Investigators linked a 2018–2019 satellite-derived registry of 6,978 brick kilns to 3,149 displaced clusters from the Bangladesh Multiple Indicator Cluster Survey (MICS) 2025. The exposure was defined as the number of kilns within a 5 km radius of each cluster, a distance chosen to accommodate the public cluster displacement applied to protect privacy. Three in four children (75.4%) in the sample lived within 5 km of at least one kiln, and about one in five (20.3%) resided near ten or more kilns. The study team used survey weights and district fixed effects in analytical models; the report did not provide household-level exact coordinates because MICS cluster locations are displaced in the public dataset.
The primary respiratory outcomes were caregiver-reported cough and caregiver-reported acute respiratory infection (ARI) symptoms among 23,357 children under five. Comparing clusters with ten or more kilns to clusters with no kilns, the adjusted prevalence ratio (PR) for cough was 1.05 (95% CI 0.96–1.16). For ARI symptoms the adjusted PR for the same comparison was 1.21 (0.90–1.62). Per 10 additional kilns in the 5 km buffer the ARI PR was 1.11 (1.01–1.21) in a linear model; however, the authors fitted a spline across kiln counts and reported no consistent association (spline P = 0.153), indicating that the apparent linear effect may not hold across the distribution. Models included district fixed effects to account for broad geographic confounding; the source did not report additional individual-level covariates in the abstract, so readers should consult the full manuscript for the complete model specification.
Venous blood lead concentration was measured in 10,622 children aged 12–59 months. The investigators reported the proportion of children with blood lead ≥5 µg/dL and estimated associations with kiln density. In crude analyses, higher kiln counts were associated with increased prevalence of elevated blood lead (crude PR 1.27, 1.14–1.41 comparing ten or more kilns to none). After adjustment, the association attenuated and crossed the null (adjusted PR 0.94, 0.85–1.03 for ten or more kilns versus none). The lead result remained null across the described sensitivity analyses. The source did not report alternative lead thresholds or detailed biomarker distributions in the abstract; those elements may appear in the full text or supplement.
The study exploited timing of interviews relative to the kiln firing season as an internal check. Cough prevalence showed no association with kiln density during the reported peak firing period (PR per 10 kilns 1.00, 0.96–1.05) but did show an association in May–June (PR per 10 kilns 1.10, 1.04–1.17). Investigators used diarrhoea and interview timing as negative controls. These patterns—respiratory signals that are absent at peak firing yet present in other months—led the authors to interpret the findings as more consistent with confounding by place (for example, unmeasured community-level factors that correlate with kiln presence) than with a direct, acute smoke effect on respiratory symptoms measured by caregiver report.
This analysis leveraged a large, nationally representative survey with venous blood lead measures and an independently generated kiln registry, enabling geospatial linkage at scale. The sample sizes reported in the abstract were 23,357 children for respiratory outcomes and 10,622 children for venous blood lead. Key limitations arise from the public MICS cluster displacement and the 5 km exposure buffer used to accommodate that displacement; such choices dilute spatial precision and reduce ability to detect highly localized effects. The authors also reported that exposure frequency increased with household wealth, which complicates simple socioeconomic confounding adjustments. The abstract did not list detailed individual-level covariates or model covariate sets; those were not reported in the source abstract and require consultation of the full preprint.
The investigators concluded that, at the spatial resolution available in MICS public data, living near more kilns was not associated with higher venous blood lead and produced only weak respiratory associations that did not align with peak firing seasons. They recommended that future studies obtain exact household locations and evaluate exposures within narrower buffers—on the order of 1 to 2 km—to better capture local pollutant gradients and acute smoke impacts. That specific recommendation about household-level coordinates and 1–2 km buffers is stated directly by the authors and represents the primary methodological gap identified by the analysis.