Spatial heterogeneity and drivers of pulmonary tuberculosis in Guangzhou: a street-level analysis of residents and migrant subgroups from 2015 to 2023.

OBJECTIVE

Rapid urbanisation and internal migration may reshape fine-scale tuberculosis (TB) risk within cities, yet street-level evidence by migrant subgroup remains limited. We quantified the street-level spatial heterogeneity of pulmonary TB in Guangzhou, China, and assessed individual- and area-level drivers to inform risk-stratified active case finding and targeted prevention strategies.

METHODS

We analysed 74,449 pulmonary TB cases from Guangzhou's TB surveillance system (2015-2023). Cases were categorised according to household registration as registered residents, intra-provincial migrants (within Guangdong), or inter-provincial migrants (outside Guangdong). We calculated street-level incidence and mapped spatial patterns. Hotspots were identified using the local Getis-Ord Gi* statistic. Subgroup risk patterns were compared using adaptive-kernel log-relative risk surfaces. A multilevel Bayesian logistic regression model was used to identify individual-level factors associated with hotspot residence, accounting for street-level clustering. Alternative random-effects structures, including spatial specifications, were compared during model selection. Street-level environmental determinants were assessed using hierarchical Bayesian negative binomial models fitted with INLA and multi-source indicators.

RESULTS

TB incidence showed substantial street-level spatial heterogeneity, with hotspots concentrated in the central urban core. Age-standardised incidence was highest among inter-provincial migrants (47.24 per 100,000), followed by residents (42.69 per 100,000) and intra-provincial migrants (28.86 per 100,000). However, inter‑provincial migrants were less likely to reside in hotspots compared with residents (aOR = 0.850, 95% CrI: 0.740, 0.987). Cases detected via active screening were also less likely to live in hotspots relative to those identified through symptom‑based consultation (aOR = 0.396, 95% CrI: 0.168, 0.931). In the Bayesian model, a higher street-level per capita GDP was associated with an increased TB risk (relative risk [RR] per one standard deviation [SD] increase = 1.184, approximately equivalent to a 1,000 Chinese yuan increase), while a higher proportion of intra-provincial migrants was associated with a decreased risk (RR per one SD increase = 0.895, approximately equivalent to a 10 percentage-point increase). Incidence among inter-provincial migrants correlated with the TB burden in provinces of origin (ρ = 0.56).

CONCLUSION

Pulmonary TB in Guangzhou exhibits pronounced street-level spatial heterogeneity. This study found that active screening was associated with a lower likelihood of residing in a TB hotspot after accounting for street-level clustering, suggesting that estimates may be biased if this geographic context is ignored. Two priority profiles emerged from the analysis: older adults living in central hotspot communities, and inter-provincial migrants who had the highest overall incidence and a clear ecological linkage to origin-province burden. We therefore recommend risk-stratified, spatially targeted interventions, including intensified active screening for the elderly in hotspot streets, and enhanced cross-provincial coordination for TB control among inter-provincial migrants.