and Sanitation and Hygiene Practices. The strong path coefficients indicate that availability is an important factor in supporting this research hypothesis. Effect sizes were large (f² = 0.33–0.65), emphasizing the primacy of a diverse and steady food supply.

In contrast, utilization (UT) shows a more complex picture. The relationship between UT and EBT is not significant (original sample = 0.127, t-value = 1.183; p = 0.238), indicating that utilization does not directly influence eating behaviour. This suggests that although resources are available, the way they are utilized may not be optimal in influencing eating habits. Then, the relationship between UT and IPSs shows a negative coefficient (−0.109) with a t-value of 1.005 (p = 0.315), indicating that utilization is not significant for stunting prevention interventions. Finally, the relationship between UT and SHPs was not significant (original sample = 0.178, t-value = 1.598; p = 0.111), indicating that although utilization influences Sanitation and Hygiene Practices, this influence is not strong enough to be concluded statistically. A possible explanation is that utilization operates through unmeasured variables—namely nutrition education and cultural feeding norms—suggesting a need for extended models in future longitudinal work.

Overall, these findings confirm that food security plays an important role in supporting efforts to address stunting, with variations in the specific impacts of various factors. The significant contribution of availability (AV) and access (AC) highlights the importance of these variables in stunting prevention. Health Prevention Practices (HPPs) and Interventions to Prevent Stunting (IPSs) show a strong contribution to food security, while Toddler Eating Behaviour (EBT) and Sanitation and Hygiene Practices (SHPs) show more varied influences.

The research results show that the hypothesis about the positive impact of food security on stunting prevention is largely confirmed. Access (AC) has a significant influence on Toddler Eating Behaviour (EBT), which supports stunting prevention by ensuring toddlers receive adequate nutrition. Availability (AV) shows a strong influence on EBT, Health Prevention Practices (HPPs), Interventions to Prevent Stunting (IPSs), and Sanitation and Hygiene Practices (SHPs), highlighting the importance of adequate food supplies. Meanwhile, utilization (UT) shows a varied but significant influence on HPPs, indicating the importance of the appropriate use of resources in health practices. Overall, access to and availability of food resources play a key role in stunting prevention, and effective utilization is also important to achieve optimal results.

Figure 4 shows the effect of food security on stunting prevention behaviour in Bondowoso. AC exerted a medium effect (β = 0.26, p < 0.001) on Toddler Eating Behaviour, emphasizing the need for regular, affordable access to nutrient-dense foods for children under five years old. AV produced the largest path coefficients (β = 0.58–0.86), underscoring that a stable and diversified local food supply is the strongest lever for diet diversity, hygiene actions, and the uptake of community nutrition services. UT influenced only Health Prevention Practices (β = 0.23) and did not reach significance for diet or WASH outcomes, suggesting that utilization may be mediated by caregiver knowledge and cultural norms not captured in this model. Overall, our SEM results align with recent Indonesian and global studies, reinforcing food security as a foundation for multisectoral stunting strategies.

5. Discussion

The prevalence of stunting in Bondowoso (32.1%) is significantly higher than the national average (24.4%), underscoring the need for interventions tailored to local contexts. Consistent with multi-district studies in Java [13] and Sulawesi [1], our model confirms that improved household food security is associated with increased multidomain prevention practices and a reduced risk of stunting.

This study highlights the critical role of food security in preventing stunting in the Bondowoso community, Indonesia. Stunting is a severe public health concern that affects children’s growth and development. Our findings support the research hypothesis by demonstrating a significant relationship between food security and the effectiveness of stunting prevention efforts. Improved food security is associated with lower stunting rates, emphasizing the importance of stable access to nutritious food. In this context, food security includes the availability of, and access to, diverse and safe food, along with public education and awareness about the importance of nutrition. These findings reinforce earlier empirical evidence, such as that presented by Ruel et al. [33], which highlights the potential of food security interventions to reduce stunting in low- and middle-income countries.

Our findings align with Dewi et al. [27] who found that a one-unit increase in the food access score reduced the odds of stunting by 19%. The significant AC → EBT and AV → IPSs paths (β = 0.26 and 0.86, respectively) suggest that economic access and physical availability influence different behavioural pathways—dietary diversity and program participation. Complementary breastfeeding training, delivered by Posyandu cadres as recommended by Surmita et al. [20], remains an effective strategy in rural agricultural areas. The emphasis on empowering mothers as primary caregivers in feeding and hygiene practices is consistent with the findings of Bhutta et al. [34], who showed that family-based nutrition interventions are most effective when mothers are provided with adequate knowledge and access to resources.

Strengthening food security at both the household and community levels has proven essential in reducing stunting rates. Economic factors, such as household income, significantly influence food security and the prevalence of stunting [13,30]. Training programs for parents on complementary food preparation have shown promising outcomes in improving nutritional intake and reducing stunting [20,31]. Education and public awareness are also crucial in enhancing food security. Informing communities about the importance of nutrition ensures that children have access to nutrient-rich foods, thereby lowering the risk of stunting [9]. Numerous studies have confirmed the strong relationship between household food security and stunting [15]. This study also aligns with the findings of Pakaya, Kadir, and Kasim, [35] who emphasized that a multisectoral approach integrating nutrition, sanitation, and food security interventions is more effective than sectoral strategies alone in reducing stunting prevalence.

While AC and AV demonstrated medium-to-large effect sizes (f² ≥ 0.15), UT’s influence was more nuanced: it predicted Health Prevention Practices but did not significantly affect toddler diet or sanitation behaviours. Similar attenuation patterns have been documented in Ethiopian cohorts, where utilization effects were mediated by maternal nutrition knowledge [16]. This pattern suggests that cognitive and behavioural factors—absent from our current model—shape how households convert resources into child-focused actions. As noted by [36], effective stunting interventions must incorporate social and behavioural dimensions at both the family and community levels as a critical component of stunting reduction strategies.

AV exerted the strongest total effect (Σβ = 2.67), reinforcing calls for the village-level diversification of staple and protein sources. A recent quasi-experiment in West Java showed that maize–soybean intercropping combined with small-livestock support reduced moderate stunting by five percentage points in one year [28]. AV shows a strong influence on several variables such as EBT, COGS, IPSs, and SHPs. This emphasizes the importance of ensuring a stable supply of nutritious food to support healthy eating patterns and good health practices.

Meanwhile, resource utilization (UT) only has a significant influence on HPPs, which shows the importance of using appropriate resources in supporting the health of children under five. The non-significant UT → IPSs/SHPs paths may reflect measurement limits (self-report bias) or the absence of cultural variables (e.g., food taboos). Future longitudinal designs should incorporate objective diet-diversity scores and qualitative probes into feeding norms.

Overall, this research confirms that comprehensive and integrated factors, including food availability, access, and utilization, are critical to improving the health and well-being of children in communities vulnerable to stunting. Academically, this study contributes to strengthening the conceptual framework regarding food security and stunting at the microlevel (household), which has so far mostly been studied at the macro- or national scale.

Policy implications that can be implemented based on the results of this research include the following: a three-year action package for Wonosari could combine (i) food availability—subsidized orange-fleshed sweet-potato seed and small-ruminant starter kits; (ii) food access—conditional cash transfers for households below the poverty line tied to growth-monitoring visits; and (iii) food utilization—monthly cooking demos and WASH clinics led by Posyandu cadres. The indicative costs (USD 75,000 per village) and key performance indicators (≥25% increase in diet-diversity score; stunting reduction ≥3 pp y⁻¹) align with the East Java nutrition-convergence roadmap.

The limitations and future work related to this research include a reliance on caregiver self-reports, the moderate sample size, and the absence of cultural moderators, which may limit external validity. Triangulation with 24 h dietary recalls and a planned 2026 follow-up survey will address these gaps. Despite these caveats, our findings strengthen the evidence that boosting availability and access—while embedding utilization within culturally attuned education—offers the most realistic route to cutting stunting in rural East Java.

Overall, food security is key in preventing stunting. By ensuring good availability of, utilization of, and access to nutritious food, we can create an environment that supports the healthy growth and development of children, as well as reduce the stunting rates in society.

6. Conclusions

This study confirms that the “three-pillar” construct of food security—availability (AV), access (AC), and utilization (UT)—explains 57% of the variance in stunting prevention behaviour among Bondowoso households and supports the hypothesis that stronger food security predicts lower stunting risk. The hypothesis proposed in this research, which states that food security has a positive influence on stunting prevention, is largely confirmed through findings showing a significant relationship between food security and a reduction in stunting rates.

Overall, this research confirms that food security, which includes the availability of, access to, and utilization of food, is a key factor in supporting stunting prevention. Good access to food and the availability of sufficient food resources are essential to ensuring that children obtain the nutrition they need for optimal growth. Apart from that, the effective use of food also plays a role in supporting preventive health practices which can strengthen efforts to reduce stunting rates.

Policies that focus on increasing food access for families of children under five and providing sufficient nutritious food must be a priority. Nutrition education programs targeting parents of toddlers are also needed to ensure the optimal use of food. In addition, it is important to diversify local food and increase awareness about sanitation and hygiene as part of efforts to prevent stunting. The policy implications are as follows: AV—provide subsidized seed and small-livestock packages and establish village food banks to buffer lean-season shortages. AC—expand conditional cash/food-voucher schemes tied to growth-monitoring visits for families below the poverty line. UT—scale up Posyandu-led cooking demonstrations and hygiene clinics to translate resources into safe, diverse diets. The key performance target is as follows: ≥3 percentage-point annual reduction in stunting prevalence.

Limitations and future research: Self-reported behaviours, the moderate sample size, and the omission of cultural moderators may limit external validity. Future longitudinal and mixed-methods studies should triangulate with 24 h dietary recalls, the direct observation of WASH practices, and qualitative probes into feeding norms to test causal pathways and context-specific barriers.