Objective Urban park green spaces are important spatial elements that may affect respiratory health, but their associations with the acute incidence risk of different types of respiratory diseases remain unclear. Previous studies have reported protective, adverse, or non-significant relationships between urban green space and respiratory health, partly because of differences in disease outcomes, green space indicators, spatial and temporal scales, and model structures. Taking the central urban area of Nanjing, Jiangsu Province, as the study area, this study evaluates the associations between park green space exposure and the acute incidence risk of respiratory diseases, with particular attention to differences between infectious and non-infectious respiratory diseases. By using high-spatiotemporal-resolution emergency medical service data, this study aims to provide empirical evidence for health-oriented urban park green space planning and landscape architecture interventions.

Methods This study used 120 emergency medical service (EMS) GPS data from the central urban area of Nanjing from May 1, 2023 to May 5, 2024. Respiratory disease cases were identified from emergency records and classified into infectious and non-infectious respiratory diseases according to the International Statistical Classification of Diseases and Related Health Problems, Tenth Revision (ICD-10). The study area was divided into hexagonal grids of approximately 4 km², and emergency events and environmental variables were aggregated at the weekly grid scale. The weekly number of respiratory emergency events in each grid was used to construct an indicator of acute incidence risk. Park green space exposure indicators included NDVI, green space ratio, tree canopy cover, service coverage rates of comprehensive parks and community parks, and minimum distances from each grid to the nearest comprehensive park and community park. Meteorological factors, air pollution factors, and socio-built environmental factors were included as control variables. Kernel density estimation and global spatial autocorrelation analysis were used to identify the spatial clustering characteristics of infectious and non-infectious respiratory diseases. A Bayesian spatiotemporal model was then constructed to estimate the associations between environmental variables and the acute incidence risk of respiratory diseases. Multiple submodels with different random-effect structures were compared using DIC, WAIC, R², MSE, and MAE. In addition, interaction terms between NDVI and selected environmental factors, including weekly mean temperature, PM₂.₅, and NO₂, were introduced to examine whether the association between NDVI and respiratory risk varied under different meteorological and air pollution conditions.

Results The acute incidence risk of respiratory diseases showed significant spatiotemporal clustering in the central urban area of Nanjing. Infectious and non-infectious respiratory diseases both exhibited a monocentric and uneven spatial distribution pattern, with higher-density areas mainly concentrated in the old urban core and surrounding high-density urban areas. The global Moran’s I values for infectious and non-infectious respiratory diseases were 0.61 and 0.58, respectively, indicating significant positive spatial autocorrelation. Temporally, infectious respiratory diseases showed more evident short-term fluctuations, whereas non-infectious respiratory diseases showed a relatively smoother temporal pattern, with higher levels in late autumn and winter. Model comparison showed that Bayesian spatiotemporal models incorporating both spatial and temporal random effects generally performed better than models with only independent, temporal, or spatial effects, suggesting that spatiotemporal dependence should be considered when modeling acute respiratory risk at the intra-urban scale.Environmental effect estimates differed between the two disease types. NDVI was negatively associated with the acute incidence risk of both infectious and non-infectious respiratory diseases, with a stronger association for infectious diseases. The RR was 0.79 with a 95% credible interval of 0.74−0.84 for infectious diseases, and 0.92 with a 95% credible interval of 0.87−0.97 for non-infectious diseases. Green space ratio and tree canopy cover did not show consistent protective associations. Among park accessibility indicators, minimum distance to community parks was positively associated with both disease types, whereas comprehensive park-related accessibility indicators did not show stable significant associations. Meteorological, air pollution, and socio-built environmental factors also showed disease-specific associations. PM₂.₅ was positively associated with both disease types, while residential land-use density and road network density were positively associated with non-infectious respiratory diseases. Interaction analysis suggested that the association between NDVI and respiratory disease risk may vary with weekly mean temperature, PM₂.₅, and NO₂ levels, but these results should be interpreted as statistical interactions rather than direct evidence of causal pathways.

Conclusion The health effects of park green spaces on respiratory diseases are not simply a result of increasing green quantity. Instead, they depend on the specific green space indicator used, including NDVI, green space ratio, tree canopy cover, service coverage rates, and proximity indicators of different types of parks, as well as disease type and spatiotemporal scale. In the central urban area of Nanjing, NDVI showed a relatively stable protective association with the acute incidence risk of both infectious and non-infectious respiratory diseases, whereas green space ratio, tree canopy cover, and comprehensive park-related accessibility indicators did not show consistent protective effects. Community park proximity showed a more stable association with lower respiratory risk, highlighting the potential importance of neighborhood-scale park access. The findings suggest that health-oriented urban park planning should move beyond simple increases in green space area and pay more attention to vegetation condition, community-level accessibility, air pollution exposure, and high-density built environments. This study provides empirical evidence for optimizing urban park green space allocation and improving the precision of landscape architecture interventions for public health.