Environmental and system-level factors associated with emergency medical service utilization and immediate on-scene symptom response: a population-based study

Abstract

Suspected cardiopulmonary emergencies represent a major component of EMS demand and are highly sensitive to environmental exposures. However, the combined impact of these dual macro-environmental and micro-operational stressors on prehospital treatment effectiveness remains poorly understood. We aimed to evaluate how spatiotemporal, meteorological, and operational factors are jointly associated with EMS demand and immediate on-scene symptom response in a large metropolitan system. This retrospective cohort study analyzed dispatch records from the centralized EMS system of Nanning, southern China (2021–2025). From 38,562 initial activations screened via cardiopulmonary symptom keywords, 15,884 eligible EMS missions recorded from August 24, 2021 to December 31, 2025, were included after excluding non-medical or incomplete records. Objective meteorological data were integrated using generalized additive models (GAM). Multivariable logistic regression incorporating categorical congestion levels and spatial-operational variables was used to identify factors independently associated with unfavorable immediate on-scene symptom response, defined statistically as prehospital treatment failure. Exploratory restricted cubic splines (RCS) evaluated the association between continuous on-scene treatment duration and clinical outcomes. GAM analysis revealed that extreme relative humidity ( P  = 0.003) and extreme ambient temperatures ( P  = 0.028) non-linearly were associated with population-level surges in EMS demand. In the integrated multivariable model, severe traffic congestion (CI > 2.0) was associated with unfavorable immediate on-scene symptom response compared with free-flow conditions (aOR 1.19, 95% CI 1.08–1.32), after adjustment for spatial location and travel distance, although potential residual confounding by dispatch distance cannot be fully excluded. Restricted cubic spline analysis further demonstrated a significant non-linear association between congestion index and symptom response (Overall P  = 0.0001). Furthermore, RCS analysis showed no statistically significant overall independent association between continuous on-scene time and immediate symptom response (Overall P  = 0.075). In line with this, categorical analysis similarly indicated that prolonged on-scene time (> 30 min) was not significantly associated with increased treatment failure risk (aOR 1.23, 95% CI 0.98–1.55). While environmental extremes were associated with increased EMS demand, urban spatial characteristics and extreme traffic congestion were associated with unfavorable immediate on-scene symptom response. Furthermore, prolonged on-scene time showed limited independent clinical relevance after adjusting for severity. Integrating spatiotemporal and operational metrics provides a public health framework for optimizing emergency care resource allocation, improving system resilience, and guiding urban health policy in rapidly growing metropolitan regions.