Mapping paediatric acute gastroenteritis in Talavera, Nueva Ecija, Philippines: a cross-sectional study on prevalence and community risk factors

Abstract

Aim

To map the prevalence of paediatric acute gastroenteritis (AGE) in Talavera, Nueva Ecija, Philippines and to examine how community-level risk factors shape its spatial distribution.

Methods

We conducted a single-centre cross-sectional study with spatial analysis at Talavera General Hospital in 2023. Children aged 0–17 years diagnosed with AGE were included. Barangay-level prevalence was mapped, and spatial autocorrelation was assessed using Moran’s I and Local Indicators of Spatial Association. Ordinary least squares (OLS) and geographically weighted regression (GWR) were used to evaluate associations between community risk factors and AGE prevalence.

Results

A total of 398 paediatric AGE cases were included, mainly among male children under 5 years of age. Paediatric AGE prevalence ranged from 1.74 to 22.34 cases per 1000 children, with clear geographic variability. Significant spatial clustering was observed (Moran’s I=0.268, p=0.001), with high-risk clusters in southwestern barangays and low-risk clusters in the north. GWR outperformed OLS and revealed localised associations: household income below the poverty line (25 barangays) and population density (13 barangays) showed positive effects, while distance to the hospital showed a negative association across the municipality (t-values: −2.11 to −3.35).

Conclusion

This study demonstrates significant spatial heterogeneity in the paediatric AGE burden in Talavera, with distinct high-risk clusters and spatially varying associations. The findings support geographically targeted interventions addressing local poverty and population density as well as strengthening access to healthcare in remote areas to improve case detection and reduce under-reporting.

WHAT IS ALREADY KNOWN ON THIS TOPIC

• Paediatric acute gastroenteritis (AGE) remains a major cause of morbidity in low and middle-income countries, including the Philippines. Diarrhoeal disease risks are shaped by community-level socioeconomic and environmental conditions; however, spatial patterns often vary across local settings. Despite this, barangay-level spatial analyses are rarely conducted to inform municipal-level public health planning, limiting geographically targeted interventions.

WHAT THIS STUDY ADDS

• This study provides the first barangay-level spatial analysis of paediatric AGE in Talavera, Nueva Ecija, identifying significant geographic clustering and variation in prevalence within the municipality. It demonstrates that healthcare access, poverty and population density are key spatial predictors of AGE prevalence. Contrary to expectations, environmental factors, such as flood risk, water sources and toilet access, were not found to be significant in spatial regression analyses.

HOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE OR POLICY

• This study demonstrates the importance of integrating local spatial analysis into paediatric infectious disease research and municipal health planning. It supports geographically targeted interventions over uniform approaches. It also provides evidence for strengthening healthcare systems in remote areas and offers a framework for community-level risk assessment applicable to other diseases.

Introduction

Acute gastroenteritis (AGE) among children under five remains a major public health concern globally. It is predominantly caused by viral, bacterial or parasitic infections, leading to diarrhoea, vomiting and fever.^{1 2} In low-resource settings, AGE contributes substantially to childhood morbidity and mortality, particularly where structural, environmental and socioeconomic vulnerabilities persist.2,4

Globally, the epidemiological profile and seasonality of AGE vary across regions. In countries such as Iraq and Italy, paediatric AGE persists as a recurrent health issue, with seasonal surges placing strain on health systems. In Iraq, seasonal peaks are linked to environmental and infrastructural factors,⁵ while in Italy, rotavirus accounts for over half of paediatric AGE hospitalisations in young children, most prominently during winter.⁶

Multiple factors shape the global burden of paediatric AGE. These include poor water and sanitation infrastructure, undernutrition, suboptimal immunisation coverage and delayed access to healthcare. Socioeconomic inequalities further compound these risks. Malnourished children, particularly those who are stunted or wasted, exhibit weakened immune responses, increasing their susceptibility to prolonged or severe illness.⁷ Environmental exposures such as flooding, inadequate drainage and a lack of safe drinking water also play a significant role.^{3 8} Similar patterns are observed in marginalised communities in LMICs, such as rural Chiapas in Mexico, where high rates of stunting and parasitic infections are strongly associated with extreme poverty.⁹

In the Philippines, AGE remains one of the leading causes of childhood morbidity,¹⁰ with diarrhoeal diseases are reported at morbidity rates of approximately 515 per 100 000 and affect about 8.39% of children under 5 years of age nationwide.¹¹ Despite the availability of oral dehydration therapy and efforts to scale-up rotavirus vaccination, uneven uptake and coverage contribute to geographic disparities in disease outcomes.^{12 13} Postdisaster settings, such as those following Typhoon Haiyan, have seen AGE outbreaks linked to disrupted water and sanitation services.¹⁴ Health service delivery in rural areas is often compromised, particularly in times of flooding, when physical access to care and continuity of essential supplies are compromised.^{15 16}

However, the current understanding of AGE in the Philippines remains limited at the subnational level. Existing data are mostly facility-based and do not account for community-level variation in risk. Health infrastructure varies, with barangay health stations differing in staffing, supplies and resilience to climate-related disruptions. Caregivers often delay seeking treatment due to cost, distance or low trust in the health system.^{15 16} Although the environmental and social determinants of AGE are widely recognised, their spatial interactions and localised effects remain poorly characterised. Geospatial techniques such as spatial autocorrelation and geographically weighted regression (GWR), widely applied in epidemiological research elsewhere,1217,19 remain underutilised in public health planning in the Philippines.

Therefore, this study aims to map the spatial distribution and prevalence of paediatric AGE across the 53 barangays of Talavera, Nueva Ecija, Philippines, in 2023, and to identify community-level risk factors associated with spatial variation in disease burden. Findings are expected to identify high-risk barangay clusters and inform geographically targeted strategies that respond to the community factors shaping paediatric risk in the municipality.

Methods

Study design and population

We conducted a single-centre cross-sectional study with spatial analysis to examine the 2023 distribution of paediatric AGE and its community-level risk factors across the 53 barangays of Talavera, Nueva Ecija, Philippines.

This study was carried out at Talavera General Hospital (TGH), an ISO-certified Level II public hospital and the sole hospital serving the municipality, providing both direct access and referral care from other health facilities that require inpatient management.

All children aged 0–17 years diagnosed with AGE from January to December 2023 were considered for inclusion. AGE was defined based on the clinical diagnosis recorded in hospital medical charts, consistent with standard paediatric diagnostic practice at TGH. We excluded non-residents of Talavera, as well as readmissions, emergency revisits or follow-up encounters occurring within 30 days of a previous AGE-related discharge, to avoid duplication and ensure inclusion of only new acute episodes, consistent with standard diarrhoeal episode definitions used in epidemiologic studies and clinical classifications distinguishing acute from persistent and chronic diarrhoea.^{20 21} A unique hospital identification number was used to identify and exclude duplicate records. These identifiers were accessible only during extraction and were removed before analysis to ensure complete anonymisation. A total of 398 unique AGE cases met the criteria. Figure 1 summarises the case selection process.

Figure 1. Flow diagram of AGE cases included in the study. Consultations occurring within 30 days of a previous AGE-related discharge were excluded. AGE, acute gastroenteritis; TGH, Talavera General Hospital.

Data collection

Talavera is a landlocked, predominantly agricultural municipality in the province of Nueva Ecija, located in Central Luzon (Region III), Philippines, on the island of Luzon (figure 2). It has a total land area of 140.92 km², representing approximately 2.48% of the total land area in Nueva Ecija and is administratively divided into 53 barangays. In 2023, Talavera had a paediatric population (<18 years) of 41043, comprising 19897 females and 21043 males, and a total of 34820 households, according to the Municipal Development and Planning Office (MDPO)-Talavera. This paediatric population served as the denominator for calculating barangay-level AGE prevalence.

Figure 2. Study area: Talavera, Nueva Ecija, Philippines. Talavera is located in Central Luzon (Region III) on the island of Luzon.

Hospital records from TGH were reviewed to obtain the characteristics of all eligible cases. Extracted variables included age, sex, barangay of residence, symptoms, onset of illness and hospital stay duration. Barangay-level data were gathered from the MDPO, including the proportion of households living below the poverty line, those without access to sanitary toilets and the distribution of primary water sources. Only piped water and borehole categories were retained in the analysis due to data completeness issues.

Population density was calculated using barangay population counts and land areas derived from official shapefiles available through the Humanitarian Data Exchange. Flood-risk data were obtained from GeoAnalyticsPH by overlaying high-hazard zones with barangay boundaries to determine the proportion of each barangay classified as highly flood-prone. The distance to the hospital (km) was calculated as the Euclidean distance between barangay centroids derived from official shapefiles and the geographic location of TGH, using spatial analysis in R.

Data analysis

Descriptive analyses were used to summarise the characteristics of paediatric AGE cases. Barangay-level AGE prevalence was calculated as the number of unique paediatric AGE cases recorded in 203 divided by the corresponding barangay paediatric population (<18 years) in 2023, expressed per 1000 paediatric population and visualised through a choropleth map. Potential reporting bias was anticipated due to the use of facility-based hospital data, particularly for children residing in barangays farther from the hospital.

All spatial analyses were performed using R (V.4.5.0). Spatial relationships across barangays were defined with a Queen contiguity spatial weights matrix. Global spatial autocorrelation was assessed using Moran’s I with permutation-derived p values (95% CI). Local Indicators of Spatial Association (LISA) were used to identify high–high, low–low and spatial outlier clusters.

Ordinary least squares (OLS) regression was used to evaluate global associations between paediatric AGE prevalence and community-level risk factors. Variance inflation factors (VIF) were calculated to detect multicollinearity, and Moran’s I was used on model residuals to check for remaining spatial dependence.

GWR was then used to capture spatially varying effects. An adaptive bisquare kernel and AICc-based bandwidth selection were applied. Local coefficients were considered statistically significant at t values greater than 1.96 or less than −1.96.

Model performance of the OLS model and GWR was compared using multiple criteria: (1) coefficient of determination (R²), (2) Akaike information criterion (AIC), (3) corrected AIC (AICc), (4) Bayesian information criterion (BIC) and (5) residual sum of squares (RSS). Lower information criteria and higher R² were interpreted as better model fit.

Patient and public involvement

Patients and members of the public were not involved in the design, conduct, reporting or dissemination plans of this research.

Results

Profile of the paediatric AGE cases in Talavera

Out of 1311 total hospital records from 2023 at TGH, 398 met the inclusion criteria. Among the 398 AGE cases involving Talavera residents, 171 (43.0%) were female, and 227 (57.0%) were male, indicating a higher burden among male children. This corresponds to an overall paediatric AGE prevalence of 9.72 cases per 1000 children (8.59 per 1000 among females and 10.73 per 1000 among males).

The mean age of paediatric AGE cases was 5.22 (±3.47) years. With more than half, 57.3%, aged <5 years. Vomiting (75.9%) was the most commonly reported symptom, followed by fever (26.6%), abdominal pain (17.3%) and body weakness (2.8%). The mean onset of symptoms before consultation was 24.7 hours. Among the 132 (33.2%) admitted patients, the mean length of stay was 79.5 (±36.22) hours, while non-admitted cases (n=266) had a mean emergency stay of 4.47 hours (table 1).

Table 1. Characteristics of paediatric AGE cases (n=398).

Characteristics	n (%)
Sex
Male	227 (57.0)
Female	171 (43.0)
Age group (mean=5.22, SD=±3.47)
<5	228 (57.3)
5–8	108 (27.1)
9–12	40 (10.1)
13–17	22 (5.5)
Signs and symptoms
Vomiting	302 (75.9)
Fever	106 (26.6)
Abdominal pain	69 (17.3)
Body weakness	11 (2.8)

AGE, acute gastroenteritis.

Paediatric AGE prevalence

Figure 3 presents the spatial distribution of paediatric AGE prevalence across the 53 barangays of Talavera in 2023. Prevalence ranged widely from 1.74 (Bagong Sikat) to 22.34 (Sibul) cases per 1000 children, showing clear geographic variability.

Figure 3. Prevalence map of paediatric AGE in Talavera, Nueva Ecija, 2023. AGE, acute gastroenteritis; Pob., Poblacion (downtown area)).

Several barangays in both central and southwestern parts of the municipality recorded prevalence above the municipal average of 9.72 per 1000, including Sampaloc (19.06), Collado (18.53), Pinagpanaan (16.54), Purok Matias (17.20), Maestrang Kikay (13.73), Mamandil (13.30), Marcos District (11.49), Esguerra District (11.46), Poblacion Sur (12.65) and Pag-asa (12.43). Lower prevalence values were observed in northern barangays, including Bagong Sikat (1.74), Dimasalang Sur (2.03), Dimasalang Norte (2.05), Bacal I (2.22), Matingkis (2.36), Kinalanguyan (2.55) and Mabuhay (3.50). The complete list of Barangay-level paediatric AGE prevalence is provided in online supplemental data 2.

Spatial autocorrelation analysis of paediatric AGE

Global spatial autocorrelation analysis demonstrated statistically significant clustering of paediatric AGE prevalence across Talavera. Global Moran’s I was 0.268 (expected=−0.019; p=0.00105), indicating that barangays with similar prevalence were more spatially proximate than expected under spatial randomness (figure 4).

Figure 4. Global Moran’s I and LISA cluster map of paediatric AGE prevalence in Talavera. AGE, acute gastroenteritis; LISA, Local Indicators of Spatial Association.

LISA identified distinct local clusters of paediatric AGE prevalence. A high–high cluster was observed in the southwestern area, consisting of adjacent barangays with similar elevated prevalence, including Andal Alino, Esguerra District, Marcos District, Maestrang Kikay District and Poblacion Sur. A low–low cluster was identified in the northern part of Talavera, represented by Barangay Bakal II.

One high–low cluster was detected in San Ricardo, where a barangay with high prevalence bordered areas with lower prevalence. All remaining barangays showed no significant spatial association.

Community risk factors

Figure 5 illustrates the spatial distribution of community-level risk factors across the 53 barangays of Talavera. Travel distance to TGH (figure 5-1) was shortest in the central barangays and increased toward the northern and southern edges of the municipality. Population density (figure 5-2) was highest in the central area and declined outward, with the lowest densities observed in the northern barangays. Household income below the poverty line (figure 5-3) varied across the municipality without a pronounced spatial concentration, with both higher and lower proportions appearing in multiple dispersed locations. Water source patterns varied across barangays.

Figure 5. Spatial distribution of community-level risk factors across the 53 barangays of Talavera. (5-1) distance to hospital, (5-2) population density, (5-3) income below poverty line, (5-4) direct piped water, (5-5) borehole reliance, (5-6) households without sanitary toilet and (5-7) flood risk.

Direct piped water coverage (figure 5-4) was highest in many central and southern barangays, while reliance on boreholes (figure 5-5) was more common in parts of the northern and northeastern areas. Access to sanitary toilets (figure 5-6) was generally high, with Pantoc Bulac standing out as the only barangay with an elevated proportion of households without access to sanitary toilets. Flood-risk distribution (figure 5-7) was greatest in the northwestern barangays and decreased gradually towards the southeast. The complete list of community risk factor variables is provided in online supplemental data 2.

OLS regression and GWR analysis of community risk factors of AGE prevalence

The OLS model showed a moderate level of explanatory power (R²=0.332; adjusted AR²=0.228; RSS=788.01; AIC=311.47; AICc=315.65; BIC=311.93). Among the predictors, distance to the hospital was the only statistically significant risk factor (β=−0.78, p<0.001). All other variables did not show significant global associations (table 2).

Table 2. Summary of GWR, OLS regression results and variance inflation factors (VIF).

Variables	GWR coefficients			OLS	VIF
	Min	Median	Max	β
Intercept	3.434	5.930	7.947	6.3939	–
Distance to hospital	−0.961	−0.023	−0.002	−0.7790	1.79
Population density	0.000066	0.00105	0.002200	0.00119	1.25
Household income below poverty line	0.088	0.148	0.231	1.1236	1.45
Direct piped water	−0.089	−0.023	−0.002	−0.0232	3.08
Borehole	0.023	0.031	0.055	0.0300	2.48
Without access to a sanitary toilet	−0.233	−0.041	−0.004	−0.0090	1.21
Flood risk	0.016	0.058	0.86	0.0644	1.09
Residual sum of squares (RSS)	670.00			788.01
R2	0.432			0.332
Adjusted R2	0.193			0.228
AIC	297.64			311.47
AICc	323.05			315.65
BIC	282.57			311.93

AIC, Akaike information criterion; AICc, corrected AIC; BIC, Bayesian information criterion; GWR, geographically weighted regression; OLS, ordinary least squares.

GWR demonstrated spatial variation in the strength and direction of associations of community risk factors. Local coefficients for population density, household income below the poverty line, borehole reliance and flood risk were consistently positive across barangays, although their magnitudes varied across barangays. In contrast, the coefficient of distance to hospital, direct piped water and households without access to a sanitary toilet remained negative throughout the municipality.

The GWR model captured more spatial structure with a higher R² (0.43) and lower RSS (670.0) than the OLS model. Model-selection criteria pointed in slightly different directions. AIC and BIC were both lower for GWR, while AICc is higher. The higher AICc value for GWR reflects the penalty applied for increased model complexity due to barangay-specific parameter estimation. Although GWR improved model fit, the improvement was not sufficient to fully offset the penalty for additional parameters, indicating the trade-off between increased model complexity and explanatory gain.

VIF values ranged from 1.09 to 3.08, showing no serious multicollinearity, and OLS estimates were reasonably stable. The presence of spatial heterogeneity in OLS residuals justified applying GWR to explore how community risk factors operate differently across barangays and was used as an explanatory tool to identify spatial heterogeneity rather than as a strictly predictive model.

Figure 6 illustrates the spatial distribution of local GWR coefficients and their statistical significance for each community-level predictor of paediatric AGE prevalence across 53 barangays of Talavera.

Figure 6. GWR coefficient map of the community-level risk factors of AGE in Talavera, Nueva Ecija. Red dots (∙) indicate statistically significant negative associations (t<−1.96); blue dots (∙) indicate statistically significant positive associations (t>1.96). AGE, acute gastroenteritis; GWR, geographically weighted regression.

Distance to the hospital (figure 6-1) showed a consistent spatial pattern. All barangays exhibited statistically significant negative local coefficients, with t-values ranging from −2.11 to −3.35. Although the negative magnitude of the effect varied geographically, the direction remained uniformly negative across the municipality.

Population density (figure 6-2) demonstrated significant positive local effects in 13 barangays located mostly in the northeastern part of Talavera. These barangays had t-values above 1.96, indicating that higher population density was associated with higher AGE prevalence in localised areas rather than uniformly across the municipality.

For income below the poverty line (figure 6-3), spatially variable effects were observed. A total of 25 barangays displayed statistically significant local associations, forming clusters primarily in the central and northeastern areas of the municipality. All significant coefficients were positive (blue dots), indicating that higher proportions of households with income below the poverty line were associated with increased paediatric AGE prevalence in these areas, reflecting the locally varying influence of poverty across the municipality.

Predictors related to water source (figure 6-4,5), households without a sanitary toilet (figure 6-6), and flood risks (figure 6-7) did not exhibit statistically significant local effects. While their coefficients varied across space, all t-values remained within non-significant ranges, indicating limited evidence of spatially varying relationships for these factors. The complete list of GWR t-value is provided in online supplemental data 1.

Discussion

This study provides the first spatial analysis of paediatric AGE burden across Talavera, Nueva Ecija, Philippines, revealing significant geographic heterogeneity in disease distribution and community-level risk factors. Our findings demonstrate four key insights: (1) the characteristics of paediatric AGE cases in Talavera are consistent with global patterns, including higher burden in children under 5 years and male predominance; (2) substantial geographic variation in paediatric AGE prevalence across barangays (1.74–22.34 per 1000 children) with evidence of significant spatial clustering; (3) spatially varying associations between community risk factors and AGE prevalence, with poverty and population density showing localised effects while distance to hospital consistently showed negative associations; and (4) evidence of healthcare access disparities that may contribute to spatial clustering patterns.

The predominance of cases in children under 5 years (57.3%) is consistent with global surveillance data from LMICs, where children under five bear a disproportionate AGE burden.⁴ This increased vulnerability reflects a combination of biological and behavioural factors, including immature function, lack of prior pathogen exposure and greater susceptibility to dehydration. Young children also have higher exposure risk due to frequent hand-to-mouth behaviours and dependence on caregivers for feeding and hygiene, particularly in settings with suboptimal sanitation.^{22 23} The higher proportion among males (57%) supports the evidence of sex-based differences in acute paediatric diarrhoea.²⁴ The common occurrence of vomiting as the presenting symptom (75.9%) reflects established AGE pathophysiological patterns. Clinical studies have consistently identified vomiting as the most common presenting symptom in paediatric gastroenteritis, with duration patterns similar to our findings.²² Although rotavirus vaccination has been introduced in the Philippines, coverage remains variable across regions, and incomplete uptake may contribute to the continued burden of AGE among young children, as rotavirus remains a leading cause of AGE in this age group.²⁵

The study findings collectively indicate that the timing of care-seeking and subsequent hospitalisation for paediatric AGE imposes a considerable burden on both households and the healthcare system. The mean duration of 24.7 hours before consultation suggests that many caregivers seek care only after symptoms persist, a pattern linked to increased illness severity and transmission as observed in low-resource settings.²⁶ This timing still falls within the typical course of AGE, where vomiting often resolves within 1–2 days and diarrhoea within 5–7 days and aligns with National Institute for Health and Clinical Excellence guidance on gastroenteritis in young children.²⁷ The observed admission rate and multiday hospital stays among paediatric AGE cases reflect the substantial healthcare burden associated with severe presentations requiring inpatient management. These findings are consistent with the evidence showing that inpatient paediatric AGE in LMICs is associated with multiday admissions, substantial direct costs to both health systems and households.²⁸ In the Philippines, this pattern of precautionary consultation and admission reflects limited home monitoring capacity and heightened vulnerability of young children, which often prompts earlier medical care once symptoms progress.¹⁵ This burden is commonly observed in children under 5 years, who are more susceptible to rapid dehydration and complications, often necessitating closer clinical observation and supportive care.

We confirm that paediatric AGE exhibits significant spatial clustering, with a Moran’s I of 0.268 (p=0.001), indicating a non-random distribution across barangays. This finding aligns with global evidence demonstrating spatial heterogeneity in diarrhoeal disease burden. Studies from South Asia have similarly documented spatial clustering patterns in rotavirus transmission, highlighting the importance of local environmental conditions in disease distribution.¹⁸ Research from India has also identified spatial risk patterns for AGE, emphasising the role of local geographic factors in disease mapping and intervention targeting.¹⁹

The identification of high-high prevalence clusters in southwestern barangays contrasts with low–low clusters in northern areas, suggesting that AGE risk is shaped by local conditions rather than uniform exposure patterns. This geographic heterogeneity is consistent with findings from European outbreak investigations, where municipal water supply contamination created distinct spatial patterns of gastroenteritis incidence.²⁹ Our results extend these observations by demonstrating that even in endemic settings, spatial clustering reflects local environmental and infrastructural conditions.

OLS regression offered an initial view of the global association between community risk factors and AGE prevalence. The model showed a moderate level of explanatory power (R²=0.332), with the distance to the hospital emerging as the only significant predictor. The GWR model demonstrated improved explanatory power and was considered superior compared with the global OLS model to capture spatially varying relationships that global OLS cannot detect. However, its AICc was higher than that of OLS, indicating that the added model complexity is not fully justified by an improvement in overall model fit. Even so, the spatially varying coefficients produced by GWR offer valuable local insights, demonstrating that geographic heterogeneity is essential in understanding AGE distribution in Talavera.

Our GWR analysis revealed spatially varying associations that offer important insight into community-level AGE risk factors. Poverty demonstrated a significant positive association with AGE prevalence in 47.2% of barangays (25 of 53), forming clear clusters primarily in the central and northeastern zones of Talavera. This pattern aligns with extensive global evidence identifying poverty as a fundamental determinant of diarrhoeal disease burden.³ The localised nature of these associations suggests that socioeconomic disadvantage influences AGE risk through area-specific pathways rather than uniform exposure across the municipality. The clustering of poverty-related AGE burden likely reflects intersecting vulnerabilities, including limited access to clean water and sanitation, overcrowded living environments conducive to transmission and financial barriers that reduce timely healthcare-seeking behaviour. Similar spatially concentrated poverty effects have been documented in Southeast Asia, where socioeconomic deprivation creates hotspots of infectious disease transmission.³ These findings indicate that targeted poverty alleviation efforts, paired with WASH and infrastructure improvements in high-risk clusters, may be more effective than municipality-wide approaches in addressing the socioeconomic determinants of paediatric AGE. The spatial heterogeneity observed indicates that AGE risk is shaped by localised conditions rather than uniform municipal-level effects. This signifies the importance of spatial analysis in informing geographically targeted public health interventions, community surveillance, resource allocation and health education to be prioritised in barangays with the greatest localised vulnerability.

The variation in population density effects across barangays reflects complex interactions between urbanisation, environmental conditions and disease transmission. Studies from Southeast Asia have documented similar spatially varying effects of population density on infectious disease transmission and emphasise the importance of localised risk assessment.¹⁷ Our findings suggest that population density effects are not uniform but depend on local environmental and infrastructural conditions. In Talavera, disparities in access to direct piped water, sanitary toilets and flood exposure likely interact to shape localised AGE risk. Barangays with limited piped water access may rely on alternative water sources more vulnerable to contamination, particularly during flooding events, while inadequate sanitation infrastructure can facilitate faecal-oral transmission through environmental pathways. Flood-prone areas may further amplify these risks by disrupting water and sanitation systems and increasing exposure to contaminated surface water. These interconnected environmental vulnerabilities suggest the importance of an integrated water, sanitation and flood mitigation strategy in reducing AGE transmission at the community level.

The consistent negative association between distance to hospital and AGE prevalence across all barangays likely reflects healthcare access barriers, leading to under-reporting rather than a true protective effect. This pattern aligns with the critical importance of geographic accessibility in healthcare utilisation, particularly in rural and remote areas of the Philippines, where healthcare infrastructure is limited.¹⁵ Research in the Philippines from similar settings has documented barriers related to travel distance, transportation and cost, particularly for paediatric conditions requiring immediate care.¹⁶ These findings suggest the importance of community-based surveillance in remote barangays to address potential reporting gaps.

Limitations include the reliance on facility-based data from TGH, which introduces reporting bias because cases in remote barangays may be under-represented.¹⁶ This is especially relevant to the negative association between distance to the hospital and AGE prevalence, which likely reflects access barriers rather than a true protective effect. This under-reporting may also influence the observed spatial clustering patterns. Future incorporation of community-based sentinel surveillance could improve case detection and provide a more accurate spatial presentation. The cross-sectional design also limits causal interpretation, and using only 2023 data may overlook seasonal and longer term trends. AGE exhibits seasonal and interannual variability influenced by climate and pathogen circulation, which may limit the generalisability of single-year findings. Future multiyear analyses would strengthen temporal and spatial interpretation. Environmental variables were measured in broad categories that may miss more nuanced transmission pathways. The absence of individual-level socioeconomic and behavioural data leaves room for unmeasured confounding. In addition, AGE classification was based on physician-documented clinical diagnoses recorded in hospital charts, as a laboratory confirmation and standardised surveillance case definitions were not consistently available in routine hospital records. Also, the specific aetiologic agents responsible for AGE were not identified, limiting the ability to assess pathogen-specific transmission patterns, including rotavirus and other common viral or bacterial causes. Age standardisation was not performed due to the unavailability of age-stratified population data at the barangay level. Given the higher burden of AGE among children under 5 years, variation in age structure across barangays may have influenced the observed patterns. Sensitivity analysis to assess the stability of identified spatial clusters in barangays with small case counts was not conducted. Future studies incorporating age-standardised rates and cluster stability assessments help improve the robustness and interpretability of spatial patterns.

The results indicate the need for community-based surveillance in barangays far from the hospital to reduce reporting gaps and enhance case detection. Expanding barangay-level outreach activities, particularly in remote communities, may help bridge access limitations that contribute to under-reporting. Improving healthcare access in remote areas through transport enhancements and mobile services may help reduce delays in care seeking.¹⁵ Interventions should focus on high-high clusters in southwestern barangays, linking water, sanitation and hygiene improvements with poverty alleviation strategies.³ Future work would benefit from longitudinal designs and more detailed environmental and individual-level data. Finally, incorporating geospatial analysis into routine health planning can support more targeted resource allocation that reflects local variations in disease burden. Although the findings are specific to Talavera, Nueva Ecija, the spatial analytical approach and observed community-level patterns may apply to similar rural municipalities in the Philippines and other LMICs. Thus, this study confirms the presence of localised socioeconomic and healthcare access disparities influencing paediatric AGE distribution and gives importance to spatial analysis for targeted public health interventions to improve disease surveillance in underserved communities.

Data availability statement

Data are available upon reasonable request.