Spatiotemporal dynamics and driving factors of human resources for health in traditional Chinese medicine in China

The variation trend of HRH in TCM at provincial level in China

Analysis of the health resource aggregation

To investigate the spatial concentration of HRH in TCM, this study employs the agglomeration degree to evaluate the allocation of HRH, as illustrated in (Fig. 3). Overall, from 2008 to 2021, the distribution of HRH in TCM has been markedly uneven, with significant disparities observed among provinces. Beijing exhibits the highest agglomeration degree of HRH in TCM in the country, with an average value of 2.3, significantly exceeding the national average. As the capital of China, Beijing benefits from a strategic location that positions it as a central hub for resource allocation, thereby attracting a substantial number of TCM professionals. In contrast, the agglomeration degree of HRH in TCM in the central and western regions, including Chongqing, Sichuan, Inner Mongolia, Gansu, Tibet, and Qinghai, averages around 1.5, indicating a relatively high ranking. This can be largely attributed to the abundance of ethnic medical resources and a comparatively lower population in the Midwest region. In 2021, the agglomeration degree of HRH in TCM in 17 provinces, including Liaoning, Heilongjiang, Shanghai, and Jiangsu, fell below 1, signaling an insufficient supply of TCM resources relative to their populations. Among these, Hainan recorded the lowest agglomeration degree at 0.71, remaining below the national average for HRH in TCM. Additionally, we observed some improvement of HRH in TCM in certain previously uneven provinces. From 2008 to 2021, the agglomeration degree of HRH in TCM in regions such as Shanghai, Guangxi, Anhui, and Guizhou have increased annually, approaching a more balanced state near 1.

However, it is noteworthy that the agglomeration degree of HRH in TCM in developed central regions such as Beijing and Tianjin, as well as in coastal eastern regions like Zhejiang, Fujian, and Guangdong, and in northeastern industrial areas including Heilongjiang, Jilin, and Liaoning, has been declining annually. This trend indicates a significant loss of HRH in TCM in these areas. In summary, there is considerable variation in the concentration and distribution patterns of HRH in TCM across Chinese provinces. The key factors contributing to these disparities warrant further investigation.

Analysis of HRH per thousand population

This study employs the number of TCM practitioners per 1000 people from 2008 to 2021 to create a trend chart that illustrates the changes in HRH across Chinese provinces, as depicted in (Fig. 4). The total number of HRH in China increased from 253,200 in 2008 to 731,700 in 2021, reflecting a growth of 188.9%. Over this period, the overall trend in HRH across Chinese provinces has demonstrated a steady annual increase. In 2021, there were 0.52 TCM practitioners per 1,000 people in China, which remains below the 2025 target of 0.62 practitioners per 1,000 people established by the 14th Five-Year Plan of China. Figure 4 illustrates that the growth rate of HRH in TCM per 1000 people varies significantly across provinces in China, underscoring considerable regional disparities in both the total number of HRH in TCM and the number of TCM practitioners per 1000 people.

Analysis of geographical density of HRH

The measurement of Fig. 5 reflects the degree of change in HRH in each province over the years through geographic density. It should be noted that this comparison method may exaggerate the degree of imbalance between provinces due to ignoring the objective laws followed by the flow of talents. This discovery showed that the number of HRH is increasing year by year, meanwhile the relative level difference in the allocation of health resources has decreased but not obvious.

Comparisons of different HRH results

There are some similarities and differences in the overall changes of the provinces. The degree of change in HRH in different provinces is basically the same in the three models. While maintaining the growth in the number of HRH, the agglomeration of Beijing and Shanghai have the most obvious decline. Interestingly, the HRH growth rate in Guizhou is the fastest among the provinces. There is a wide gap between these three models, which is shown in (Fig. 6). by comparing the total degree of change. The results demonstrated that agglomeration density performed well in dynamic changes of HRH in provincial level of China.

Temporal and spatial changes of HRH in TCM

To further investigate the spatial distribution and evolution of HRH in China, this study utilizes cross-sectional data from 2008, along with average data from three distinct periods: 2009–2013, 2014–2018, and 2019–2021. These datasets are employed to create maps that depict the spatial distribution of HRH in TCM across the country. Figure 7 illustrates a significant imbalance in the distribution pattern of these resources, which are predominantly concentrated in the North China region, particularly around Beijing and Tianjin, the eastern coastal region, including Shanghai and Zhejiang, and the southwestern region, primarily comprising Tibet and Sichuan. Conversely, areas with limited resources are mainly found in the central region, a distribution pattern closely associated with the availability of medicinal resources and levels of economic development. In 2008, only Beijing and Tianjin were classified as high-density areas for HRH. However, during the period from 2019 to 2021, despite a notable increase in HRH throughout China, the number of high-density areas further declined, with only Beijing remaining classified as such, and its agglomeration degree experiencing a decrease of 18.82% compared to 2008. From 2008 to 2021, Inner Mongolia and Sichuan consistently ranked among the semi-dense areas. In contrast, Anhui and Yunnan remained in the sparse resource areas, with their agglomeration degree consistently below 1. Additionally, neighboring provinces such as Guizhou, Guangxi, Jiangxi, and Hunan fluctuated between the sparse and secondary sparse areas. In comparison, provinces like Qinghai, Guizhou, and Guangxi experienced the most significant increases in agglomeration degree of HRH; however, they still fell within the balanced and secondary sparse areas.

Although Beijing, Tianjin, and Chongqing have experienced the most significant declines in agglomeration degree of HRH, these regions remain classified as dense or semi-dense areas. This disparity highlights a pronounced ‘Matthew Effect’ in the distribution of HRH across China. Economically advanced regions exhibit a stronger ‘siphoning effect’ for TCM resources, enabling them to continue attracting high-quality TCM professionals despite rapid population growth. In contrast, economically less developed regions are becoming increasingly less appealing for TCM resources. This polarized distribution of HRH in TCM has a substantial impact on the accessibility and equity of healthcare services.

Moran’s I spatial autocorrelation analysis

Through extensive Monte Carlo simulations, the Moran’s I index for health technicians per 10,00 people in 2008 was calculated to be 0.254 (p = 0.02), indicating a significant spatial agglomeration effect. In 2018, the index increased to 0.257 (p = 0.03), demonstrating continued significant clustering. By 2021, the index rose further to 0.382 (p = 0.007), reflecting a highly significant spatial agglomeration effect. This data illustrates a notable and increasing trend in the spatial clustering of HRH in China. To explore regional spatial effects, scatter plots of the Moran’s I index for 2008, 2018, and 2021 were generated (Fig. 8). In these years, most provinces were positioned in the upper right and lower left quadrants. Notably, Tianjin, Beijing, and Inner Mongolia consistently appeared in the upper-right quadrant, indicating that these provinces possess high levels of HRH and are surrounded by provinces with dense distributions.

Spatial-temporal clustering characteristics of HRH in TCM

Using the kernel density analysis tool in ArcGIS 10.2, kernel density estimates for HRH in TCM were conducted for the representative years 2008, 2018, and 2021, resulting in the generation of a HRH density plot, as shown in (Fig. 9). The overall trend in kernel density distribution from 2008 to 2021 remained consistent, indicating the formation of a relatively stable spatial pattern of HRH in TCM. The kernel density of HRH in TCM varied across different regions, generally exhibiting a pattern characterized by ‘higher in the east and lower in the west’. The highest density centers of HRH in TCM were concentrated in areas such as Beijing, Tianjin, and Hebei, which served as central points from which density spread outward in layers. This pattern is closely associated with the coordinated development policy of TCM health resources in the Beijing-Tianjin-Hebei region. Additionally, these areas experienced a spillover effect, contributing to the concentration of HRH in TCM in neighboring provinces such as Shanxi, Henan, and Shandong. High density was also observed at the borders of Jiangsu and Zhejiang, as well as Sichuan and Chongqing, where a siphoning effect inhibited the concentration of HRH in TCM in adjacent provinces. This indicates a gradual concentration of HRH in TCM in economically developed regions.

Spatial-temporal variation characteristics of HRH in TCM

The standard deviation ellipse was utilized to illustrate the center of distribution and the diffusion trend of HRH in China from 2008 to 2021, as depicted in (Fig. 10). This figure indicates that the spatial distribution of these resources follows a pattern extending from the east (slightly north) to the west (slightly south). The standard deviation ellipse encompasses regions primarily in eastern and central China, including Beijing, Tianjin, Hebei, Henan, Shandong, Shanxi, Shanghai, Sichuan, Chongqing, Hunan, Hubei, Anhui, Jiangsu, and Shaanxi. Compared to 2008, the ellipse in 2021 has shifted southwestward and decreased in size. In 2021, the centripetal force of the spatial distribution of HRH in China became increasingly evident, accompanied by a rise in the degree of dispersion. This suggests that while HRH have expanded across regions, the spatial agglomeration effect has intensified. The changes in HRH are increasingly correlated with the economic development of each region. The central positions of the spatial standard deviation ellipses for 2008 and 2021 are relatively consistent, indicating that the mean distribution across regions has remained stable, while the overall distribution pattern of HRH in China has become more stable.

Model fitting

Evaluating and comparing the model fit and performance, a corrected Akaike Information Criterion (AICC) was used to select the optimal bandwidth. We evaluated the performance of the OLS, GWR, and MGWR models across four stages, with results consistently demonstrating that the MGWR model achieved the highest accuracy, as illustrated in (Table 3). In each of the four stages, the goodness-of-fit (R²) and adjusted R² values for the MGWR model exceeded those of the OLS and GWR models. The best model fit is indicating by a larger adjusted R and a smaller AIC value. Consequently, the MGWR model is deemed the most appropriate for investigating the spatial heterogeneity of factors influencing HRH in TCM.

Spatial heterogeneity analysis of influencing factors of HRH based on the MGWR model

The analysis results in 2008

In 2008, the primary factors influencing the distribution of HRH in TCM included the unemployment rate (p < 0.05), average wages (p < 0.05), education funding (p < 0.01), the number of TCM institutions (p < 0.01), and per-capita healthcare expenditure (p < 0.05). Among these factors, average wage had the most significant impact, followed by per-capita healthcare expenditure. Detailed results are presented in (Table 4). The regression coefficient for unemployment demonstrated a notable decrease from northern to southern regions, indicating a more substantial negative impact in the north compared to the south. Similarly, the regression coefficient for average wage exhibited a significant decline from northeast to southwest, suggesting a considerably stronger positive impact in the northeast relative to the southwest. The coefficient for education funding revealed a significant decrease from west to east, indicating a greater negative impact in the western regions compared to the eastern regions. Furthermore, the coefficient for the number of TCM institutions also decreased from west to east, implying that the positive impact of these institutions was much stronger in the western regions than in the eastern regions. High-value areas for the coefficient of per-capita healthcare expenditure were identified in North China and Northeast China, while low-value areas were found in Xinjiang and Tibet. This suggests that the negative impact of per-capita healthcare expenditure was more pronounced in high-value areas and less so in low-value areas. For further details, please refer to (Fig. 11).

The analysis results from 2009 to 2013

From 2009 to 2013, the number of TCM institutions was the only significant factor influencing the distribution of HRH in TCM. This finding suggests that regions with a higher density of TCM institution were more likely to see an increase in HRH. Detailed results can be found in (Table 5). Figure 12 illustrates that the coefficient for the number of TCM institutions displays a spatial distribution pattern that gradually decreases from northwest to southeast. Compared to 2008, areas of high value have shifted toward the northeast, while areas of low value have migrated toward the southwest. During this period, Xinjiang demonstrated particular sensitivity to the number of TCM institutions, whereas the eastern coastal regions, including Zhejiang, Fujian, and Guangdong, exhibited lower sensitivity, likely reflecting the relatively abundant health resources available in these areas at that time.

The analysis results from 2014 to 2018

From 2014 to 2018, average wage of healthcare personnel (p < 0.01), education funding (p < 0.1), the number of TCM institutions (p < 0.001), and per-capita healthcare expenditure (p < 0.1) significantly influenced the number of HRH in TCM (Table 6). Average wage of healthcare personnel had a significant positive impact on the number of HRH, indicating that regions with higher wages were more likely to experience an increase in HRH. High-value areas for medical wages were primarily located in the northeastern regions, including Heilongjiang, Jilin, and Liaoning, while low-value areas were found in parts of the southwest (Tibet, Yunnan, Guizhou) and South China (Guangxi, Guangdong, Hainan). This suggests that the quantity of HRH in TCM in high-value areas was more responsive to wage fluctuations, whereas low-value areas exhibited less sensitivity. Conversely, education funding had a significant negative impact on the number of HRH in TCM, indicating that higher local education funding corresponded with a less pronounced increase in HRH. The spatial distribution of education funding coefficients revealed a general pattern low east high west, with high-value areas concentrated in Xinjiang and low-value areas in Shanghai, Zhejiang, and Fujian. This indicates that the negative impact of education funding was considerably greater in high-value areas compared to low-value areas. Compared to 2008, both high-value and low-value areas saw a decrease, while the number of average-value areas increased. The coefficients for the number of TCM institutions exhibited a significant spatial pattern, gradually decreasing from northwest to southeast, consistent with the spatial pattern observed during the 2009–2013 period. The positive impact of TCM institutions was greater in the northwest than in the southeast.

The coefficients for per-capita healthcare expenditure shifted from a significant negative impact in 2008 to a significant positive impact, indicating that during this period, per-capita healthcare expenditure was aligned with the growth of HRH in TCM. However, the spatial distribution pattern of these coefficients remained consistent with that of 2008, demonstrating a significant decrease from the northeast to the southwest. This suggests that per-capita healthcare expenditure had a considerably greater positive impact in northeastern regions compared to southwestern regions, as shown in (Fig. 13).

The analysis results from 2019 to 2021

From 2019 to 2021, the primary factors influencing HRH in TCM remained consistent with those identified in the previous period. These factors included average wage of healthcare personnel (p < 0.05), education funding (p < 0.001), the number of TCM institutions (p < 0.001), and per-capita healthcare expenditure (p < 0.05). The specific effects of these factors are detailed in (Table 7). However, the spatial distribution of regression coefficients for these factors changed significantly compared to the previous period. High-value areas for average wage of healthcare personnel shifted from the three northeastern provinces to Xinjiang, while low-value areas transitioned to parts of the southeastern coastal regions, including Guangxi, Guangdong, Fujian, and Hainan. During this period, the positive impact of average wage of healthcare personnel on HRH was more pronounced in high-value areas. Education funding continued to exert a significant negative impact on the number of HRH in TCM, indicating that regions with higher education spending experienced less noticeable increases in HRH. The high-value areas for education funding were concentrated in the western regions, gradually decreasing towards the east. Low-value areas shifted from the eastern coastal regions of Shanghai, Zhejiang, and Fujian to the northeastern regions of Heilongjiang, Liaoning, and Jilin. Throughout this period, the number of TCM institutions continued to have a significant positive impact on the number of HRH in TCM.

The spatial distribution pattern of the coefficients for the number of TCM institutions has remained largely consistent with the previous period, with high-value areas concentrated in the western regions, particularly in Xinjiang. In contrast, low-value areas have shown a slight shift to the southwest. The coefficients for per-capita healthcare expenditure exhibit a significant positive distribution from north to south, characterized by an increase in high-value areas and a decrease in low-value areas compared to the previous period. This indicates that HRH in northern regions have become increasingly responsive to per-capita healthcare expenditure. For further details, please refer to (Fig. 14).