Air Pollution and Vulnerability Combine to Raise Toddler Pneumonia Risk Across Java

A new study of Indonesia’s most densely populated island has revealed a stark pattern: the places where young children face the greatest threat of pneumonia are not simply the places with the dirtiest air, but the places where heavy pollution exposure collides with heightened social and industrial sensitivity. Analyzing data from across Java Island for 2023, researchers found that regions classified as moderately or highly vulnerable to air pollution were 2.67 times more likely to experience elevated prevalence of toddler pneumonia than regions in low-vulnerability clusters. The findings, published in the journal Air Quality, Atmosphere & Health, offer one of the clearest pictures yet of how environmental hazard and structural susceptibility interact to shape childhood respiratory disease in a rapidly urbanizing tropical setting.

The research team, led by Muhammad Fauzan Azima A and Robert Kurniawan of the Polytechnic of Statistics STIS in Jakarta, together with colleagues from Universitas Sumatera Utara and several Jakarta and Surabaya universities, chose Java for a reason that may seem paradoxical at first glance. Despite recording relatively low readings on the Air Quality Index compared with some of the world’s most notoriously polluted regions, Java carries an exceptionally high burden of pneumonia among children under five. That mismatch suggested to the authors that pollution concentrations alone cannot explain the island’s pediatric respiratory crisis, and that the vulnerability of the population, driven by dense settlement and industrial activity, must be built into any credible assessment of risk.

To capture that vulnerability, the researchers assembled a multilayered dataset covering five major air pollutants: fine particulate matter known as PM2.5, nitrogen dioxide, ground-level ozone, sulfur dioxide, and methane. Alongside these they incorporated climatic variables that influence how pollutants disperse, accumulate, or linger over the landscape, and two structural sensitivity factors, industrial density and population density, that describe how many people and how much polluting economic activity are concentrated in a given area. The conceptual framework draws on established vulnerability science, in which risk emerges not only from the hazard itself but from the exposure of populations and their sensitivity to harm, a perspective rooted in widely cited frameworks for vulnerability analysis in sustainability science.

The methodological heart of the study is a family of geographic clustering techniques designed to group districts and cities according to their overall air pollution vulnerability. Conventional clustering methods treat every observation as independent, which can scatter related regions into different categories and produce maps that make little spatial sense. The authors therefore tested a spatially constrained approach known as the Spatial Generalized Fuzzy C-Means method, or SGFCM. Fuzzy clustering, unlike hard classification, allows each region to hold partial membership in several clusters at once, reflecting the reality that vulnerability grades smoothly across administrative boundaries. Adding the spatial component penalizes classifications that place neighboring regions into wildly different categories, yielding contiguous, interpretable zones of risk.

After comparing clustering configurations, the team found that the SGFCM method with three clusters produced the most favorable evaluation results, striking the best balance between statistical fit and spatial coherence. The resulting map divided Java into low, moderate, and high pollution vulnerability zones. The geographic pattern was striking: clusters with elevated and moderate air pollution vulnerability were concentrated predominantly in the western and eastern portions of the island. Western Java encompasses the sprawling Jakarta metropolitan area, one of the largest urban agglomerations on Earth, while eastern Java hosts major industrial corridors and densely populated urban centers around Surabaya. The central corridor of the island, by contrast, contains more of the lower-vulnerability territory, much of it rural and less industrialized.

With the vulnerability zones established, the researchers turned to the central question of the study: does membership in a high or moderate vulnerability cluster actually correspond to higher rates of toddler pneumonia? They employed association analysis to test the relationship between cluster membership and pneumonia incidence categories, and then applied ordinal logistic regression, a statistical technique suited to outcome variables that fall into ordered categories, to quantify the strength of the relationship while accounting for the ordered nature of the pneumonia prevalence data. Ordinal regression has proven valuable in earlier health research for modeling progressive outcomes, and it allowed the team to estimate how the odds of falling into a higher pneumonia prevalence category changed with the vulnerability classification of a region.

The answer was unambiguous. Regions falling within the moderate or high air pollution vulnerability clusters were 2.67 times more susceptible to elevated toddler pneumonia prevalence than regions in the low pollution clusters. The study’s authors emphasize that this association does not single out one pollutant as the culprit; rather, it demonstrates that the composite condition of high exposure combined with high sensitivity, the double burden referenced in the study’s title, is what tracks most closely with the disease burden carried by the island’s youngest residents. Fine particulates such as PM2.5 are small enough to penetrate deep into developing lungs and have been linked in prior research to pneumonia hospitalizations in children across China, Southeast Asia, and beyond. Nitrogen dioxide, largely a byproduct of traffic and combustion, has previously been associated with acute respiratory tract infections in Indonesian children in earlier studies, giving biological plausibility to the spatial patterns the team uncovered.

What distinguishes this study from much of the existing literature is its treatment of place rather than individual exposure. By clustering whole districts and cities, the approach speaks directly to the geography of risk, identifying which territories deserve priority intervention. That matters for policy in a country where air quality management is coordinated nationally but experienced locally. Indonesia’s Ministry of Environment and Forestry has maintained an Air Pollutant Standard Index regime and a strategic plan targeting air pollution control, and the Ministry of Health publishes national health profiles tracking childhood pneumonia. Yet the new findings suggest that policies calibrated to AQI readings alone may systematically underestimate risk in places where dense populations and heavy industry amplify the consequences of a given pollution load. A moderate pollutant concentration in a district packed with families and factories may exact a far greater toll in pediatric pneumonia than a nominally higher concentration in a sparsely settled area.

The authors are careful about the limitations inherent in an ecological design. The analysis links regional-level environmental and structural indicators with regional-level disease statistics, so it cannot determine which individual children fell ill or disentangle every pathway connecting pollution to infection. Pneumonia in toddlers is multifactorial, influenced by nutrition, vaccination coverage, housing conditions, healthcare access, and indoor air quality from cooking fuels, factors that operate alongside ambient pollution. Data availability is also constrained; the team notes that the underlying data will be made available on reasonable request, reflecting the reliance on official statistics and remote sensing-derived environmental measurements. Still, the consistency of the association, with nearly a threefold elevation in the odds of high pneumonia prevalence in vulnerable clusters, is difficult to dismiss as statistical noise.

The implications stretch well beyond Java. Rapidly urbanizing regions across South and Southeast Asia face the same configuration of dense population, expanding industry, and rising pollutant loads, and the Health Effects Institute’s State of Global Air reporting has repeatedly flagged children’s health as a central casualty of the region’s air quality crisis. The Java study offers those regions a transferable template: spatially constrained fuzzy clustering of exposure and sensitivity data can generate actionable vulnerability maps even where dense networks of ground-level pollution monitors are lacking, because satellite-derived pollutant estimates can stand in for sparse instrumentation. For Indonesian policymakers, the immediate message is concrete. The western and eastern vulnerability belts identified by the study, home to tens of millions of people and countless children under five, are the territories where targeted measures, from industrial emission controls and traffic management to strengthened pediatric respiratory services, are most likely to reduce the island’s stubborn burden of toddler pneumonia. The study’s authors frame their work as a direct aid to stakeholders crafting air pollution management policy, and with a nearly threefold difference in disease risk separating the most and least vulnerable zones, the map they have drawn is one that public health officials in Jakarta, Surabaya, and beyond can hardly afford to ignore.

Subject of Research: Spatial association between air pollution exposure-sensitivity clusters and toddler pneumonia incidence on Java Island, Indonesia

Article Title: The double burden of air pollution exposure and sensitivity: association with toddler pneumonia in high-density areas of Java

Article References: Azima. A, M. F., Kurniawan, R., Gio, P. U., Hikmah, H., Wongsonadi, S. K., Lestari, G. D., & Yuniarto, B. (2026). The double burden of air pollution exposure and sensitivity: association with toddler pneumonia in high-density areas of Java. Air Quality, Atmosphere & Health, 19(9), Article 203. https://doi.org/10.1007/s11869-026-02088-0

Image Credits: AI Generated

DOI: 10.1007/s11869-026-02088-0

Keywords: air pollution, toddler pneumonia, spatial clustering, fuzzy c-means, PM2.5, nitrogen dioxide, Java Island, vulnerability, ordinal logistic regression, Indonesia, public health, urban air quality