Spatial neighborhood patterns of pulmonary tuberculosis in a large urban area: the case of Santiago, Chile

Abstract

In recent years, global concern about tuberculosis (TB) has grown due to the slow progress in achieving control goals, with a record number of cases reported worldwide in 2022. This study analyzed the spatial distribution of all pulmonary TB cases reported between 2016 and 2020 in the Gran Santiago (n = 3348), Chile’s largest urban area. Cases were geocoded using an automated cascade method (Bing Maps, Google Maps, and manual verification) and aggregated at the neighborhood level. To detect spatial clusters, we estimated gender-adjusted rates, assessed spatial autocorrelation using Moran’s I, and identified spatial clusters with Flexible Scan Statistics. Bivariate analyses were used to explore associations with demographic and residential variables. We observed higher incidence rates in males compared to females (13.23 vs. 7.03 per 100,000), with no statistically significant changes over time. A total of 29.4% (n = 984) of all cases were concentrated in 11 significant spatial clusters. These clusters were characterized by a higher proportion of immigrant, Indigenous, and older populations, along with indicators of precarious living conditions, including overcrowding and tenement housing. Identifying spatial patterns of pulmonary TB can support the design of targeted, community-based case-finding strategies in urban settings, particularly in populations at higher social and structural risk.

Supplementary Information

The online version contains supplementary material available at 10.1038/s41598-026-36462-w.

Introduction

Tuberculosis (TB) remains a critical global health threat, significantly impacting public health systems, particularly in densely populated and low-income urban settings. TB is primarily transmitted through airborne droplets expelled by infected individuals through coughing, sneezing, or spitting¹. Despite global control efforts, the World Health Organization (WHO) has reported slow progress in achieving its 2025 targets, which aim for a 50% reduction in TB incidence and a 75% reduction in mortality. By 2023, the incidence rate had only decreased by 8.3%, and mortality by 23%².

According to the most recent Global Tuberculosis Report², TB incidence remains alarmingly high, with 10.8 million cases (95% CI, 10.1–11.7) diagnosed in 2023, one of the highest numbers reported since WHO began systematic monitoring in 1995³. Moreover, with 1.25 million (95% CI, 1.13–1.37) deaths in 2023², TB is the leading cause from a single agent, replacing COVID-19 and nearly twice that of HIV/AIDS².

In this context, spatial analysis has emerged as a valuable tool for identifying geographic patterns in TB incidence and their relationship with social determinants, especially in urban environments^4–6. Neighborhoods, as administrative units, are crucial for such analyses, as they reflect residents’ socioeconomic conditions, social structures, and access to healthcare services, factors that may contribute to health disparities and TB risk^7,8. Identifying neighborhood-level spatial patterns of TB provides essential information for designing targeted case-finding strategies based on the specific conditions of each community, offering evidence to guide public health interventions^9–11.

Neighborhoods, as a spatial context, are intrinsically linked to population health, not only because of shared social and cultural characteristics, but also because they reflect spatially embedded forms of inequality and marginalization^7,8. Studying these local-level patterns is crucial for addressing health disparities. Currently, 57% of the global population, and 82% in Latin America and the Caribbean, live in urban areas, with a continuous increase over the years¹². Understanding the dynamics of communicable diseases in urban settings is crucial for planning better community-based prevention strategies.

In Chile, the TB incidence rate increased by 1.4% in 2023 compared to 2022, reaching 15.8 cases per 100,000 inhabitants (n = 2,973), with pulmonary TB accounting for 81.8% of reported cases—a rise from the previous year¹³. Recent reports highlight shifts in risk profiles, with TB increasingly concentrated in urban areas and associated with low income, overcrowding, and social vulnerability¹⁴.

Currently, TB case-finding is conducted by local health teams through passive or active approaches¹⁵. However, territorially defined strategies for community-based active case-finding are lacking, limiting early detection and effective control. The primary objective of this study is to identify the spatial patterns of pulmonary TB in Chile’s largest urban area, along with its key sociodemographic correlates, to provide evidence supporting the development of targeted, community-based case-finding strategies in urban settings.

Methods

Study area and design

We conducted an ecological study using all registered cases (n = 3,348) from the National Program for the Control and Elimination of Tuberculosis (PROCET) of the Chilean Ministry of Health in the Gran Santiago (GS) area from 2016 to 2020. TB is subject to universal surveillance in Chile, with mandatory daily notification for both the public and private sectors. All confirmed cases are recorded in the PROCET database.

The Metropolitan Region of Santiago (RM, for its acronym in Spanish) is the largest of Chile’s 16 administrative regions and is located in the country’s central zone. It accounts for 40.5% of the national population (n = 17,574,003)¹⁶. The Gran Santiago area (GS) is the largest urban area in the country, comprising 34 of the 52 communes that form the RM (Fig. 1), and concentrates 86.31% of the RM’s population¹⁷. We included all 1,083 neighborhoods within the GS. Neighborhoods represent one of the smaller units within a commune, and are defined by physical continuity, shared social interests, and other factors that create community dynamics, aiming to enhance citizen participation and local management¹⁸. Due to their population homogeneity, neighborhoods are key units for planning activities at the primary healthcare level.

Fig. 1.

Delimitation and location of the study area: Gran Santiago and included communes. (A) Urban area of the Gran Santiago (red area), municipal administrative boundaries in solid black lines; (B) Metropolitan Region (RM) boundary (gray area); (C) Location of Chile (gray area), regional boundaries in solid black line. The map was created using QGIS version 3.26.3-Buenos Aires software (https://qgis.org/). The background map is from CARTO (https://carto.com/). All national boundaries data and shapefiles are sourced from IDE-Chile (https://www.ide.cl/).

Tuberculoses cases

We included all pulmonary tuberculosis (TB) cases within the study area and classified them according to International Classification of Diseases (ICD)-10 codes. We excluded all cases from prisons and homeless individuals due to their lack of neighborhood connection and inability to register a residential address. The diagnosis included both laboratory-confirmed cases with positive biological samples (smear microscopy, culture, or diagnostic test) and clinical cases (radiography, histology, and other methods)¹⁵. The dataset included diagnostic information, age, sex, and residential address for each case.

Population indicators

Population and housing data were obtained from the most recent national census (2017), disaggregated at the block level, the basic spatial unit for collecting population statistics in urban areas in Chile. Subsequently, we aggregated this data at the neighborhood level. For each neighborhood, we calculated a set of indicators potentially associated with TB prevalence, according evidence: masculinity index (ratio between the number of men and women per 100); aged population (percentage of the population over or equal to 65 years); immigrant population (percentage of international immigrant population); Indigenous population (percentage of the population belonging to indigenous or native groups); health facilities (number of primary health care facilities per 1000 inhabitants); overcrowding (percentage of dwellings with 2.5 or more persons per room used exclusively for sleeping); population density (number of persons per neighborhood per square kilometer); tenements (number of dwellings composed of rooms that form an independent quarter per 1000 inhabitants); and total population.

Data analysis

Geocoding

We geocoded all pulmonary TB cases using individuals’ residential addresses through an automated cascade process, initially employing Bing Maps, followed by Google Maps using the Tidygeocoder (version 1.0.6) R package. We applied a modified version of the quality filters proposed by Quinteros et al.¹⁹ for Chilean cities, whose results showed a mean positional error of 11.1 m for Bing Maps and 83.2 m for Google Maps. For Bing Maps, we used the following criteria: BingMatchCode (Good), BingConfidence (High), BingEntityType (Address or RoadBlock), BingMethod (Interpolation or Rooftop), and concordance between the identified and declared commune. For Google Maps, we used type (establishment, street address, route, premise, subpremise, or convenience store), loctype (rooftop or range interpolated), and concordance of commune. To ensure higher positional accuracy, all quality criteria had to be simultaneously satisfied in each geocoding processes.

All addresses that could not be geocoded automatically were manually geocoded using Google Maps and local landmarks. This process resulted in the acquisition of geographic coordinates (latitude and longitude) for each case.

Using the quality criteria described above, 97.6% of patients with addresses in the Gran Santiago were successfully geocoded. Patients who could not be geocoded (n = 82) had incomplete or problematic addresses, or no address recorded. Among the geocoded cases, 44% were located using Bing Maps, 42% using Google Maps, and 14% through a manual search. Finally, all cases were aggregated at the neighborhood level to calculate the frequency of TB per neighborhood.

Descriptive analysis

An initial exploratory descriptive analysis was conducted, detailing the total number and frequency of pulmonary TB cases, as well as the crude incidence rates per 100,000 inhabitants. Analyses were stratified by sex, major age groups (< 15 years, 15–64 years, and ≥ 65 years), and year of diagnosis across the study period. Period incidence rates were calculated using the mean number of pulmonary TB cases in the period as the numerator and the mid-period population (year 2018) as the denominator for crude rates by total period, sex, and age group. The annual incidence rate was calculated using the total number of pulmonary TB cases and the official population projection for each year. All population denominators used to calculate crude incidence rates were obtained from the official population projections of Chile’s National Institute of Statistics.

Global spatial autocorrelation

Moran’s Global Index (Moran’s I) was used to assess spatial autocorrelation of TB incidence rates for each year and the entire study period. This index measures the degree of similarity between TB incidence rates in each neighborhood and those of its surrounding neighborhoods, applying a first-order queen contiguity structure. Moran’s I values range from − 1 to + 1, where values closer to + 1 indicate strong positive spatial autocorrelation. A hypothesis test was conducted, where the null hypothesis assumed a random distribution of pulmonary TB incidence with no spatial dependency^20,21.

Spatial cluster detection analysis

Tango’s flexible spatial scan statistic (FlexScan) was applied to identify high-risk spatial clusters of pulmonary TB using a Poisson model and gender-adjusted incidence rates^22,23. Due to marked variability in pulmonary TB case counts across neighborhoods, the adjusted rates were calculated using indirect standardization, based on national pulmonary TB rates by sex and the official country-level population as reference values, and the average number of cases during the specified time period. The expected number of cases was estimated using the 2017 census population data (the only source with a spatial distribution below the commune level) for each neighborhood. This approach allows the detection of spatial clusters with irregular shapes and determines whether observed incidence within a cluster is significantly higher than expected under spatial randomness.

We estimated different window sizes (K) and calculated the number of significant clusters, observed and expected cases, average relative risk, and the number of neighborhoods included for each window size (see supplementary materials 1). Based on graphical evaluation, we specify a maximum spatial scanning window size of 20 neighborhoods for spatial cluster detection using a Poisson probability model. The alternative hypothesis tested was that the relative risk within the window was higher than outside, assessed by a log-likelihood ratio test with Monte Carlo replications (n = 999). Only high-risk clusters with statistical significance (p-value < 0.05) adjusted for multiple comparisons using the Bonferroni correction were considered for further analysis. All analyses were conducted using the rflexscan R package²⁴.

The flexible spatial scan statistic proposed by Tango and Takahashi’s (FlexScan) was chosen due to its ability to detect irregularly shaped clusters, a key advantage in heterogeneous urban environments^22,23. Although Kulldorff’s SaTScan is widely used in spatial epidemiology²⁵, it has been shown to overestimate cluster size because of its reliance on circular scanning windows²⁶. The choice between these methods should be based on the study design, the spatial characteristics of the study area, and the availability of computational resources. FlexScan is more adaptable in urban environments where disease risk is heterogeneously distributed. However, it requires higher computational demands.

Finally, we mapped each statistically significant cluster according to the relative risk by age groups and sex. Also, we compared population and residential indicators between areas inside and outside the identified spatial clusters for the total population, specific age groups, and sex. Categorical variables were compared using Chi-square tests, and continuous variables (log-transformed) using Student’s t-tests. Statistical significance was set at p < 0.05, and differences were also assessed using 95% confidence intervals.

Maps were produced using QGIS version 3.26.3-Buenos Aires (Fig. 1) and R software version 4.0.0 (Fig. 2). Freely accessible shapefiles of the country’s administrative boundaries and neighborhood units corresponding to 2022 were obtained from the National Spatial Data Infrastructure (IDE-Chile). All spatial data processed and analyzed using the WGS84 coordinate reference system (EPSG = 4326). Statistical analyses and table generation were performed using R software²⁷.

Fig. 2.

Spatial distribution of statistically significant high-risk clusters of pulmonary TB stratified by age groups and gender in the Gran Santiago area. Chile 2016–2020. (A) All age groups; (B) 0–14 age group; (C) 15–64 age group; (D) 65 + age group; (E) Male; (F) Female; Black lines show the original clusters for all groups. The dotted gray line shows the municipal boundary. The solid white line shows the boundary between neighborhoods. The light gray area indicates the Gran Santiago area. The map was created using R software version 4.0.0 (https://www.r-project.org/), RStudio version 2025.09.1 (https://posit.co/), and “ggplot2” package version 4.0.0 (https://ggplot2.tidyverse.org/). All national boundaries data and shapefiles are sourced from IDE-Chile (https://www.ide.cl/).

Results

A total of 3,348 pulmonary TB cases were confirmed during the study period, with 64.52% occurring in men. The incidence rates were nearly twice as high in men as in women (13.23 vs. 7.03 per 100,000 inhabitants, respectively). When stratified by age group, the incidence increases progressively with age, reaching 14.82 cases per 100,000 inhabitants among individuals aged 65 years and older. The annual trend showed a slight but non-significant increase at the beginning of the period, followed by stable rates in 2017 and 2018. Subsequently, in 2019, the incidence declined to 8.36 cases per 100,000 inhabitants, but rose again in 2020 to 10.66 cases per 100,000 (Table 1).

Table 1.

Number of cases and crude incidence rate per 100,000 inhabitants stratified by sex, age, and years in the Gran Santiago area, 2016–2020.

Variable		Cases (N [%])	Incidence rate
Total	Number of cases	3348 [-]	10.12
Sex/gender	Male	2160 [64.52]	13.23
	Female	1186 [35.42]	7.03
	Indeterminate	1 [0.03]	–
	Not reported	1 [0.03]	–
Age (years)	0–14	83 [2.48]	1.35
	15–64	2720 [81.24]	11.64
	65+	544 [16.25]	14.82
	Not reported	1 [0.03]	–
Years	2016	609 [18.19]	9.59
	2017	710 [21.21]	10.99
	2018	717 [21.42]	10.83
	2019	568 [16.97]	8.36
	2020	744 [22.22]	10.66

Not calculated

Moran’s I revealed a positive and statistically significant spatial autocorrelation pattern in four of the five years analyzed (2016, 2017, 2018, and 2020). For the entire study period, the index confirmed a significant clustering of pulmonary TB cases (Moran’s I = 0.175; p-value: <0.001), indicating non-random geographic concentration within specific neighborhoods of the Gran Santiago (GS). Additionally, a gradual increase in Moran’s I over time suggests a strengthening spatial pattern (Table 2).

Table 2.

Results of the global spatial auto-correlation tests on the incident rate in the Gran Santiago area, Chile (2016–2020).

Year	Neighborhoods (N)	Moran’s I	P-value
2016	1,083	0.058	< 0.001
2017	1,083	0.090	< 0.001
2018	1,083	0.069	< 0.001
2019	1,083	0.022	0.051
2020	1,083	0.091	< 0.001
Study period (2016–2020)	1,083	0.175	< 0.001

In our spatial analysis, we identified 11 statistically significant spatial clusters with elevated pulmonary TB risk, each with a relative risk (RR) greater than 2.0. The primary cluster (ID1) included 12 neighborhoods, accounting for 142 observed cases and an RR of 3.88 (p < 0.001). The secondary cluster (ID2), composed of eight geographically isolated neighborhoods, exhibited the highest RR at 6.12 (p < 0.001) (Fig. 2; Supplementary Materials 2).

In the under-15 age group, we identified two small clusters with 14 neighborhoods, with RRs of 21.43 and 22.93 for the primary and secondary clusters, respectively. Among individuals aged 15–64 years, we observed geographic patterns consistent with the overall population, identifying eight clusters with RRs ranging from 2.74 to 6.44. In the ≥ 65 age group, a single high-risk cluster was detected in the southern area of the GS, with an RR of 6.04 (p < 0.001) (Fig. 2; Supplementary Materials 2).

Sex-stratified analysis revealed seven clusters among males and six among females. In males, the spatial distribution of elevated RR closely mirrored that observed in the 15–64 age group, forming a continuous high-risk corridor in the central zone of the GS. In contrast, the female clusters showed more dispersed and discontinuous spatial patterns. Notably, female cluster 4 (RR = 2.63) overlapped with the primary cluster identified among those aged ≥ 65 years (Fig. 2; Supplementary Materials 2).

Overall, 11.4% (n = 123) of the GS neighborhoods and 29.4% (n = 984) of diagnosed pulmonary TB cases were located within statistically significant clusters. Comparative analysis of demographic and residential characteristics between clustered and non-clustered areas revealed higher proportions of men (masculinity index), immigrants, Indigenous populations, health facilities, overcrowded households, tenement dwellings, and population density in clustered areas (Table 3).

Table 3.

Stratified age and gender comparison of principal summary statistics by population data and cluster identification in the Gran Santiago area. Chile 2016–2020.

Variable	Total	All age groups		0–14 years		15–64 years		65 + years		Male		Female
		Cluster	No Cluster	Cluster	No cluster	Cluster	No cluster	Cluster	No Cluster	Cluster	No cluster	cluster	No cluster
Neighborhood (N)	1.083	123	960	14	1.069	93	990	11	1.072	79	1.004	65	1.018
Cases (N)	3.348	984	2.364	17	66	650	2.070	22	522	471	1.689	252	934
Masculinity index (%)‡	96 (94.3–97.7)	99.0 (97.4-100.5)	95.6 (93-7-97.6)	96.2 (93.0–99.4)	96 (94.2–97.8)	99.2 (98.0-100.5)	95.7 (93.8–97.6)	97.7 (94.7-100.7)	96.0 (94.2–97.7)	100.1 (98.0-102.2)	95.7 (93.8–97.5)	99.8 (98.0-101.5)	95.8 (93.9–97.6)
Elderly population (%)δ	12.0 (11.7–12.3)	11.0 (10.3–11.8)	12.1 (11.8–12.4)	11.4 (9.4–13.3)	12.0 (11.7–12.3)	10.8 (10.0-11.5)	12.1 (11.8–12.4)	7.7 (5.2–10.1)	12.0 (11.7–12.3)	10.4 (9.5–11.3)	12.1 (11.8–12.4)	9.8 (9.0-10.7)	12.1 (11.8–12.4)
Immigrants (%) ‡	6.8 (6.3–7.3)	12.4 (9.7–15.1)	6.1 (5.7–6.5)	17.5 (7.0–27.8)	6.7 (6.2–7.2)	16 (12.8–19.2)	6.0 (5.6–6.4)	3.0 (1.7–4.4)	6.9 (6.4–7.4)	16.2 (12.5–20.0)	6.1 (5.7–6.5)	20.5 (16.4–24.6)	6.0 (5.6–6.4)
Indigenous population (%) ‡	10.0 (9.8–10.3)	11.4 (10.7–12.1)	9.9 (9.6–10.1)	9.7 (8.4–11.0)	10.0 (9.8–10.3)	11.3 (10.5–12.2)	9.9 (9.7–10.1)	14.5 (12.3–16.7)	10.0 (9.8–10.2)	10.5 (9.8–11.1)	10 (9.8–10.2)	10.2 (9.6–10.8)	10.0 (9.8–10.3)
Health facilities (per 1000) ‡	0.04 (0.03–0.05)	0.07 (0.01–0.12)	0.04 (0.03–0.04)	0.04 (0.02–0.10)	0.04 (0.03–0.05)	0.05 (0.02–0.07)	0.04 (0.03–0.05)	0.06 (0.02–0.15)	0.04 (0.03–0.05)	0.03 (0.01–0.05)	0.04 (0.03–0.05)	0.05 (0.01–0.09)	0.04 (0.03–0.05)
Overcrowding (%) ‡	8.8 (8.5–9.1)	13.4 (12.6–14.2)	8.2 (7.9–8.5)	11.7 (9.1–14.4)	8.8 (8.4–9.1)	14.6 (13.7–15.6)	8.2 (7.9–8.5)	14.2 (11.2–17.2)	8.7 (8.4-9.0)	14.2 (13.0-15.3)	8.4 (8.1–8.7)	14.5 (13.2–15.8)	8.4 (8.1–8.7)
Tenement-Houses (per 1000) ‡	6.0 (5.3–6.8)	16.1 (11.4–20.9)	4.7 (4.2–5.3)	15.3 (6.3–24.3)	5.9 (5.2–6.7)	20.2 (14.3–26.2)	4.7 (4.2–5.2)	1.2 (0.3-2.0)	6.1 (5.3–6.8)	21.0 (14.3–27.8)	4.9 (4.3–5.4)	24.9 (16.6–33.2)	4.8 (4.3–5.4)
Population density (per km2) ‡	12233.6 (11803.2–12664.0)	14135.8 (12805.7–15466.0)	11989.9 (11536.6–12433.1)	12628.3 (8919.2–16338.5)	12228.4 (11794.5–12662.4)	13729.2 (12344.4–15114.1)	12093.1 (11640.9–12545.4)	20342.7 (18072.9–22612.5)	12150.4 (11718.8–12582.0)	13459.6 (11788.0–15131.3)	12137.1 (11691.6–12582.7)	15085.9 (13101.3–17070.5)	12051.5 (11612.9–12490.1)
Population (N)	6.130.628	622.951	5.507.677	64.293	6.066.335	401.511	5.729.117	93.711	6.036.917	377.742	5.752.886	355.390	5.775.238

Significant differences were observed between cluster and non-cluster areas across several variables. The masculinity index was consistently higher in clustered areas across all demographic groups. A higher proportion of immigrant populations was observed, particularly among the under-15 age group and females. The Indigenous population was more prevalent in clusters involving individuals aged ≥ 65. Overcrowding was significantly more prevalent across all age groups, with the highest rates observed in the 15–64 and ≥ 65 age groups. Tenement housing also showed substantial differences, especially in female clusters. Although less pronounced, differences in the density of health facilities and population were also evident, particularly among older adults (Table 3).

Discussion

This study is the first to analyze the spatial distribution of pulmonary tuberculosis (TB) at the neighborhood level in Chile and one of the first to employ this methodology with a high number of unit areas worldwide, offering a novel contribution to understanding the local epidemiology of TB in urban contexts. Our findings reveal a significant spatial concentration of pulmonary TB cases in specific neighborhoods of the GS, with nearly one-third (29.4%) of all reported cases occurring within statistically significant spatial clusters. These clusters were predominantly located in areas with higher proportions of vulnerable populations, including immigrants, Indigenous groups, and residents living in overcrowded or tenement housing.

Throughout the study period, incidence rates of pulmonary TB remained relatively stable, ranging from 8.36 to 10.99 cases per 100,000 inhabitants. These patterns are consistent with previous reports from Chile’s Metropolitan Region¹⁴. The observed decline in 2019 may be attributed to nationwide social unrest, which disrupted primary healthcare services and likely affected TB case detection and reporting.

Positive and statistically significant spatial autocorrelation was observed in four of the five years studied, confirming the persistence of spatial clustering. These clusters were mainly distributed along a central corridor of the GS, with isolated high-risk areas in the southern and southeastern zones. This central corridor includes four major supply markets—La Vega, Matadero Franklin, Lo Valledor, and Terminal Pesquero—which are characterized by high foot traffic and large populations of economically active individuals. The sociodemographic profile of these neighborhoods suggests that TB transmission may be linked to occupational and residential exposures in densely populated and precarious urban environments.

Our findings are consistent with international evidence. In Peru and Brazil, TB clusters have been identified in urban areas with high poverty, overcrowding, and even air pollution^28,29. In Asia, spatial clusters have been documented at both local and national levels in regions such as Sichuan and Yunnan in China, highlighting the global importance of spatial heterogeneity in TB transmission^30–32. Unlike rural clusters observed in African countries such as Ethiopia, Zimbabwe, and Kenya—associated with limited healthcare access and extreme poverty^10,33,34 —our study found exclusively urban clusters. This reinforces the need for locally tailored TB control strategies. While spatial clustering is a global phenomenon, its distribution and drivers are shaped by regional socioeconomic and geographic contexts.

Clustered neighborhoods had higher proportions of known TB risk factors, such as male predominance, immigrant and Indigenous populations, overcrowding, and tenement housing¹³. These characteristics align with national surveillance data, reinforcing the need to focus public health interventions on vulnerable groups. A total of 123 neighborhoods were identified within pulmonary TB clusters, many of which overlapped across age and sex-specific analyses. However, a pediatric cluster (ID2) did not coincide with clusters identified for all ages. This may be explained by the low number of pediatric TB cases and the unique demographic characteristics of the affected areas. Pediatric spatial clusters require special targeted strategies, considering the specific vulnerabilities of this age group.

We identified a key challenge related to the fragmentation of pulmonary TB clusters across administrative boundaries. Several spatial clusters were located along the borders of communes and health service jurisdictions, hindering coordinated intervention efforts. In Chile, TB control and active case-finding are primarily managed at the municipal level, in coordination with regional health services. However, urban mobility frequently extends beyond these administrative limits, diminishing the effectiveness of territorially bounded strategies. These findings underscore the need for supra-communal approaches to TB control that account for the fluid dynamics of urban populations, particularly in densely populated metropolitan areas.

This study examined the spatial distribution of pulmonary TB at the neighborhood level using individual and geocoded case data and advanced spatial analysis methods. One of its main strengths lies in the application of Tango and Takahashi’s flexible spatial scan statistic, a robust method particularly suited for identifying irregularly shaped clusters in urban areas. However, some limitations should be noted. In larger study areas, testing optimal window sizes requires substantial computational capacity; we propose exploring different window sizes until the number of statistically significant clusters stabilizes, within the limits of available processing power. Additionally, while individual-level covariates such as socioeconomic status or comorbidities were not available, we incorporated neighborhood-level contextual variables that serve as proxies for social determinants. Despite these limitations, the study provides valuable evidence to support spatially targeted TB control strategies in complex and heterogeneous urban environments.

Conclusions

Tuberculosis (TB) spatial clusters have been identified across a wide range of geographic regions, countries, and contexts. Our findings reinforce the importance of integrating the geographic and social characteristics of neighborhoods into public health planning³⁵. Recognizing neighborhoods as critical social determinants of TB risk is essential for tailoring prevention strategies to local realities and aligning health services more closely with the population’s needs^9,10,32,36. Future research should aim to identify additional predictors of TB spatial distribution to better guide community-based targeting strategies, with the flexibility to update interventions over time.

In 2022, the Chilean Ministry of Health revised its technical regulations for TB surveillance, incorporating active case-finding beyond healthcare facilities. Our results provide locally grounded evidence to support this new direction, offering insights for implementing more targeted and effective TB control strategies in Chile. Moreover, the integration of geocoded epidemiological surveillance enables the development of innovative practices in the spatial targeting of health problems, including the early detection of outbreaks, case isolation, and improved contact management for infectious diseases^37–39. These findings also serve as a practical example for potential adoption in other countries or metropolitan areas across Latin America and globally. Identifying spatial clusters offers valuable guidance for designing community-based active case-finding programs, which have proven effective in detecting cases early and reducing transmission^40–45. Such strategies should be adapted to the specific conditions and contexts of each local territory⁴⁵, thereby optimizing resource use and contributing to a sustained reduction in TB incidence over the medium and long term.

Supplementary Information

Below is the link to the electronic supplementary material.

Author contributions

AS, CM, and EN contributed to the construction of the database. AS performed the statistical analyses. All authors participated in the review and interpretation of the analytical results and contributed to the writing and critical revision of the final manuscript. All authors approved the final version of the manuscript for submission.

Funding

Open access for this publication was supported by the Agencia Nacional de Investigación y Desarrollo (ANID) through Fondecyt Regular Grant No. 1251995. Salvador Ayala received funding from ANID’s National Doctoral Fellowship, Grant No. N21191111.

Data availability

The data on pulmonary tuberculosis (TB) cases analyzed in this study are not publicly available due to privacy and data protection regulations. Access to these data must be requested directly from the Chilean Ministry of Health. Population and residential indicators used in the analysis are publicly available through the National Institute of Statistics of Chile (INE) at [www.ine.cl](http:/www.ine.cl) .

Declarations

Competing interests

The authors declare no competing interests.

Ethical approval

This study was approved by the Human Research Ethics Committee of the Faculty of Medicine, University of Chile (Santiago, Chile).