This study utilized panel data spanning 2001 to 2021 from 365 Chinese cities, including three healthcare resource indicators, namely hospitals, hospital beds, and licensed (assistant) physicians, and 34 socioeconomic and environmental influencing factors (24, 27), which were provided in Supplementary Table S1. Healthcare resource and socioeconomic data derived from China City Statistical Yearbook (2000–2021),¹ while environmental metrics sourced from validated remote sensing platforms. For example, SO₂, PM_2.5, and normalized difference vegetation index (NDVI) were obtained from the Giovanni platform,² PM₁ and PM₁₀ were obtained from the Zenodo platform,³ and nighttime light data were obtained from the Figshare platform (47).

We employed the Healthcare Resource Density Index (HRDI) as the dependent variable for single indicators, mitigating single-metric bias inherent in population- or area-exclusive denominators (12). The HRDI is defined as:

where i, j and t refer to city i, healthcare indicator j and year t, y_ijt refers to the quantity of city i’s healthcare resource j in year t, P_it refers to the number of city i’s total population in year t, and A_it is city i’s land area in year t.

The composite healthcare resource score, derived via the entropy weighting method (48), serves as the main dependent variable. This objective weighting technique applies information entropy principles to calculate indicator-specific weights through Equation (1) based on dispersion magnitude, thereby eliminating subjective bias and quantifying the relative importance of healthcare resource indicators.

The independent variables included socioeconomic and environmental factors related to population and economy, industry and employment, education and human resources, and environment and infrastructure. This study employed variance inflation factor (VIF) and node purity indicator for variable selection. We set a VIF threshold of 5 to identify variables with lower multicollinearity, and used node purity indicator to quantify explanatory significance through complex nonlinear relationships and interaction effects.

The composite index of healthcare resources calculated through entropy weighting of hospitals, beds, and physicians, with respective weights of 0.25, 0.46, and 0.29, indicating that hospital beds’ main influence on resource allocation. The spatiotemporal analysis revealed that, the growth of city healthcare resources during the study period was relatively small and exhibited an inequitable pattern, characterized by lower resource levels in the west compared to the east, and in the north compared to the south. The composite score exhibited minimal average annual growth, demonstrating a modest increasing trend since 2007 with annual growth rates not exceeding 5%. Hospital density displayed a U-shaped trajectory, transitioning from initial decline to gradual increase after 2013, with growth rates remaining below 3%. Bed and physician densities followed similar growth patterns, both showing sustained increases since 2004 and reaching maximum annual growth rates of 10% (Figure 2).

Figure 3A presents the average annual changes in spatial Gini coefficients for the composite index and three single indicators during the study period. The composite index showed an overall decline in spatial equity, with an average annual increase of 0.34% in its spatial Gini coefficient. Regional analysis revealed that 18 of 31 provinces experienced reduced equity (maximum increase: 2.77% in spatial Gini coefficient), while the remaining 13 provinces showed improved equity (maximum decrease: 1.49%). These trends correlated with declining equity in bed indicator (0.34% annual increase) and physician indicator (0.26% annual increase).

Analysis of single indicators revealed distinct spatial equity patterns: bed indicator showed equity deterioration in 15 provinces (maximum annual increase: 3.86% in spatial Gini coefficient) versus improvement in 16 provinces (maximum decrease: 1.46%), while physician indicator exhibited declines in 18 provinces (maximum increase: 5.66%) and gains in 13 provinces (maximum reduction: 4.44%). In contrast, hospital indicator demonstrated nationwide improvement with an annual spatial Gini coefficient decrease of 0.32%. As shown in Figure 3B and listed in Supplementary Table S2, there were strong concordance between composite index equity and bed and physician equity. Regions with lower composite index equity consistently displayed poorer bed and physician equity, whereas high-equity regions exhibited smaller inter-indicator disparities.

Figure 4 shows spatiotemporal hotspot and coldspot distributions of city healthcare resources in China during the study period, with hotspots representing statistically significant resource-abundant clusters and coldspots indicating resource-deficient clusters. This study identified eleven distinct spatiotemporal agglomeration patterns. Supplementary Table S3 lists the detailed numbers of spatiotemporal agglomeration patterns for composite index and three single indicators. Supplementary Text S1 clarifies the specific meanings of these hotspots and coldspots.

The city-level composite index revealed that northwest China and coastal East China were predominantly characterized by intensifying and consecutive hotspots, while northeast and southwest regions manifested primarily as intensifying, consecutive, and persistent coldspots. Above patterns were substantially consistent with single indicator analysis. Nationally, 72 hotspot cities (19.73%) and 90 coldspot cities (24.66%) were identified. Quantitatively, the comparison of hotspot distributions across indicators revealed distinct patterns. The composite score exhibited fewer hotspots than hospital indicator (n = 105), but more than physician (n = 68) and bed indicator (n = 45). Conversely, coldspot counts for the composite score significantly exceeded all single indicators (hospital = 76, bed = 82, physician = 29). Structurally, 91.67% of the composite score hotspots cities overlapped with hotspots in at least one single indicator (hospital, bed, or physician), while 70.0% of the composite score coldspots cities coincided with coldspots in at least one specific resource indicator (hospital, bed, or physician). The observed spatiotemporal polarization of composite index demonstrated an intensifying Matthew effect in China’s city healthcare resource allocation, where the disparities between resource-rich and resource-deficient regions had progressively widened over time, with disadvantaged areas facing increasing challenges in closing the gap. The composite index exhibited strong representativeness of all three single indicators, as evidenced by both hotspot and coldspot quantification and structural alignment.

This study’s joint spatiotemporal evaluation framework advanced multi-dimensional equity evaluation by simultaneously capturing localized asynchronous variations and systemic constraints of healthcare indicators while mapping macroscale allocation patterns. The results methodological diverged from single-indicator temporal analyses and nonspatial multi-dimensional studies (12, 24, 38), yet substantively aligned with United Nations statistical conclusions (3), confirming the critical role of spatiotemporal integration in comprehensive evaluation. Furthermore, the framework revealed the operational logic of allocation mechanisms through distribution pattern analytics, thereby providing robust theoretical support for policymakers to balance macro-level equilibrium with micro-level precise adjustments.

Specifically, this study systematically examined the spatiotemporal evolution and inter-regional coordination of China’s city healthcare resources from 2000 to 2021, revealing three critical findings. First, while aggregate healthcare resources measured via the composite index exhibited sustained increases, over half of Chinese regions concurrently experienced considerable equity declines, underscoring persistent systemic challenges in achieving equitable distribution. Second, the analysis revealed systemic interdependencies between composite and single indicators, highlighting the necessity for policymakers to sustain the stabilizing effects of incremental hospital equity improvements while prioritizing bed availability as a critical determinant of system-wide equity. Such a dual-strategy approach enabled targeted interventions to address allocation imbalances. Third, the framework identified polarized equity clusters, enabling dual-track policy responses: strategic replication of best practices from high-equity clusters and targeted interventions in low-equity areas exhibiting critical bottlenecks. These findings operationalized a multiscale governance model that integrated macroscale equity diagnostics with precision resource targeting, offering empirically validated pathways to reconcile aggregate growth with localized equity optimization in China’s healthcare system.

This study advanced a multi-dimensional spatiotemporal evaluation framework to empirically identify critical healthcare allocation patterns and priority intervention targets. The composite index functioned as a macroscale diagnostic tool, exposing systemic allocation deficiencies and geospatial resource disparities, while parallel single indicator analyses revealed intra-regional inequities through comparative validation. This dual-scale analytical architecture provided policymakers with dual-capacity decision-support. The composite index dynamics guided macroscale intervention zoning, whereas single indicator evaluations served to identify short-board effects in specific resource categories at operational levels. Methodologically, the framework transcended descriptive spatiotemporal equity and spatiotemporal agglomeration analysis to a dynamic decision support system that bridged macro-equity monitoring with precision governance strategies, enabling simultaneous pursuit of system-wide equilibrium through strategic resource reallocation and localized optimization via context-specific short-board remediation.

This study utilized spatial Gini coefficient trajectories derived from composite index and single indicators to identify asynchronous dynamics in the equity of healthcare resource distribution across Chinese provinces, pinpointing regions with pronounced short-board effects. For example, Tibet achieved sustained improvement in equity through integrated policy interventions including centralized fiscal transfers (63), targeted aid programs (64), population redistribution policies (65), and the development of traditional healthcare system (66), thereby establishing a replicable governance model for sparsely populated and economically underdeveloped regions (67). In contrast, Sichuan experienced a decline in equity, primarily driven by resource concentration induced by population-economic gradients (68), and patients’ cross-regional health seeking behaviors (69). Above structural contradictions constrained equity improvements despite intervention efforts (70). These contrasting cases highlighted three critical policy imperatives, including enhancing the monitoring mechanism for the spatial impacts of population-economic gradients, optimizing allocation mechanisms to mitigate excessive resource polarization and implementing context-specific governance to address systemic bottlenecks, thereby identifying strategic priorities for achieving nationwide equitable healthcare distribution.

The spatiotemporal hotspot analysis identified 11 distinct agglomeration patterns of healthcare resource disparities across China, revealing systemic rather than isolated regional imbalances. Three geographically stratified paradigms were distinguished. Northeastern industrial bases exhibited resource depletion driven by economic contraction-induced fiscal austerity and population outflow-related service attrition (71). Yunnan-Guizhou plateau regions displayed suboptimal efficiency in central fiscal transfers due to dispersed settlements that impaired population-resource spatial matching. Yangtze River Delta regions manifested market-driven structural inequities where economic advancement coexisted with healthcare accessibility deficits despite resource abundance (72). These patterns collectively demonstrated the multiscale complexity of national healthcare imbalances, integrating macroeconomic volatility, mesoscale demographic-geographic mismatches, and market-mediated allocation mechanisms. These findings necessitate policy frameworks that reconcile macro-level equity objectives with micro-scale precision governance, addressing both fiscal constraints from economic cycles and population distribution impacts on allocation efficiency. The developed analytical model provides a methodological foundation for designing spatiotemporally adaptive interventions that balance systemic equity with context-specific effectiveness in complex resource allocation systems.

The spatiotemporal analysis revealed distinct temporal dynamics and spatial patterns in healthcare resource allocation determinants. Temporally, economic development exerted immediate regulatory effects on resource distribution through rapid policy feedback mechanisms (73), while urbanization exhibited a three-phase U-shaped trajectory in hospital allocation, potentially reflecting municipal prioritization of economic growth over public service infrastructure development (44, 74). Spatially, the evaluation discrepancy between single indicators and composite index was attributed to spatial spillover effects that inter-city policy diffusion enhanced resource concentration (75), though regional heterogeneity in socioeconomic contexts and policy enforcement shaped final composite outcomes. Five principal determinants emerged with combined explanatory power of 83.64–88.16% variance, including population density (30.05–43.05% contribution), workforce concentration (30.69–33.83%), per capita fiscal expenditure, primary industry GDP share, and per capita secondary education infrastructure. The predominance of demographic and economic density factors underscored their critical role in shaping healthcare resource allocation patterns, necessitating priority consideration in policy optimization frameworks.

The analysis identified five structural determinants governing spatial healthcare disparities. Population density demonstrated polarization effects, with hyperdense city clusters exhibiting facility congestion while hypodense peripheries faced service scarcity and accessibility deficits (20, 76). Fiscal capacity, operationalized through per capita public expenditure, regulated healthcare investment magnitude, where increased allocations corresponded to infrastructure expansion, service optimization, and measurable outcome improvements including mortality reduction and longevity extension (77). Primary industry-dependent regions encountered compounded accessibility barriers from geographic isolation, structurally limiting advanced care availability while increasing reliance on primary care system to address occupation-environmental health risks (78, 79). Educational attainment mediated healthcare utilization efficiency through health literacy enhancement (80), establishing quantifiable associations between academic achievement and service optimization (81). Workforce concentration generated dual effects. It might exacerbate demand–supply imbalances through accelerated urbanization (82, 83) while paradoxically strengthen healthcare workforce retention in industrial clusters through economic attractiveness (84). These interdependent mechanisms collectively explained the structural foundations of healthcare inequities, supporting the implementation of differentiated allocation frameworks that addressed regional specificities through targeted policy interventions (85).

This study proposed a spatiotemporally informed policy framework to optimize healthcare equity through three governance strategies. First, the spatiotemporal evaluation framework could serve as a dynamic decision-support mechanism. The evaluation system would enable real-time identification of regions with equity deterioration through spatial Gini coefficient trends of composite scores, guide evidence-based resource reallocation via spatiotemporal hotspot and coldspot analysis, and further facilitate iterative policy adjustments through systematic monitoring of allocation outcomes. Second, the implementation of spatiotemporally adaptive governance protocols is highly recommended to address regional heterogeneities through differentiated interventions. For example, declining equity areas require prioritized establishment of cross-regional medical alliances to reduce access gaps, while physician-deficient areas necessitate incentive systems integrating service-credit portability, accelerated career pathways, and tax-benefit package to stabilize workforce allocation. Third, healthcare resource prioritization should be stratified according to economic structure and urbanization levels. Primary industry regions need infrastructure investments for basic care capacity enhancement and disease-specific competency building, while city healthcare development areas focus on specialized service networks and medical technology innovation to address advanced care demands. This framework bridges macroscale equity monitoring with context-specific implementation, offering a scalable model for precision health governance aligned with SDGs.

The proposed spatiotemporal evaluation framework highlights the critical role of multiple resource interplay in driving healthcare equity. A co-evolutionary analysis of hospital infrastructure and physician workforce reallocation reveals that hospitals unlock substantially greater allocation efficiencies when their capacity expansions are synchronized with targeted physician distribution adjustments, an effect most pronounced in rapidly urbanizing zones (86). These intertwined dynamics underscore the necessity of integrated policy design: hospital bed-and-facility growth must be coupled with robust physician incentive and mobility schemes to forestall local resource agglomeration and access gaps. The combined use of the composite index and its constituent indicators equips decision-makers with a dual-perspective diagnostic system to locate resource-complementary regions. In areas exhibiting bed abundance but physician scarcity, for instance, rotational clinical training programs and telemedicine platforms can be strategically introduced to optimize utilization and redress short-board effects (87).

Furthermore, the bidirectional feedback loop between composite and individual metrics refines precision governance across scales. The composite index charts macroscale allocation trajectories and, through spatial Gini coefficient trends, identifies priority intervention zones, whereas single indicators spotlight granular resource deficits demanding immediate operational attention (88). This dual-scale diagnostic architecture enables a stratified implementation pathway: high-scoring regions can concentrate on optimizing specialized service networks, while low-scoring areas should prioritize foundational infrastructure build-out. Crucially, monitoring the influence of single-indicator improvements (e.g., increases in physician density) on composite index dynamics creates an iterative, spatiotemporally adaptive mechanism for fine-tuning resource redistribution, thereby enhancing allocation accuracy and sustaining equity over time (89, 90).

This study has limitations. First, the selected healthcare resource indicators were hospitals, hospital beds, and physicians. We did not include other indicators such as primary healthcare facilities, nursing staff, and medical equipment due do data unavailability, potentially constraining the granularity of multi-dimensional characterization. Second, the city-level analytical scale prevented granular analysis of intra-city resource distribution patterns across sub-city administrative units, thereby limiting insights into localized disparities in healthcare accessibility. Third, the current evaluation framework primarily relied on quantitative metrics for resource distribution, neglecting critical dimensions such as service quality assessments, facility upgrading indices, and patient accessibility metrics, which resulted in an incomplete examination of operational effectiveness. Fourth, the analysis did not incorporate cross-provincial healthcare utilization due to the absence of comprehensive interprovincial patient flow data and the jurisdictional basis of resource allocation, leaving spatiotemporal equity considerations only partially addressed. These methodological constraints underscore the need for future research to expand indicator systems, conduct multiscale spatial analyses, integrate quality-effectiveness and patient flow metrics, and accommodate cross-boundary utilization patterns to enhance the comprehensiveness and applicability of healthcare resource evaluations. Finally, it is worth noting that a common practice in current mainstream research, namely, first reducing multiple target variables into a single composite indicator, has significant drawbacks, as it loses crucial multi-target information. Future methodological studies on BSTVC should aim to jointly model the spatiotemporal nonstationarity of multiple targets within a unified framework (91, 92).